{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 4764,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00020990764063811922,
      "grad_norm": 0.8211116790771484,
      "learning_rate": 2.09643605870021e-07,
      "loss": 0.9775,
      "step": 1
    },
    {
      "epoch": 0.00041981528127623844,
      "grad_norm": 0.7017123103141785,
      "learning_rate": 4.19287211740042e-07,
      "loss": 0.868,
      "step": 2
    },
    {
      "epoch": 0.0006297229219143577,
      "grad_norm": 0.6900752782821655,
      "learning_rate": 6.28930817610063e-07,
      "loss": 0.8862,
      "step": 3
    },
    {
      "epoch": 0.0008396305625524769,
      "grad_norm": 0.7597951292991638,
      "learning_rate": 8.38574423480084e-07,
      "loss": 0.923,
      "step": 4
    },
    {
      "epoch": 0.0010495382031905961,
      "grad_norm": 0.8734421730041504,
      "learning_rate": 1.048218029350105e-06,
      "loss": 1.0745,
      "step": 5
    },
    {
      "epoch": 0.0012594458438287153,
      "grad_norm": 0.8313158750534058,
      "learning_rate": 1.257861635220126e-06,
      "loss": 0.9817,
      "step": 6
    },
    {
      "epoch": 0.0014693534844668346,
      "grad_norm": 0.656913697719574,
      "learning_rate": 1.467505241090147e-06,
      "loss": 0.8544,
      "step": 7
    },
    {
      "epoch": 0.0016792611251049538,
      "grad_norm": 0.7147523164749146,
      "learning_rate": 1.677148846960168e-06,
      "loss": 0.8955,
      "step": 8
    },
    {
      "epoch": 0.001889168765743073,
      "grad_norm": 0.753795325756073,
      "learning_rate": 1.8867924528301887e-06,
      "loss": 0.881,
      "step": 9
    },
    {
      "epoch": 0.0020990764063811922,
      "grad_norm": 0.7141973972320557,
      "learning_rate": 2.09643605870021e-06,
      "loss": 0.9229,
      "step": 10
    },
    {
      "epoch": 0.0023089840470193117,
      "grad_norm": 0.8047632575035095,
      "learning_rate": 2.306079664570231e-06,
      "loss": 0.9438,
      "step": 11
    },
    {
      "epoch": 0.0025188916876574307,
      "grad_norm": 0.8492773175239563,
      "learning_rate": 2.515723270440252e-06,
      "loss": 0.9977,
      "step": 12
    },
    {
      "epoch": 0.00272879932829555,
      "grad_norm": 0.744103729724884,
      "learning_rate": 2.7253668763102727e-06,
      "loss": 0.9049,
      "step": 13
    },
    {
      "epoch": 0.002938706968933669,
      "grad_norm": 0.7073290944099426,
      "learning_rate": 2.935010482180294e-06,
      "loss": 0.8824,
      "step": 14
    },
    {
      "epoch": 0.0031486146095717885,
      "grad_norm": 0.6808626651763916,
      "learning_rate": 3.1446540880503146e-06,
      "loss": 0.8854,
      "step": 15
    },
    {
      "epoch": 0.0033585222502099076,
      "grad_norm": 0.6526097655296326,
      "learning_rate": 3.354297693920336e-06,
      "loss": 0.8219,
      "step": 16
    },
    {
      "epoch": 0.003568429890848027,
      "grad_norm": 0.8246333003044128,
      "learning_rate": 3.563941299790356e-06,
      "loss": 0.9948,
      "step": 17
    },
    {
      "epoch": 0.003778337531486146,
      "grad_norm": 0.8494473099708557,
      "learning_rate": 3.7735849056603773e-06,
      "loss": 1.0187,
      "step": 18
    },
    {
      "epoch": 0.003988245172124265,
      "grad_norm": 0.9259452223777771,
      "learning_rate": 3.9832285115303985e-06,
      "loss": 1.0864,
      "step": 19
    },
    {
      "epoch": 0.0041981528127623844,
      "grad_norm": 0.8235483765602112,
      "learning_rate": 4.19287211740042e-06,
      "loss": 1.0255,
      "step": 20
    },
    {
      "epoch": 0.004408060453400504,
      "grad_norm": 0.8101536631584167,
      "learning_rate": 4.40251572327044e-06,
      "loss": 0.9892,
      "step": 21
    },
    {
      "epoch": 0.004617968094038623,
      "grad_norm": 0.7406070232391357,
      "learning_rate": 4.612159329140462e-06,
      "loss": 0.9266,
      "step": 22
    },
    {
      "epoch": 0.004827875734676742,
      "grad_norm": 0.7151230573654175,
      "learning_rate": 4.821802935010482e-06,
      "loss": 0.902,
      "step": 23
    },
    {
      "epoch": 0.005037783375314861,
      "grad_norm": 0.6921148896217346,
      "learning_rate": 5.031446540880504e-06,
      "loss": 0.8621,
      "step": 24
    },
    {
      "epoch": 0.005247691015952981,
      "grad_norm": 0.6169761419296265,
      "learning_rate": 5.241090146750524e-06,
      "loss": 0.8296,
      "step": 25
    },
    {
      "epoch": 0.0054575986565911,
      "grad_norm": 0.679093599319458,
      "learning_rate": 5.4507337526205454e-06,
      "loss": 0.8593,
      "step": 26
    },
    {
      "epoch": 0.005667506297229219,
      "grad_norm": 0.8058671951293945,
      "learning_rate": 5.660377358490566e-06,
      "loss": 0.9504,
      "step": 27
    },
    {
      "epoch": 0.005877413937867338,
      "grad_norm": 0.8705667853355408,
      "learning_rate": 5.870020964360588e-06,
      "loss": 1.0245,
      "step": 28
    },
    {
      "epoch": 0.006087321578505458,
      "grad_norm": 0.7334048748016357,
      "learning_rate": 6.079664570230608e-06,
      "loss": 0.9172,
      "step": 29
    },
    {
      "epoch": 0.006297229219143577,
      "grad_norm": 0.7490406632423401,
      "learning_rate": 6.289308176100629e-06,
      "loss": 0.9329,
      "step": 30
    },
    {
      "epoch": 0.006507136859781696,
      "grad_norm": 0.8739404678344727,
      "learning_rate": 6.49895178197065e-06,
      "loss": 0.9584,
      "step": 31
    },
    {
      "epoch": 0.006717044500419815,
      "grad_norm": 0.7686614990234375,
      "learning_rate": 6.708595387840672e-06,
      "loss": 0.9493,
      "step": 32
    },
    {
      "epoch": 0.0069269521410579345,
      "grad_norm": 0.9184572100639343,
      "learning_rate": 6.918238993710692e-06,
      "loss": 1.0678,
      "step": 33
    },
    {
      "epoch": 0.007136859781696054,
      "grad_norm": 0.8303453326225281,
      "learning_rate": 7.127882599580712e-06,
      "loss": 0.9834,
      "step": 34
    },
    {
      "epoch": 0.0073467674223341725,
      "grad_norm": 0.6670796871185303,
      "learning_rate": 7.337526205450735e-06,
      "loss": 0.8434,
      "step": 35
    },
    {
      "epoch": 0.007556675062972292,
      "grad_norm": 0.6518784761428833,
      "learning_rate": 7.547169811320755e-06,
      "loss": 0.8445,
      "step": 36
    },
    {
      "epoch": 0.007766582703610411,
      "grad_norm": 0.8396414518356323,
      "learning_rate": 7.756813417190776e-06,
      "loss": 0.9229,
      "step": 37
    },
    {
      "epoch": 0.00797649034424853,
      "grad_norm": 0.9019722938537598,
      "learning_rate": 7.966457023060797e-06,
      "loss": 0.9545,
      "step": 38
    },
    {
      "epoch": 0.00818639798488665,
      "grad_norm": 0.6634522676467896,
      "learning_rate": 8.176100628930818e-06,
      "loss": 0.8279,
      "step": 39
    },
    {
      "epoch": 0.008396305625524769,
      "grad_norm": 0.86155104637146,
      "learning_rate": 8.38574423480084e-06,
      "loss": 0.9326,
      "step": 40
    },
    {
      "epoch": 0.008606213266162888,
      "grad_norm": 0.6448208689689636,
      "learning_rate": 8.59538784067086e-06,
      "loss": 0.8261,
      "step": 41
    },
    {
      "epoch": 0.008816120906801008,
      "grad_norm": 0.7442598938941956,
      "learning_rate": 8.80503144654088e-06,
      "loss": 0.9352,
      "step": 42
    },
    {
      "epoch": 0.009026028547439127,
      "grad_norm": 0.8854546546936035,
      "learning_rate": 9.014675052410902e-06,
      "loss": 1.017,
      "step": 43
    },
    {
      "epoch": 0.009235936188077247,
      "grad_norm": 0.66485196352005,
      "learning_rate": 9.224318658280923e-06,
      "loss": 0.8096,
      "step": 44
    },
    {
      "epoch": 0.009445843828715366,
      "grad_norm": 0.5998132824897766,
      "learning_rate": 9.433962264150944e-06,
      "loss": 0.7932,
      "step": 45
    },
    {
      "epoch": 0.009655751469353484,
      "grad_norm": 0.6402536034584045,
      "learning_rate": 9.643605870020965e-06,
      "loss": 0.8146,
      "step": 46
    },
    {
      "epoch": 0.009865659109991603,
      "grad_norm": 0.6665769219398499,
      "learning_rate": 9.853249475890985e-06,
      "loss": 0.8435,
      "step": 47
    },
    {
      "epoch": 0.010075566750629723,
      "grad_norm": 0.6947203874588013,
      "learning_rate": 1.0062893081761008e-05,
      "loss": 0.8712,
      "step": 48
    },
    {
      "epoch": 0.010285474391267842,
      "grad_norm": 0.6707759499549866,
      "learning_rate": 1.0272536687631027e-05,
      "loss": 0.8382,
      "step": 49
    },
    {
      "epoch": 0.010495382031905962,
      "grad_norm": 0.6716253161430359,
      "learning_rate": 1.0482180293501048e-05,
      "loss": 0.8887,
      "step": 50
    },
    {
      "epoch": 0.010705289672544081,
      "grad_norm": 0.6954374313354492,
      "learning_rate": 1.069182389937107e-05,
      "loss": 0.895,
      "step": 51
    },
    {
      "epoch": 0.0109151973131822,
      "grad_norm": 0.5633372068405151,
      "learning_rate": 1.0901467505241091e-05,
      "loss": 0.7971,
      "step": 52
    },
    {
      "epoch": 0.01112510495382032,
      "grad_norm": 0.5513401031494141,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 0.7848,
      "step": 53
    },
    {
      "epoch": 0.011335012594458438,
      "grad_norm": 0.6735419034957886,
      "learning_rate": 1.1320754716981132e-05,
      "loss": 0.8558,
      "step": 54
    },
    {
      "epoch": 0.011544920235096557,
      "grad_norm": 0.5273690223693848,
      "learning_rate": 1.1530398322851153e-05,
      "loss": 0.7531,
      "step": 55
    },
    {
      "epoch": 0.011754827875734676,
      "grad_norm": 0.612610936164856,
      "learning_rate": 1.1740041928721176e-05,
      "loss": 0.8009,
      "step": 56
    },
    {
      "epoch": 0.011964735516372796,
      "grad_norm": 0.5250133275985718,
      "learning_rate": 1.1949685534591196e-05,
      "loss": 0.7411,
      "step": 57
    },
    {
      "epoch": 0.012174643157010915,
      "grad_norm": 0.6210602521896362,
      "learning_rate": 1.2159329140461215e-05,
      "loss": 0.814,
      "step": 58
    },
    {
      "epoch": 0.012384550797649035,
      "grad_norm": 0.5767892003059387,
      "learning_rate": 1.2368972746331238e-05,
      "loss": 0.7512,
      "step": 59
    },
    {
      "epoch": 0.012594458438287154,
      "grad_norm": 0.5112836360931396,
      "learning_rate": 1.2578616352201259e-05,
      "loss": 0.7405,
      "step": 60
    },
    {
      "epoch": 0.012804366078925274,
      "grad_norm": 0.5214811563491821,
      "learning_rate": 1.2788259958071281e-05,
      "loss": 0.7152,
      "step": 61
    },
    {
      "epoch": 0.013014273719563391,
      "grad_norm": 0.5820367932319641,
      "learning_rate": 1.29979035639413e-05,
      "loss": 0.7359,
      "step": 62
    },
    {
      "epoch": 0.01322418136020151,
      "grad_norm": 0.5728296637535095,
      "learning_rate": 1.320754716981132e-05,
      "loss": 0.7096,
      "step": 63
    },
    {
      "epoch": 0.01343408900083963,
      "grad_norm": 0.514997124671936,
      "learning_rate": 1.3417190775681343e-05,
      "loss": 0.6945,
      "step": 64
    },
    {
      "epoch": 0.01364399664147775,
      "grad_norm": 0.5707572102546692,
      "learning_rate": 1.3626834381551362e-05,
      "loss": 0.7065,
      "step": 65
    },
    {
      "epoch": 0.013853904282115869,
      "grad_norm": 0.5674712657928467,
      "learning_rate": 1.3836477987421385e-05,
      "loss": 0.6759,
      "step": 66
    },
    {
      "epoch": 0.014063811922753989,
      "grad_norm": 0.5445975661277771,
      "learning_rate": 1.4046121593291406e-05,
      "loss": 0.6487,
      "step": 67
    },
    {
      "epoch": 0.014273719563392108,
      "grad_norm": 0.5629355311393738,
      "learning_rate": 1.4255765199161425e-05,
      "loss": 0.6635,
      "step": 68
    },
    {
      "epoch": 0.014483627204030227,
      "grad_norm": 0.47151610255241394,
      "learning_rate": 1.4465408805031447e-05,
      "loss": 0.5976,
      "step": 69
    },
    {
      "epoch": 0.014693534844668345,
      "grad_norm": 0.44633767008781433,
      "learning_rate": 1.467505241090147e-05,
      "loss": 0.5738,
      "step": 70
    },
    {
      "epoch": 0.014903442485306465,
      "grad_norm": 0.48507657647132874,
      "learning_rate": 1.488469601677149e-05,
      "loss": 0.5859,
      "step": 71
    },
    {
      "epoch": 0.015113350125944584,
      "grad_norm": 0.4147733151912689,
      "learning_rate": 1.509433962264151e-05,
      "loss": 0.5386,
      "step": 72
    },
    {
      "epoch": 0.015323257766582703,
      "grad_norm": 0.635608434677124,
      "learning_rate": 1.530398322851153e-05,
      "loss": 0.6529,
      "step": 73
    },
    {
      "epoch": 0.015533165407220823,
      "grad_norm": 0.5556919574737549,
      "learning_rate": 1.5513626834381552e-05,
      "loss": 0.5935,
      "step": 74
    },
    {
      "epoch": 0.015743073047858942,
      "grad_norm": 0.5627433657646179,
      "learning_rate": 1.572327044025157e-05,
      "loss": 0.5677,
      "step": 75
    },
    {
      "epoch": 0.01595298068849706,
      "grad_norm": 0.5727344155311584,
      "learning_rate": 1.5932914046121594e-05,
      "loss": 0.5701,
      "step": 76
    },
    {
      "epoch": 0.01616288832913518,
      "grad_norm": 0.5192092657089233,
      "learning_rate": 1.6142557651991616e-05,
      "loss": 0.537,
      "step": 77
    },
    {
      "epoch": 0.0163727959697733,
      "grad_norm": 0.6583610773086548,
      "learning_rate": 1.6352201257861635e-05,
      "loss": 0.5718,
      "step": 78
    },
    {
      "epoch": 0.01658270361041142,
      "grad_norm": 0.4762994050979614,
      "learning_rate": 1.6561844863731658e-05,
      "loss": 0.5016,
      "step": 79
    },
    {
      "epoch": 0.016792611251049538,
      "grad_norm": 0.7903013825416565,
      "learning_rate": 1.677148846960168e-05,
      "loss": 0.6123,
      "step": 80
    },
    {
      "epoch": 0.01700251889168766,
      "grad_norm": 0.6027877330780029,
      "learning_rate": 1.69811320754717e-05,
      "loss": 0.5085,
      "step": 81
    },
    {
      "epoch": 0.017212426532325777,
      "grad_norm": 0.6400225162506104,
      "learning_rate": 1.719077568134172e-05,
      "loss": 0.5325,
      "step": 82
    },
    {
      "epoch": 0.017422334172963894,
      "grad_norm": 0.5193424224853516,
      "learning_rate": 1.740041928721174e-05,
      "loss": 0.476,
      "step": 83
    },
    {
      "epoch": 0.017632241813602016,
      "grad_norm": 0.5318325757980347,
      "learning_rate": 1.761006289308176e-05,
      "loss": 0.4574,
      "step": 84
    },
    {
      "epoch": 0.017842149454240133,
      "grad_norm": 0.5530166029930115,
      "learning_rate": 1.7819706498951782e-05,
      "loss": 0.4331,
      "step": 85
    },
    {
      "epoch": 0.018052057094878254,
      "grad_norm": 0.5483909845352173,
      "learning_rate": 1.8029350104821805e-05,
      "loss": 0.4254,
      "step": 86
    },
    {
      "epoch": 0.018261964735516372,
      "grad_norm": 0.4599871039390564,
      "learning_rate": 1.8238993710691824e-05,
      "loss": 0.3828,
      "step": 87
    },
    {
      "epoch": 0.018471872376154493,
      "grad_norm": 0.47489672899246216,
      "learning_rate": 1.8448637316561846e-05,
      "loss": 0.3989,
      "step": 88
    },
    {
      "epoch": 0.01868178001679261,
      "grad_norm": 0.6532279253005981,
      "learning_rate": 1.865828092243187e-05,
      "loss": 0.3968,
      "step": 89
    },
    {
      "epoch": 0.018891687657430732,
      "grad_norm": 0.707245945930481,
      "learning_rate": 1.8867924528301888e-05,
      "loss": 0.4099,
      "step": 90
    },
    {
      "epoch": 0.01910159529806885,
      "grad_norm": 0.4290582835674286,
      "learning_rate": 1.9077568134171907e-05,
      "loss": 0.3546,
      "step": 91
    },
    {
      "epoch": 0.019311502938706968,
      "grad_norm": 0.4370197057723999,
      "learning_rate": 1.928721174004193e-05,
      "loss": 0.3519,
      "step": 92
    },
    {
      "epoch": 0.01952141057934509,
      "grad_norm": 0.4862878918647766,
      "learning_rate": 1.9496855345911952e-05,
      "loss": 0.3494,
      "step": 93
    },
    {
      "epoch": 0.019731318219983206,
      "grad_norm": 0.514576256275177,
      "learning_rate": 1.970649895178197e-05,
      "loss": 0.326,
      "step": 94
    },
    {
      "epoch": 0.019941225860621328,
      "grad_norm": 0.3785364031791687,
      "learning_rate": 1.9916142557651993e-05,
      "loss": 0.3196,
      "step": 95
    },
    {
      "epoch": 0.020151133501259445,
      "grad_norm": 0.4459572732448578,
      "learning_rate": 2.0125786163522016e-05,
      "loss": 0.3245,
      "step": 96
    },
    {
      "epoch": 0.020361041141897566,
      "grad_norm": 0.3634876310825348,
      "learning_rate": 2.0335429769392035e-05,
      "loss": 0.2999,
      "step": 97
    },
    {
      "epoch": 0.020570948782535684,
      "grad_norm": 0.39789989590644836,
      "learning_rate": 2.0545073375262054e-05,
      "loss": 0.2908,
      "step": 98
    },
    {
      "epoch": 0.020780856423173802,
      "grad_norm": 0.3628767430782318,
      "learning_rate": 2.0754716981132076e-05,
      "loss": 0.278,
      "step": 99
    },
    {
      "epoch": 0.020990764063811923,
      "grad_norm": 0.3945654332637787,
      "learning_rate": 2.0964360587002095e-05,
      "loss": 0.2902,
      "step": 100
    },
    {
      "epoch": 0.02120067170445004,
      "grad_norm": 0.2995467185974121,
      "learning_rate": 2.1174004192872118e-05,
      "loss": 0.2748,
      "step": 101
    },
    {
      "epoch": 0.021410579345088162,
      "grad_norm": 0.2776371240615845,
      "learning_rate": 2.138364779874214e-05,
      "loss": 0.2772,
      "step": 102
    },
    {
      "epoch": 0.02162048698572628,
      "grad_norm": 0.292316734790802,
      "learning_rate": 2.159329140461216e-05,
      "loss": 0.2732,
      "step": 103
    },
    {
      "epoch": 0.0218303946263644,
      "grad_norm": 0.26235565543174744,
      "learning_rate": 2.1802935010482182e-05,
      "loss": 0.2592,
      "step": 104
    },
    {
      "epoch": 0.02204030226700252,
      "grad_norm": 0.2782291769981384,
      "learning_rate": 2.2012578616352204e-05,
      "loss": 0.2586,
      "step": 105
    },
    {
      "epoch": 0.02225020990764064,
      "grad_norm": 0.24781855940818787,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.2571,
      "step": 106
    },
    {
      "epoch": 0.022460117548278757,
      "grad_norm": 0.3645837604999542,
      "learning_rate": 2.2431865828092242e-05,
      "loss": 0.247,
      "step": 107
    },
    {
      "epoch": 0.022670025188916875,
      "grad_norm": 0.4096992313861847,
      "learning_rate": 2.2641509433962265e-05,
      "loss": 0.2648,
      "step": 108
    },
    {
      "epoch": 0.022879932829554996,
      "grad_norm": 0.36801740527153015,
      "learning_rate": 2.2851153039832284e-05,
      "loss": 0.2677,
      "step": 109
    },
    {
      "epoch": 0.023089840470193114,
      "grad_norm": 0.221563920378685,
      "learning_rate": 2.3060796645702306e-05,
      "loss": 0.2544,
      "step": 110
    },
    {
      "epoch": 0.023299748110831235,
      "grad_norm": 0.17734010517597198,
      "learning_rate": 2.327044025157233e-05,
      "loss": 0.2458,
      "step": 111
    },
    {
      "epoch": 0.023509655751469353,
      "grad_norm": 0.23815220594406128,
      "learning_rate": 2.348008385744235e-05,
      "loss": 0.2249,
      "step": 112
    },
    {
      "epoch": 0.023719563392107474,
      "grad_norm": 0.24534103274345398,
      "learning_rate": 2.368972746331237e-05,
      "loss": 0.2283,
      "step": 113
    },
    {
      "epoch": 0.02392947103274559,
      "grad_norm": 0.17044654488563538,
      "learning_rate": 2.3899371069182393e-05,
      "loss": 0.2543,
      "step": 114
    },
    {
      "epoch": 0.02413937867338371,
      "grad_norm": 0.22411420941352844,
      "learning_rate": 2.4109014675052412e-05,
      "loss": 0.2537,
      "step": 115
    },
    {
      "epoch": 0.02434928631402183,
      "grad_norm": 0.17107880115509033,
      "learning_rate": 2.431865828092243e-05,
      "loss": 0.2477,
      "step": 116
    },
    {
      "epoch": 0.02455919395465995,
      "grad_norm": 0.19663883745670319,
      "learning_rate": 2.4528301886792453e-05,
      "loss": 0.2503,
      "step": 117
    },
    {
      "epoch": 0.02476910159529807,
      "grad_norm": 0.2755718231201172,
      "learning_rate": 2.4737945492662476e-05,
      "loss": 0.201,
      "step": 118
    },
    {
      "epoch": 0.024979009235936187,
      "grad_norm": 0.16535572707653046,
      "learning_rate": 2.4947589098532495e-05,
      "loss": 0.2364,
      "step": 119
    },
    {
      "epoch": 0.02518891687657431,
      "grad_norm": 0.15798090398311615,
      "learning_rate": 2.5157232704402517e-05,
      "loss": 0.214,
      "step": 120
    },
    {
      "epoch": 0.025398824517212426,
      "grad_norm": 0.1900860220193863,
      "learning_rate": 2.5366876310272536e-05,
      "loss": 0.2182,
      "step": 121
    },
    {
      "epoch": 0.025608732157850547,
      "grad_norm": 0.18855144083499908,
      "learning_rate": 2.5576519916142562e-05,
      "loss": 0.2123,
      "step": 122
    },
    {
      "epoch": 0.025818639798488665,
      "grad_norm": 0.1780049353837967,
      "learning_rate": 2.578616352201258e-05,
      "loss": 0.2197,
      "step": 123
    },
    {
      "epoch": 0.026028547439126783,
      "grad_norm": 0.27040061354637146,
      "learning_rate": 2.59958071278826e-05,
      "loss": 0.2395,
      "step": 124
    },
    {
      "epoch": 0.026238455079764904,
      "grad_norm": 0.1983417570590973,
      "learning_rate": 2.6205450733752623e-05,
      "loss": 0.2197,
      "step": 125
    },
    {
      "epoch": 0.02644836272040302,
      "grad_norm": 0.1723383665084839,
      "learning_rate": 2.641509433962264e-05,
      "loss": 0.2183,
      "step": 126
    },
    {
      "epoch": 0.026658270361041143,
      "grad_norm": 0.24477605521678925,
      "learning_rate": 2.662473794549266e-05,
      "loss": 0.2293,
      "step": 127
    },
    {
      "epoch": 0.02686817800167926,
      "grad_norm": 0.15110079944133759,
      "learning_rate": 2.6834381551362687e-05,
      "loss": 0.2245,
      "step": 128
    },
    {
      "epoch": 0.02707808564231738,
      "grad_norm": 0.22366492450237274,
      "learning_rate": 2.7044025157232706e-05,
      "loss": 0.2031,
      "step": 129
    },
    {
      "epoch": 0.0272879932829555,
      "grad_norm": 0.32891571521759033,
      "learning_rate": 2.7253668763102725e-05,
      "loss": 0.1821,
      "step": 130
    },
    {
      "epoch": 0.02749790092359362,
      "grad_norm": 0.20027081668376923,
      "learning_rate": 2.746331236897275e-05,
      "loss": 0.223,
      "step": 131
    },
    {
      "epoch": 0.027707808564231738,
      "grad_norm": 0.2269366830587387,
      "learning_rate": 2.767295597484277e-05,
      "loss": 0.2246,
      "step": 132
    },
    {
      "epoch": 0.027917716204869856,
      "grad_norm": 0.1355280727148056,
      "learning_rate": 2.788259958071279e-05,
      "loss": 0.2183,
      "step": 133
    },
    {
      "epoch": 0.028127623845507977,
      "grad_norm": 0.29291626811027527,
      "learning_rate": 2.809224318658281e-05,
      "loss": 0.2049,
      "step": 134
    },
    {
      "epoch": 0.028337531486146095,
      "grad_norm": 0.1778186410665512,
      "learning_rate": 2.830188679245283e-05,
      "loss": 0.2095,
      "step": 135
    },
    {
      "epoch": 0.028547439126784216,
      "grad_norm": 0.23263931274414062,
      "learning_rate": 2.851153039832285e-05,
      "loss": 0.2372,
      "step": 136
    },
    {
      "epoch": 0.028757346767422334,
      "grad_norm": 0.2121749222278595,
      "learning_rate": 2.8721174004192875e-05,
      "loss": 0.2107,
      "step": 137
    },
    {
      "epoch": 0.028967254408060455,
      "grad_norm": 0.19954991340637207,
      "learning_rate": 2.8930817610062894e-05,
      "loss": 0.2181,
      "step": 138
    },
    {
      "epoch": 0.029177162048698572,
      "grad_norm": 0.15431921184062958,
      "learning_rate": 2.9140461215932913e-05,
      "loss": 0.2198,
      "step": 139
    },
    {
      "epoch": 0.02938706968933669,
      "grad_norm": 0.17603729665279388,
      "learning_rate": 2.935010482180294e-05,
      "loss": 0.2339,
      "step": 140
    },
    {
      "epoch": 0.02959697732997481,
      "grad_norm": 0.1604471355676651,
      "learning_rate": 2.9559748427672958e-05,
      "loss": 0.2064,
      "step": 141
    },
    {
      "epoch": 0.02980688497061293,
      "grad_norm": 0.17169184982776642,
      "learning_rate": 2.976939203354298e-05,
      "loss": 0.1987,
      "step": 142
    },
    {
      "epoch": 0.03001679261125105,
      "grad_norm": 0.1285230815410614,
      "learning_rate": 2.9979035639413e-05,
      "loss": 0.2342,
      "step": 143
    },
    {
      "epoch": 0.030226700251889168,
      "grad_norm": 0.1755395084619522,
      "learning_rate": 3.018867924528302e-05,
      "loss": 0.222,
      "step": 144
    },
    {
      "epoch": 0.03043660789252729,
      "grad_norm": 0.15474987030029297,
      "learning_rate": 3.0398322851153044e-05,
      "loss": 0.2159,
      "step": 145
    },
    {
      "epoch": 0.030646515533165407,
      "grad_norm": 0.12986472249031067,
      "learning_rate": 3.060796645702306e-05,
      "loss": 0.2124,
      "step": 146
    },
    {
      "epoch": 0.030856423173803528,
      "grad_norm": 0.1458188146352768,
      "learning_rate": 3.081761006289308e-05,
      "loss": 0.214,
      "step": 147
    },
    {
      "epoch": 0.031066330814441646,
      "grad_norm": 0.1323792040348053,
      "learning_rate": 3.1027253668763105e-05,
      "loss": 0.2153,
      "step": 148
    },
    {
      "epoch": 0.03127623845507976,
      "grad_norm": 0.16542711853981018,
      "learning_rate": 3.1236897274633124e-05,
      "loss": 0.2154,
      "step": 149
    },
    {
      "epoch": 0.031486146095717885,
      "grad_norm": 0.17730407416820526,
      "learning_rate": 3.144654088050314e-05,
      "loss": 0.2202,
      "step": 150
    },
    {
      "epoch": 0.031696053736356006,
      "grad_norm": 0.15039502084255219,
      "learning_rate": 3.165618448637317e-05,
      "loss": 0.2056,
      "step": 151
    },
    {
      "epoch": 0.03190596137699412,
      "grad_norm": 0.20309150218963623,
      "learning_rate": 3.186582809224319e-05,
      "loss": 0.2175,
      "step": 152
    },
    {
      "epoch": 0.03211586901763224,
      "grad_norm": 0.16652604937553406,
      "learning_rate": 3.207547169811321e-05,
      "loss": 0.2076,
      "step": 153
    },
    {
      "epoch": 0.03232577665827036,
      "grad_norm": 0.14530467987060547,
      "learning_rate": 3.228511530398323e-05,
      "loss": 0.2167,
      "step": 154
    },
    {
      "epoch": 0.032535684298908484,
      "grad_norm": 0.13003528118133545,
      "learning_rate": 3.249475890985325e-05,
      "loss": 0.2089,
      "step": 155
    },
    {
      "epoch": 0.0327455919395466,
      "grad_norm": 0.16985855996608734,
      "learning_rate": 3.270440251572327e-05,
      "loss": 0.1994,
      "step": 156
    },
    {
      "epoch": 0.03295549958018472,
      "grad_norm": 0.18479777872562408,
      "learning_rate": 3.29140461215933e-05,
      "loss": 0.2127,
      "step": 157
    },
    {
      "epoch": 0.03316540722082284,
      "grad_norm": 0.1541491150856018,
      "learning_rate": 3.3123689727463316e-05,
      "loss": 0.2208,
      "step": 158
    },
    {
      "epoch": 0.033375314861460954,
      "grad_norm": 0.13511165976524353,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.218,
      "step": 159
    },
    {
      "epoch": 0.033585222502099076,
      "grad_norm": 0.1392865628004074,
      "learning_rate": 3.354297693920336e-05,
      "loss": 0.226,
      "step": 160
    },
    {
      "epoch": 0.0337951301427372,
      "grad_norm": 0.1614847183227539,
      "learning_rate": 3.375262054507338e-05,
      "loss": 0.2072,
      "step": 161
    },
    {
      "epoch": 0.03400503778337532,
      "grad_norm": 0.12186679244041443,
      "learning_rate": 3.39622641509434e-05,
      "loss": 0.2193,
      "step": 162
    },
    {
      "epoch": 0.03421494542401343,
      "grad_norm": 0.1777278333902359,
      "learning_rate": 3.417190775681342e-05,
      "loss": 0.202,
      "step": 163
    },
    {
      "epoch": 0.03442485306465155,
      "grad_norm": 0.13323499262332916,
      "learning_rate": 3.438155136268344e-05,
      "loss": 0.1983,
      "step": 164
    },
    {
      "epoch": 0.034634760705289674,
      "grad_norm": 0.22301942110061646,
      "learning_rate": 3.4591194968553456e-05,
      "loss": 0.2252,
      "step": 165
    },
    {
      "epoch": 0.03484466834592779,
      "grad_norm": 0.1538372039794922,
      "learning_rate": 3.480083857442348e-05,
      "loss": 0.2071,
      "step": 166
    },
    {
      "epoch": 0.03505457598656591,
      "grad_norm": 0.128251314163208,
      "learning_rate": 3.50104821802935e-05,
      "loss": 0.214,
      "step": 167
    },
    {
      "epoch": 0.03526448362720403,
      "grad_norm": 0.26556313037872314,
      "learning_rate": 3.522012578616352e-05,
      "loss": 0.1863,
      "step": 168
    },
    {
      "epoch": 0.03547439126784215,
      "grad_norm": 0.14938302338123322,
      "learning_rate": 3.5429769392033546e-05,
      "loss": 0.2152,
      "step": 169
    },
    {
      "epoch": 0.035684298908480266,
      "grad_norm": 0.19645771384239197,
      "learning_rate": 3.5639412997903565e-05,
      "loss": 0.1977,
      "step": 170
    },
    {
      "epoch": 0.03589420654911839,
      "grad_norm": 0.1725340038537979,
      "learning_rate": 3.5849056603773584e-05,
      "loss": 0.1981,
      "step": 171
    },
    {
      "epoch": 0.03610411418975651,
      "grad_norm": 0.16048069298267365,
      "learning_rate": 3.605870020964361e-05,
      "loss": 0.1983,
      "step": 172
    },
    {
      "epoch": 0.03631402183039462,
      "grad_norm": 0.21855585277080536,
      "learning_rate": 3.626834381551363e-05,
      "loss": 0.2321,
      "step": 173
    },
    {
      "epoch": 0.036523929471032744,
      "grad_norm": 0.1136787012219429,
      "learning_rate": 3.647798742138365e-05,
      "loss": 0.1973,
      "step": 174
    },
    {
      "epoch": 0.036733837111670865,
      "grad_norm": 0.15145623683929443,
      "learning_rate": 3.6687631027253674e-05,
      "loss": 0.1947,
      "step": 175
    },
    {
      "epoch": 0.036943744752308987,
      "grad_norm": 0.21631890535354614,
      "learning_rate": 3.689727463312369e-05,
      "loss": 0.2151,
      "step": 176
    },
    {
      "epoch": 0.0371536523929471,
      "grad_norm": 0.2623152434825897,
      "learning_rate": 3.710691823899371e-05,
      "loss": 0.1814,
      "step": 177
    },
    {
      "epoch": 0.03736356003358522,
      "grad_norm": 0.1753605455160141,
      "learning_rate": 3.731656184486374e-05,
      "loss": 0.203,
      "step": 178
    },
    {
      "epoch": 0.03757346767422334,
      "grad_norm": 0.10878176242113113,
      "learning_rate": 3.752620545073376e-05,
      "loss": 0.2052,
      "step": 179
    },
    {
      "epoch": 0.037783375314861464,
      "grad_norm": 0.13699688017368317,
      "learning_rate": 3.7735849056603776e-05,
      "loss": 0.2004,
      "step": 180
    },
    {
      "epoch": 0.03799328295549958,
      "grad_norm": 0.14288806915283203,
      "learning_rate": 3.7945492662473795e-05,
      "loss": 0.1745,
      "step": 181
    },
    {
      "epoch": 0.0382031905961377,
      "grad_norm": 0.12457548081874847,
      "learning_rate": 3.8155136268343814e-05,
      "loss": 0.194,
      "step": 182
    },
    {
      "epoch": 0.03841309823677582,
      "grad_norm": 0.167145237326622,
      "learning_rate": 3.836477987421384e-05,
      "loss": 0.2027,
      "step": 183
    },
    {
      "epoch": 0.038623005877413935,
      "grad_norm": 0.12857979536056519,
      "learning_rate": 3.857442348008386e-05,
      "loss": 0.2126,
      "step": 184
    },
    {
      "epoch": 0.038832913518052056,
      "grad_norm": 0.16190126538276672,
      "learning_rate": 3.878406708595388e-05,
      "loss": 0.2037,
      "step": 185
    },
    {
      "epoch": 0.03904282115869018,
      "grad_norm": 0.168744757771492,
      "learning_rate": 3.8993710691823904e-05,
      "loss": 0.2108,
      "step": 186
    },
    {
      "epoch": 0.0392527287993283,
      "grad_norm": 0.1676539033651352,
      "learning_rate": 3.920335429769392e-05,
      "loss": 0.216,
      "step": 187
    },
    {
      "epoch": 0.03946263643996641,
      "grad_norm": 0.13556820154190063,
      "learning_rate": 3.941299790356394e-05,
      "loss": 0.2085,
      "step": 188
    },
    {
      "epoch": 0.039672544080604534,
      "grad_norm": 0.1797979772090912,
      "learning_rate": 3.962264150943397e-05,
      "loss": 0.1933,
      "step": 189
    },
    {
      "epoch": 0.039882451721242655,
      "grad_norm": 0.20826327800750732,
      "learning_rate": 3.983228511530399e-05,
      "loss": 0.2214,
      "step": 190
    },
    {
      "epoch": 0.04009235936188077,
      "grad_norm": 0.19972363114356995,
      "learning_rate": 4.0041928721174006e-05,
      "loss": 0.1941,
      "step": 191
    },
    {
      "epoch": 0.04030226700251889,
      "grad_norm": 0.149556502699852,
      "learning_rate": 4.025157232704403e-05,
      "loss": 0.1956,
      "step": 192
    },
    {
      "epoch": 0.04051217464315701,
      "grad_norm": 0.22496013343334198,
      "learning_rate": 4.046121593291405e-05,
      "loss": 0.1973,
      "step": 193
    },
    {
      "epoch": 0.04072208228379513,
      "grad_norm": 0.16132576763629913,
      "learning_rate": 4.067085953878407e-05,
      "loss": 0.203,
      "step": 194
    },
    {
      "epoch": 0.04093198992443325,
      "grad_norm": 0.17156128585338593,
      "learning_rate": 4.088050314465409e-05,
      "loss": 0.1942,
      "step": 195
    },
    {
      "epoch": 0.04114189756507137,
      "grad_norm": 0.14846180379390717,
      "learning_rate": 4.109014675052411e-05,
      "loss": 0.2005,
      "step": 196
    },
    {
      "epoch": 0.04135180520570949,
      "grad_norm": 0.20252752304077148,
      "learning_rate": 4.129979035639413e-05,
      "loss": 0.186,
      "step": 197
    },
    {
      "epoch": 0.041561712846347604,
      "grad_norm": 0.16286462545394897,
      "learning_rate": 4.150943396226415e-05,
      "loss": 0.1997,
      "step": 198
    },
    {
      "epoch": 0.041771620486985725,
      "grad_norm": 0.13008786737918854,
      "learning_rate": 4.171907756813417e-05,
      "loss": 0.2052,
      "step": 199
    },
    {
      "epoch": 0.041981528127623846,
      "grad_norm": 0.13853180408477783,
      "learning_rate": 4.192872117400419e-05,
      "loss": 0.1907,
      "step": 200
    },
    {
      "epoch": 0.04219143576826197,
      "grad_norm": 0.20382314920425415,
      "learning_rate": 4.213836477987422e-05,
      "loss": 0.1931,
      "step": 201
    },
    {
      "epoch": 0.04240134340890008,
      "grad_norm": 0.23206844925880432,
      "learning_rate": 4.2348008385744236e-05,
      "loss": 0.1849,
      "step": 202
    },
    {
      "epoch": 0.0426112510495382,
      "grad_norm": 0.2456827312707901,
      "learning_rate": 4.2557651991614255e-05,
      "loss": 0.2088,
      "step": 203
    },
    {
      "epoch": 0.042821158690176324,
      "grad_norm": 0.15247821807861328,
      "learning_rate": 4.276729559748428e-05,
      "loss": 0.1944,
      "step": 204
    },
    {
      "epoch": 0.043031066330814445,
      "grad_norm": 0.174981027841568,
      "learning_rate": 4.29769392033543e-05,
      "loss": 0.2161,
      "step": 205
    },
    {
      "epoch": 0.04324097397145256,
      "grad_norm": 0.20193031430244446,
      "learning_rate": 4.318658280922432e-05,
      "loss": 0.2084,
      "step": 206
    },
    {
      "epoch": 0.04345088161209068,
      "grad_norm": 0.20125791430473328,
      "learning_rate": 4.3396226415094345e-05,
      "loss": 0.189,
      "step": 207
    },
    {
      "epoch": 0.0436607892527288,
      "grad_norm": 0.16958673298358917,
      "learning_rate": 4.3605870020964364e-05,
      "loss": 0.1952,
      "step": 208
    },
    {
      "epoch": 0.043870696893366916,
      "grad_norm": 0.20714177191257477,
      "learning_rate": 4.381551362683438e-05,
      "loss": 0.1832,
      "step": 209
    },
    {
      "epoch": 0.04408060453400504,
      "grad_norm": 0.14229562878608704,
      "learning_rate": 4.402515723270441e-05,
      "loss": 0.1806,
      "step": 210
    },
    {
      "epoch": 0.04429051217464316,
      "grad_norm": 0.1985626220703125,
      "learning_rate": 4.423480083857443e-05,
      "loss": 0.1973,
      "step": 211
    },
    {
      "epoch": 0.04450041981528128,
      "grad_norm": 0.13714846968650818,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.2018,
      "step": 212
    },
    {
      "epoch": 0.044710327455919394,
      "grad_norm": 0.25591611862182617,
      "learning_rate": 4.4654088050314466e-05,
      "loss": 0.2001,
      "step": 213
    },
    {
      "epoch": 0.044920235096557515,
      "grad_norm": 0.29017260670661926,
      "learning_rate": 4.4863731656184485e-05,
      "loss": 0.2278,
      "step": 214
    },
    {
      "epoch": 0.045130142737195636,
      "grad_norm": 0.12594066560268402,
      "learning_rate": 4.5073375262054504e-05,
      "loss": 0.2023,
      "step": 215
    },
    {
      "epoch": 0.04534005037783375,
      "grad_norm": 0.1602821797132492,
      "learning_rate": 4.528301886792453e-05,
      "loss": 0.2101,
      "step": 216
    },
    {
      "epoch": 0.04554995801847187,
      "grad_norm": 0.36664336919784546,
      "learning_rate": 4.549266247379455e-05,
      "loss": 0.1665,
      "step": 217
    },
    {
      "epoch": 0.04575986565910999,
      "grad_norm": 0.20665952563285828,
      "learning_rate": 4.570230607966457e-05,
      "loss": 0.1965,
      "step": 218
    },
    {
      "epoch": 0.045969773299748114,
      "grad_norm": 0.15664128959178925,
      "learning_rate": 4.5911949685534594e-05,
      "loss": 0.2104,
      "step": 219
    },
    {
      "epoch": 0.04617968094038623,
      "grad_norm": 0.14733830094337463,
      "learning_rate": 4.612159329140461e-05,
      "loss": 0.1955,
      "step": 220
    },
    {
      "epoch": 0.04638958858102435,
      "grad_norm": 0.19135522842407227,
      "learning_rate": 4.633123689727464e-05,
      "loss": 0.1947,
      "step": 221
    },
    {
      "epoch": 0.04659949622166247,
      "grad_norm": 0.24508413672447205,
      "learning_rate": 4.654088050314466e-05,
      "loss": 0.2005,
      "step": 222
    },
    {
      "epoch": 0.046809403862300585,
      "grad_norm": 0.16794531047344208,
      "learning_rate": 4.6750524109014677e-05,
      "loss": 0.1958,
      "step": 223
    },
    {
      "epoch": 0.047019311502938706,
      "grad_norm": 0.1870536506175995,
      "learning_rate": 4.69601677148847e-05,
      "loss": 0.1868,
      "step": 224
    },
    {
      "epoch": 0.04722921914357683,
      "grad_norm": 0.16214439272880554,
      "learning_rate": 4.716981132075472e-05,
      "loss": 0.2003,
      "step": 225
    },
    {
      "epoch": 0.04743912678421495,
      "grad_norm": 0.24978841841220856,
      "learning_rate": 4.737945492662474e-05,
      "loss": 0.1909,
      "step": 226
    },
    {
      "epoch": 0.04764903442485306,
      "grad_norm": 0.163265198469162,
      "learning_rate": 4.7589098532494766e-05,
      "loss": 0.2056,
      "step": 227
    },
    {
      "epoch": 0.04785894206549118,
      "grad_norm": 0.24885287880897522,
      "learning_rate": 4.7798742138364785e-05,
      "loss": 0.2035,
      "step": 228
    },
    {
      "epoch": 0.048068849706129305,
      "grad_norm": 0.1393681764602661,
      "learning_rate": 4.8008385744234804e-05,
      "loss": 0.1976,
      "step": 229
    },
    {
      "epoch": 0.04827875734676742,
      "grad_norm": 0.17042241990566254,
      "learning_rate": 4.8218029350104823e-05,
      "loss": 0.1824,
      "step": 230
    },
    {
      "epoch": 0.04848866498740554,
      "grad_norm": 0.1625502109527588,
      "learning_rate": 4.842767295597484e-05,
      "loss": 0.1904,
      "step": 231
    },
    {
      "epoch": 0.04869857262804366,
      "grad_norm": 0.14994169771671295,
      "learning_rate": 4.863731656184486e-05,
      "loss": 0.1926,
      "step": 232
    },
    {
      "epoch": 0.04890848026868178,
      "grad_norm": 0.15602821111679077,
      "learning_rate": 4.884696016771489e-05,
      "loss": 0.1828,
      "step": 233
    },
    {
      "epoch": 0.0491183879093199,
      "grad_norm": 0.13405688107013702,
      "learning_rate": 4.9056603773584906e-05,
      "loss": 0.1989,
      "step": 234
    },
    {
      "epoch": 0.04932829554995802,
      "grad_norm": 0.18559689819812775,
      "learning_rate": 4.9266247379454926e-05,
      "loss": 0.2131,
      "step": 235
    },
    {
      "epoch": 0.04953820319059614,
      "grad_norm": 0.1557319462299347,
      "learning_rate": 4.947589098532495e-05,
      "loss": 0.19,
      "step": 236
    },
    {
      "epoch": 0.04974811083123426,
      "grad_norm": 0.2162303477525711,
      "learning_rate": 4.968553459119497e-05,
      "loss": 0.196,
      "step": 237
    },
    {
      "epoch": 0.049958018471872374,
      "grad_norm": 0.17403477430343628,
      "learning_rate": 4.989517819706499e-05,
      "loss": 0.1997,
      "step": 238
    },
    {
      "epoch": 0.050167926112510496,
      "grad_norm": 0.11738390475511551,
      "learning_rate": 5.010482180293501e-05,
      "loss": 0.1955,
      "step": 239
    },
    {
      "epoch": 0.05037783375314862,
      "grad_norm": 0.15942999720573425,
      "learning_rate": 5.0314465408805034e-05,
      "loss": 0.1899,
      "step": 240
    },
    {
      "epoch": 0.05058774139378673,
      "grad_norm": 0.14695511758327484,
      "learning_rate": 5.052410901467506e-05,
      "loss": 0.2032,
      "step": 241
    },
    {
      "epoch": 0.05079764903442485,
      "grad_norm": 0.16291062533855438,
      "learning_rate": 5.073375262054507e-05,
      "loss": 0.1929,
      "step": 242
    },
    {
      "epoch": 0.05100755667506297,
      "grad_norm": 0.18514905869960785,
      "learning_rate": 5.09433962264151e-05,
      "loss": 0.1899,
      "step": 243
    },
    {
      "epoch": 0.051217464315701094,
      "grad_norm": 0.2196233868598938,
      "learning_rate": 5.1153039832285124e-05,
      "loss": 0.1741,
      "step": 244
    },
    {
      "epoch": 0.05142737195633921,
      "grad_norm": 0.19183433055877686,
      "learning_rate": 5.1362683438155136e-05,
      "loss": 0.1985,
      "step": 245
    },
    {
      "epoch": 0.05163727959697733,
      "grad_norm": 0.1604142189025879,
      "learning_rate": 5.157232704402516e-05,
      "loss": 0.1865,
      "step": 246
    },
    {
      "epoch": 0.05184718723761545,
      "grad_norm": 0.18311725556850433,
      "learning_rate": 5.178197064989518e-05,
      "loss": 0.2005,
      "step": 247
    },
    {
      "epoch": 0.052057094878253565,
      "grad_norm": 0.21732251346111298,
      "learning_rate": 5.19916142557652e-05,
      "loss": 0.2032,
      "step": 248
    },
    {
      "epoch": 0.052267002518891686,
      "grad_norm": 0.2600694000720978,
      "learning_rate": 5.220125786163522e-05,
      "loss": 0.1838,
      "step": 249
    },
    {
      "epoch": 0.05247691015952981,
      "grad_norm": 0.18634290993213654,
      "learning_rate": 5.2410901467505245e-05,
      "loss": 0.1809,
      "step": 250
    },
    {
      "epoch": 0.05268681780016793,
      "grad_norm": 0.14735780656337738,
      "learning_rate": 5.262054507337526e-05,
      "loss": 0.1981,
      "step": 251
    },
    {
      "epoch": 0.05289672544080604,
      "grad_norm": 0.16118381917476654,
      "learning_rate": 5.283018867924528e-05,
      "loss": 0.1836,
      "step": 252
    },
    {
      "epoch": 0.053106633081444164,
      "grad_norm": 0.17707999050617218,
      "learning_rate": 5.303983228511531e-05,
      "loss": 0.2044,
      "step": 253
    },
    {
      "epoch": 0.053316540722082285,
      "grad_norm": 0.25897523760795593,
      "learning_rate": 5.324947589098532e-05,
      "loss": 0.1753,
      "step": 254
    },
    {
      "epoch": 0.0535264483627204,
      "grad_norm": 0.1371389776468277,
      "learning_rate": 5.345911949685535e-05,
      "loss": 0.1839,
      "step": 255
    },
    {
      "epoch": 0.05373635600335852,
      "grad_norm": 0.21849682927131653,
      "learning_rate": 5.366876310272537e-05,
      "loss": 0.1938,
      "step": 256
    },
    {
      "epoch": 0.05394626364399664,
      "grad_norm": 0.16861748695373535,
      "learning_rate": 5.3878406708595385e-05,
      "loss": 0.1751,
      "step": 257
    },
    {
      "epoch": 0.05415617128463476,
      "grad_norm": 0.19400931894779205,
      "learning_rate": 5.408805031446541e-05,
      "loss": 0.2117,
      "step": 258
    },
    {
      "epoch": 0.05436607892527288,
      "grad_norm": 0.18074113130569458,
      "learning_rate": 5.429769392033544e-05,
      "loss": 0.1949,
      "step": 259
    },
    {
      "epoch": 0.054575986565911,
      "grad_norm": 0.17707990109920502,
      "learning_rate": 5.450733752620545e-05,
      "loss": 0.199,
      "step": 260
    },
    {
      "epoch": 0.05478589420654912,
      "grad_norm": 0.16568966209888458,
      "learning_rate": 5.4716981132075475e-05,
      "loss": 0.2027,
      "step": 261
    },
    {
      "epoch": 0.05499580184718724,
      "grad_norm": 0.24486149847507477,
      "learning_rate": 5.49266247379455e-05,
      "loss": 0.2042,
      "step": 262
    },
    {
      "epoch": 0.055205709487825355,
      "grad_norm": 0.20431551337242126,
      "learning_rate": 5.513626834381551e-05,
      "loss": 0.2013,
      "step": 263
    },
    {
      "epoch": 0.055415617128463476,
      "grad_norm": 0.21347559988498688,
      "learning_rate": 5.534591194968554e-05,
      "loss": 0.1687,
      "step": 264
    },
    {
      "epoch": 0.0556255247691016,
      "grad_norm": 0.22354485094547272,
      "learning_rate": 5.555555555555556e-05,
      "loss": 0.1819,
      "step": 265
    },
    {
      "epoch": 0.05583543240973971,
      "grad_norm": 0.2891826033592224,
      "learning_rate": 5.576519916142558e-05,
      "loss": 0.1927,
      "step": 266
    },
    {
      "epoch": 0.05604534005037783,
      "grad_norm": 0.2465565800666809,
      "learning_rate": 5.5974842767295596e-05,
      "loss": 0.1852,
      "step": 267
    },
    {
      "epoch": 0.056255247691015954,
      "grad_norm": 0.1743604689836502,
      "learning_rate": 5.618448637316562e-05,
      "loss": 0.1722,
      "step": 268
    },
    {
      "epoch": 0.056465155331654075,
      "grad_norm": 0.2461417019367218,
      "learning_rate": 5.6394129979035634e-05,
      "loss": 0.1816,
      "step": 269
    },
    {
      "epoch": 0.05667506297229219,
      "grad_norm": 0.21032604575157166,
      "learning_rate": 5.660377358490566e-05,
      "loss": 0.182,
      "step": 270
    },
    {
      "epoch": 0.05688497061293031,
      "grad_norm": 0.2683754861354828,
      "learning_rate": 5.6813417190775686e-05,
      "loss": 0.1718,
      "step": 271
    },
    {
      "epoch": 0.05709487825356843,
      "grad_norm": 0.18455228209495544,
      "learning_rate": 5.70230607966457e-05,
      "loss": 0.1954,
      "step": 272
    },
    {
      "epoch": 0.057304785894206546,
      "grad_norm": 0.22255247831344604,
      "learning_rate": 5.7232704402515724e-05,
      "loss": 0.1922,
      "step": 273
    },
    {
      "epoch": 0.05751469353484467,
      "grad_norm": 0.22789119184017181,
      "learning_rate": 5.744234800838575e-05,
      "loss": 0.1882,
      "step": 274
    },
    {
      "epoch": 0.05772460117548279,
      "grad_norm": 0.20874802768230438,
      "learning_rate": 5.765199161425576e-05,
      "loss": 0.1859,
      "step": 275
    },
    {
      "epoch": 0.05793450881612091,
      "grad_norm": 0.17554089426994324,
      "learning_rate": 5.786163522012579e-05,
      "loss": 0.1793,
      "step": 276
    },
    {
      "epoch": 0.058144416456759024,
      "grad_norm": 0.2011173665523529,
      "learning_rate": 5.8071278825995814e-05,
      "loss": 0.2009,
      "step": 277
    },
    {
      "epoch": 0.058354324097397145,
      "grad_norm": 0.28261420130729675,
      "learning_rate": 5.8280922431865826e-05,
      "loss": 0.2002,
      "step": 278
    },
    {
      "epoch": 0.058564231738035266,
      "grad_norm": 0.2356766015291214,
      "learning_rate": 5.849056603773585e-05,
      "loss": 0.2095,
      "step": 279
    },
    {
      "epoch": 0.05877413937867338,
      "grad_norm": 0.15072722733020782,
      "learning_rate": 5.870020964360588e-05,
      "loss": 0.1781,
      "step": 280
    },
    {
      "epoch": 0.0589840470193115,
      "grad_norm": 0.1446981132030487,
      "learning_rate": 5.89098532494759e-05,
      "loss": 0.1813,
      "step": 281
    },
    {
      "epoch": 0.05919395465994962,
      "grad_norm": 0.16516901552677155,
      "learning_rate": 5.9119496855345916e-05,
      "loss": 0.1946,
      "step": 282
    },
    {
      "epoch": 0.059403862300587744,
      "grad_norm": 0.16732774674892426,
      "learning_rate": 5.9329140461215935e-05,
      "loss": 0.1813,
      "step": 283
    },
    {
      "epoch": 0.05961376994122586,
      "grad_norm": 0.2000836730003357,
      "learning_rate": 5.953878406708596e-05,
      "loss": 0.1918,
      "step": 284
    },
    {
      "epoch": 0.05982367758186398,
      "grad_norm": 0.15576116740703583,
      "learning_rate": 5.974842767295597e-05,
      "loss": 0.1814,
      "step": 285
    },
    {
      "epoch": 0.0600335852225021,
      "grad_norm": 0.18421867489814758,
      "learning_rate": 5.9958071278826e-05,
      "loss": 0.1745,
      "step": 286
    },
    {
      "epoch": 0.06024349286314022,
      "grad_norm": 0.2663988471031189,
      "learning_rate": 6.0167714884696025e-05,
      "loss": 0.1889,
      "step": 287
    },
    {
      "epoch": 0.060453400503778336,
      "grad_norm": 0.18096649646759033,
      "learning_rate": 6.037735849056604e-05,
      "loss": 0.191,
      "step": 288
    },
    {
      "epoch": 0.06066330814441646,
      "grad_norm": 0.24025796353816986,
      "learning_rate": 6.058700209643606e-05,
      "loss": 0.1907,
      "step": 289
    },
    {
      "epoch": 0.06087321578505458,
      "grad_norm": 0.1682557910680771,
      "learning_rate": 6.079664570230609e-05,
      "loss": 0.1747,
      "step": 290
    },
    {
      "epoch": 0.06108312342569269,
      "grad_norm": 0.2801767587661743,
      "learning_rate": 6.10062893081761e-05,
      "loss": 0.1813,
      "step": 291
    },
    {
      "epoch": 0.061293031066330814,
      "grad_norm": 0.2288123071193695,
      "learning_rate": 6.121593291404612e-05,
      "loss": 0.1681,
      "step": 292
    },
    {
      "epoch": 0.061502938706968935,
      "grad_norm": 0.14664186537265778,
      "learning_rate": 6.142557651991615e-05,
      "loss": 0.1723,
      "step": 293
    },
    {
      "epoch": 0.061712846347607056,
      "grad_norm": 0.15858514606952667,
      "learning_rate": 6.163522012578616e-05,
      "loss": 0.198,
      "step": 294
    },
    {
      "epoch": 0.06192275398824517,
      "grad_norm": 0.24689258635044098,
      "learning_rate": 6.184486373165618e-05,
      "loss": 0.1865,
      "step": 295
    },
    {
      "epoch": 0.06213266162888329,
      "grad_norm": 0.19141900539398193,
      "learning_rate": 6.205450733752621e-05,
      "loss": 0.2103,
      "step": 296
    },
    {
      "epoch": 0.06234256926952141,
      "grad_norm": 0.22445173561573029,
      "learning_rate": 6.226415094339622e-05,
      "loss": 0.1778,
      "step": 297
    },
    {
      "epoch": 0.06255247691015953,
      "grad_norm": 0.1805533766746521,
      "learning_rate": 6.247379454926625e-05,
      "loss": 0.1752,
      "step": 298
    },
    {
      "epoch": 0.06276238455079765,
      "grad_norm": 0.17849349975585938,
      "learning_rate": 6.268343815513627e-05,
      "loss": 0.1735,
      "step": 299
    },
    {
      "epoch": 0.06297229219143577,
      "grad_norm": 0.2601464092731476,
      "learning_rate": 6.289308176100629e-05,
      "loss": 0.1868,
      "step": 300
    },
    {
      "epoch": 0.06318219983207389,
      "grad_norm": 0.2865089178085327,
      "learning_rate": 6.310272536687631e-05,
      "loss": 0.2058,
      "step": 301
    },
    {
      "epoch": 0.06339210747271201,
      "grad_norm": 0.1764407902956009,
      "learning_rate": 6.331236897274634e-05,
      "loss": 0.1828,
      "step": 302
    },
    {
      "epoch": 0.06360201511335013,
      "grad_norm": 0.136027991771698,
      "learning_rate": 6.352201257861635e-05,
      "loss": 0.1841,
      "step": 303
    },
    {
      "epoch": 0.06381192275398824,
      "grad_norm": 0.26960527896881104,
      "learning_rate": 6.373165618448638e-05,
      "loss": 0.1806,
      "step": 304
    },
    {
      "epoch": 0.06402183039462636,
      "grad_norm": 0.2371356189250946,
      "learning_rate": 6.39412997903564e-05,
      "loss": 0.1755,
      "step": 305
    },
    {
      "epoch": 0.06423173803526448,
      "grad_norm": 0.16067345440387726,
      "learning_rate": 6.415094339622641e-05,
      "loss": 0.1899,
      "step": 306
    },
    {
      "epoch": 0.0644416456759026,
      "grad_norm": 0.1733190417289734,
      "learning_rate": 6.436058700209644e-05,
      "loss": 0.1829,
      "step": 307
    },
    {
      "epoch": 0.06465155331654072,
      "grad_norm": 0.19170600175857544,
      "learning_rate": 6.457023060796647e-05,
      "loss": 0.1777,
      "step": 308
    },
    {
      "epoch": 0.06486146095717885,
      "grad_norm": 0.17290905117988586,
      "learning_rate": 6.477987421383648e-05,
      "loss": 0.187,
      "step": 309
    },
    {
      "epoch": 0.06507136859781697,
      "grad_norm": 0.3149113059043884,
      "learning_rate": 6.49895178197065e-05,
      "loss": 0.2037,
      "step": 310
    },
    {
      "epoch": 0.06528127623845507,
      "grad_norm": 0.15404744446277618,
      "learning_rate": 6.519916142557653e-05,
      "loss": 0.1909,
      "step": 311
    },
    {
      "epoch": 0.0654911838790932,
      "grad_norm": 0.157347172498703,
      "learning_rate": 6.540880503144654e-05,
      "loss": 0.191,
      "step": 312
    },
    {
      "epoch": 0.06570109151973132,
      "grad_norm": 0.23342733085155487,
      "learning_rate": 6.561844863731657e-05,
      "loss": 0.1867,
      "step": 313
    },
    {
      "epoch": 0.06591099916036944,
      "grad_norm": 0.2597595155239105,
      "learning_rate": 6.58280922431866e-05,
      "loss": 0.1949,
      "step": 314
    },
    {
      "epoch": 0.06612090680100756,
      "grad_norm": 0.23665842413902283,
      "learning_rate": 6.60377358490566e-05,
      "loss": 0.1898,
      "step": 315
    },
    {
      "epoch": 0.06633081444164568,
      "grad_norm": 0.18460237979888916,
      "learning_rate": 6.624737945492663e-05,
      "loss": 0.1824,
      "step": 316
    },
    {
      "epoch": 0.0665407220822838,
      "grad_norm": 0.1677280068397522,
      "learning_rate": 6.645702306079666e-05,
      "loss": 0.1871,
      "step": 317
    },
    {
      "epoch": 0.06675062972292191,
      "grad_norm": 0.1769377440214157,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.1779,
      "step": 318
    },
    {
      "epoch": 0.06696053736356003,
      "grad_norm": 0.1881011724472046,
      "learning_rate": 6.68763102725367e-05,
      "loss": 0.1711,
      "step": 319
    },
    {
      "epoch": 0.06717044500419815,
      "grad_norm": 0.19110549986362457,
      "learning_rate": 6.708595387840672e-05,
      "loss": 0.1823,
      "step": 320
    },
    {
      "epoch": 0.06738035264483627,
      "grad_norm": 0.26796162128448486,
      "learning_rate": 6.729559748427673e-05,
      "loss": 0.1863,
      "step": 321
    },
    {
      "epoch": 0.0675902602854744,
      "grad_norm": 0.17290090024471283,
      "learning_rate": 6.750524109014676e-05,
      "loss": 0.1816,
      "step": 322
    },
    {
      "epoch": 0.06780016792611251,
      "grad_norm": 0.2324109524488449,
      "learning_rate": 6.771488469601677e-05,
      "loss": 0.1944,
      "step": 323
    },
    {
      "epoch": 0.06801007556675064,
      "grad_norm": 0.24944299459457397,
      "learning_rate": 6.79245283018868e-05,
      "loss": 0.1965,
      "step": 324
    },
    {
      "epoch": 0.06821998320738874,
      "grad_norm": 0.2102229744195938,
      "learning_rate": 6.813417190775681e-05,
      "loss": 0.221,
      "step": 325
    },
    {
      "epoch": 0.06842989084802686,
      "grad_norm": 0.22497773170471191,
      "learning_rate": 6.834381551362684e-05,
      "loss": 0.1818,
      "step": 326
    },
    {
      "epoch": 0.06863979848866499,
      "grad_norm": 0.19047041237354279,
      "learning_rate": 6.855345911949685e-05,
      "loss": 0.1817,
      "step": 327
    },
    {
      "epoch": 0.0688497061293031,
      "grad_norm": 0.19890040159225464,
      "learning_rate": 6.876310272536687e-05,
      "loss": 0.2054,
      "step": 328
    },
    {
      "epoch": 0.06905961376994123,
      "grad_norm": 0.18274420499801636,
      "learning_rate": 6.89727463312369e-05,
      "loss": 0.1732,
      "step": 329
    },
    {
      "epoch": 0.06926952141057935,
      "grad_norm": 0.20556879043579102,
      "learning_rate": 6.918238993710691e-05,
      "loss": 0.1773,
      "step": 330
    },
    {
      "epoch": 0.06947942905121747,
      "grad_norm": 0.19632075726985931,
      "learning_rate": 6.939203354297694e-05,
      "loss": 0.1841,
      "step": 331
    },
    {
      "epoch": 0.06968933669185558,
      "grad_norm": 0.25757917761802673,
      "learning_rate": 6.960167714884696e-05,
      "loss": 0.1861,
      "step": 332
    },
    {
      "epoch": 0.0698992443324937,
      "grad_norm": 0.1654757410287857,
      "learning_rate": 6.981132075471698e-05,
      "loss": 0.1797,
      "step": 333
    },
    {
      "epoch": 0.07010915197313182,
      "grad_norm": 0.1424175500869751,
      "learning_rate": 7.0020964360587e-05,
      "loss": 0.1659,
      "step": 334
    },
    {
      "epoch": 0.07031905961376994,
      "grad_norm": 0.21559248864650726,
      "learning_rate": 7.023060796645703e-05,
      "loss": 0.175,
      "step": 335
    },
    {
      "epoch": 0.07052896725440806,
      "grad_norm": 0.3100188076496124,
      "learning_rate": 7.044025157232704e-05,
      "loss": 0.1644,
      "step": 336
    },
    {
      "epoch": 0.07073887489504618,
      "grad_norm": 0.2530849277973175,
      "learning_rate": 7.064989517819707e-05,
      "loss": 0.1697,
      "step": 337
    },
    {
      "epoch": 0.0709487825356843,
      "grad_norm": 0.2111438810825348,
      "learning_rate": 7.085953878406709e-05,
      "loss": 0.202,
      "step": 338
    },
    {
      "epoch": 0.07115869017632241,
      "grad_norm": 0.21798165142536163,
      "learning_rate": 7.10691823899371e-05,
      "loss": 0.1832,
      "step": 339
    },
    {
      "epoch": 0.07136859781696053,
      "grad_norm": 0.24779516458511353,
      "learning_rate": 7.127882599580713e-05,
      "loss": 0.1934,
      "step": 340
    },
    {
      "epoch": 0.07157850545759865,
      "grad_norm": 0.21718356013298035,
      "learning_rate": 7.148846960167716e-05,
      "loss": 0.1951,
      "step": 341
    },
    {
      "epoch": 0.07178841309823678,
      "grad_norm": 0.22320568561553955,
      "learning_rate": 7.169811320754717e-05,
      "loss": 0.1619,
      "step": 342
    },
    {
      "epoch": 0.0719983207388749,
      "grad_norm": 0.19393590092658997,
      "learning_rate": 7.19077568134172e-05,
      "loss": 0.201,
      "step": 343
    },
    {
      "epoch": 0.07220822837951302,
      "grad_norm": 0.1626208871603012,
      "learning_rate": 7.211740041928722e-05,
      "loss": 0.1848,
      "step": 344
    },
    {
      "epoch": 0.07241813602015114,
      "grad_norm": 0.2256711721420288,
      "learning_rate": 7.232704402515723e-05,
      "loss": 0.192,
      "step": 345
    },
    {
      "epoch": 0.07262804366078925,
      "grad_norm": 0.2225414514541626,
      "learning_rate": 7.253668763102726e-05,
      "loss": 0.1907,
      "step": 346
    },
    {
      "epoch": 0.07283795130142737,
      "grad_norm": 0.1786690205335617,
      "learning_rate": 7.274633123689728e-05,
      "loss": 0.1756,
      "step": 347
    },
    {
      "epoch": 0.07304785894206549,
      "grad_norm": 0.2155577391386032,
      "learning_rate": 7.29559748427673e-05,
      "loss": 0.1739,
      "step": 348
    },
    {
      "epoch": 0.07325776658270361,
      "grad_norm": 0.32533329725265503,
      "learning_rate": 7.316561844863732e-05,
      "loss": 0.1988,
      "step": 349
    },
    {
      "epoch": 0.07346767422334173,
      "grad_norm": 0.1870083510875702,
      "learning_rate": 7.337526205450735e-05,
      "loss": 0.1615,
      "step": 350
    },
    {
      "epoch": 0.07367758186397985,
      "grad_norm": 0.2160840779542923,
      "learning_rate": 7.358490566037736e-05,
      "loss": 0.1901,
      "step": 351
    },
    {
      "epoch": 0.07388748950461797,
      "grad_norm": 0.19049416482448578,
      "learning_rate": 7.379454926624739e-05,
      "loss": 0.1763,
      "step": 352
    },
    {
      "epoch": 0.0740973971452561,
      "grad_norm": 0.15733250975608826,
      "learning_rate": 7.400419287211741e-05,
      "loss": 0.1812,
      "step": 353
    },
    {
      "epoch": 0.0743073047858942,
      "grad_norm": 0.22470858693122864,
      "learning_rate": 7.421383647798742e-05,
      "loss": 0.194,
      "step": 354
    },
    {
      "epoch": 0.07451721242653232,
      "grad_norm": 0.3271860182285309,
      "learning_rate": 7.442348008385745e-05,
      "loss": 0.1708,
      "step": 355
    },
    {
      "epoch": 0.07472712006717044,
      "grad_norm": 0.17839424312114716,
      "learning_rate": 7.463312368972748e-05,
      "loss": 0.182,
      "step": 356
    },
    {
      "epoch": 0.07493702770780857,
      "grad_norm": 0.1907908171415329,
      "learning_rate": 7.484276729559749e-05,
      "loss": 0.163,
      "step": 357
    },
    {
      "epoch": 0.07514693534844669,
      "grad_norm": 0.20342503488063812,
      "learning_rate": 7.505241090146751e-05,
      "loss": 0.2029,
      "step": 358
    },
    {
      "epoch": 0.07535684298908481,
      "grad_norm": 0.21872438490390778,
      "learning_rate": 7.526205450733753e-05,
      "loss": 0.192,
      "step": 359
    },
    {
      "epoch": 0.07556675062972293,
      "grad_norm": 0.22313977777957916,
      "learning_rate": 7.547169811320755e-05,
      "loss": 0.1823,
      "step": 360
    },
    {
      "epoch": 0.07577665827036104,
      "grad_norm": 0.1931924819946289,
      "learning_rate": 7.568134171907756e-05,
      "loss": 0.1992,
      "step": 361
    },
    {
      "epoch": 0.07598656591099916,
      "grad_norm": 0.2859954535961151,
      "learning_rate": 7.589098532494759e-05,
      "loss": 0.1684,
      "step": 362
    },
    {
      "epoch": 0.07619647355163728,
      "grad_norm": 0.18601499497890472,
      "learning_rate": 7.610062893081762e-05,
      "loss": 0.1871,
      "step": 363
    },
    {
      "epoch": 0.0764063811922754,
      "grad_norm": 0.26345667243003845,
      "learning_rate": 7.631027253668763e-05,
      "loss": 0.1883,
      "step": 364
    },
    {
      "epoch": 0.07661628883291352,
      "grad_norm": 0.24455974996089935,
      "learning_rate": 7.651991614255765e-05,
      "loss": 0.1991,
      "step": 365
    },
    {
      "epoch": 0.07682619647355164,
      "grad_norm": 0.1787412464618683,
      "learning_rate": 7.672955974842768e-05,
      "loss": 0.1895,
      "step": 366
    },
    {
      "epoch": 0.07703610411418976,
      "grad_norm": 0.2711624205112457,
      "learning_rate": 7.693920335429769e-05,
      "loss": 0.1889,
      "step": 367
    },
    {
      "epoch": 0.07724601175482787,
      "grad_norm": 0.2764052152633667,
      "learning_rate": 7.714884696016772e-05,
      "loss": 0.1895,
      "step": 368
    },
    {
      "epoch": 0.07745591939546599,
      "grad_norm": 0.15490169823169708,
      "learning_rate": 7.735849056603774e-05,
      "loss": 0.1812,
      "step": 369
    },
    {
      "epoch": 0.07766582703610411,
      "grad_norm": 0.18659183382987976,
      "learning_rate": 7.756813417190776e-05,
      "loss": 0.1871,
      "step": 370
    },
    {
      "epoch": 0.07787573467674223,
      "grad_norm": 0.19188903272151947,
      "learning_rate": 7.777777777777778e-05,
      "loss": 0.1716,
      "step": 371
    },
    {
      "epoch": 0.07808564231738035,
      "grad_norm": 0.22174161672592163,
      "learning_rate": 7.798742138364781e-05,
      "loss": 0.1795,
      "step": 372
    },
    {
      "epoch": 0.07829554995801848,
      "grad_norm": 0.1882723569869995,
      "learning_rate": 7.819706498951782e-05,
      "loss": 0.1874,
      "step": 373
    },
    {
      "epoch": 0.0785054575986566,
      "grad_norm": 0.1762145459651947,
      "learning_rate": 7.840670859538785e-05,
      "loss": 0.1852,
      "step": 374
    },
    {
      "epoch": 0.0787153652392947,
      "grad_norm": 0.17931701242923737,
      "learning_rate": 7.861635220125787e-05,
      "loss": 0.1817,
      "step": 375
    },
    {
      "epoch": 0.07892527287993283,
      "grad_norm": 0.1833990216255188,
      "learning_rate": 7.882599580712788e-05,
      "loss": 0.1852,
      "step": 376
    },
    {
      "epoch": 0.07913518052057095,
      "grad_norm": 0.2758026421070099,
      "learning_rate": 7.903563941299791e-05,
      "loss": 0.1868,
      "step": 377
    },
    {
      "epoch": 0.07934508816120907,
      "grad_norm": 0.24222204089164734,
      "learning_rate": 7.924528301886794e-05,
      "loss": 0.186,
      "step": 378
    },
    {
      "epoch": 0.07955499580184719,
      "grad_norm": 0.17609156668186188,
      "learning_rate": 7.945492662473795e-05,
      "loss": 0.1807,
      "step": 379
    },
    {
      "epoch": 0.07976490344248531,
      "grad_norm": 0.23695167899131775,
      "learning_rate": 7.966457023060797e-05,
      "loss": 0.1799,
      "step": 380
    },
    {
      "epoch": 0.07997481108312343,
      "grad_norm": 0.25356245040893555,
      "learning_rate": 7.9874213836478e-05,
      "loss": 0.1899,
      "step": 381
    },
    {
      "epoch": 0.08018471872376154,
      "grad_norm": 0.23144365847110748,
      "learning_rate": 8.008385744234801e-05,
      "loss": 0.192,
      "step": 382
    },
    {
      "epoch": 0.08039462636439966,
      "grad_norm": 0.1521812379360199,
      "learning_rate": 8.029350104821804e-05,
      "loss": 0.1864,
      "step": 383
    },
    {
      "epoch": 0.08060453400503778,
      "grad_norm": 0.16725748777389526,
      "learning_rate": 8.050314465408806e-05,
      "loss": 0.202,
      "step": 384
    },
    {
      "epoch": 0.0808144416456759,
      "grad_norm": 0.21173058450222015,
      "learning_rate": 8.071278825995808e-05,
      "loss": 0.1751,
      "step": 385
    },
    {
      "epoch": 0.08102434928631402,
      "grad_norm": 0.15676653385162354,
      "learning_rate": 8.09224318658281e-05,
      "loss": 0.1946,
      "step": 386
    },
    {
      "epoch": 0.08123425692695214,
      "grad_norm": 0.21838362514972687,
      "learning_rate": 8.113207547169813e-05,
      "loss": 0.1889,
      "step": 387
    },
    {
      "epoch": 0.08144416456759027,
      "grad_norm": 0.19586238265037537,
      "learning_rate": 8.134171907756814e-05,
      "loss": 0.1884,
      "step": 388
    },
    {
      "epoch": 0.08165407220822837,
      "grad_norm": 0.21012739837169647,
      "learning_rate": 8.155136268343817e-05,
      "loss": 0.1822,
      "step": 389
    },
    {
      "epoch": 0.0818639798488665,
      "grad_norm": 0.2092917114496231,
      "learning_rate": 8.176100628930818e-05,
      "loss": 0.1783,
      "step": 390
    },
    {
      "epoch": 0.08207388748950462,
      "grad_norm": 0.3745954930782318,
      "learning_rate": 8.19706498951782e-05,
      "loss": 0.1976,
      "step": 391
    },
    {
      "epoch": 0.08228379513014274,
      "grad_norm": 0.2579379081726074,
      "learning_rate": 8.218029350104822e-05,
      "loss": 0.197,
      "step": 392
    },
    {
      "epoch": 0.08249370277078086,
      "grad_norm": 0.18806852400302887,
      "learning_rate": 8.238993710691824e-05,
      "loss": 0.1873,
      "step": 393
    },
    {
      "epoch": 0.08270361041141898,
      "grad_norm": 0.24592849612236023,
      "learning_rate": 8.259958071278825e-05,
      "loss": 0.2083,
      "step": 394
    },
    {
      "epoch": 0.0829135180520571,
      "grad_norm": 0.2678208649158478,
      "learning_rate": 8.280922431865828e-05,
      "loss": 0.1878,
      "step": 395
    },
    {
      "epoch": 0.08312342569269521,
      "grad_norm": 0.2023075520992279,
      "learning_rate": 8.30188679245283e-05,
      "loss": 0.1861,
      "step": 396
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 0.2390558272600174,
      "learning_rate": 8.322851153039832e-05,
      "loss": 0.1802,
      "step": 397
    },
    {
      "epoch": 0.08354324097397145,
      "grad_norm": 0.175222247838974,
      "learning_rate": 8.343815513626834e-05,
      "loss": 0.1674,
      "step": 398
    },
    {
      "epoch": 0.08375314861460957,
      "grad_norm": 0.2340380698442459,
      "learning_rate": 8.364779874213837e-05,
      "loss": 0.17,
      "step": 399
    },
    {
      "epoch": 0.08396305625524769,
      "grad_norm": 0.20814655721187592,
      "learning_rate": 8.385744234800838e-05,
      "loss": 0.1513,
      "step": 400
    },
    {
      "epoch": 0.08417296389588581,
      "grad_norm": 0.34867948293685913,
      "learning_rate": 8.406708595387841e-05,
      "loss": 0.1644,
      "step": 401
    },
    {
      "epoch": 0.08438287153652393,
      "grad_norm": 0.21573619544506073,
      "learning_rate": 8.427672955974843e-05,
      "loss": 0.1785,
      "step": 402
    },
    {
      "epoch": 0.08459277917716204,
      "grad_norm": 0.21437713503837585,
      "learning_rate": 8.448637316561845e-05,
      "loss": 0.1523,
      "step": 403
    },
    {
      "epoch": 0.08480268681780016,
      "grad_norm": 0.2250152826309204,
      "learning_rate": 8.469601677148847e-05,
      "loss": 0.1728,
      "step": 404
    },
    {
      "epoch": 0.08501259445843828,
      "grad_norm": 0.2514733672142029,
      "learning_rate": 8.49056603773585e-05,
      "loss": 0.1755,
      "step": 405
    },
    {
      "epoch": 0.0852225020990764,
      "grad_norm": 0.16003377735614777,
      "learning_rate": 8.511530398322851e-05,
      "loss": 0.1935,
      "step": 406
    },
    {
      "epoch": 0.08543240973971453,
      "grad_norm": 0.1792365163564682,
      "learning_rate": 8.532494758909854e-05,
      "loss": 0.1991,
      "step": 407
    },
    {
      "epoch": 0.08564231738035265,
      "grad_norm": 0.16854703426361084,
      "learning_rate": 8.553459119496856e-05,
      "loss": 0.1904,
      "step": 408
    },
    {
      "epoch": 0.08585222502099077,
      "grad_norm": 0.19401603937149048,
      "learning_rate": 8.574423480083857e-05,
      "loss": 0.1681,
      "step": 409
    },
    {
      "epoch": 0.08606213266162889,
      "grad_norm": 0.15639828145503998,
      "learning_rate": 8.59538784067086e-05,
      "loss": 0.1628,
      "step": 410
    },
    {
      "epoch": 0.086272040302267,
      "grad_norm": 0.15357258915901184,
      "learning_rate": 8.616352201257863e-05,
      "loss": 0.1905,
      "step": 411
    },
    {
      "epoch": 0.08648194794290512,
      "grad_norm": 0.256944477558136,
      "learning_rate": 8.637316561844864e-05,
      "loss": 0.2015,
      "step": 412
    },
    {
      "epoch": 0.08669185558354324,
      "grad_norm": 0.13482192158699036,
      "learning_rate": 8.658280922431866e-05,
      "loss": 0.1718,
      "step": 413
    },
    {
      "epoch": 0.08690176322418136,
      "grad_norm": 0.16525831818580627,
      "learning_rate": 8.679245283018869e-05,
      "loss": 0.1778,
      "step": 414
    },
    {
      "epoch": 0.08711167086481948,
      "grad_norm": 0.2145531326532364,
      "learning_rate": 8.70020964360587e-05,
      "loss": 0.1714,
      "step": 415
    },
    {
      "epoch": 0.0873215785054576,
      "grad_norm": 0.20481255650520325,
      "learning_rate": 8.721174004192873e-05,
      "loss": 0.1832,
      "step": 416
    },
    {
      "epoch": 0.08753148614609572,
      "grad_norm": 0.14865756034851074,
      "learning_rate": 8.742138364779875e-05,
      "loss": 0.1659,
      "step": 417
    },
    {
      "epoch": 0.08774139378673383,
      "grad_norm": 0.14713706076145172,
      "learning_rate": 8.763102725366877e-05,
      "loss": 0.1674,
      "step": 418
    },
    {
      "epoch": 0.08795130142737195,
      "grad_norm": 0.1728464812040329,
      "learning_rate": 8.784067085953879e-05,
      "loss": 0.1727,
      "step": 419
    },
    {
      "epoch": 0.08816120906801007,
      "grad_norm": 0.2646033465862274,
      "learning_rate": 8.805031446540882e-05,
      "loss": 0.1973,
      "step": 420
    },
    {
      "epoch": 0.0883711167086482,
      "grad_norm": 0.2262433022260666,
      "learning_rate": 8.825995807127883e-05,
      "loss": 0.1817,
      "step": 421
    },
    {
      "epoch": 0.08858102434928632,
      "grad_norm": 0.16398945450782776,
      "learning_rate": 8.846960167714886e-05,
      "loss": 0.1628,
      "step": 422
    },
    {
      "epoch": 0.08879093198992444,
      "grad_norm": 0.15976634621620178,
      "learning_rate": 8.867924528301888e-05,
      "loss": 0.1678,
      "step": 423
    },
    {
      "epoch": 0.08900083963056256,
      "grad_norm": 0.19314904510974884,
      "learning_rate": 8.888888888888889e-05,
      "loss": 0.1753,
      "step": 424
    },
    {
      "epoch": 0.08921074727120067,
      "grad_norm": 0.21701818704605103,
      "learning_rate": 8.909853249475892e-05,
      "loss": 0.19,
      "step": 425
    },
    {
      "epoch": 0.08942065491183879,
      "grad_norm": 0.22768795490264893,
      "learning_rate": 8.930817610062893e-05,
      "loss": 0.1613,
      "step": 426
    },
    {
      "epoch": 0.08963056255247691,
      "grad_norm": 0.13172288239002228,
      "learning_rate": 8.951781970649896e-05,
      "loss": 0.1749,
      "step": 427
    },
    {
      "epoch": 0.08984047019311503,
      "grad_norm": 0.2015480250120163,
      "learning_rate": 8.972746331236897e-05,
      "loss": 0.1857,
      "step": 428
    },
    {
      "epoch": 0.09005037783375315,
      "grad_norm": 0.20239531993865967,
      "learning_rate": 8.9937106918239e-05,
      "loss": 0.1804,
      "step": 429
    },
    {
      "epoch": 0.09026028547439127,
      "grad_norm": 0.17528317868709564,
      "learning_rate": 9.014675052410901e-05,
      "loss": 0.1729,
      "step": 430
    },
    {
      "epoch": 0.0904701931150294,
      "grad_norm": 0.16663801670074463,
      "learning_rate": 9.035639412997903e-05,
      "loss": 0.1896,
      "step": 431
    },
    {
      "epoch": 0.0906801007556675,
      "grad_norm": 0.18777558207511902,
      "learning_rate": 9.056603773584906e-05,
      "loss": 0.1801,
      "step": 432
    },
    {
      "epoch": 0.09089000839630562,
      "grad_norm": 0.1440989226102829,
      "learning_rate": 9.077568134171907e-05,
      "loss": 0.179,
      "step": 433
    },
    {
      "epoch": 0.09109991603694374,
      "grad_norm": 0.1786854863166809,
      "learning_rate": 9.09853249475891e-05,
      "loss": 0.1799,
      "step": 434
    },
    {
      "epoch": 0.09130982367758186,
      "grad_norm": 0.20794442296028137,
      "learning_rate": 9.119496855345912e-05,
      "loss": 0.1731,
      "step": 435
    },
    {
      "epoch": 0.09151973131821999,
      "grad_norm": 0.16377133131027222,
      "learning_rate": 9.140461215932914e-05,
      "loss": 0.1815,
      "step": 436
    },
    {
      "epoch": 0.0917296389588581,
      "grad_norm": 0.174666628241539,
      "learning_rate": 9.161425576519916e-05,
      "loss": 0.1798,
      "step": 437
    },
    {
      "epoch": 0.09193954659949623,
      "grad_norm": 0.2127188891172409,
      "learning_rate": 9.182389937106919e-05,
      "loss": 0.1805,
      "step": 438
    },
    {
      "epoch": 0.09214945424013433,
      "grad_norm": 0.1936446726322174,
      "learning_rate": 9.203354297693921e-05,
      "loss": 0.1723,
      "step": 439
    },
    {
      "epoch": 0.09235936188077246,
      "grad_norm": 0.18736332654953003,
      "learning_rate": 9.224318658280923e-05,
      "loss": 0.1849,
      "step": 440
    },
    {
      "epoch": 0.09256926952141058,
      "grad_norm": 0.16817238926887512,
      "learning_rate": 9.245283018867925e-05,
      "loss": 0.1746,
      "step": 441
    },
    {
      "epoch": 0.0927791771620487,
      "grad_norm": 0.18249107897281647,
      "learning_rate": 9.266247379454928e-05,
      "loss": 0.1628,
      "step": 442
    },
    {
      "epoch": 0.09298908480268682,
      "grad_norm": 0.1728898137807846,
      "learning_rate": 9.287211740041929e-05,
      "loss": 0.203,
      "step": 443
    },
    {
      "epoch": 0.09319899244332494,
      "grad_norm": 0.16144797205924988,
      "learning_rate": 9.308176100628931e-05,
      "loss": 0.1843,
      "step": 444
    },
    {
      "epoch": 0.09340890008396306,
      "grad_norm": 0.19680747389793396,
      "learning_rate": 9.329140461215934e-05,
      "loss": 0.168,
      "step": 445
    },
    {
      "epoch": 0.09361880772460117,
      "grad_norm": 0.16198395192623138,
      "learning_rate": 9.350104821802935e-05,
      "loss": 0.1598,
      "step": 446
    },
    {
      "epoch": 0.09382871536523929,
      "grad_norm": 0.17398878931999207,
      "learning_rate": 9.371069182389938e-05,
      "loss": 0.17,
      "step": 447
    },
    {
      "epoch": 0.09403862300587741,
      "grad_norm": 0.18602675199508667,
      "learning_rate": 9.39203354297694e-05,
      "loss": 0.1716,
      "step": 448
    },
    {
      "epoch": 0.09424853064651553,
      "grad_norm": 0.18403322994709015,
      "learning_rate": 9.412997903563942e-05,
      "loss": 0.1948,
      "step": 449
    },
    {
      "epoch": 0.09445843828715365,
      "grad_norm": 0.18783587217330933,
      "learning_rate": 9.433962264150944e-05,
      "loss": 0.1812,
      "step": 450
    },
    {
      "epoch": 0.09466834592779177,
      "grad_norm": 0.20252300798892975,
      "learning_rate": 9.454926624737947e-05,
      "loss": 0.1666,
      "step": 451
    },
    {
      "epoch": 0.0948782535684299,
      "grad_norm": 0.20129899680614471,
      "learning_rate": 9.475890985324948e-05,
      "loss": 0.1695,
      "step": 452
    },
    {
      "epoch": 0.095088161209068,
      "grad_norm": 0.17035968601703644,
      "learning_rate": 9.496855345911951e-05,
      "loss": 0.1678,
      "step": 453
    },
    {
      "epoch": 0.09529806884970612,
      "grad_norm": 0.20403030514717102,
      "learning_rate": 9.517819706498953e-05,
      "loss": 0.1843,
      "step": 454
    },
    {
      "epoch": 0.09550797649034425,
      "grad_norm": 0.17489562928676605,
      "learning_rate": 9.538784067085954e-05,
      "loss": 0.1657,
      "step": 455
    },
    {
      "epoch": 0.09571788413098237,
      "grad_norm": 0.19699983298778534,
      "learning_rate": 9.559748427672957e-05,
      "loss": 0.1634,
      "step": 456
    },
    {
      "epoch": 0.09592779177162049,
      "grad_norm": 0.17113354802131653,
      "learning_rate": 9.58071278825996e-05,
      "loss": 0.1879,
      "step": 457
    },
    {
      "epoch": 0.09613769941225861,
      "grad_norm": 0.2290397435426712,
      "learning_rate": 9.601677148846961e-05,
      "loss": 0.1784,
      "step": 458
    },
    {
      "epoch": 0.09634760705289673,
      "grad_norm": 0.2173147350549698,
      "learning_rate": 9.622641509433963e-05,
      "loss": 0.1673,
      "step": 459
    },
    {
      "epoch": 0.09655751469353484,
      "grad_norm": 0.18280835449695587,
      "learning_rate": 9.643605870020965e-05,
      "loss": 0.1806,
      "step": 460
    },
    {
      "epoch": 0.09676742233417296,
      "grad_norm": 0.14638672769069672,
      "learning_rate": 9.664570230607967e-05,
      "loss": 0.1833,
      "step": 461
    },
    {
      "epoch": 0.09697732997481108,
      "grad_norm": 0.16228064894676208,
      "learning_rate": 9.685534591194969e-05,
      "loss": 0.1762,
      "step": 462
    },
    {
      "epoch": 0.0971872376154492,
      "grad_norm": 0.1836690902709961,
      "learning_rate": 9.706498951781971e-05,
      "loss": 0.1782,
      "step": 463
    },
    {
      "epoch": 0.09739714525608732,
      "grad_norm": 0.19470515847206116,
      "learning_rate": 9.727463312368972e-05,
      "loss": 0.185,
      "step": 464
    },
    {
      "epoch": 0.09760705289672544,
      "grad_norm": 0.1833791732788086,
      "learning_rate": 9.748427672955975e-05,
      "loss": 0.1753,
      "step": 465
    },
    {
      "epoch": 0.09781696053736356,
      "grad_norm": 0.22608265280723572,
      "learning_rate": 9.769392033542977e-05,
      "loss": 0.1792,
      "step": 466
    },
    {
      "epoch": 0.09802686817800169,
      "grad_norm": 0.16552825272083282,
      "learning_rate": 9.790356394129979e-05,
      "loss": 0.1808,
      "step": 467
    },
    {
      "epoch": 0.0982367758186398,
      "grad_norm": 0.2294851690530777,
      "learning_rate": 9.811320754716981e-05,
      "loss": 0.1935,
      "step": 468
    },
    {
      "epoch": 0.09844668345927791,
      "grad_norm": 0.26589101552963257,
      "learning_rate": 9.832285115303984e-05,
      "loss": 0.1676,
      "step": 469
    },
    {
      "epoch": 0.09865659109991604,
      "grad_norm": 0.22315791249275208,
      "learning_rate": 9.853249475890985e-05,
      "loss": 0.1806,
      "step": 470
    },
    {
      "epoch": 0.09886649874055416,
      "grad_norm": 0.16855137050151825,
      "learning_rate": 9.874213836477988e-05,
      "loss": 0.1816,
      "step": 471
    },
    {
      "epoch": 0.09907640638119228,
      "grad_norm": 0.19197392463684082,
      "learning_rate": 9.89517819706499e-05,
      "loss": 0.1923,
      "step": 472
    },
    {
      "epoch": 0.0992863140218304,
      "grad_norm": 0.18722014129161835,
      "learning_rate": 9.916142557651992e-05,
      "loss": 0.1823,
      "step": 473
    },
    {
      "epoch": 0.09949622166246852,
      "grad_norm": 0.15668706595897675,
      "learning_rate": 9.937106918238994e-05,
      "loss": 0.1896,
      "step": 474
    },
    {
      "epoch": 0.09970612930310663,
      "grad_norm": 0.17297013103961945,
      "learning_rate": 9.958071278825997e-05,
      "loss": 0.1907,
      "step": 475
    },
    {
      "epoch": 0.09991603694374475,
      "grad_norm": 0.23546694219112396,
      "learning_rate": 9.979035639412998e-05,
      "loss": 0.1647,
      "step": 476
    },
    {
      "epoch": 0.10012594458438287,
      "grad_norm": 0.1627054661512375,
      "learning_rate": 0.0001,
      "loss": 0.1686,
      "step": 477
    },
    {
      "epoch": 0.10033585222502099,
      "grad_norm": 0.21043647825717926,
      "learning_rate": 9.999998657442895e-05,
      "loss": 0.1865,
      "step": 478
    },
    {
      "epoch": 0.10054575986565911,
      "grad_norm": 0.19615764915943146,
      "learning_rate": 9.999994629772298e-05,
      "loss": 0.1683,
      "step": 479
    },
    {
      "epoch": 0.10075566750629723,
      "grad_norm": 0.2884671092033386,
      "learning_rate": 9.999987916990372e-05,
      "loss": 0.1858,
      "step": 480
    },
    {
      "epoch": 0.10096557514693535,
      "grad_norm": 0.2381323128938675,
      "learning_rate": 9.999978519100723e-05,
      "loss": 0.1879,
      "step": 481
    },
    {
      "epoch": 0.10117548278757346,
      "grad_norm": 0.19187557697296143,
      "learning_rate": 9.999966436108398e-05,
      "loss": 0.1808,
      "step": 482
    },
    {
      "epoch": 0.10138539042821158,
      "grad_norm": 0.19443491101264954,
      "learning_rate": 9.999951668019887e-05,
      "loss": 0.1659,
      "step": 483
    },
    {
      "epoch": 0.1015952980688497,
      "grad_norm": 0.20151716470718384,
      "learning_rate": 9.999934214843116e-05,
      "loss": 0.1679,
      "step": 484
    },
    {
      "epoch": 0.10180520570948783,
      "grad_norm": 0.2896507978439331,
      "learning_rate": 9.999914076587464e-05,
      "loss": 0.1734,
      "step": 485
    },
    {
      "epoch": 0.10201511335012595,
      "grad_norm": 0.21598441898822784,
      "learning_rate": 9.999891253263741e-05,
      "loss": 0.1779,
      "step": 486
    },
    {
      "epoch": 0.10222502099076407,
      "grad_norm": 0.196011021733284,
      "learning_rate": 9.999865744884207e-05,
      "loss": 0.1815,
      "step": 487
    },
    {
      "epoch": 0.10243492863140219,
      "grad_norm": 0.20962318778038025,
      "learning_rate": 9.999837551462558e-05,
      "loss": 0.1727,
      "step": 488
    },
    {
      "epoch": 0.1026448362720403,
      "grad_norm": 0.18339572846889496,
      "learning_rate": 9.999806673013935e-05,
      "loss": 0.1689,
      "step": 489
    },
    {
      "epoch": 0.10285474391267842,
      "grad_norm": 0.20994813740253448,
      "learning_rate": 9.999773109554922e-05,
      "loss": 0.1768,
      "step": 490
    },
    {
      "epoch": 0.10306465155331654,
      "grad_norm": 0.163935124874115,
      "learning_rate": 9.999736861103541e-05,
      "loss": 0.1777,
      "step": 491
    },
    {
      "epoch": 0.10327455919395466,
      "grad_norm": 0.1465967446565628,
      "learning_rate": 9.99969792767926e-05,
      "loss": 0.1846,
      "step": 492
    },
    {
      "epoch": 0.10348446683459278,
      "grad_norm": 0.21443922817707062,
      "learning_rate": 9.999656309302987e-05,
      "loss": 0.1807,
      "step": 493
    },
    {
      "epoch": 0.1036943744752309,
      "grad_norm": 0.18504248559474945,
      "learning_rate": 9.999612005997071e-05,
      "loss": 0.1548,
      "step": 494
    },
    {
      "epoch": 0.10390428211586902,
      "grad_norm": 0.15490441024303436,
      "learning_rate": 9.999565017785305e-05,
      "loss": 0.1696,
      "step": 495
    },
    {
      "epoch": 0.10411418975650713,
      "grad_norm": 0.1881389319896698,
      "learning_rate": 9.999515344692923e-05,
      "loss": 0.1852,
      "step": 496
    },
    {
      "epoch": 0.10432409739714525,
      "grad_norm": 0.16337451338768005,
      "learning_rate": 9.999462986746598e-05,
      "loss": 0.1834,
      "step": 497
    },
    {
      "epoch": 0.10453400503778337,
      "grad_norm": 0.16641898453235626,
      "learning_rate": 9.99940794397445e-05,
      "loss": 0.187,
      "step": 498
    },
    {
      "epoch": 0.1047439126784215,
      "grad_norm": 0.15948446094989777,
      "learning_rate": 9.999350216406038e-05,
      "loss": 0.1835,
      "step": 499
    },
    {
      "epoch": 0.10495382031905962,
      "grad_norm": 0.1550200879573822,
      "learning_rate": 9.999289804072363e-05,
      "loss": 0.1706,
      "step": 500
    },
    {
      "epoch": 0.10516372795969774,
      "grad_norm": 0.16597698628902435,
      "learning_rate": 9.999226707005867e-05,
      "loss": 0.1811,
      "step": 501
    },
    {
      "epoch": 0.10537363560033586,
      "grad_norm": 0.17551501095294952,
      "learning_rate": 9.999160925240434e-05,
      "loss": 0.1677,
      "step": 502
    },
    {
      "epoch": 0.10558354324097396,
      "grad_norm": 0.15515847504138947,
      "learning_rate": 9.999092458811393e-05,
      "loss": 0.1789,
      "step": 503
    },
    {
      "epoch": 0.10579345088161209,
      "grad_norm": 0.18121638894081116,
      "learning_rate": 9.99902130775551e-05,
      "loss": 0.1684,
      "step": 504
    },
    {
      "epoch": 0.10600335852225021,
      "grad_norm": 0.1853945255279541,
      "learning_rate": 9.998947472110994e-05,
      "loss": 0.1907,
      "step": 505
    },
    {
      "epoch": 0.10621326616288833,
      "grad_norm": 0.21303139626979828,
      "learning_rate": 9.998870951917496e-05,
      "loss": 0.1712,
      "step": 506
    },
    {
      "epoch": 0.10642317380352645,
      "grad_norm": 0.16773764789104462,
      "learning_rate": 9.998791747216113e-05,
      "loss": 0.1756,
      "step": 507
    },
    {
      "epoch": 0.10663308144416457,
      "grad_norm": 0.18033501505851746,
      "learning_rate": 9.998709858049376e-05,
      "loss": 0.1654,
      "step": 508
    },
    {
      "epoch": 0.10684298908480269,
      "grad_norm": 0.14199328422546387,
      "learning_rate": 9.998625284461263e-05,
      "loss": 0.1587,
      "step": 509
    },
    {
      "epoch": 0.1070528967254408,
      "grad_norm": 0.19968685507774353,
      "learning_rate": 9.998538026497192e-05,
      "loss": 0.1796,
      "step": 510
    },
    {
      "epoch": 0.10726280436607892,
      "grad_norm": 0.17311611771583557,
      "learning_rate": 9.998448084204021e-05,
      "loss": 0.1864,
      "step": 511
    },
    {
      "epoch": 0.10747271200671704,
      "grad_norm": 0.20124119520187378,
      "learning_rate": 9.998355457630053e-05,
      "loss": 0.1829,
      "step": 512
    },
    {
      "epoch": 0.10768261964735516,
      "grad_norm": 0.12473297864198685,
      "learning_rate": 9.998260146825029e-05,
      "loss": 0.175,
      "step": 513
    },
    {
      "epoch": 0.10789252728799328,
      "grad_norm": 0.1696644425392151,
      "learning_rate": 9.998162151840135e-05,
      "loss": 0.1762,
      "step": 514
    },
    {
      "epoch": 0.1081024349286314,
      "grad_norm": 0.1781477928161621,
      "learning_rate": 9.998061472727996e-05,
      "loss": 0.1679,
      "step": 515
    },
    {
      "epoch": 0.10831234256926953,
      "grad_norm": 0.19112960994243622,
      "learning_rate": 9.997958109542675e-05,
      "loss": 0.1553,
      "step": 516
    },
    {
      "epoch": 0.10852225020990765,
      "grad_norm": 0.1417030394077301,
      "learning_rate": 9.997852062339685e-05,
      "loss": 0.1737,
      "step": 517
    },
    {
      "epoch": 0.10873215785054575,
      "grad_norm": 0.15080858767032623,
      "learning_rate": 9.997743331175976e-05,
      "loss": 0.1595,
      "step": 518
    },
    {
      "epoch": 0.10894206549118388,
      "grad_norm": 0.2046668380498886,
      "learning_rate": 9.997631916109937e-05,
      "loss": 0.1839,
      "step": 519
    },
    {
      "epoch": 0.109151973131822,
      "grad_norm": 0.19941595196723938,
      "learning_rate": 9.997517817201401e-05,
      "loss": 0.1718,
      "step": 520
    },
    {
      "epoch": 0.10936188077246012,
      "grad_norm": 0.15989692509174347,
      "learning_rate": 9.997401034511642e-05,
      "loss": 0.1613,
      "step": 521
    },
    {
      "epoch": 0.10957178841309824,
      "grad_norm": 0.1697997897863388,
      "learning_rate": 9.997281568103374e-05,
      "loss": 0.1603,
      "step": 522
    },
    {
      "epoch": 0.10978169605373636,
      "grad_norm": 0.1840822696685791,
      "learning_rate": 9.997159418040754e-05,
      "loss": 0.1735,
      "step": 523
    },
    {
      "epoch": 0.10999160369437448,
      "grad_norm": 0.20991730690002441,
      "learning_rate": 9.99703458438938e-05,
      "loss": 0.154,
      "step": 524
    },
    {
      "epoch": 0.11020151133501259,
      "grad_norm": 0.16802968084812164,
      "learning_rate": 9.99690706721629e-05,
      "loss": 0.1761,
      "step": 525
    },
    {
      "epoch": 0.11041141897565071,
      "grad_norm": 0.18329255282878876,
      "learning_rate": 9.996776866589962e-05,
      "loss": 0.1609,
      "step": 526
    },
    {
      "epoch": 0.11062132661628883,
      "grad_norm": 0.18645748496055603,
      "learning_rate": 9.996643982580318e-05,
      "loss": 0.1793,
      "step": 527
    },
    {
      "epoch": 0.11083123425692695,
      "grad_norm": 0.1966720074415207,
      "learning_rate": 9.996508415258722e-05,
      "loss": 0.1714,
      "step": 528
    },
    {
      "epoch": 0.11104114189756507,
      "grad_norm": 0.18155452609062195,
      "learning_rate": 9.996370164697974e-05,
      "loss": 0.1673,
      "step": 529
    },
    {
      "epoch": 0.1112510495382032,
      "grad_norm": 0.2004195600748062,
      "learning_rate": 9.996229230972317e-05,
      "loss": 0.1865,
      "step": 530
    },
    {
      "epoch": 0.11146095717884132,
      "grad_norm": 0.15521694719791412,
      "learning_rate": 9.996085614157438e-05,
      "loss": 0.1757,
      "step": 531
    },
    {
      "epoch": 0.11167086481947942,
      "grad_norm": 0.1686578243970871,
      "learning_rate": 9.995939314330462e-05,
      "loss": 0.1768,
      "step": 532
    },
    {
      "epoch": 0.11188077246011754,
      "grad_norm": 0.20034368336200714,
      "learning_rate": 9.995790331569954e-05,
      "loss": 0.1823,
      "step": 533
    },
    {
      "epoch": 0.11209068010075567,
      "grad_norm": 0.1494702696800232,
      "learning_rate": 9.995638665955922e-05,
      "loss": 0.175,
      "step": 534
    },
    {
      "epoch": 0.11230058774139379,
      "grad_norm": 0.16365233063697815,
      "learning_rate": 9.995484317569814e-05,
      "loss": 0.1716,
      "step": 535
    },
    {
      "epoch": 0.11251049538203191,
      "grad_norm": 0.19227434694766998,
      "learning_rate": 9.995327286494521e-05,
      "loss": 0.1605,
      "step": 536
    },
    {
      "epoch": 0.11272040302267003,
      "grad_norm": 0.21946166455745697,
      "learning_rate": 9.995167572814365e-05,
      "loss": 0.182,
      "step": 537
    },
    {
      "epoch": 0.11293031066330815,
      "grad_norm": 0.2211793065071106,
      "learning_rate": 9.995005176615124e-05,
      "loss": 0.1783,
      "step": 538
    },
    {
      "epoch": 0.11314021830394626,
      "grad_norm": 0.2154102325439453,
      "learning_rate": 9.994840097984006e-05,
      "loss": 0.1888,
      "step": 539
    },
    {
      "epoch": 0.11335012594458438,
      "grad_norm": 0.20600587129592896,
      "learning_rate": 9.994672337009658e-05,
      "loss": 0.1871,
      "step": 540
    },
    {
      "epoch": 0.1135600335852225,
      "grad_norm": 0.22028079628944397,
      "learning_rate": 9.994501893782176e-05,
      "loss": 0.1855,
      "step": 541
    },
    {
      "epoch": 0.11376994122586062,
      "grad_norm": 0.23957398533821106,
      "learning_rate": 9.99432876839309e-05,
      "loss": 0.1616,
      "step": 542
    },
    {
      "epoch": 0.11397984886649874,
      "grad_norm": 0.14516577124595642,
      "learning_rate": 9.994152960935375e-05,
      "loss": 0.1864,
      "step": 543
    },
    {
      "epoch": 0.11418975650713686,
      "grad_norm": 0.14327426254749298,
      "learning_rate": 9.99397447150344e-05,
      "loss": 0.1596,
      "step": 544
    },
    {
      "epoch": 0.11439966414777498,
      "grad_norm": 0.15387804806232452,
      "learning_rate": 9.99379330019314e-05,
      "loss": 0.1638,
      "step": 545
    },
    {
      "epoch": 0.11460957178841309,
      "grad_norm": 0.14998720586299896,
      "learning_rate": 9.993609447101767e-05,
      "loss": 0.1456,
      "step": 546
    },
    {
      "epoch": 0.11481947942905121,
      "grad_norm": 0.21649526059627533,
      "learning_rate": 9.993422912328054e-05,
      "loss": 0.1656,
      "step": 547
    },
    {
      "epoch": 0.11502938706968933,
      "grad_norm": 0.19370022416114807,
      "learning_rate": 9.993233695972175e-05,
      "loss": 0.173,
      "step": 548
    },
    {
      "epoch": 0.11523929471032746,
      "grad_norm": 0.20101432502269745,
      "learning_rate": 9.993041798135745e-05,
      "loss": 0.1914,
      "step": 549
    },
    {
      "epoch": 0.11544920235096558,
      "grad_norm": 0.18047718703746796,
      "learning_rate": 9.992847218921816e-05,
      "loss": 0.2084,
      "step": 550
    },
    {
      "epoch": 0.1156591099916037,
      "grad_norm": 0.18755191564559937,
      "learning_rate": 9.99264995843488e-05,
      "loss": 0.1646,
      "step": 551
    },
    {
      "epoch": 0.11586901763224182,
      "grad_norm": 0.2605830430984497,
      "learning_rate": 9.992450016780876e-05,
      "loss": 0.1767,
      "step": 552
    },
    {
      "epoch": 0.11607892527287993,
      "grad_norm": 0.1701487898826599,
      "learning_rate": 9.99224739406717e-05,
      "loss": 0.173,
      "step": 553
    },
    {
      "epoch": 0.11628883291351805,
      "grad_norm": 0.19497598707675934,
      "learning_rate": 9.99204209040258e-05,
      "loss": 0.1857,
      "step": 554
    },
    {
      "epoch": 0.11649874055415617,
      "grad_norm": 0.16379636526107788,
      "learning_rate": 9.991834105897356e-05,
      "loss": 0.154,
      "step": 555
    },
    {
      "epoch": 0.11670864819479429,
      "grad_norm": 0.18630164861679077,
      "learning_rate": 9.991623440663192e-05,
      "loss": 0.1871,
      "step": 556
    },
    {
      "epoch": 0.11691855583543241,
      "grad_norm": 0.219542995095253,
      "learning_rate": 9.991410094813221e-05,
      "loss": 0.1877,
      "step": 557
    },
    {
      "epoch": 0.11712846347607053,
      "grad_norm": 0.1855912059545517,
      "learning_rate": 9.991194068462011e-05,
      "loss": 0.1771,
      "step": 558
    },
    {
      "epoch": 0.11733837111670865,
      "grad_norm": 0.15919265151023865,
      "learning_rate": 9.990975361725577e-05,
      "loss": 0.1684,
      "step": 559
    },
    {
      "epoch": 0.11754827875734676,
      "grad_norm": 0.14570386707782745,
      "learning_rate": 9.990753974721366e-05,
      "loss": 0.1656,
      "step": 560
    },
    {
      "epoch": 0.11775818639798488,
      "grad_norm": 0.14153516292572021,
      "learning_rate": 9.990529907568272e-05,
      "loss": 0.1642,
      "step": 561
    },
    {
      "epoch": 0.117968094038623,
      "grad_norm": 0.14015786349773407,
      "learning_rate": 9.99030316038662e-05,
      "loss": 0.1704,
      "step": 562
    },
    {
      "epoch": 0.11817800167926112,
      "grad_norm": 0.2615254819393158,
      "learning_rate": 9.99007373329818e-05,
      "loss": 0.1601,
      "step": 563
    },
    {
      "epoch": 0.11838790931989925,
      "grad_norm": 0.2077956199645996,
      "learning_rate": 9.989841626426162e-05,
      "loss": 0.1715,
      "step": 564
    },
    {
      "epoch": 0.11859781696053737,
      "grad_norm": 0.1743435561656952,
      "learning_rate": 9.989606839895208e-05,
      "loss": 0.1725,
      "step": 565
    },
    {
      "epoch": 0.11880772460117549,
      "grad_norm": 0.14333437383174896,
      "learning_rate": 9.989369373831407e-05,
      "loss": 0.1731,
      "step": 566
    },
    {
      "epoch": 0.1190176322418136,
      "grad_norm": 0.16472546756267548,
      "learning_rate": 9.989129228362284e-05,
      "loss": 0.1641,
      "step": 567
    },
    {
      "epoch": 0.11922753988245172,
      "grad_norm": 0.19595298171043396,
      "learning_rate": 9.988886403616802e-05,
      "loss": 0.1789,
      "step": 568
    },
    {
      "epoch": 0.11943744752308984,
      "grad_norm": 0.15337203443050385,
      "learning_rate": 9.988640899725361e-05,
      "loss": 0.166,
      "step": 569
    },
    {
      "epoch": 0.11964735516372796,
      "grad_norm": 0.13052469491958618,
      "learning_rate": 9.988392716819806e-05,
      "loss": 0.1762,
      "step": 570
    },
    {
      "epoch": 0.11985726280436608,
      "grad_norm": 0.17160564661026,
      "learning_rate": 9.988141855033415e-05,
      "loss": 0.1742,
      "step": 571
    },
    {
      "epoch": 0.1200671704450042,
      "grad_norm": 0.15600250661373138,
      "learning_rate": 9.987888314500906e-05,
      "loss": 0.1819,
      "step": 572
    },
    {
      "epoch": 0.12027707808564232,
      "grad_norm": 0.17436926066875458,
      "learning_rate": 9.987632095358437e-05,
      "loss": 0.154,
      "step": 573
    },
    {
      "epoch": 0.12048698572628044,
      "grad_norm": 0.1583249568939209,
      "learning_rate": 9.987373197743603e-05,
      "loss": 0.1812,
      "step": 574
    },
    {
      "epoch": 0.12069689336691855,
      "grad_norm": 0.14877773821353912,
      "learning_rate": 9.987111621795437e-05,
      "loss": 0.1591,
      "step": 575
    },
    {
      "epoch": 0.12090680100755667,
      "grad_norm": 0.13348308205604553,
      "learning_rate": 9.986847367654414e-05,
      "loss": 0.164,
      "step": 576
    },
    {
      "epoch": 0.12111670864819479,
      "grad_norm": 0.16104738414287567,
      "learning_rate": 9.986580435462443e-05,
      "loss": 0.1568,
      "step": 577
    },
    {
      "epoch": 0.12132661628883291,
      "grad_norm": 0.1393367350101471,
      "learning_rate": 9.98631082536287e-05,
      "loss": 0.1504,
      "step": 578
    },
    {
      "epoch": 0.12153652392947104,
      "grad_norm": 0.17166094481945038,
      "learning_rate": 9.986038537500488e-05,
      "loss": 0.1746,
      "step": 579
    },
    {
      "epoch": 0.12174643157010916,
      "grad_norm": 0.19518902897834778,
      "learning_rate": 9.985763572021516e-05,
      "loss": 0.1961,
      "step": 580
    },
    {
      "epoch": 0.12195633921074728,
      "grad_norm": 0.17867891490459442,
      "learning_rate": 9.985485929073619e-05,
      "loss": 0.1836,
      "step": 581
    },
    {
      "epoch": 0.12216624685138538,
      "grad_norm": 0.2270008772611618,
      "learning_rate": 9.9852056088059e-05,
      "loss": 0.1847,
      "step": 582
    },
    {
      "epoch": 0.1223761544920235,
      "grad_norm": 0.16426697373390198,
      "learning_rate": 9.984922611368892e-05,
      "loss": 0.1587,
      "step": 583
    },
    {
      "epoch": 0.12258606213266163,
      "grad_norm": 0.16188107430934906,
      "learning_rate": 9.984636936914575e-05,
      "loss": 0.1783,
      "step": 584
    },
    {
      "epoch": 0.12279596977329975,
      "grad_norm": 0.16157647967338562,
      "learning_rate": 9.984348585596361e-05,
      "loss": 0.1612,
      "step": 585
    },
    {
      "epoch": 0.12300587741393787,
      "grad_norm": 0.17590083181858063,
      "learning_rate": 9.984057557569104e-05,
      "loss": 0.1852,
      "step": 586
    },
    {
      "epoch": 0.12321578505457599,
      "grad_norm": 0.13868169486522675,
      "learning_rate": 9.983763852989088e-05,
      "loss": 0.1915,
      "step": 587
    },
    {
      "epoch": 0.12342569269521411,
      "grad_norm": 0.15027277171611786,
      "learning_rate": 9.983467472014043e-05,
      "loss": 0.1924,
      "step": 588
    },
    {
      "epoch": 0.12363560033585222,
      "grad_norm": 0.16492144763469696,
      "learning_rate": 9.983168414803132e-05,
      "loss": 0.1768,
      "step": 589
    },
    {
      "epoch": 0.12384550797649034,
      "grad_norm": 0.16169002652168274,
      "learning_rate": 9.982866681516954e-05,
      "loss": 0.1704,
      "step": 590
    },
    {
      "epoch": 0.12405541561712846,
      "grad_norm": 0.16292813420295715,
      "learning_rate": 9.982562272317546e-05,
      "loss": 0.1687,
      "step": 591
    },
    {
      "epoch": 0.12426532325776658,
      "grad_norm": 0.16432535648345947,
      "learning_rate": 9.982255187368386e-05,
      "loss": 0.1702,
      "step": 592
    },
    {
      "epoch": 0.1244752308984047,
      "grad_norm": 0.13967812061309814,
      "learning_rate": 9.981945426834382e-05,
      "loss": 0.1858,
      "step": 593
    },
    {
      "epoch": 0.12468513853904283,
      "grad_norm": 0.16663575172424316,
      "learning_rate": 9.981632990881885e-05,
      "loss": 0.158,
      "step": 594
    },
    {
      "epoch": 0.12489504617968095,
      "grad_norm": 0.17746829986572266,
      "learning_rate": 9.981317879678679e-05,
      "loss": 0.1679,
      "step": 595
    },
    {
      "epoch": 0.12510495382031905,
      "grad_norm": 0.15006081759929657,
      "learning_rate": 9.981000093393986e-05,
      "loss": 0.1659,
      "step": 596
    },
    {
      "epoch": 0.1253148614609572,
      "grad_norm": 0.15521980822086334,
      "learning_rate": 9.980679632198466e-05,
      "loss": 0.1587,
      "step": 597
    },
    {
      "epoch": 0.1255247691015953,
      "grad_norm": 0.17882917821407318,
      "learning_rate": 9.980356496264212e-05,
      "loss": 0.1571,
      "step": 598
    },
    {
      "epoch": 0.12573467674223343,
      "grad_norm": 0.17295823991298676,
      "learning_rate": 9.980030685764754e-05,
      "loss": 0.1936,
      "step": 599
    },
    {
      "epoch": 0.12594458438287154,
      "grad_norm": 0.1699322760105133,
      "learning_rate": 9.979702200875065e-05,
      "loss": 0.1555,
      "step": 600
    },
    {
      "epoch": 0.12615449202350965,
      "grad_norm": 0.15549197793006897,
      "learning_rate": 9.979371041771543e-05,
      "loss": 0.1569,
      "step": 601
    },
    {
      "epoch": 0.12636439966414778,
      "grad_norm": 0.16207100450992584,
      "learning_rate": 9.979037208632034e-05,
      "loss": 0.1609,
      "step": 602
    },
    {
      "epoch": 0.1265743073047859,
      "grad_norm": 0.156686931848526,
      "learning_rate": 9.978700701635807e-05,
      "loss": 0.165,
      "step": 603
    },
    {
      "epoch": 0.12678421494542402,
      "grad_norm": 0.16895289719104767,
      "learning_rate": 9.97836152096358e-05,
      "loss": 0.1623,
      "step": 604
    },
    {
      "epoch": 0.12699412258606213,
      "grad_norm": 0.14077980816364288,
      "learning_rate": 9.978019666797498e-05,
      "loss": 0.1573,
      "step": 605
    },
    {
      "epoch": 0.12720403022670027,
      "grad_norm": 0.14448733627796173,
      "learning_rate": 9.977675139321146e-05,
      "loss": 0.1671,
      "step": 606
    },
    {
      "epoch": 0.12741393786733837,
      "grad_norm": 0.1879337728023529,
      "learning_rate": 9.977327938719541e-05,
      "loss": 0.1644,
      "step": 607
    },
    {
      "epoch": 0.12762384550797648,
      "grad_norm": 0.22603319585323334,
      "learning_rate": 9.976978065179138e-05,
      "loss": 0.1817,
      "step": 608
    },
    {
      "epoch": 0.12783375314861462,
      "grad_norm": 0.19939488172531128,
      "learning_rate": 9.976625518887828e-05,
      "loss": 0.1925,
      "step": 609
    },
    {
      "epoch": 0.12804366078925272,
      "grad_norm": 0.17629611492156982,
      "learning_rate": 9.976270300034936e-05,
      "loss": 0.1603,
      "step": 610
    },
    {
      "epoch": 0.12825356842989086,
      "grad_norm": 0.17778918147087097,
      "learning_rate": 9.975912408811223e-05,
      "loss": 0.1849,
      "step": 611
    },
    {
      "epoch": 0.12846347607052896,
      "grad_norm": 0.2002590447664261,
      "learning_rate": 9.975551845408886e-05,
      "loss": 0.166,
      "step": 612
    },
    {
      "epoch": 0.1286733837111671,
      "grad_norm": 0.20324808359146118,
      "learning_rate": 9.975188610021553e-05,
      "loss": 0.1673,
      "step": 613
    },
    {
      "epoch": 0.1288832913518052,
      "grad_norm": 0.20010024309158325,
      "learning_rate": 9.974822702844291e-05,
      "loss": 0.1529,
      "step": 614
    },
    {
      "epoch": 0.12909319899244331,
      "grad_norm": 0.18944194912910461,
      "learning_rate": 9.974454124073603e-05,
      "loss": 0.1652,
      "step": 615
    },
    {
      "epoch": 0.12930310663308145,
      "grad_norm": 0.16956883668899536,
      "learning_rate": 9.974082873907418e-05,
      "loss": 0.1812,
      "step": 616
    },
    {
      "epoch": 0.12951301427371956,
      "grad_norm": 0.21157757937908173,
      "learning_rate": 9.973708952545111e-05,
      "loss": 0.1862,
      "step": 617
    },
    {
      "epoch": 0.1297229219143577,
      "grad_norm": 0.20554983615875244,
      "learning_rate": 9.973332360187486e-05,
      "loss": 0.1705,
      "step": 618
    },
    {
      "epoch": 0.1299328295549958,
      "grad_norm": 0.21964187920093536,
      "learning_rate": 9.97295309703678e-05,
      "loss": 0.1807,
      "step": 619
    },
    {
      "epoch": 0.13014273719563393,
      "grad_norm": 0.15438182651996613,
      "learning_rate": 9.972571163296666e-05,
      "loss": 0.1951,
      "step": 620
    },
    {
      "epoch": 0.13035264483627204,
      "grad_norm": 0.17805755138397217,
      "learning_rate": 9.972186559172253e-05,
      "loss": 0.1543,
      "step": 621
    },
    {
      "epoch": 0.13056255247691015,
      "grad_norm": 0.1914507895708084,
      "learning_rate": 9.97179928487008e-05,
      "loss": 0.1815,
      "step": 622
    },
    {
      "epoch": 0.13077246011754828,
      "grad_norm": 0.18952740728855133,
      "learning_rate": 9.971409340598123e-05,
      "loss": 0.1601,
      "step": 623
    },
    {
      "epoch": 0.1309823677581864,
      "grad_norm": 0.1550108641386032,
      "learning_rate": 9.971016726565791e-05,
      "loss": 0.1726,
      "step": 624
    },
    {
      "epoch": 0.13119227539882453,
      "grad_norm": 0.15205375850200653,
      "learning_rate": 9.970621442983929e-05,
      "loss": 0.1536,
      "step": 625
    },
    {
      "epoch": 0.13140218303946263,
      "grad_norm": 0.15574775636196136,
      "learning_rate": 9.970223490064809e-05,
      "loss": 0.167,
      "step": 626
    },
    {
      "epoch": 0.13161209068010077,
      "grad_norm": 0.18050767481327057,
      "learning_rate": 9.969822868022143e-05,
      "loss": 0.1752,
      "step": 627
    },
    {
      "epoch": 0.13182199832073888,
      "grad_norm": 0.1507614701986313,
      "learning_rate": 9.969419577071076e-05,
      "loss": 0.1712,
      "step": 628
    },
    {
      "epoch": 0.13203190596137698,
      "grad_norm": 0.20433206856250763,
      "learning_rate": 9.96901361742818e-05,
      "loss": 0.1782,
      "step": 629
    },
    {
      "epoch": 0.13224181360201512,
      "grad_norm": 0.14705337584018707,
      "learning_rate": 9.968604989311467e-05,
      "loss": 0.1554,
      "step": 630
    },
    {
      "epoch": 0.13245172124265323,
      "grad_norm": 0.1573822945356369,
      "learning_rate": 9.968193692940382e-05,
      "loss": 0.1762,
      "step": 631
    },
    {
      "epoch": 0.13266162888329136,
      "grad_norm": 0.17816203832626343,
      "learning_rate": 9.967779728535797e-05,
      "loss": 0.1783,
      "step": 632
    },
    {
      "epoch": 0.13287153652392947,
      "grad_norm": 0.1705409288406372,
      "learning_rate": 9.967363096320022e-05,
      "loss": 0.1734,
      "step": 633
    },
    {
      "epoch": 0.1330814441645676,
      "grad_norm": 0.14574241638183594,
      "learning_rate": 9.966943796516798e-05,
      "loss": 0.1599,
      "step": 634
    },
    {
      "epoch": 0.1332913518052057,
      "grad_norm": 0.1341760754585266,
      "learning_rate": 9.966521829351297e-05,
      "loss": 0.1833,
      "step": 635
    },
    {
      "epoch": 0.13350125944584382,
      "grad_norm": 0.1680106520652771,
      "learning_rate": 9.966097195050128e-05,
      "loss": 0.1701,
      "step": 636
    },
    {
      "epoch": 0.13371116708648195,
      "grad_norm": 0.1687798947095871,
      "learning_rate": 9.965669893841326e-05,
      "loss": 0.1669,
      "step": 637
    },
    {
      "epoch": 0.13392107472712006,
      "grad_norm": 0.1792893260717392,
      "learning_rate": 9.965239925954364e-05,
      "loss": 0.1705,
      "step": 638
    },
    {
      "epoch": 0.1341309823677582,
      "grad_norm": 0.15732963383197784,
      "learning_rate": 9.964807291620144e-05,
      "loss": 0.176,
      "step": 639
    },
    {
      "epoch": 0.1343408900083963,
      "grad_norm": 0.14176061749458313,
      "learning_rate": 9.964371991070999e-05,
      "loss": 0.1755,
      "step": 640
    },
    {
      "epoch": 0.13455079764903444,
      "grad_norm": 0.1526860147714615,
      "learning_rate": 9.963934024540698e-05,
      "loss": 0.1589,
      "step": 641
    },
    {
      "epoch": 0.13476070528967254,
      "grad_norm": 0.15608391165733337,
      "learning_rate": 9.963493392264435e-05,
      "loss": 0.1652,
      "step": 642
    },
    {
      "epoch": 0.13497061293031065,
      "grad_norm": 0.14473848044872284,
      "learning_rate": 9.963050094478845e-05,
      "loss": 0.1742,
      "step": 643
    },
    {
      "epoch": 0.1351805205709488,
      "grad_norm": 0.1382717341184616,
      "learning_rate": 9.962604131421984e-05,
      "loss": 0.1713,
      "step": 644
    },
    {
      "epoch": 0.1353904282115869,
      "grad_norm": 0.14345118403434753,
      "learning_rate": 9.962155503333348e-05,
      "loss": 0.1648,
      "step": 645
    },
    {
      "epoch": 0.13560033585222503,
      "grad_norm": 0.1398243010044098,
      "learning_rate": 9.96170421045386e-05,
      "loss": 0.1889,
      "step": 646
    },
    {
      "epoch": 0.13581024349286314,
      "grad_norm": 0.14956693351268768,
      "learning_rate": 9.96125025302587e-05,
      "loss": 0.1744,
      "step": 647
    },
    {
      "epoch": 0.13602015113350127,
      "grad_norm": 0.15064063668251038,
      "learning_rate": 9.96079363129317e-05,
      "loss": 0.1625,
      "step": 648
    },
    {
      "epoch": 0.13623005877413938,
      "grad_norm": 0.15413698554039001,
      "learning_rate": 9.960334345500974e-05,
      "loss": 0.1745,
      "step": 649
    },
    {
      "epoch": 0.13643996641477749,
      "grad_norm": 0.18636789917945862,
      "learning_rate": 9.959872395895929e-05,
      "loss": 0.1639,
      "step": 650
    },
    {
      "epoch": 0.13664987405541562,
      "grad_norm": 0.2444021999835968,
      "learning_rate": 9.959407782726108e-05,
      "loss": 0.1826,
      "step": 651
    },
    {
      "epoch": 0.13685978169605373,
      "grad_norm": 0.16845685243606567,
      "learning_rate": 9.958940506241026e-05,
      "loss": 0.1793,
      "step": 652
    },
    {
      "epoch": 0.13706968933669186,
      "grad_norm": 0.16233941912651062,
      "learning_rate": 9.958470566691618e-05,
      "loss": 0.18,
      "step": 653
    },
    {
      "epoch": 0.13727959697732997,
      "grad_norm": 0.18665330111980438,
      "learning_rate": 9.95799796433025e-05,
      "loss": 0.1662,
      "step": 654
    },
    {
      "epoch": 0.1374895046179681,
      "grad_norm": 0.18732531368732452,
      "learning_rate": 9.957522699410723e-05,
      "loss": 0.1646,
      "step": 655
    },
    {
      "epoch": 0.1376994122586062,
      "grad_norm": 0.12580807507038116,
      "learning_rate": 9.957044772188266e-05,
      "loss": 0.1616,
      "step": 656
    },
    {
      "epoch": 0.13790931989924432,
      "grad_norm": 0.17410576343536377,
      "learning_rate": 9.956564182919535e-05,
      "loss": 0.1628,
      "step": 657
    },
    {
      "epoch": 0.13811922753988246,
      "grad_norm": 0.1753510683774948,
      "learning_rate": 9.95608093186262e-05,
      "loss": 0.1867,
      "step": 658
    },
    {
      "epoch": 0.13832913518052056,
      "grad_norm": 0.17664988338947296,
      "learning_rate": 9.955595019277032e-05,
      "loss": 0.1819,
      "step": 659
    },
    {
      "epoch": 0.1385390428211587,
      "grad_norm": 0.1787986010313034,
      "learning_rate": 9.955106445423722e-05,
      "loss": 0.1668,
      "step": 660
    },
    {
      "epoch": 0.1387489504617968,
      "grad_norm": 0.14035004377365112,
      "learning_rate": 9.954615210565065e-05,
      "loss": 0.1651,
      "step": 661
    },
    {
      "epoch": 0.13895885810243494,
      "grad_norm": 0.20008955895900726,
      "learning_rate": 9.954121314964864e-05,
      "loss": 0.179,
      "step": 662
    },
    {
      "epoch": 0.13916876574307305,
      "grad_norm": 0.22378800809383392,
      "learning_rate": 9.953624758888352e-05,
      "loss": 0.1729,
      "step": 663
    },
    {
      "epoch": 0.13937867338371115,
      "grad_norm": 0.18687045574188232,
      "learning_rate": 9.953125542602193e-05,
      "loss": 0.1756,
      "step": 664
    },
    {
      "epoch": 0.1395885810243493,
      "grad_norm": 0.1506877988576889,
      "learning_rate": 9.952623666374475e-05,
      "loss": 0.1616,
      "step": 665
    },
    {
      "epoch": 0.1397984886649874,
      "grad_norm": 0.20918136835098267,
      "learning_rate": 9.95211913047472e-05,
      "loss": 0.1888,
      "step": 666
    },
    {
      "epoch": 0.14000839630562553,
      "grad_norm": 0.17521612346172333,
      "learning_rate": 9.951611935173872e-05,
      "loss": 0.1886,
      "step": 667
    },
    {
      "epoch": 0.14021830394626364,
      "grad_norm": 0.14788690209388733,
      "learning_rate": 9.951102080744308e-05,
      "loss": 0.162,
      "step": 668
    },
    {
      "epoch": 0.14042821158690177,
      "grad_norm": 0.18560314178466797,
      "learning_rate": 9.950589567459832e-05,
      "loss": 0.1573,
      "step": 669
    },
    {
      "epoch": 0.14063811922753988,
      "grad_norm": 0.16913674771785736,
      "learning_rate": 9.950074395595675e-05,
      "loss": 0.1713,
      "step": 670
    },
    {
      "epoch": 0.140848026868178,
      "grad_norm": 0.13768184185028076,
      "learning_rate": 9.949556565428496e-05,
      "loss": 0.1733,
      "step": 671
    },
    {
      "epoch": 0.14105793450881612,
      "grad_norm": 0.19472239911556244,
      "learning_rate": 9.949036077236382e-05,
      "loss": 0.1638,
      "step": 672
    },
    {
      "epoch": 0.14126784214945423,
      "grad_norm": 0.17684867978096008,
      "learning_rate": 9.948512931298846e-05,
      "loss": 0.1686,
      "step": 673
    },
    {
      "epoch": 0.14147774979009237,
      "grad_norm": 0.20061515271663666,
      "learning_rate": 9.94798712789683e-05,
      "loss": 0.1921,
      "step": 674
    },
    {
      "epoch": 0.14168765743073047,
      "grad_norm": 0.17213481664657593,
      "learning_rate": 9.9474586673127e-05,
      "loss": 0.1636,
      "step": 675
    },
    {
      "epoch": 0.1418975650713686,
      "grad_norm": 0.16144217550754547,
      "learning_rate": 9.946927549830258e-05,
      "loss": 0.1594,
      "step": 676
    },
    {
      "epoch": 0.14210747271200672,
      "grad_norm": 0.16045495867729187,
      "learning_rate": 9.946393775734719e-05,
      "loss": 0.1585,
      "step": 677
    },
    {
      "epoch": 0.14231738035264482,
      "grad_norm": 0.168419748544693,
      "learning_rate": 9.945857345312735e-05,
      "loss": 0.1618,
      "step": 678
    },
    {
      "epoch": 0.14252728799328296,
      "grad_norm": 0.16631141304969788,
      "learning_rate": 9.945318258852383e-05,
      "loss": 0.1648,
      "step": 679
    },
    {
      "epoch": 0.14273719563392107,
      "grad_norm": 0.17133933305740356,
      "learning_rate": 9.944776516643161e-05,
      "loss": 0.1902,
      "step": 680
    },
    {
      "epoch": 0.1429471032745592,
      "grad_norm": 0.144994854927063,
      "learning_rate": 9.944232118976e-05,
      "loss": 0.1645,
      "step": 681
    },
    {
      "epoch": 0.1431570109151973,
      "grad_norm": 0.13521502912044525,
      "learning_rate": 9.943685066143252e-05,
      "loss": 0.1679,
      "step": 682
    },
    {
      "epoch": 0.14336691855583544,
      "grad_norm": 0.1505574733018875,
      "learning_rate": 9.943135358438698e-05,
      "loss": 0.1497,
      "step": 683
    },
    {
      "epoch": 0.14357682619647355,
      "grad_norm": 0.1701841652393341,
      "learning_rate": 9.942582996157544e-05,
      "loss": 0.141,
      "step": 684
    },
    {
      "epoch": 0.14378673383711166,
      "grad_norm": 0.16892337799072266,
      "learning_rate": 9.94202797959642e-05,
      "loss": 0.1845,
      "step": 685
    },
    {
      "epoch": 0.1439966414777498,
      "grad_norm": 0.1322741061449051,
      "learning_rate": 9.941470309053384e-05,
      "loss": 0.1635,
      "step": 686
    },
    {
      "epoch": 0.1442065491183879,
      "grad_norm": 0.18180270493030548,
      "learning_rate": 9.940909984827915e-05,
      "loss": 0.1521,
      "step": 687
    },
    {
      "epoch": 0.14441645675902604,
      "grad_norm": 0.17136745154857635,
      "learning_rate": 9.940347007220924e-05,
      "loss": 0.1694,
      "step": 688
    },
    {
      "epoch": 0.14462636439966414,
      "grad_norm": 0.16122983396053314,
      "learning_rate": 9.93978137653474e-05,
      "loss": 0.1726,
      "step": 689
    },
    {
      "epoch": 0.14483627204030228,
      "grad_norm": 0.15024663507938385,
      "learning_rate": 9.939213093073118e-05,
      "loss": 0.1703,
      "step": 690
    },
    {
      "epoch": 0.14504617968094038,
      "grad_norm": 0.14193399250507355,
      "learning_rate": 9.938642157141245e-05,
      "loss": 0.1837,
      "step": 691
    },
    {
      "epoch": 0.1452560873215785,
      "grad_norm": 0.15650463104248047,
      "learning_rate": 9.938068569045721e-05,
      "loss": 0.1665,
      "step": 692
    },
    {
      "epoch": 0.14546599496221663,
      "grad_norm": 0.14254000782966614,
      "learning_rate": 9.937492329094577e-05,
      "loss": 0.175,
      "step": 693
    },
    {
      "epoch": 0.14567590260285473,
      "grad_norm": 0.17051447927951813,
      "learning_rate": 9.93691343759727e-05,
      "loss": 0.1664,
      "step": 694
    },
    {
      "epoch": 0.14588581024349287,
      "grad_norm": 0.16692955791950226,
      "learning_rate": 9.936331894864677e-05,
      "loss": 0.1682,
      "step": 695
    },
    {
      "epoch": 0.14609571788413098,
      "grad_norm": 0.20158237218856812,
      "learning_rate": 9.935747701209096e-05,
      "loss": 0.1574,
      "step": 696
    },
    {
      "epoch": 0.1463056255247691,
      "grad_norm": 0.1351911425590515,
      "learning_rate": 9.935160856944257e-05,
      "loss": 0.1618,
      "step": 697
    },
    {
      "epoch": 0.14651553316540722,
      "grad_norm": 0.1611570417881012,
      "learning_rate": 9.934571362385305e-05,
      "loss": 0.164,
      "step": 698
    },
    {
      "epoch": 0.14672544080604533,
      "grad_norm": 0.22555968165397644,
      "learning_rate": 9.933979217848815e-05,
      "loss": 0.2044,
      "step": 699
    },
    {
      "epoch": 0.14693534844668346,
      "grad_norm": 0.17471735179424286,
      "learning_rate": 9.93338442365278e-05,
      "loss": 0.1809,
      "step": 700
    },
    {
      "epoch": 0.14714525608732157,
      "grad_norm": 0.12285126745700836,
      "learning_rate": 9.93278698011662e-05,
      "loss": 0.1644,
      "step": 701
    },
    {
      "epoch": 0.1473551637279597,
      "grad_norm": 0.1364145576953888,
      "learning_rate": 9.932186887561175e-05,
      "loss": 0.1655,
      "step": 702
    },
    {
      "epoch": 0.1475650713685978,
      "grad_norm": 0.14189580082893372,
      "learning_rate": 9.931584146308708e-05,
      "loss": 0.1729,
      "step": 703
    },
    {
      "epoch": 0.14777497900923595,
      "grad_norm": 0.2254076600074768,
      "learning_rate": 9.930978756682905e-05,
      "loss": 0.1731,
      "step": 704
    },
    {
      "epoch": 0.14798488664987405,
      "grad_norm": 0.15398503839969635,
      "learning_rate": 9.930370719008875e-05,
      "loss": 0.1813,
      "step": 705
    },
    {
      "epoch": 0.1481947942905122,
      "grad_norm": 0.15267014503479004,
      "learning_rate": 9.929760033613146e-05,
      "loss": 0.1681,
      "step": 706
    },
    {
      "epoch": 0.1484047019311503,
      "grad_norm": 0.1611442267894745,
      "learning_rate": 9.929146700823671e-05,
      "loss": 0.1925,
      "step": 707
    },
    {
      "epoch": 0.1486146095717884,
      "grad_norm": 0.15641498565673828,
      "learning_rate": 9.928530720969827e-05,
      "loss": 0.1733,
      "step": 708
    },
    {
      "epoch": 0.14882451721242654,
      "grad_norm": 0.12724818289279938,
      "learning_rate": 9.927912094382403e-05,
      "loss": 0.1735,
      "step": 709
    },
    {
      "epoch": 0.14903442485306465,
      "grad_norm": 0.17570020258426666,
      "learning_rate": 9.92729082139362e-05,
      "loss": 0.1732,
      "step": 710
    },
    {
      "epoch": 0.14924433249370278,
      "grad_norm": 0.19477427005767822,
      "learning_rate": 9.926666902337115e-05,
      "loss": 0.1551,
      "step": 711
    },
    {
      "epoch": 0.1494542401343409,
      "grad_norm": 0.16798420250415802,
      "learning_rate": 9.926040337547946e-05,
      "loss": 0.1906,
      "step": 712
    },
    {
      "epoch": 0.14966414777497902,
      "grad_norm": 0.17551501095294952,
      "learning_rate": 9.925411127362594e-05,
      "loss": 0.1747,
      "step": 713
    },
    {
      "epoch": 0.14987405541561713,
      "grad_norm": 0.1842852234840393,
      "learning_rate": 9.924779272118957e-05,
      "loss": 0.1563,
      "step": 714
    },
    {
      "epoch": 0.15008396305625524,
      "grad_norm": 0.1805478185415268,
      "learning_rate": 9.924144772156358e-05,
      "loss": 0.165,
      "step": 715
    },
    {
      "epoch": 0.15029387069689337,
      "grad_norm": 0.21542084217071533,
      "learning_rate": 9.923507627815536e-05,
      "loss": 0.1469,
      "step": 716
    },
    {
      "epoch": 0.15050377833753148,
      "grad_norm": 0.15649262070655823,
      "learning_rate": 9.922867839438654e-05,
      "loss": 0.1706,
      "step": 717
    },
    {
      "epoch": 0.15071368597816961,
      "grad_norm": 0.14313052594661713,
      "learning_rate": 9.92222540736929e-05,
      "loss": 0.1737,
      "step": 718
    },
    {
      "epoch": 0.15092359361880772,
      "grad_norm": 0.13190749287605286,
      "learning_rate": 9.92158033195245e-05,
      "loss": 0.1742,
      "step": 719
    },
    {
      "epoch": 0.15113350125944586,
      "grad_norm": 0.18187177181243896,
      "learning_rate": 9.920932613534549e-05,
      "loss": 0.1748,
      "step": 720
    },
    {
      "epoch": 0.15134340890008396,
      "grad_norm": 0.14302362501621246,
      "learning_rate": 9.920282252463429e-05,
      "loss": 0.1711,
      "step": 721
    },
    {
      "epoch": 0.15155331654072207,
      "grad_norm": 0.16898708045482635,
      "learning_rate": 9.919629249088347e-05,
      "loss": 0.1616,
      "step": 722
    },
    {
      "epoch": 0.1517632241813602,
      "grad_norm": 0.15528154373168945,
      "learning_rate": 9.918973603759984e-05,
      "loss": 0.1652,
      "step": 723
    },
    {
      "epoch": 0.15197313182199831,
      "grad_norm": 0.1718195527791977,
      "learning_rate": 9.918315316830434e-05,
      "loss": 0.1758,
      "step": 724
    },
    {
      "epoch": 0.15218303946263645,
      "grad_norm": 0.15404529869556427,
      "learning_rate": 9.917654388653211e-05,
      "loss": 0.162,
      "step": 725
    },
    {
      "epoch": 0.15239294710327456,
      "grad_norm": 0.19148094952106476,
      "learning_rate": 9.916990819583252e-05,
      "loss": 0.1887,
      "step": 726
    },
    {
      "epoch": 0.1526028547439127,
      "grad_norm": 0.21298371255397797,
      "learning_rate": 9.916324609976906e-05,
      "loss": 0.1712,
      "step": 727
    },
    {
      "epoch": 0.1528127623845508,
      "grad_norm": 0.2041487991809845,
      "learning_rate": 9.915655760191944e-05,
      "loss": 0.167,
      "step": 728
    },
    {
      "epoch": 0.1530226700251889,
      "grad_norm": 0.17242367565631866,
      "learning_rate": 9.914984270587552e-05,
      "loss": 0.163,
      "step": 729
    },
    {
      "epoch": 0.15323257766582704,
      "grad_norm": 0.1646365076303482,
      "learning_rate": 9.914310141524339e-05,
      "loss": 0.1638,
      "step": 730
    },
    {
      "epoch": 0.15344248530646515,
      "grad_norm": 0.1947726458311081,
      "learning_rate": 9.913633373364324e-05,
      "loss": 0.1828,
      "step": 731
    },
    {
      "epoch": 0.15365239294710328,
      "grad_norm": 0.17962804436683655,
      "learning_rate": 9.912953966470948e-05,
      "loss": 0.1699,
      "step": 732
    },
    {
      "epoch": 0.1538623005877414,
      "grad_norm": 0.17121249437332153,
      "learning_rate": 9.912271921209068e-05,
      "loss": 0.1605,
      "step": 733
    },
    {
      "epoch": 0.15407220822837953,
      "grad_norm": 0.17192313075065613,
      "learning_rate": 9.911587237944959e-05,
      "loss": 0.1746,
      "step": 734
    },
    {
      "epoch": 0.15428211586901763,
      "grad_norm": 0.1262722760438919,
      "learning_rate": 9.910899917046311e-05,
      "loss": 0.1677,
      "step": 735
    },
    {
      "epoch": 0.15449202350965574,
      "grad_norm": 0.15364903211593628,
      "learning_rate": 9.910209958882231e-05,
      "loss": 0.1792,
      "step": 736
    },
    {
      "epoch": 0.15470193115029388,
      "grad_norm": 0.17038469016551971,
      "learning_rate": 9.909517363823241e-05,
      "loss": 0.1588,
      "step": 737
    },
    {
      "epoch": 0.15491183879093198,
      "grad_norm": 0.14676110446453094,
      "learning_rate": 9.908822132241281e-05,
      "loss": 0.1639,
      "step": 738
    },
    {
      "epoch": 0.15512174643157012,
      "grad_norm": 0.15300148725509644,
      "learning_rate": 9.908124264509707e-05,
      "loss": 0.1752,
      "step": 739
    },
    {
      "epoch": 0.15533165407220823,
      "grad_norm": 0.1372271180152893,
      "learning_rate": 9.90742376100329e-05,
      "loss": 0.1728,
      "step": 740
    },
    {
      "epoch": 0.15554156171284636,
      "grad_norm": 0.13885709643363953,
      "learning_rate": 9.906720622098215e-05,
      "loss": 0.1625,
      "step": 741
    },
    {
      "epoch": 0.15575146935348447,
      "grad_norm": 0.12582066655158997,
      "learning_rate": 9.906014848172086e-05,
      "loss": 0.17,
      "step": 742
    },
    {
      "epoch": 0.15596137699412257,
      "grad_norm": 0.13245654106140137,
      "learning_rate": 9.905306439603918e-05,
      "loss": 0.1711,
      "step": 743
    },
    {
      "epoch": 0.1561712846347607,
      "grad_norm": 0.12488420307636261,
      "learning_rate": 9.904595396774142e-05,
      "loss": 0.1725,
      "step": 744
    },
    {
      "epoch": 0.15638119227539882,
      "grad_norm": 0.1577821522951126,
      "learning_rate": 9.903881720064606e-05,
      "loss": 0.1707,
      "step": 745
    },
    {
      "epoch": 0.15659109991603695,
      "grad_norm": 0.16098785400390625,
      "learning_rate": 9.903165409858567e-05,
      "loss": 0.1608,
      "step": 746
    },
    {
      "epoch": 0.15680100755667506,
      "grad_norm": 0.14498507976531982,
      "learning_rate": 9.902446466540707e-05,
      "loss": 0.1776,
      "step": 747
    },
    {
      "epoch": 0.1570109151973132,
      "grad_norm": 0.15581083297729492,
      "learning_rate": 9.901724890497109e-05,
      "loss": 0.1746,
      "step": 748
    },
    {
      "epoch": 0.1572208228379513,
      "grad_norm": 0.16081508994102478,
      "learning_rate": 9.901000682115276e-05,
      "loss": 0.1869,
      "step": 749
    },
    {
      "epoch": 0.1574307304785894,
      "grad_norm": 0.16249604523181915,
      "learning_rate": 9.900273841784126e-05,
      "loss": 0.1629,
      "step": 750
    },
    {
      "epoch": 0.15764063811922754,
      "grad_norm": 0.1613830178976059,
      "learning_rate": 9.899544369893992e-05,
      "loss": 0.1588,
      "step": 751
    },
    {
      "epoch": 0.15785054575986565,
      "grad_norm": 0.15741245448589325,
      "learning_rate": 9.898812266836613e-05,
      "loss": 0.1708,
      "step": 752
    },
    {
      "epoch": 0.1580604534005038,
      "grad_norm": 0.12306373566389084,
      "learning_rate": 9.898077533005144e-05,
      "loss": 0.1751,
      "step": 753
    },
    {
      "epoch": 0.1582703610411419,
      "grad_norm": 0.15071968734264374,
      "learning_rate": 9.897340168794155e-05,
      "loss": 0.1686,
      "step": 754
    },
    {
      "epoch": 0.15848026868178003,
      "grad_norm": 0.14102348685264587,
      "learning_rate": 9.896600174599632e-05,
      "loss": 0.1701,
      "step": 755
    },
    {
      "epoch": 0.15869017632241814,
      "grad_norm": 0.16288180649280548,
      "learning_rate": 9.895857550818963e-05,
      "loss": 0.1652,
      "step": 756
    },
    {
      "epoch": 0.15890008396305624,
      "grad_norm": 0.17342409491539001,
      "learning_rate": 9.895112297850956e-05,
      "loss": 0.1702,
      "step": 757
    },
    {
      "epoch": 0.15910999160369438,
      "grad_norm": 0.13485445082187653,
      "learning_rate": 9.894364416095829e-05,
      "loss": 0.1705,
      "step": 758
    },
    {
      "epoch": 0.15931989924433249,
      "grad_norm": 0.17127734422683716,
      "learning_rate": 9.893613905955211e-05,
      "loss": 0.1831,
      "step": 759
    },
    {
      "epoch": 0.15952980688497062,
      "grad_norm": 0.13848379254341125,
      "learning_rate": 9.892860767832144e-05,
      "loss": 0.161,
      "step": 760
    },
    {
      "epoch": 0.15973971452560873,
      "grad_norm": 0.12154096364974976,
      "learning_rate": 9.892105002131081e-05,
      "loss": 0.1637,
      "step": 761
    },
    {
      "epoch": 0.15994962216624686,
      "grad_norm": 0.15433917939662933,
      "learning_rate": 9.891346609257882e-05,
      "loss": 0.1765,
      "step": 762
    },
    {
      "epoch": 0.16015952980688497,
      "grad_norm": 0.16946150362491608,
      "learning_rate": 9.890585589619825e-05,
      "loss": 0.1766,
      "step": 763
    },
    {
      "epoch": 0.16036943744752308,
      "grad_norm": 0.14503143727779388,
      "learning_rate": 9.889821943625594e-05,
      "loss": 0.1458,
      "step": 764
    },
    {
      "epoch": 0.1605793450881612,
      "grad_norm": 0.16491486132144928,
      "learning_rate": 9.889055671685283e-05,
      "loss": 0.1861,
      "step": 765
    },
    {
      "epoch": 0.16078925272879932,
      "grad_norm": 0.15087951719760895,
      "learning_rate": 9.888286774210398e-05,
      "loss": 0.1745,
      "step": 766
    },
    {
      "epoch": 0.16099916036943746,
      "grad_norm": 0.19266510009765625,
      "learning_rate": 9.887515251613857e-05,
      "loss": 0.1711,
      "step": 767
    },
    {
      "epoch": 0.16120906801007556,
      "grad_norm": 0.19419905543327332,
      "learning_rate": 9.886741104309981e-05,
      "loss": 0.1705,
      "step": 768
    },
    {
      "epoch": 0.1614189756507137,
      "grad_norm": 0.13408306241035461,
      "learning_rate": 9.885964332714508e-05,
      "loss": 0.1629,
      "step": 769
    },
    {
      "epoch": 0.1616288832913518,
      "grad_norm": 0.17757900059223175,
      "learning_rate": 9.885184937244581e-05,
      "loss": 0.1643,
      "step": 770
    },
    {
      "epoch": 0.1618387909319899,
      "grad_norm": 0.16376204788684845,
      "learning_rate": 9.884402918318754e-05,
      "loss": 0.1869,
      "step": 771
    },
    {
      "epoch": 0.16204869857262805,
      "grad_norm": 0.1362319141626358,
      "learning_rate": 9.883618276356988e-05,
      "loss": 0.1736,
      "step": 772
    },
    {
      "epoch": 0.16225860621326615,
      "grad_norm": 0.15015824139118195,
      "learning_rate": 9.882831011780653e-05,
      "loss": 0.1598,
      "step": 773
    },
    {
      "epoch": 0.1624685138539043,
      "grad_norm": 0.1815597116947174,
      "learning_rate": 9.882041125012528e-05,
      "loss": 0.1597,
      "step": 774
    },
    {
      "epoch": 0.1626784214945424,
      "grad_norm": 0.16493958234786987,
      "learning_rate": 9.881248616476803e-05,
      "loss": 0.181,
      "step": 775
    },
    {
      "epoch": 0.16288832913518053,
      "grad_norm": 0.1213487982749939,
      "learning_rate": 9.880453486599072e-05,
      "loss": 0.1644,
      "step": 776
    },
    {
      "epoch": 0.16309823677581864,
      "grad_norm": 0.1440252661705017,
      "learning_rate": 9.879655735806337e-05,
      "loss": 0.1657,
      "step": 777
    },
    {
      "epoch": 0.16330814441645675,
      "grad_norm": 0.1406947672367096,
      "learning_rate": 9.878855364527007e-05,
      "loss": 0.172,
      "step": 778
    },
    {
      "epoch": 0.16351805205709488,
      "grad_norm": 0.17073537409305573,
      "learning_rate": 9.878052373190902e-05,
      "loss": 0.1852,
      "step": 779
    },
    {
      "epoch": 0.163727959697733,
      "grad_norm": 0.15924111008644104,
      "learning_rate": 9.877246762229247e-05,
      "loss": 0.18,
      "step": 780
    },
    {
      "epoch": 0.16393786733837112,
      "grad_norm": 0.14216449856758118,
      "learning_rate": 9.876438532074672e-05,
      "loss": 0.1575,
      "step": 781
    },
    {
      "epoch": 0.16414777497900923,
      "grad_norm": 0.12731285393238068,
      "learning_rate": 9.875627683161217e-05,
      "loss": 0.161,
      "step": 782
    },
    {
      "epoch": 0.16435768261964737,
      "grad_norm": 0.16211947798728943,
      "learning_rate": 9.874814215924324e-05,
      "loss": 0.1592,
      "step": 783
    },
    {
      "epoch": 0.16456759026028547,
      "grad_norm": 0.1463523507118225,
      "learning_rate": 9.873998130800844e-05,
      "loss": 0.1773,
      "step": 784
    },
    {
      "epoch": 0.16477749790092358,
      "grad_norm": 0.14601489901542664,
      "learning_rate": 9.873179428229033e-05,
      "loss": 0.1685,
      "step": 785
    },
    {
      "epoch": 0.16498740554156172,
      "grad_norm": 0.1590069830417633,
      "learning_rate": 9.872358108648557e-05,
      "loss": 0.1709,
      "step": 786
    },
    {
      "epoch": 0.16519731318219982,
      "grad_norm": 0.14027854800224304,
      "learning_rate": 9.871534172500479e-05,
      "loss": 0.1842,
      "step": 787
    },
    {
      "epoch": 0.16540722082283796,
      "grad_norm": 0.12902334332466125,
      "learning_rate": 9.870707620227271e-05,
      "loss": 0.1484,
      "step": 788
    },
    {
      "epoch": 0.16561712846347607,
      "grad_norm": 0.13427670300006866,
      "learning_rate": 9.869878452272812e-05,
      "loss": 0.1599,
      "step": 789
    },
    {
      "epoch": 0.1658270361041142,
      "grad_norm": 0.17016825079917908,
      "learning_rate": 9.869046669082386e-05,
      "loss": 0.1506,
      "step": 790
    },
    {
      "epoch": 0.1660369437447523,
      "grad_norm": 0.13757233321666718,
      "learning_rate": 9.868212271102678e-05,
      "loss": 0.1574,
      "step": 791
    },
    {
      "epoch": 0.16624685138539042,
      "grad_norm": 0.19485127925872803,
      "learning_rate": 9.867375258781778e-05,
      "loss": 0.1686,
      "step": 792
    },
    {
      "epoch": 0.16645675902602855,
      "grad_norm": 0.16131381690502167,
      "learning_rate": 9.866535632569182e-05,
      "loss": 0.1693,
      "step": 793
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.16035085916519165,
      "learning_rate": 9.865693392915787e-05,
      "loss": 0.1589,
      "step": 794
    },
    {
      "epoch": 0.1668765743073048,
      "grad_norm": 0.16287830471992493,
      "learning_rate": 9.864848540273897e-05,
      "loss": 0.1749,
      "step": 795
    },
    {
      "epoch": 0.1670864819479429,
      "grad_norm": 0.16353754699230194,
      "learning_rate": 9.864001075097214e-05,
      "loss": 0.1846,
      "step": 796
    },
    {
      "epoch": 0.16729638958858103,
      "grad_norm": 0.19826947152614594,
      "learning_rate": 9.863150997840849e-05,
      "loss": 0.1624,
      "step": 797
    },
    {
      "epoch": 0.16750629722921914,
      "grad_norm": 0.14296779036521912,
      "learning_rate": 9.862298308961313e-05,
      "loss": 0.1823,
      "step": 798
    },
    {
      "epoch": 0.16771620486985725,
      "grad_norm": 0.1371513456106186,
      "learning_rate": 9.861443008916517e-05,
      "loss": 0.1742,
      "step": 799
    },
    {
      "epoch": 0.16792611251049538,
      "grad_norm": 0.17043526470661163,
      "learning_rate": 9.86058509816578e-05,
      "loss": 0.1661,
      "step": 800
    },
    {
      "epoch": 0.1681360201511335,
      "grad_norm": 0.17045053839683533,
      "learning_rate": 9.859724577169815e-05,
      "loss": 0.1773,
      "step": 801
    },
    {
      "epoch": 0.16834592779177163,
      "grad_norm": 0.1564718633890152,
      "learning_rate": 9.858861446390748e-05,
      "loss": 0.1636,
      "step": 802
    },
    {
      "epoch": 0.16855583543240973,
      "grad_norm": 0.13335298001766205,
      "learning_rate": 9.857995706292092e-05,
      "loss": 0.18,
      "step": 803
    },
    {
      "epoch": 0.16876574307304787,
      "grad_norm": 0.14552432298660278,
      "learning_rate": 9.857127357338775e-05,
      "loss": 0.1599,
      "step": 804
    },
    {
      "epoch": 0.16897565071368598,
      "grad_norm": 0.1437654346227646,
      "learning_rate": 9.856256399997119e-05,
      "loss": 0.178,
      "step": 805
    },
    {
      "epoch": 0.16918555835432408,
      "grad_norm": 0.143344447016716,
      "learning_rate": 9.855382834734848e-05,
      "loss": 0.159,
      "step": 806
    },
    {
      "epoch": 0.16939546599496222,
      "grad_norm": 0.15504872798919678,
      "learning_rate": 9.854506662021085e-05,
      "loss": 0.1944,
      "step": 807
    },
    {
      "epoch": 0.16960537363560033,
      "grad_norm": 0.12894919514656067,
      "learning_rate": 9.853627882326357e-05,
      "loss": 0.1677,
      "step": 808
    },
    {
      "epoch": 0.16981528127623846,
      "grad_norm": 0.1339186728000641,
      "learning_rate": 9.852746496122587e-05,
      "loss": 0.1666,
      "step": 809
    },
    {
      "epoch": 0.17002518891687657,
      "grad_norm": 0.11661262065172195,
      "learning_rate": 9.8518625038831e-05,
      "loss": 0.1611,
      "step": 810
    },
    {
      "epoch": 0.1702350965575147,
      "grad_norm": 0.14091713726520538,
      "learning_rate": 9.85097590608262e-05,
      "loss": 0.1483,
      "step": 811
    },
    {
      "epoch": 0.1704450041981528,
      "grad_norm": 0.13658110797405243,
      "learning_rate": 9.85008670319727e-05,
      "loss": 0.1667,
      "step": 812
    },
    {
      "epoch": 0.17065491183879095,
      "grad_norm": 0.13288968801498413,
      "learning_rate": 9.849194895704575e-05,
      "loss": 0.1685,
      "step": 813
    },
    {
      "epoch": 0.17086481947942905,
      "grad_norm": 0.1870601624250412,
      "learning_rate": 9.84830048408345e-05,
      "loss": 0.1794,
      "step": 814
    },
    {
      "epoch": 0.17107472712006716,
      "grad_norm": 0.15742088854312897,
      "learning_rate": 9.84740346881422e-05,
      "loss": 0.1546,
      "step": 815
    },
    {
      "epoch": 0.1712846347607053,
      "grad_norm": 0.1625792235136032,
      "learning_rate": 9.846503850378602e-05,
      "loss": 0.1506,
      "step": 816
    },
    {
      "epoch": 0.1714945424013434,
      "grad_norm": 0.1315995752811432,
      "learning_rate": 9.845601629259708e-05,
      "loss": 0.1611,
      "step": 817
    },
    {
      "epoch": 0.17170445004198154,
      "grad_norm": 0.16079425811767578,
      "learning_rate": 9.844696805942055e-05,
      "loss": 0.1588,
      "step": 818
    },
    {
      "epoch": 0.17191435768261965,
      "grad_norm": 0.16755861043930054,
      "learning_rate": 9.843789380911554e-05,
      "loss": 0.1629,
      "step": 819
    },
    {
      "epoch": 0.17212426532325778,
      "grad_norm": 0.12923209369182587,
      "learning_rate": 9.84287935465551e-05,
      "loss": 0.1675,
      "step": 820
    },
    {
      "epoch": 0.1723341729638959,
      "grad_norm": 0.14494512975215912,
      "learning_rate": 9.84196672766263e-05,
      "loss": 0.1674,
      "step": 821
    },
    {
      "epoch": 0.172544080604534,
      "grad_norm": 0.13111597299575806,
      "learning_rate": 9.841051500423014e-05,
      "loss": 0.1835,
      "step": 822
    },
    {
      "epoch": 0.17275398824517213,
      "grad_norm": 0.1468263864517212,
      "learning_rate": 9.840133673428162e-05,
      "loss": 0.1635,
      "step": 823
    },
    {
      "epoch": 0.17296389588581024,
      "grad_norm": 0.14243565499782562,
      "learning_rate": 9.839213247170967e-05,
      "loss": 0.1617,
      "step": 824
    },
    {
      "epoch": 0.17317380352644837,
      "grad_norm": 0.14563241600990295,
      "learning_rate": 9.838290222145718e-05,
      "loss": 0.1614,
      "step": 825
    },
    {
      "epoch": 0.17338371116708648,
      "grad_norm": 0.15182636678218842,
      "learning_rate": 9.837364598848102e-05,
      "loss": 0.1712,
      "step": 826
    },
    {
      "epoch": 0.17359361880772461,
      "grad_norm": 0.15270951390266418,
      "learning_rate": 9.8364363777752e-05,
      "loss": 0.1721,
      "step": 827
    },
    {
      "epoch": 0.17380352644836272,
      "grad_norm": 0.14734511077404022,
      "learning_rate": 9.835505559425487e-05,
      "loss": 0.1751,
      "step": 828
    },
    {
      "epoch": 0.17401343408900083,
      "grad_norm": 0.13699984550476074,
      "learning_rate": 9.834572144298834e-05,
      "loss": 0.1895,
      "step": 829
    },
    {
      "epoch": 0.17422334172963896,
      "grad_norm": 0.1422858089208603,
      "learning_rate": 9.833636132896505e-05,
      "loss": 0.1757,
      "step": 830
    },
    {
      "epoch": 0.17443324937027707,
      "grad_norm": 0.1300913542509079,
      "learning_rate": 9.832697525721161e-05,
      "loss": 0.1774,
      "step": 831
    },
    {
      "epoch": 0.1746431570109152,
      "grad_norm": 0.16247855126857758,
      "learning_rate": 9.831756323276856e-05,
      "loss": 0.1499,
      "step": 832
    },
    {
      "epoch": 0.1748530646515533,
      "grad_norm": 0.13687850534915924,
      "learning_rate": 9.830812526069036e-05,
      "loss": 0.1739,
      "step": 833
    },
    {
      "epoch": 0.17506297229219145,
      "grad_norm": 0.16242316365242004,
      "learning_rate": 9.829866134604543e-05,
      "loss": 0.1703,
      "step": 834
    },
    {
      "epoch": 0.17527287993282956,
      "grad_norm": 0.1983039677143097,
      "learning_rate": 9.82891714939161e-05,
      "loss": 0.1726,
      "step": 835
    },
    {
      "epoch": 0.17548278757346766,
      "grad_norm": 0.17525269091129303,
      "learning_rate": 9.827965570939861e-05,
      "loss": 0.1839,
      "step": 836
    },
    {
      "epoch": 0.1756926952141058,
      "grad_norm": 0.1517772525548935,
      "learning_rate": 9.827011399760319e-05,
      "loss": 0.17,
      "step": 837
    },
    {
      "epoch": 0.1759026028547439,
      "grad_norm": 0.17186671495437622,
      "learning_rate": 9.826054636365396e-05,
      "loss": 0.1526,
      "step": 838
    },
    {
      "epoch": 0.17611251049538204,
      "grad_norm": 0.15066854655742645,
      "learning_rate": 9.825095281268894e-05,
      "loss": 0.1442,
      "step": 839
    },
    {
      "epoch": 0.17632241813602015,
      "grad_norm": 0.14937272667884827,
      "learning_rate": 9.82413333498601e-05,
      "loss": 0.1641,
      "step": 840
    },
    {
      "epoch": 0.17653232577665828,
      "grad_norm": 0.1843566745519638,
      "learning_rate": 9.823168798033328e-05,
      "loss": 0.1674,
      "step": 841
    },
    {
      "epoch": 0.1767422334172964,
      "grad_norm": 0.15555419027805328,
      "learning_rate": 9.82220167092883e-05,
      "loss": 0.1551,
      "step": 842
    },
    {
      "epoch": 0.1769521410579345,
      "grad_norm": 0.1585703343153,
      "learning_rate": 9.821231954191885e-05,
      "loss": 0.1578,
      "step": 843
    },
    {
      "epoch": 0.17716204869857263,
      "grad_norm": 0.1662077009677887,
      "learning_rate": 9.82025964834325e-05,
      "loss": 0.1778,
      "step": 844
    },
    {
      "epoch": 0.17737195633921074,
      "grad_norm": 0.1385970562696457,
      "learning_rate": 9.819284753905078e-05,
      "loss": 0.1653,
      "step": 845
    },
    {
      "epoch": 0.17758186397984888,
      "grad_norm": 0.14429986476898193,
      "learning_rate": 9.81830727140091e-05,
      "loss": 0.1801,
      "step": 846
    },
    {
      "epoch": 0.17779177162048698,
      "grad_norm": 0.16140583157539368,
      "learning_rate": 9.817327201355675e-05,
      "loss": 0.1808,
      "step": 847
    },
    {
      "epoch": 0.17800167926112512,
      "grad_norm": 0.13548845052719116,
      "learning_rate": 9.816344544295692e-05,
      "loss": 0.1567,
      "step": 848
    },
    {
      "epoch": 0.17821158690176322,
      "grad_norm": 0.14740432798862457,
      "learning_rate": 9.815359300748674e-05,
      "loss": 0.1737,
      "step": 849
    },
    {
      "epoch": 0.17842149454240133,
      "grad_norm": 0.15687352418899536,
      "learning_rate": 9.814371471243715e-05,
      "loss": 0.1775,
      "step": 850
    },
    {
      "epoch": 0.17863140218303947,
      "grad_norm": 0.14518170058727264,
      "learning_rate": 9.813381056311307e-05,
      "loss": 0.1536,
      "step": 851
    },
    {
      "epoch": 0.17884130982367757,
      "grad_norm": 0.1569397896528244,
      "learning_rate": 9.812388056483319e-05,
      "loss": 0.1667,
      "step": 852
    },
    {
      "epoch": 0.1790512174643157,
      "grad_norm": 0.13371489942073822,
      "learning_rate": 9.81139247229302e-05,
      "loss": 0.175,
      "step": 853
    },
    {
      "epoch": 0.17926112510495382,
      "grad_norm": 0.15128138661384583,
      "learning_rate": 9.810394304275058e-05,
      "loss": 0.1601,
      "step": 854
    },
    {
      "epoch": 0.17947103274559195,
      "grad_norm": 0.12941020727157593,
      "learning_rate": 9.809393552965476e-05,
      "loss": 0.1579,
      "step": 855
    },
    {
      "epoch": 0.17968094038623006,
      "grad_norm": 0.1644691675901413,
      "learning_rate": 9.808390218901696e-05,
      "loss": 0.1737,
      "step": 856
    },
    {
      "epoch": 0.17989084802686817,
      "grad_norm": 0.16271840035915375,
      "learning_rate": 9.807384302622533e-05,
      "loss": 0.1846,
      "step": 857
    },
    {
      "epoch": 0.1801007556675063,
      "grad_norm": 0.15787339210510254,
      "learning_rate": 9.806375804668189e-05,
      "loss": 0.1733,
      "step": 858
    },
    {
      "epoch": 0.1803106633081444,
      "grad_norm": 0.16740775108337402,
      "learning_rate": 9.805364725580248e-05,
      "loss": 0.1625,
      "step": 859
    },
    {
      "epoch": 0.18052057094878254,
      "grad_norm": 0.11970685422420502,
      "learning_rate": 9.804351065901682e-05,
      "loss": 0.1554,
      "step": 860
    },
    {
      "epoch": 0.18073047858942065,
      "grad_norm": 0.21359467506408691,
      "learning_rate": 9.803334826176852e-05,
      "loss": 0.1658,
      "step": 861
    },
    {
      "epoch": 0.1809403862300588,
      "grad_norm": 0.1699121594429016,
      "learning_rate": 9.8023160069515e-05,
      "loss": 0.1608,
      "step": 862
    },
    {
      "epoch": 0.1811502938706969,
      "grad_norm": 0.15073060989379883,
      "learning_rate": 9.801294608772755e-05,
      "loss": 0.156,
      "step": 863
    },
    {
      "epoch": 0.181360201511335,
      "grad_norm": 0.17342859506607056,
      "learning_rate": 9.800270632189133e-05,
      "loss": 0.1829,
      "step": 864
    },
    {
      "epoch": 0.18157010915197314,
      "grad_norm": 0.20463357865810394,
      "learning_rate": 9.799244077750531e-05,
      "loss": 0.1839,
      "step": 865
    },
    {
      "epoch": 0.18178001679261124,
      "grad_norm": 0.17843082547187805,
      "learning_rate": 9.798214946008234e-05,
      "loss": 0.1707,
      "step": 866
    },
    {
      "epoch": 0.18198992443324938,
      "grad_norm": 0.14360542595386505,
      "learning_rate": 9.797183237514907e-05,
      "loss": 0.1687,
      "step": 867
    },
    {
      "epoch": 0.18219983207388749,
      "grad_norm": 0.15701772272586823,
      "learning_rate": 9.796148952824603e-05,
      "loss": 0.1584,
      "step": 868
    },
    {
      "epoch": 0.18240973971452562,
      "grad_norm": 0.12531954050064087,
      "learning_rate": 9.795112092492755e-05,
      "loss": 0.1467,
      "step": 869
    },
    {
      "epoch": 0.18261964735516373,
      "grad_norm": 0.16143764555454254,
      "learning_rate": 9.794072657076182e-05,
      "loss": 0.1687,
      "step": 870
    },
    {
      "epoch": 0.18282955499580184,
      "grad_norm": 0.13665997982025146,
      "learning_rate": 9.793030647133084e-05,
      "loss": 0.1644,
      "step": 871
    },
    {
      "epoch": 0.18303946263643997,
      "grad_norm": 0.15797339379787445,
      "learning_rate": 9.791986063223045e-05,
      "loss": 0.1737,
      "step": 872
    },
    {
      "epoch": 0.18324937027707808,
      "grad_norm": 0.13726350665092468,
      "learning_rate": 9.790938905907027e-05,
      "loss": 0.1842,
      "step": 873
    },
    {
      "epoch": 0.1834592779177162,
      "grad_norm": 0.13706736266613007,
      "learning_rate": 9.78988917574738e-05,
      "loss": 0.1699,
      "step": 874
    },
    {
      "epoch": 0.18366918555835432,
      "grad_norm": 0.13394543528556824,
      "learning_rate": 9.788836873307835e-05,
      "loss": 0.1757,
      "step": 875
    },
    {
      "epoch": 0.18387909319899245,
      "grad_norm": 0.13909773528575897,
      "learning_rate": 9.7877819991535e-05,
      "loss": 0.1761,
      "step": 876
    },
    {
      "epoch": 0.18408900083963056,
      "grad_norm": 0.1416359543800354,
      "learning_rate": 9.786724553850865e-05,
      "loss": 0.1657,
      "step": 877
    },
    {
      "epoch": 0.18429890848026867,
      "grad_norm": 0.13720481097698212,
      "learning_rate": 9.785664537967806e-05,
      "loss": 0.173,
      "step": 878
    },
    {
      "epoch": 0.1845088161209068,
      "grad_norm": 0.14136558771133423,
      "learning_rate": 9.784601952073573e-05,
      "loss": 0.1625,
      "step": 879
    },
    {
      "epoch": 0.1847187237615449,
      "grad_norm": 0.1424713134765625,
      "learning_rate": 9.783536796738802e-05,
      "loss": 0.1676,
      "step": 880
    },
    {
      "epoch": 0.18492863140218305,
      "grad_norm": 0.163283571600914,
      "learning_rate": 9.782469072535502e-05,
      "loss": 0.1574,
      "step": 881
    },
    {
      "epoch": 0.18513853904282115,
      "grad_norm": 0.16645042598247528,
      "learning_rate": 9.781398780037067e-05,
      "loss": 0.1576,
      "step": 882
    },
    {
      "epoch": 0.1853484466834593,
      "grad_norm": 0.17722882330417633,
      "learning_rate": 9.780325919818268e-05,
      "loss": 0.1815,
      "step": 883
    },
    {
      "epoch": 0.1855583543240974,
      "grad_norm": 0.1435573697090149,
      "learning_rate": 9.779250492455257e-05,
      "loss": 0.168,
      "step": 884
    },
    {
      "epoch": 0.1857682619647355,
      "grad_norm": 0.13388660550117493,
      "learning_rate": 9.778172498525559e-05,
      "loss": 0.1493,
      "step": 885
    },
    {
      "epoch": 0.18597816960537364,
      "grad_norm": 0.15198923647403717,
      "learning_rate": 9.777091938608088e-05,
      "loss": 0.1759,
      "step": 886
    },
    {
      "epoch": 0.18618807724601175,
      "grad_norm": 0.14781691133975983,
      "learning_rate": 9.776008813283125e-05,
      "loss": 0.1521,
      "step": 887
    },
    {
      "epoch": 0.18639798488664988,
      "grad_norm": 0.18348653614521027,
      "learning_rate": 9.774923123132332e-05,
      "loss": 0.1522,
      "step": 888
    },
    {
      "epoch": 0.186607892527288,
      "grad_norm": 0.15346892178058624,
      "learning_rate": 9.773834868738752e-05,
      "loss": 0.1593,
      "step": 889
    },
    {
      "epoch": 0.18681780016792612,
      "grad_norm": 0.1563442200422287,
      "learning_rate": 9.7727440506868e-05,
      "loss": 0.1569,
      "step": 890
    },
    {
      "epoch": 0.18702770780856423,
      "grad_norm": 0.12280000746250153,
      "learning_rate": 9.771650669562274e-05,
      "loss": 0.1593,
      "step": 891
    },
    {
      "epoch": 0.18723761544920234,
      "grad_norm": 0.1622755527496338,
      "learning_rate": 9.770554725952341e-05,
      "loss": 0.1736,
      "step": 892
    },
    {
      "epoch": 0.18744752308984047,
      "grad_norm": 0.19185318052768707,
      "learning_rate": 9.769456220445549e-05,
      "loss": 0.165,
      "step": 893
    },
    {
      "epoch": 0.18765743073047858,
      "grad_norm": 0.16040024161338806,
      "learning_rate": 9.768355153631822e-05,
      "loss": 0.1837,
      "step": 894
    },
    {
      "epoch": 0.18786733837111672,
      "grad_norm": 0.11211330443620682,
      "learning_rate": 9.767251526102456e-05,
      "loss": 0.152,
      "step": 895
    },
    {
      "epoch": 0.18807724601175482,
      "grad_norm": 0.16628898680210114,
      "learning_rate": 9.766145338450125e-05,
      "loss": 0.1729,
      "step": 896
    },
    {
      "epoch": 0.18828715365239296,
      "grad_norm": 0.14260315895080566,
      "learning_rate": 9.765036591268877e-05,
      "loss": 0.165,
      "step": 897
    },
    {
      "epoch": 0.18849706129303107,
      "grad_norm": 0.12734055519104004,
      "learning_rate": 9.763925285154135e-05,
      "loss": 0.1714,
      "step": 898
    },
    {
      "epoch": 0.18870696893366917,
      "grad_norm": 0.1465056985616684,
      "learning_rate": 9.762811420702693e-05,
      "loss": 0.1805,
      "step": 899
    },
    {
      "epoch": 0.1889168765743073,
      "grad_norm": 0.12369433790445328,
      "learning_rate": 9.761694998512727e-05,
      "loss": 0.1737,
      "step": 900
    },
    {
      "epoch": 0.18912678421494541,
      "grad_norm": 0.13893358409404755,
      "learning_rate": 9.760576019183775e-05,
      "loss": 0.1502,
      "step": 901
    },
    {
      "epoch": 0.18933669185558355,
      "grad_norm": 0.13517262041568756,
      "learning_rate": 9.759454483316761e-05,
      "loss": 0.1648,
      "step": 902
    },
    {
      "epoch": 0.18954659949622166,
      "grad_norm": 0.13170479238033295,
      "learning_rate": 9.75833039151397e-05,
      "loss": 0.159,
      "step": 903
    },
    {
      "epoch": 0.1897565071368598,
      "grad_norm": 0.13293495774269104,
      "learning_rate": 9.757203744379067e-05,
      "loss": 0.1667,
      "step": 904
    },
    {
      "epoch": 0.1899664147774979,
      "grad_norm": 0.12381456047296524,
      "learning_rate": 9.756074542517088e-05,
      "loss": 0.1684,
      "step": 905
    },
    {
      "epoch": 0.190176322418136,
      "grad_norm": 0.12754730880260468,
      "learning_rate": 9.75494278653444e-05,
      "loss": 0.1559,
      "step": 906
    },
    {
      "epoch": 0.19038623005877414,
      "grad_norm": 0.17824846506118774,
      "learning_rate": 9.753808477038899e-05,
      "loss": 0.1533,
      "step": 907
    },
    {
      "epoch": 0.19059613769941225,
      "grad_norm": 0.13958828151226044,
      "learning_rate": 9.752671614639619e-05,
      "loss": 0.1494,
      "step": 908
    },
    {
      "epoch": 0.19080604534005038,
      "grad_norm": 0.14593012630939484,
      "learning_rate": 9.75153219994712e-05,
      "loss": 0.1598,
      "step": 909
    },
    {
      "epoch": 0.1910159529806885,
      "grad_norm": 0.19262051582336426,
      "learning_rate": 9.750390233573293e-05,
      "loss": 0.1638,
      "step": 910
    },
    {
      "epoch": 0.19122586062132663,
      "grad_norm": 0.15087257325649261,
      "learning_rate": 9.7492457161314e-05,
      "loss": 0.1644,
      "step": 911
    },
    {
      "epoch": 0.19143576826196473,
      "grad_norm": 0.17646470665931702,
      "learning_rate": 9.748098648236072e-05,
      "loss": 0.1644,
      "step": 912
    },
    {
      "epoch": 0.19164567590260284,
      "grad_norm": 0.15336105227470398,
      "learning_rate": 9.746949030503312e-05,
      "loss": 0.1717,
      "step": 913
    },
    {
      "epoch": 0.19185558354324098,
      "grad_norm": 0.15364870429039001,
      "learning_rate": 9.745796863550492e-05,
      "loss": 0.1721,
      "step": 914
    },
    {
      "epoch": 0.19206549118387908,
      "grad_norm": 0.1769437938928604,
      "learning_rate": 9.74464214799635e-05,
      "loss": 0.1485,
      "step": 915
    },
    {
      "epoch": 0.19227539882451722,
      "grad_norm": 0.18178406357765198,
      "learning_rate": 9.743484884460993e-05,
      "loss": 0.1523,
      "step": 916
    },
    {
      "epoch": 0.19248530646515533,
      "grad_norm": 0.16841888427734375,
      "learning_rate": 9.742325073565905e-05,
      "loss": 0.148,
      "step": 917
    },
    {
      "epoch": 0.19269521410579346,
      "grad_norm": 0.13603579998016357,
      "learning_rate": 9.741162715933924e-05,
      "loss": 0.161,
      "step": 918
    },
    {
      "epoch": 0.19290512174643157,
      "grad_norm": 0.1348285973072052,
      "learning_rate": 9.739997812189265e-05,
      "loss": 0.145,
      "step": 919
    },
    {
      "epoch": 0.19311502938706968,
      "grad_norm": 0.13922441005706787,
      "learning_rate": 9.738830362957508e-05,
      "loss": 0.1607,
      "step": 920
    },
    {
      "epoch": 0.1933249370277078,
      "grad_norm": 0.12017328292131424,
      "learning_rate": 9.7376603688656e-05,
      "loss": 0.152,
      "step": 921
    },
    {
      "epoch": 0.19353484466834592,
      "grad_norm": 0.14594700932502747,
      "learning_rate": 9.736487830541853e-05,
      "loss": 0.1638,
      "step": 922
    },
    {
      "epoch": 0.19374475230898405,
      "grad_norm": 0.14390654861927032,
      "learning_rate": 9.73531274861595e-05,
      "loss": 0.1608,
      "step": 923
    },
    {
      "epoch": 0.19395465994962216,
      "grad_norm": 0.12394702434539795,
      "learning_rate": 9.734135123718933e-05,
      "loss": 0.1612,
      "step": 924
    },
    {
      "epoch": 0.1941645675902603,
      "grad_norm": 0.16754676401615143,
      "learning_rate": 9.732954956483218e-05,
      "loss": 0.1791,
      "step": 925
    },
    {
      "epoch": 0.1943744752308984,
      "grad_norm": 0.2816343605518341,
      "learning_rate": 9.731772247542576e-05,
      "loss": 0.1489,
      "step": 926
    },
    {
      "epoch": 0.19458438287153654,
      "grad_norm": 0.15909412503242493,
      "learning_rate": 9.730586997532155e-05,
      "loss": 0.1531,
      "step": 927
    },
    {
      "epoch": 0.19479429051217464,
      "grad_norm": 0.16626602411270142,
      "learning_rate": 9.729399207088457e-05,
      "loss": 0.1729,
      "step": 928
    },
    {
      "epoch": 0.19500419815281275,
      "grad_norm": 0.14581038057804108,
      "learning_rate": 9.728208876849354e-05,
      "loss": 0.1616,
      "step": 929
    },
    {
      "epoch": 0.1952141057934509,
      "grad_norm": 0.14985312521457672,
      "learning_rate": 9.727016007454079e-05,
      "loss": 0.1583,
      "step": 930
    },
    {
      "epoch": 0.195424013434089,
      "grad_norm": 0.13958559930324554,
      "learning_rate": 9.725820599543234e-05,
      "loss": 0.1646,
      "step": 931
    },
    {
      "epoch": 0.19563392107472713,
      "grad_norm": 0.1607862412929535,
      "learning_rate": 9.724622653758777e-05,
      "loss": 0.1549,
      "step": 932
    },
    {
      "epoch": 0.19584382871536524,
      "grad_norm": 0.17007960379123688,
      "learning_rate": 9.723422170744031e-05,
      "loss": 0.1718,
      "step": 933
    },
    {
      "epoch": 0.19605373635600337,
      "grad_norm": 0.1419927030801773,
      "learning_rate": 9.722219151143688e-05,
      "loss": 0.1689,
      "step": 934
    },
    {
      "epoch": 0.19626364399664148,
      "grad_norm": 0.1631292998790741,
      "learning_rate": 9.721013595603793e-05,
      "loss": 0.1611,
      "step": 935
    },
    {
      "epoch": 0.1964735516372796,
      "grad_norm": 0.19870012998580933,
      "learning_rate": 9.719805504771758e-05,
      "loss": 0.1836,
      "step": 936
    },
    {
      "epoch": 0.19668345927791772,
      "grad_norm": 0.14150285720825195,
      "learning_rate": 9.718594879296355e-05,
      "loss": 0.1718,
      "step": 937
    },
    {
      "epoch": 0.19689336691855583,
      "grad_norm": 0.1416793018579483,
      "learning_rate": 9.717381719827716e-05,
      "loss": 0.1511,
      "step": 938
    },
    {
      "epoch": 0.19710327455919396,
      "grad_norm": 0.14615508913993835,
      "learning_rate": 9.716166027017339e-05,
      "loss": 0.1599,
      "step": 939
    },
    {
      "epoch": 0.19731318219983207,
      "grad_norm": 0.13773533701896667,
      "learning_rate": 9.714947801518076e-05,
      "loss": 0.1765,
      "step": 940
    },
    {
      "epoch": 0.1975230898404702,
      "grad_norm": 0.11306725442409515,
      "learning_rate": 9.713727043984143e-05,
      "loss": 0.1678,
      "step": 941
    },
    {
      "epoch": 0.1977329974811083,
      "grad_norm": 0.14150364696979523,
      "learning_rate": 9.712503755071115e-05,
      "loss": 0.1535,
      "step": 942
    },
    {
      "epoch": 0.19794290512174642,
      "grad_norm": 0.15026667714118958,
      "learning_rate": 9.711277935435925e-05,
      "loss": 0.155,
      "step": 943
    },
    {
      "epoch": 0.19815281276238456,
      "grad_norm": 0.1540324091911316,
      "learning_rate": 9.710049585736866e-05,
      "loss": 0.1866,
      "step": 944
    },
    {
      "epoch": 0.19836272040302266,
      "grad_norm": 0.12859384715557098,
      "learning_rate": 9.708818706633591e-05,
      "loss": 0.1512,
      "step": 945
    },
    {
      "epoch": 0.1985726280436608,
      "grad_norm": 0.14280495047569275,
      "learning_rate": 9.707585298787109e-05,
      "loss": 0.1558,
      "step": 946
    },
    {
      "epoch": 0.1987825356842989,
      "grad_norm": 0.14329122006893158,
      "learning_rate": 9.70634936285979e-05,
      "loss": 0.153,
      "step": 947
    },
    {
      "epoch": 0.19899244332493704,
      "grad_norm": 0.18129687011241913,
      "learning_rate": 9.705110899515359e-05,
      "loss": 0.1592,
      "step": 948
    },
    {
      "epoch": 0.19920235096557515,
      "grad_norm": 0.13191339373588562,
      "learning_rate": 9.7038699094189e-05,
      "loss": 0.1565,
      "step": 949
    },
    {
      "epoch": 0.19941225860621326,
      "grad_norm": 0.1381015032529831,
      "learning_rate": 9.702626393236849e-05,
      "loss": 0.1704,
      "step": 950
    },
    {
      "epoch": 0.1996221662468514,
      "grad_norm": 0.12249460071325302,
      "learning_rate": 9.701380351637007e-05,
      "loss": 0.1534,
      "step": 951
    },
    {
      "epoch": 0.1998320738874895,
      "grad_norm": 0.1820680946111679,
      "learning_rate": 9.700131785288525e-05,
      "loss": 0.1805,
      "step": 952
    },
    {
      "epoch": 0.20004198152812763,
      "grad_norm": 0.13920508325099945,
      "learning_rate": 9.698880694861913e-05,
      "loss": 0.1652,
      "step": 953
    },
    {
      "epoch": 0.20025188916876574,
      "grad_norm": 0.14306975901126862,
      "learning_rate": 9.697627081029033e-05,
      "loss": 0.1397,
      "step": 954
    },
    {
      "epoch": 0.20046179680940387,
      "grad_norm": 0.13642224669456482,
      "learning_rate": 9.696370944463104e-05,
      "loss": 0.174,
      "step": 955
    },
    {
      "epoch": 0.20067170445004198,
      "grad_norm": 0.12083397060632706,
      "learning_rate": 9.695112285838704e-05,
      "loss": 0.1807,
      "step": 956
    },
    {
      "epoch": 0.2008816120906801,
      "grad_norm": 0.17191384732723236,
      "learning_rate": 9.693851105831757e-05,
      "loss": 0.1656,
      "step": 957
    },
    {
      "epoch": 0.20109151973131822,
      "grad_norm": 0.1415136307477951,
      "learning_rate": 9.692587405119549e-05,
      "loss": 0.1726,
      "step": 958
    },
    {
      "epoch": 0.20130142737195633,
      "grad_norm": 0.15210871398448944,
      "learning_rate": 9.691321184380713e-05,
      "loss": 0.1614,
      "step": 959
    },
    {
      "epoch": 0.20151133501259447,
      "grad_norm": 0.14295780658721924,
      "learning_rate": 9.690052444295239e-05,
      "loss": 0.1884,
      "step": 960
    },
    {
      "epoch": 0.20172124265323257,
      "grad_norm": 0.13754574954509735,
      "learning_rate": 9.688781185544471e-05,
      "loss": 0.1719,
      "step": 961
    },
    {
      "epoch": 0.2019311502938707,
      "grad_norm": 0.1425933688879013,
      "learning_rate": 9.687507408811104e-05,
      "loss": 0.1564,
      "step": 962
    },
    {
      "epoch": 0.20214105793450882,
      "grad_norm": 0.1202697604894638,
      "learning_rate": 9.686231114779184e-05,
      "loss": 0.1584,
      "step": 963
    },
    {
      "epoch": 0.20235096557514692,
      "grad_norm": 0.12710309028625488,
      "learning_rate": 9.684952304134111e-05,
      "loss": 0.168,
      "step": 964
    },
    {
      "epoch": 0.20256087321578506,
      "grad_norm": 0.14089150726795197,
      "learning_rate": 9.683670977562633e-05,
      "loss": 0.1852,
      "step": 965
    },
    {
      "epoch": 0.20277078085642317,
      "grad_norm": 0.13602103292942047,
      "learning_rate": 9.682387135752856e-05,
      "loss": 0.1673,
      "step": 966
    },
    {
      "epoch": 0.2029806884970613,
      "grad_norm": 0.13352209329605103,
      "learning_rate": 9.68110077939423e-05,
      "loss": 0.1571,
      "step": 967
    },
    {
      "epoch": 0.2031905961376994,
      "grad_norm": 0.13534000515937805,
      "learning_rate": 9.679811909177556e-05,
      "loss": 0.1706,
      "step": 968
    },
    {
      "epoch": 0.20340050377833754,
      "grad_norm": 0.12475798279047012,
      "learning_rate": 9.67852052579499e-05,
      "loss": 0.1532,
      "step": 969
    },
    {
      "epoch": 0.20361041141897565,
      "grad_norm": 0.1589164137840271,
      "learning_rate": 9.677226629940032e-05,
      "loss": 0.1597,
      "step": 970
    },
    {
      "epoch": 0.20382031905961376,
      "grad_norm": 0.14976951479911804,
      "learning_rate": 9.675930222307537e-05,
      "loss": 0.1731,
      "step": 971
    },
    {
      "epoch": 0.2040302267002519,
      "grad_norm": 0.12664476037025452,
      "learning_rate": 9.6746313035937e-05,
      "loss": 0.1363,
      "step": 972
    },
    {
      "epoch": 0.20424013434089,
      "grad_norm": 0.12275015562772751,
      "learning_rate": 9.673329874496075e-05,
      "loss": 0.1574,
      "step": 973
    },
    {
      "epoch": 0.20445004198152814,
      "grad_norm": 0.13222196698188782,
      "learning_rate": 9.672025935713556e-05,
      "loss": 0.1535,
      "step": 974
    },
    {
      "epoch": 0.20465994962216624,
      "grad_norm": 0.15173639357089996,
      "learning_rate": 9.670719487946389e-05,
      "loss": 0.1792,
      "step": 975
    },
    {
      "epoch": 0.20486985726280438,
      "grad_norm": 0.17748403549194336,
      "learning_rate": 9.669410531896167e-05,
      "loss": 0.1891,
      "step": 976
    },
    {
      "epoch": 0.20507976490344249,
      "grad_norm": 0.14893555641174316,
      "learning_rate": 9.66809906826583e-05,
      "loss": 0.1438,
      "step": 977
    },
    {
      "epoch": 0.2052896725440806,
      "grad_norm": 0.16388140618801117,
      "learning_rate": 9.66678509775966e-05,
      "loss": 0.1641,
      "step": 978
    },
    {
      "epoch": 0.20549958018471873,
      "grad_norm": 0.17024089395999908,
      "learning_rate": 9.665468621083293e-05,
      "loss": 0.1839,
      "step": 979
    },
    {
      "epoch": 0.20570948782535683,
      "grad_norm": 0.14790277183055878,
      "learning_rate": 9.664149638943707e-05,
      "loss": 0.1701,
      "step": 980
    },
    {
      "epoch": 0.20591939546599497,
      "grad_norm": 0.14114373922348022,
      "learning_rate": 9.662828152049223e-05,
      "loss": 0.1699,
      "step": 981
    },
    {
      "epoch": 0.20612930310663308,
      "grad_norm": 0.12556609511375427,
      "learning_rate": 9.661504161109513e-05,
      "loss": 0.1607,
      "step": 982
    },
    {
      "epoch": 0.2063392107472712,
      "grad_norm": 0.11868204176425934,
      "learning_rate": 9.660177666835585e-05,
      "loss": 0.1487,
      "step": 983
    },
    {
      "epoch": 0.20654911838790932,
      "grad_norm": 0.12117776274681091,
      "learning_rate": 9.658848669939805e-05,
      "loss": 0.1668,
      "step": 984
    },
    {
      "epoch": 0.20675902602854743,
      "grad_norm": 0.126139298081398,
      "learning_rate": 9.657517171135866e-05,
      "loss": 0.1535,
      "step": 985
    },
    {
      "epoch": 0.20696893366918556,
      "grad_norm": 0.15042434632778168,
      "learning_rate": 9.656183171138818e-05,
      "loss": 0.168,
      "step": 986
    },
    {
      "epoch": 0.20717884130982367,
      "grad_norm": 0.1438342183828354,
      "learning_rate": 9.65484667066505e-05,
      "loss": 0.1628,
      "step": 987
    },
    {
      "epoch": 0.2073887489504618,
      "grad_norm": 0.1280035823583603,
      "learning_rate": 9.65350767043229e-05,
      "loss": 0.1655,
      "step": 988
    },
    {
      "epoch": 0.2075986565910999,
      "grad_norm": 0.13997893035411835,
      "learning_rate": 9.652166171159614e-05,
      "loss": 0.1766,
      "step": 989
    },
    {
      "epoch": 0.20780856423173805,
      "grad_norm": 0.1653011292219162,
      "learning_rate": 9.650822173567438e-05,
      "loss": 0.1744,
      "step": 990
    },
    {
      "epoch": 0.20801847187237615,
      "grad_norm": 0.1532358080148697,
      "learning_rate": 9.64947567837752e-05,
      "loss": 0.1529,
      "step": 991
    },
    {
      "epoch": 0.20822837951301426,
      "grad_norm": 0.13623711466789246,
      "learning_rate": 9.648126686312955e-05,
      "loss": 0.1632,
      "step": 992
    },
    {
      "epoch": 0.2084382871536524,
      "grad_norm": 0.13220278918743134,
      "learning_rate": 9.646775198098186e-05,
      "loss": 0.1462,
      "step": 993
    },
    {
      "epoch": 0.2086481947942905,
      "grad_norm": 0.16712597012519836,
      "learning_rate": 9.645421214458992e-05,
      "loss": 0.1658,
      "step": 994
    },
    {
      "epoch": 0.20885810243492864,
      "grad_norm": 0.16033975780010223,
      "learning_rate": 9.644064736122493e-05,
      "loss": 0.1523,
      "step": 995
    },
    {
      "epoch": 0.20906801007556675,
      "grad_norm": 0.13283276557922363,
      "learning_rate": 9.64270576381715e-05,
      "loss": 0.1738,
      "step": 996
    },
    {
      "epoch": 0.20927791771620488,
      "grad_norm": 0.2072938233613968,
      "learning_rate": 9.64134429827276e-05,
      "loss": 0.1589,
      "step": 997
    },
    {
      "epoch": 0.209487825356843,
      "grad_norm": 0.15322737395763397,
      "learning_rate": 9.639980340220462e-05,
      "loss": 0.1772,
      "step": 998
    },
    {
      "epoch": 0.2096977329974811,
      "grad_norm": 0.21009685099124908,
      "learning_rate": 9.638613890392734e-05,
      "loss": 0.1619,
      "step": 999
    },
    {
      "epoch": 0.20990764063811923,
      "grad_norm": 0.1521667093038559,
      "learning_rate": 9.63724494952339e-05,
      "loss": 0.1878,
      "step": 1000
    },
    {
      "epoch": 0.21011754827875734,
      "grad_norm": 0.11604970693588257,
      "learning_rate": 9.635873518347581e-05,
      "loss": 0.1621,
      "step": 1001
    },
    {
      "epoch": 0.21032745591939547,
      "grad_norm": 0.16362364590168,
      "learning_rate": 9.634499597601798e-05,
      "loss": 0.1779,
      "step": 1002
    },
    {
      "epoch": 0.21053736356003358,
      "grad_norm": 0.15217158198356628,
      "learning_rate": 9.63312318802387e-05,
      "loss": 0.1657,
      "step": 1003
    },
    {
      "epoch": 0.21074727120067172,
      "grad_norm": 0.15649117529392242,
      "learning_rate": 9.631744290352956e-05,
      "loss": 0.1817,
      "step": 1004
    },
    {
      "epoch": 0.21095717884130982,
      "grad_norm": 0.15730923414230347,
      "learning_rate": 9.63036290532956e-05,
      "loss": 0.1646,
      "step": 1005
    },
    {
      "epoch": 0.21116708648194793,
      "grad_norm": 0.1645728051662445,
      "learning_rate": 9.628979033695513e-05,
      "loss": 0.1546,
      "step": 1006
    },
    {
      "epoch": 0.21137699412258606,
      "grad_norm": 0.13902553915977478,
      "learning_rate": 9.627592676193988e-05,
      "loss": 0.157,
      "step": 1007
    },
    {
      "epoch": 0.21158690176322417,
      "grad_norm": 0.11890498548746109,
      "learning_rate": 9.626203833569491e-05,
      "loss": 0.1668,
      "step": 1008
    },
    {
      "epoch": 0.2117968094038623,
      "grad_norm": 0.14101600646972656,
      "learning_rate": 9.624812506567863e-05,
      "loss": 0.1477,
      "step": 1009
    },
    {
      "epoch": 0.21200671704450041,
      "grad_norm": 0.1529974490404129,
      "learning_rate": 9.623418695936275e-05,
      "loss": 0.1675,
      "step": 1010
    },
    {
      "epoch": 0.21221662468513855,
      "grad_norm": 0.15540415048599243,
      "learning_rate": 9.622022402423239e-05,
      "loss": 0.1797,
      "step": 1011
    },
    {
      "epoch": 0.21242653232577666,
      "grad_norm": 0.16966544091701508,
      "learning_rate": 9.620623626778593e-05,
      "loss": 0.1642,
      "step": 1012
    },
    {
      "epoch": 0.21263643996641476,
      "grad_norm": 0.1261100172996521,
      "learning_rate": 9.619222369753515e-05,
      "loss": 0.1633,
      "step": 1013
    },
    {
      "epoch": 0.2128463476070529,
      "grad_norm": 0.15353427827358246,
      "learning_rate": 9.617818632100508e-05,
      "loss": 0.1638,
      "step": 1014
    },
    {
      "epoch": 0.213056255247691,
      "grad_norm": 0.17301282286643982,
      "learning_rate": 9.616412414573416e-05,
      "loss": 0.1637,
      "step": 1015
    },
    {
      "epoch": 0.21326616288832914,
      "grad_norm": 0.14541825652122498,
      "learning_rate": 9.615003717927406e-05,
      "loss": 0.1677,
      "step": 1016
    },
    {
      "epoch": 0.21347607052896725,
      "grad_norm": 0.13302357494831085,
      "learning_rate": 9.613592542918981e-05,
      "loss": 0.1595,
      "step": 1017
    },
    {
      "epoch": 0.21368597816960538,
      "grad_norm": 0.17005443572998047,
      "learning_rate": 9.612178890305976e-05,
      "loss": 0.1799,
      "step": 1018
    },
    {
      "epoch": 0.2138958858102435,
      "grad_norm": 0.20354509353637695,
      "learning_rate": 9.610762760847553e-05,
      "loss": 0.1549,
      "step": 1019
    },
    {
      "epoch": 0.2141057934508816,
      "grad_norm": 0.16336257755756378,
      "learning_rate": 9.609344155304206e-05,
      "loss": 0.1404,
      "step": 1020
    },
    {
      "epoch": 0.21431570109151973,
      "grad_norm": 0.14682699739933014,
      "learning_rate": 9.607923074437758e-05,
      "loss": 0.1666,
      "step": 1021
    },
    {
      "epoch": 0.21452560873215784,
      "grad_norm": 0.15628460049629211,
      "learning_rate": 9.606499519011367e-05,
      "loss": 0.1723,
      "step": 1022
    },
    {
      "epoch": 0.21473551637279598,
      "grad_norm": 0.18369454145431519,
      "learning_rate": 9.605073489789506e-05,
      "loss": 0.1622,
      "step": 1023
    },
    {
      "epoch": 0.21494542401343408,
      "grad_norm": 0.14383664727210999,
      "learning_rate": 9.603644987537994e-05,
      "loss": 0.1776,
      "step": 1024
    },
    {
      "epoch": 0.21515533165407222,
      "grad_norm": 0.12850511074066162,
      "learning_rate": 9.602214013023963e-05,
      "loss": 0.1552,
      "step": 1025
    },
    {
      "epoch": 0.21536523929471033,
      "grad_norm": 0.12851618230342865,
      "learning_rate": 9.60078056701588e-05,
      "loss": 0.1634,
      "step": 1026
    },
    {
      "epoch": 0.21557514693534843,
      "grad_norm": 0.1252540647983551,
      "learning_rate": 9.599344650283544e-05,
      "loss": 0.1553,
      "step": 1027
    },
    {
      "epoch": 0.21578505457598657,
      "grad_norm": 0.14331619441509247,
      "learning_rate": 9.597906263598067e-05,
      "loss": 0.1681,
      "step": 1028
    },
    {
      "epoch": 0.21599496221662468,
      "grad_norm": 0.13295383751392365,
      "learning_rate": 9.5964654077319e-05,
      "loss": 0.1748,
      "step": 1029
    },
    {
      "epoch": 0.2162048698572628,
      "grad_norm": 0.14106275141239166,
      "learning_rate": 9.595022083458815e-05,
      "loss": 0.1642,
      "step": 1030
    },
    {
      "epoch": 0.21641477749790092,
      "grad_norm": 0.21775981783866882,
      "learning_rate": 9.59357629155391e-05,
      "loss": 0.1465,
      "step": 1031
    },
    {
      "epoch": 0.21662468513853905,
      "grad_norm": 0.19797000288963318,
      "learning_rate": 9.592128032793607e-05,
      "loss": 0.1559,
      "step": 1032
    },
    {
      "epoch": 0.21683459277917716,
      "grad_norm": 0.14408251643180847,
      "learning_rate": 9.590677307955656e-05,
      "loss": 0.1696,
      "step": 1033
    },
    {
      "epoch": 0.2170445004198153,
      "grad_norm": 0.14695291221141815,
      "learning_rate": 9.589224117819128e-05,
      "loss": 0.1415,
      "step": 1034
    },
    {
      "epoch": 0.2172544080604534,
      "grad_norm": 0.1755986511707306,
      "learning_rate": 9.58776846316442e-05,
      "loss": 0.1565,
      "step": 1035
    },
    {
      "epoch": 0.2174643157010915,
      "grad_norm": 0.2155688852071762,
      "learning_rate": 9.58631034477325e-05,
      "loss": 0.1631,
      "step": 1036
    },
    {
      "epoch": 0.21767422334172964,
      "grad_norm": 0.16168251633644104,
      "learning_rate": 9.584849763428665e-05,
      "loss": 0.174,
      "step": 1037
    },
    {
      "epoch": 0.21788413098236775,
      "grad_norm": 0.14332053065299988,
      "learning_rate": 9.583386719915026e-05,
      "loss": 0.1733,
      "step": 1038
    },
    {
      "epoch": 0.2180940386230059,
      "grad_norm": 0.1143033355474472,
      "learning_rate": 9.581921215018023e-05,
      "loss": 0.1431,
      "step": 1039
    },
    {
      "epoch": 0.218303946263644,
      "grad_norm": 0.16438762843608856,
      "learning_rate": 9.580453249524667e-05,
      "loss": 0.1731,
      "step": 1040
    },
    {
      "epoch": 0.21851385390428213,
      "grad_norm": 0.16054236888885498,
      "learning_rate": 9.578982824223285e-05,
      "loss": 0.1706,
      "step": 1041
    },
    {
      "epoch": 0.21872376154492024,
      "grad_norm": 0.14396212995052338,
      "learning_rate": 9.577509939903535e-05,
      "loss": 0.1694,
      "step": 1042
    },
    {
      "epoch": 0.21893366918555834,
      "grad_norm": 0.14557866752147675,
      "learning_rate": 9.576034597356382e-05,
      "loss": 0.1692,
      "step": 1043
    },
    {
      "epoch": 0.21914357682619648,
      "grad_norm": 0.1378258913755417,
      "learning_rate": 9.574556797374124e-05,
      "loss": 0.1616,
      "step": 1044
    },
    {
      "epoch": 0.2193534844668346,
      "grad_norm": 0.11917531490325928,
      "learning_rate": 9.573076540750371e-05,
      "loss": 0.1523,
      "step": 1045
    },
    {
      "epoch": 0.21956339210747272,
      "grad_norm": 0.12740926444530487,
      "learning_rate": 9.571593828280058e-05,
      "loss": 0.1581,
      "step": 1046
    },
    {
      "epoch": 0.21977329974811083,
      "grad_norm": 0.12773928046226501,
      "learning_rate": 9.570108660759432e-05,
      "loss": 0.1657,
      "step": 1047
    },
    {
      "epoch": 0.21998320738874896,
      "grad_norm": 0.15594597160816193,
      "learning_rate": 9.568621038986061e-05,
      "loss": 0.175,
      "step": 1048
    },
    {
      "epoch": 0.22019311502938707,
      "grad_norm": 0.1410975605249405,
      "learning_rate": 9.567130963758834e-05,
      "loss": 0.1527,
      "step": 1049
    },
    {
      "epoch": 0.22040302267002518,
      "grad_norm": 0.1446082442998886,
      "learning_rate": 9.565638435877957e-05,
      "loss": 0.1553,
      "step": 1050
    },
    {
      "epoch": 0.2206129303106633,
      "grad_norm": 0.14107242226600647,
      "learning_rate": 9.564143456144949e-05,
      "loss": 0.1628,
      "step": 1051
    },
    {
      "epoch": 0.22082283795130142,
      "grad_norm": 0.15437398850917816,
      "learning_rate": 9.562646025362649e-05,
      "loss": 0.1704,
      "step": 1052
    },
    {
      "epoch": 0.22103274559193956,
      "grad_norm": 0.13233573734760284,
      "learning_rate": 9.561146144335212e-05,
      "loss": 0.1588,
      "step": 1053
    },
    {
      "epoch": 0.22124265323257766,
      "grad_norm": 0.13578198850154877,
      "learning_rate": 9.559643813868107e-05,
      "loss": 0.1649,
      "step": 1054
    },
    {
      "epoch": 0.2214525608732158,
      "grad_norm": 0.17958106100559235,
      "learning_rate": 9.55813903476812e-05,
      "loss": 0.1707,
      "step": 1055
    },
    {
      "epoch": 0.2216624685138539,
      "grad_norm": 0.14567284286022186,
      "learning_rate": 9.556631807843352e-05,
      "loss": 0.1622,
      "step": 1056
    },
    {
      "epoch": 0.221872376154492,
      "grad_norm": 0.12384296208620071,
      "learning_rate": 9.55512213390322e-05,
      "loss": 0.1763,
      "step": 1057
    },
    {
      "epoch": 0.22208228379513015,
      "grad_norm": 0.14242222905158997,
      "learning_rate": 9.553610013758453e-05,
      "loss": 0.1695,
      "step": 1058
    },
    {
      "epoch": 0.22229219143576825,
      "grad_norm": 0.13062936067581177,
      "learning_rate": 9.552095448221093e-05,
      "loss": 0.1631,
      "step": 1059
    },
    {
      "epoch": 0.2225020990764064,
      "grad_norm": 0.158433735370636,
      "learning_rate": 9.550578438104494e-05,
      "loss": 0.1679,
      "step": 1060
    },
    {
      "epoch": 0.2227120067170445,
      "grad_norm": 0.14844875037670135,
      "learning_rate": 9.549058984223329e-05,
      "loss": 0.1606,
      "step": 1061
    },
    {
      "epoch": 0.22292191435768263,
      "grad_norm": 0.1360979825258255,
      "learning_rate": 9.547537087393579e-05,
      "loss": 0.1528,
      "step": 1062
    },
    {
      "epoch": 0.22313182199832074,
      "grad_norm": 0.1500774621963501,
      "learning_rate": 9.546012748432535e-05,
      "loss": 0.1639,
      "step": 1063
    },
    {
      "epoch": 0.22334172963895885,
      "grad_norm": 0.1147661879658699,
      "learning_rate": 9.544485968158803e-05,
      "loss": 0.1729,
      "step": 1064
    },
    {
      "epoch": 0.22355163727959698,
      "grad_norm": 0.13184726238250732,
      "learning_rate": 9.542956747392299e-05,
      "loss": 0.1733,
      "step": 1065
    },
    {
      "epoch": 0.2237615449202351,
      "grad_norm": 0.14958997070789337,
      "learning_rate": 9.541425086954251e-05,
      "loss": 0.1621,
      "step": 1066
    },
    {
      "epoch": 0.22397145256087322,
      "grad_norm": 0.13268455862998962,
      "learning_rate": 9.539890987667193e-05,
      "loss": 0.1542,
      "step": 1067
    },
    {
      "epoch": 0.22418136020151133,
      "grad_norm": 0.12983419001102448,
      "learning_rate": 9.538354450354974e-05,
      "loss": 0.1608,
      "step": 1068
    },
    {
      "epoch": 0.22439126784214947,
      "grad_norm": 0.11506141722202301,
      "learning_rate": 9.536815475842747e-05,
      "loss": 0.1521,
      "step": 1069
    },
    {
      "epoch": 0.22460117548278757,
      "grad_norm": 0.16174156963825226,
      "learning_rate": 9.535274064956976e-05,
      "loss": 0.176,
      "step": 1070
    },
    {
      "epoch": 0.22481108312342568,
      "grad_norm": 0.14904946088790894,
      "learning_rate": 9.533730218525438e-05,
      "loss": 0.1743,
      "step": 1071
    },
    {
      "epoch": 0.22502099076406382,
      "grad_norm": 0.1392981857061386,
      "learning_rate": 9.53218393737721e-05,
      "loss": 0.1619,
      "step": 1072
    },
    {
      "epoch": 0.22523089840470192,
      "grad_norm": 0.1343872994184494,
      "learning_rate": 9.530635222342682e-05,
      "loss": 0.1436,
      "step": 1073
    },
    {
      "epoch": 0.22544080604534006,
      "grad_norm": 0.13077078759670258,
      "learning_rate": 9.529084074253549e-05,
      "loss": 0.1501,
      "step": 1074
    },
    {
      "epoch": 0.22565071368597817,
      "grad_norm": 0.13800117373466492,
      "learning_rate": 9.527530493942812e-05,
      "loss": 0.1634,
      "step": 1075
    },
    {
      "epoch": 0.2258606213266163,
      "grad_norm": 0.15503515303134918,
      "learning_rate": 9.525974482244782e-05,
      "loss": 0.1682,
      "step": 1076
    },
    {
      "epoch": 0.2260705289672544,
      "grad_norm": 0.13676904141902924,
      "learning_rate": 9.524416039995068e-05,
      "loss": 0.1648,
      "step": 1077
    },
    {
      "epoch": 0.22628043660789252,
      "grad_norm": 0.1415819525718689,
      "learning_rate": 9.522855168030595e-05,
      "loss": 0.1732,
      "step": 1078
    },
    {
      "epoch": 0.22649034424853065,
      "grad_norm": 0.14641550183296204,
      "learning_rate": 9.521291867189581e-05,
      "loss": 0.1535,
      "step": 1079
    },
    {
      "epoch": 0.22670025188916876,
      "grad_norm": 0.13009630143642426,
      "learning_rate": 9.519726138311558e-05,
      "loss": 0.1624,
      "step": 1080
    },
    {
      "epoch": 0.2269101595298069,
      "grad_norm": 0.12057724595069885,
      "learning_rate": 9.51815798223736e-05,
      "loss": 0.1626,
      "step": 1081
    },
    {
      "epoch": 0.227120067170445,
      "grad_norm": 0.13820677995681763,
      "learning_rate": 9.516587399809116e-05,
      "loss": 0.1566,
      "step": 1082
    },
    {
      "epoch": 0.22732997481108314,
      "grad_norm": 0.19474509358406067,
      "learning_rate": 9.515014391870269e-05,
      "loss": 0.1591,
      "step": 1083
    },
    {
      "epoch": 0.22753988245172124,
      "grad_norm": 0.13335439562797546,
      "learning_rate": 9.513438959265561e-05,
      "loss": 0.1559,
      "step": 1084
    },
    {
      "epoch": 0.22774979009235935,
      "grad_norm": 0.16238383948802948,
      "learning_rate": 9.511861102841036e-05,
      "loss": 0.1701,
      "step": 1085
    },
    {
      "epoch": 0.22795969773299748,
      "grad_norm": 0.129498690366745,
      "learning_rate": 9.510280823444034e-05,
      "loss": 0.1542,
      "step": 1086
    },
    {
      "epoch": 0.2281696053736356,
      "grad_norm": 0.13683976233005524,
      "learning_rate": 9.508698121923204e-05,
      "loss": 0.1647,
      "step": 1087
    },
    {
      "epoch": 0.22837951301427373,
      "grad_norm": 0.15396280586719513,
      "learning_rate": 9.507112999128493e-05,
      "loss": 0.1519,
      "step": 1088
    },
    {
      "epoch": 0.22858942065491183,
      "grad_norm": 0.12210803478956223,
      "learning_rate": 9.50552545591115e-05,
      "loss": 0.1579,
      "step": 1089
    },
    {
      "epoch": 0.22879932829554997,
      "grad_norm": 0.1511714905500412,
      "learning_rate": 9.503935493123719e-05,
      "loss": 0.155,
      "step": 1090
    },
    {
      "epoch": 0.22900923593618808,
      "grad_norm": 0.14200669527053833,
      "learning_rate": 9.502343111620047e-05,
      "loss": 0.174,
      "step": 1091
    },
    {
      "epoch": 0.22921914357682618,
      "grad_norm": 0.12680545449256897,
      "learning_rate": 9.500748312255281e-05,
      "loss": 0.1844,
      "step": 1092
    },
    {
      "epoch": 0.22942905121746432,
      "grad_norm": 0.14180205762386322,
      "learning_rate": 9.499151095885861e-05,
      "loss": 0.1619,
      "step": 1093
    },
    {
      "epoch": 0.22963895885810243,
      "grad_norm": 0.12461036443710327,
      "learning_rate": 9.497551463369532e-05,
      "loss": 0.1558,
      "step": 1094
    },
    {
      "epoch": 0.22984886649874056,
      "grad_norm": 0.17455211281776428,
      "learning_rate": 9.495949415565332e-05,
      "loss": 0.172,
      "step": 1095
    },
    {
      "epoch": 0.23005877413937867,
      "grad_norm": 0.15062375366687775,
      "learning_rate": 9.494344953333599e-05,
      "loss": 0.1656,
      "step": 1096
    },
    {
      "epoch": 0.2302686817800168,
      "grad_norm": 0.14742077887058258,
      "learning_rate": 9.492738077535963e-05,
      "loss": 0.1751,
      "step": 1097
    },
    {
      "epoch": 0.2304785894206549,
      "grad_norm": 0.1343587338924408,
      "learning_rate": 9.491128789035355e-05,
      "loss": 0.16,
      "step": 1098
    },
    {
      "epoch": 0.23068849706129302,
      "grad_norm": 0.13116249442100525,
      "learning_rate": 9.489517088696e-05,
      "loss": 0.1495,
      "step": 1099
    },
    {
      "epoch": 0.23089840470193115,
      "grad_norm": 0.17240563035011292,
      "learning_rate": 9.487902977383415e-05,
      "loss": 0.171,
      "step": 1100
    },
    {
      "epoch": 0.23110831234256926,
      "grad_norm": 0.1725616306066513,
      "learning_rate": 9.486286455964417e-05,
      "loss": 0.1672,
      "step": 1101
    },
    {
      "epoch": 0.2313182199832074,
      "grad_norm": 0.16195368766784668,
      "learning_rate": 9.484667525307116e-05,
      "loss": 0.1697,
      "step": 1102
    },
    {
      "epoch": 0.2315281276238455,
      "grad_norm": 0.15474824607372284,
      "learning_rate": 9.483046186280912e-05,
      "loss": 0.1757,
      "step": 1103
    },
    {
      "epoch": 0.23173803526448364,
      "grad_norm": 0.13763071596622467,
      "learning_rate": 9.481422439756503e-05,
      "loss": 0.1545,
      "step": 1104
    },
    {
      "epoch": 0.23194794290512175,
      "grad_norm": 0.15999558568000793,
      "learning_rate": 9.479796286605876e-05,
      "loss": 0.168,
      "step": 1105
    },
    {
      "epoch": 0.23215785054575985,
      "grad_norm": 0.13550569117069244,
      "learning_rate": 9.478167727702315e-05,
      "loss": 0.1642,
      "step": 1106
    },
    {
      "epoch": 0.232367758186398,
      "grad_norm": 0.1440218985080719,
      "learning_rate": 9.47653676392039e-05,
      "loss": 0.1763,
      "step": 1107
    },
    {
      "epoch": 0.2325776658270361,
      "grad_norm": 0.13457611203193665,
      "learning_rate": 9.47490339613597e-05,
      "loss": 0.1604,
      "step": 1108
    },
    {
      "epoch": 0.23278757346767423,
      "grad_norm": 0.13920889794826508,
      "learning_rate": 9.473267625226207e-05,
      "loss": 0.158,
      "step": 1109
    },
    {
      "epoch": 0.23299748110831234,
      "grad_norm": 0.1562851071357727,
      "learning_rate": 9.47162945206955e-05,
      "loss": 0.1647,
      "step": 1110
    },
    {
      "epoch": 0.23320738874895047,
      "grad_norm": 0.13289715349674225,
      "learning_rate": 9.469988877545735e-05,
      "loss": 0.1535,
      "step": 1111
    },
    {
      "epoch": 0.23341729638958858,
      "grad_norm": 0.13001225888729095,
      "learning_rate": 9.468345902535787e-05,
      "loss": 0.1513,
      "step": 1112
    },
    {
      "epoch": 0.2336272040302267,
      "grad_norm": 0.1498699188232422,
      "learning_rate": 9.466700527922021e-05,
      "loss": 0.1711,
      "step": 1113
    },
    {
      "epoch": 0.23383711167086482,
      "grad_norm": 0.14534981548786163,
      "learning_rate": 9.465052754588042e-05,
      "loss": 0.1663,
      "step": 1114
    },
    {
      "epoch": 0.23404701931150293,
      "grad_norm": 0.18299245834350586,
      "learning_rate": 9.463402583418741e-05,
      "loss": 0.1478,
      "step": 1115
    },
    {
      "epoch": 0.23425692695214106,
      "grad_norm": 0.12957628071308136,
      "learning_rate": 9.461750015300296e-05,
      "loss": 0.1528,
      "step": 1116
    },
    {
      "epoch": 0.23446683459277917,
      "grad_norm": 0.13237446546554565,
      "learning_rate": 9.460095051120179e-05,
      "loss": 0.1749,
      "step": 1117
    },
    {
      "epoch": 0.2346767422334173,
      "grad_norm": 0.12967370450496674,
      "learning_rate": 9.458437691767138e-05,
      "loss": 0.1663,
      "step": 1118
    },
    {
      "epoch": 0.23488664987405541,
      "grad_norm": 0.14041563868522644,
      "learning_rate": 9.456777938131216e-05,
      "loss": 0.1581,
      "step": 1119
    },
    {
      "epoch": 0.23509655751469352,
      "grad_norm": 0.1613098829984665,
      "learning_rate": 9.455115791103736e-05,
      "loss": 0.1581,
      "step": 1120
    },
    {
      "epoch": 0.23530646515533166,
      "grad_norm": 0.12044283002614975,
      "learning_rate": 9.453451251577312e-05,
      "loss": 0.1667,
      "step": 1121
    },
    {
      "epoch": 0.23551637279596976,
      "grad_norm": 0.11583617329597473,
      "learning_rate": 9.451784320445838e-05,
      "loss": 0.1703,
      "step": 1122
    },
    {
      "epoch": 0.2357262804366079,
      "grad_norm": 0.13826625049114227,
      "learning_rate": 9.450114998604493e-05,
      "loss": 0.1634,
      "step": 1123
    },
    {
      "epoch": 0.235936188077246,
      "grad_norm": 0.1239803358912468,
      "learning_rate": 9.448443286949745e-05,
      "loss": 0.1593,
      "step": 1124
    },
    {
      "epoch": 0.23614609571788414,
      "grad_norm": 0.11571444571018219,
      "learning_rate": 9.446769186379338e-05,
      "loss": 0.1718,
      "step": 1125
    },
    {
      "epoch": 0.23635600335852225,
      "grad_norm": 0.13045620918273926,
      "learning_rate": 9.445092697792302e-05,
      "loss": 0.1767,
      "step": 1126
    },
    {
      "epoch": 0.23656591099916036,
      "grad_norm": 0.14445173740386963,
      "learning_rate": 9.44341382208895e-05,
      "loss": 0.1518,
      "step": 1127
    },
    {
      "epoch": 0.2367758186397985,
      "grad_norm": 0.13226744532585144,
      "learning_rate": 9.441732560170879e-05,
      "loss": 0.1575,
      "step": 1128
    },
    {
      "epoch": 0.2369857262804366,
      "grad_norm": 0.12810592353343964,
      "learning_rate": 9.440048912940961e-05,
      "loss": 0.1772,
      "step": 1129
    },
    {
      "epoch": 0.23719563392107473,
      "grad_norm": 0.12598370015621185,
      "learning_rate": 9.438362881303358e-05,
      "loss": 0.1663,
      "step": 1130
    },
    {
      "epoch": 0.23740554156171284,
      "grad_norm": 0.11832118779420853,
      "learning_rate": 9.436674466163503e-05,
      "loss": 0.1495,
      "step": 1131
    },
    {
      "epoch": 0.23761544920235098,
      "grad_norm": 0.13778379559516907,
      "learning_rate": 9.434983668428116e-05,
      "loss": 0.1637,
      "step": 1132
    },
    {
      "epoch": 0.23782535684298908,
      "grad_norm": 0.12428466975688934,
      "learning_rate": 9.433290489005193e-05,
      "loss": 0.1552,
      "step": 1133
    },
    {
      "epoch": 0.2380352644836272,
      "grad_norm": 0.11398124694824219,
      "learning_rate": 9.431594928804011e-05,
      "loss": 0.1814,
      "step": 1134
    },
    {
      "epoch": 0.23824517212426533,
      "grad_norm": 0.12184252589941025,
      "learning_rate": 9.429896988735123e-05,
      "loss": 0.1618,
      "step": 1135
    },
    {
      "epoch": 0.23845507976490343,
      "grad_norm": 0.12840035557746887,
      "learning_rate": 9.428196669710363e-05,
      "loss": 0.1858,
      "step": 1136
    },
    {
      "epoch": 0.23866498740554157,
      "grad_norm": 0.12335941940546036,
      "learning_rate": 9.42649397264284e-05,
      "loss": 0.1605,
      "step": 1137
    },
    {
      "epoch": 0.23887489504617967,
      "grad_norm": 0.13650208711624146,
      "learning_rate": 9.42478889844694e-05,
      "loss": 0.1811,
      "step": 1138
    },
    {
      "epoch": 0.2390848026868178,
      "grad_norm": 0.14578799903392792,
      "learning_rate": 9.423081448038334e-05,
      "loss": 0.1781,
      "step": 1139
    },
    {
      "epoch": 0.23929471032745592,
      "grad_norm": 0.13492166996002197,
      "learning_rate": 9.421371622333952e-05,
      "loss": 0.1644,
      "step": 1140
    },
    {
      "epoch": 0.23950461796809405,
      "grad_norm": 0.12785859405994415,
      "learning_rate": 9.419659422252017e-05,
      "loss": 0.1614,
      "step": 1141
    },
    {
      "epoch": 0.23971452560873216,
      "grad_norm": 0.1437363475561142,
      "learning_rate": 9.417944848712014e-05,
      "loss": 0.1737,
      "step": 1142
    },
    {
      "epoch": 0.23992443324937027,
      "grad_norm": 0.12084683775901794,
      "learning_rate": 9.416227902634711e-05,
      "loss": 0.16,
      "step": 1143
    },
    {
      "epoch": 0.2401343408900084,
      "grad_norm": 0.13587456941604614,
      "learning_rate": 9.414508584942147e-05,
      "loss": 0.1666,
      "step": 1144
    },
    {
      "epoch": 0.2403442485306465,
      "grad_norm": 0.1418214589357376,
      "learning_rate": 9.412786896557638e-05,
      "loss": 0.1526,
      "step": 1145
    },
    {
      "epoch": 0.24055415617128464,
      "grad_norm": 0.13639256358146667,
      "learning_rate": 9.411062838405762e-05,
      "loss": 0.1739,
      "step": 1146
    },
    {
      "epoch": 0.24076406381192275,
      "grad_norm": 0.1474551260471344,
      "learning_rate": 9.409336411412386e-05,
      "loss": 0.1626,
      "step": 1147
    },
    {
      "epoch": 0.2409739714525609,
      "grad_norm": 0.14461228251457214,
      "learning_rate": 9.407607616504635e-05,
      "loss": 0.1693,
      "step": 1148
    },
    {
      "epoch": 0.241183879093199,
      "grad_norm": 0.15202075242996216,
      "learning_rate": 9.405876454610915e-05,
      "loss": 0.1607,
      "step": 1149
    },
    {
      "epoch": 0.2413937867338371,
      "grad_norm": 0.14336565136909485,
      "learning_rate": 9.404142926660897e-05,
      "loss": 0.1363,
      "step": 1150
    },
    {
      "epoch": 0.24160369437447524,
      "grad_norm": 0.15390141308307648,
      "learning_rate": 9.402407033585527e-05,
      "loss": 0.1666,
      "step": 1151
    },
    {
      "epoch": 0.24181360201511334,
      "grad_norm": 0.16369935870170593,
      "learning_rate": 9.400668776317017e-05,
      "loss": 0.1584,
      "step": 1152
    },
    {
      "epoch": 0.24202350965575148,
      "grad_norm": 0.13057595491409302,
      "learning_rate": 9.398928155788853e-05,
      "loss": 0.1638,
      "step": 1153
    },
    {
      "epoch": 0.24223341729638959,
      "grad_norm": 0.13949787616729736,
      "learning_rate": 9.397185172935787e-05,
      "loss": 0.1558,
      "step": 1154
    },
    {
      "epoch": 0.24244332493702772,
      "grad_norm": 0.1358940303325653,
      "learning_rate": 9.395439828693841e-05,
      "loss": 0.1482,
      "step": 1155
    },
    {
      "epoch": 0.24265323257766583,
      "grad_norm": 0.1706434041261673,
      "learning_rate": 9.393692124000305e-05,
      "loss": 0.1663,
      "step": 1156
    },
    {
      "epoch": 0.24286314021830394,
      "grad_norm": 0.16532990336418152,
      "learning_rate": 9.391942059793734e-05,
      "loss": 0.169,
      "step": 1157
    },
    {
      "epoch": 0.24307304785894207,
      "grad_norm": 0.13570407032966614,
      "learning_rate": 9.390189637013958e-05,
      "loss": 0.1546,
      "step": 1158
    },
    {
      "epoch": 0.24328295549958018,
      "grad_norm": 0.12366579473018646,
      "learning_rate": 9.388434856602061e-05,
      "loss": 0.1517,
      "step": 1159
    },
    {
      "epoch": 0.2434928631402183,
      "grad_norm": 0.13216842710971832,
      "learning_rate": 9.386677719500406e-05,
      "loss": 0.1765,
      "step": 1160
    },
    {
      "epoch": 0.24370277078085642,
      "grad_norm": 0.15686704218387604,
      "learning_rate": 9.38491822665261e-05,
      "loss": 0.1783,
      "step": 1161
    },
    {
      "epoch": 0.24391267842149456,
      "grad_norm": 0.12082766741514206,
      "learning_rate": 9.383156379003566e-05,
      "loss": 0.1667,
      "step": 1162
    },
    {
      "epoch": 0.24412258606213266,
      "grad_norm": 0.12626822292804718,
      "learning_rate": 9.381392177499426e-05,
      "loss": 0.1589,
      "step": 1163
    },
    {
      "epoch": 0.24433249370277077,
      "grad_norm": 0.13412562012672424,
      "learning_rate": 9.379625623087603e-05,
      "loss": 0.1548,
      "step": 1164
    },
    {
      "epoch": 0.2445424013434089,
      "grad_norm": 0.14654812216758728,
      "learning_rate": 9.37785671671678e-05,
      "loss": 0.1521,
      "step": 1165
    },
    {
      "epoch": 0.244752308984047,
      "grad_norm": 0.14660955965518951,
      "learning_rate": 9.376085459336898e-05,
      "loss": 0.173,
      "step": 1166
    },
    {
      "epoch": 0.24496221662468515,
      "grad_norm": 0.1366211324930191,
      "learning_rate": 9.374311851899166e-05,
      "loss": 0.1795,
      "step": 1167
    },
    {
      "epoch": 0.24517212426532325,
      "grad_norm": 0.1468217670917511,
      "learning_rate": 9.37253589535605e-05,
      "loss": 0.1606,
      "step": 1168
    },
    {
      "epoch": 0.2453820319059614,
      "grad_norm": 0.13741639256477356,
      "learning_rate": 9.370757590661277e-05,
      "loss": 0.1653,
      "step": 1169
    },
    {
      "epoch": 0.2455919395465995,
      "grad_norm": 0.14487077295780182,
      "learning_rate": 9.368976938769839e-05,
      "loss": 0.1638,
      "step": 1170
    },
    {
      "epoch": 0.2458018471872376,
      "grad_norm": 0.14959833025932312,
      "learning_rate": 9.367193940637988e-05,
      "loss": 0.1839,
      "step": 1171
    },
    {
      "epoch": 0.24601175482787574,
      "grad_norm": 0.14737145602703094,
      "learning_rate": 9.365408597223233e-05,
      "loss": 0.1737,
      "step": 1172
    },
    {
      "epoch": 0.24622166246851385,
      "grad_norm": 0.1258353590965271,
      "learning_rate": 9.363620909484345e-05,
      "loss": 0.1574,
      "step": 1173
    },
    {
      "epoch": 0.24643157010915198,
      "grad_norm": 0.12692254781723022,
      "learning_rate": 9.361830878381352e-05,
      "loss": 0.1676,
      "step": 1174
    },
    {
      "epoch": 0.2466414777497901,
      "grad_norm": 0.13535474240779877,
      "learning_rate": 9.360038504875545e-05,
      "loss": 0.162,
      "step": 1175
    },
    {
      "epoch": 0.24685138539042822,
      "grad_norm": 0.1385834813117981,
      "learning_rate": 9.358243789929465e-05,
      "loss": 0.1611,
      "step": 1176
    },
    {
      "epoch": 0.24706129303106633,
      "grad_norm": 0.15537822246551514,
      "learning_rate": 9.356446734506918e-05,
      "loss": 0.1561,
      "step": 1177
    },
    {
      "epoch": 0.24727120067170444,
      "grad_norm": 0.13240060210227966,
      "learning_rate": 9.354647339572961e-05,
      "loss": 0.1566,
      "step": 1178
    },
    {
      "epoch": 0.24748110831234257,
      "grad_norm": 0.12547266483306885,
      "learning_rate": 9.352845606093914e-05,
      "loss": 0.1743,
      "step": 1179
    },
    {
      "epoch": 0.24769101595298068,
      "grad_norm": 0.132894828915596,
      "learning_rate": 9.351041535037347e-05,
      "loss": 0.1733,
      "step": 1180
    },
    {
      "epoch": 0.24790092359361882,
      "grad_norm": 0.1465143859386444,
      "learning_rate": 9.349235127372086e-05,
      "loss": 0.1776,
      "step": 1181
    },
    {
      "epoch": 0.24811083123425692,
      "grad_norm": 0.15798325836658478,
      "learning_rate": 9.347426384068216e-05,
      "loss": 0.1811,
      "step": 1182
    },
    {
      "epoch": 0.24832073887489506,
      "grad_norm": 0.11733090877532959,
      "learning_rate": 9.345615306097071e-05,
      "loss": 0.158,
      "step": 1183
    },
    {
      "epoch": 0.24853064651553317,
      "grad_norm": 0.11072125285863876,
      "learning_rate": 9.343801894431244e-05,
      "loss": 0.1688,
      "step": 1184
    },
    {
      "epoch": 0.24874055415617127,
      "grad_norm": 0.11508002877235413,
      "learning_rate": 9.341986150044574e-05,
      "loss": 0.1538,
      "step": 1185
    },
    {
      "epoch": 0.2489504617968094,
      "grad_norm": 0.13166356086730957,
      "learning_rate": 9.340168073912162e-05,
      "loss": 0.1489,
      "step": 1186
    },
    {
      "epoch": 0.24916036943744752,
      "grad_norm": 0.14776529371738434,
      "learning_rate": 9.338347667010352e-05,
      "loss": 0.1459,
      "step": 1187
    },
    {
      "epoch": 0.24937027707808565,
      "grad_norm": 0.13042989373207092,
      "learning_rate": 9.336524930316749e-05,
      "loss": 0.1595,
      "step": 1188
    },
    {
      "epoch": 0.24958018471872376,
      "grad_norm": 0.11727176606655121,
      "learning_rate": 9.334699864810201e-05,
      "loss": 0.1525,
      "step": 1189
    },
    {
      "epoch": 0.2497900923593619,
      "grad_norm": 0.14481998980045319,
      "learning_rate": 9.332872471470811e-05,
      "loss": 0.1627,
      "step": 1190
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13064126670360565,
      "learning_rate": 9.331042751279929e-05,
      "loss": 0.1711,
      "step": 1191
    },
    {
      "epoch": 0.2502099076406381,
      "grad_norm": 0.1243947371840477,
      "learning_rate": 9.329210705220159e-05,
      "loss": 0.1592,
      "step": 1192
    },
    {
      "epoch": 0.2504198152812762,
      "grad_norm": 0.10476188361644745,
      "learning_rate": 9.327376334275351e-05,
      "loss": 0.1548,
      "step": 1193
    },
    {
      "epoch": 0.2506297229219144,
      "grad_norm": 0.1116849035024643,
      "learning_rate": 9.325539639430602e-05,
      "loss": 0.1621,
      "step": 1194
    },
    {
      "epoch": 0.2508396305625525,
      "grad_norm": 0.134584441781044,
      "learning_rate": 9.323700621672263e-05,
      "loss": 0.1657,
      "step": 1195
    },
    {
      "epoch": 0.2510495382031906,
      "grad_norm": 0.14008194208145142,
      "learning_rate": 9.321859281987926e-05,
      "loss": 0.1569,
      "step": 1196
    },
    {
      "epoch": 0.2512594458438287,
      "grad_norm": 0.136927530169487,
      "learning_rate": 9.320015621366432e-05,
      "loss": 0.1612,
      "step": 1197
    },
    {
      "epoch": 0.25146935348446686,
      "grad_norm": 0.12949542701244354,
      "learning_rate": 9.31816964079787e-05,
      "loss": 0.1612,
      "step": 1198
    },
    {
      "epoch": 0.25167926112510497,
      "grad_norm": 0.131092369556427,
      "learning_rate": 9.316321341273574e-05,
      "loss": 0.1663,
      "step": 1199
    },
    {
      "epoch": 0.2518891687657431,
      "grad_norm": 0.15064486861228943,
      "learning_rate": 9.314470723786121e-05,
      "loss": 0.1626,
      "step": 1200
    },
    {
      "epoch": 0.2520990764063812,
      "grad_norm": 0.14182250201702118,
      "learning_rate": 9.312617789329339e-05,
      "loss": 0.1775,
      "step": 1201
    },
    {
      "epoch": 0.2523089840470193,
      "grad_norm": 0.14778292179107666,
      "learning_rate": 9.310762538898292e-05,
      "loss": 0.1699,
      "step": 1202
    },
    {
      "epoch": 0.25251889168765745,
      "grad_norm": 0.13292577862739563,
      "learning_rate": 9.308904973489292e-05,
      "loss": 0.1598,
      "step": 1203
    },
    {
      "epoch": 0.25272879932829556,
      "grad_norm": 0.14361746609210968,
      "learning_rate": 9.307045094099898e-05,
      "loss": 0.1701,
      "step": 1204
    },
    {
      "epoch": 0.25293870696893367,
      "grad_norm": 0.18234290182590485,
      "learning_rate": 9.305182901728904e-05,
      "loss": 0.1454,
      "step": 1205
    },
    {
      "epoch": 0.2531486146095718,
      "grad_norm": 0.13576486706733704,
      "learning_rate": 9.30331839737635e-05,
      "loss": 0.1751,
      "step": 1206
    },
    {
      "epoch": 0.2533585222502099,
      "grad_norm": 0.14371106028556824,
      "learning_rate": 9.30145158204352e-05,
      "loss": 0.1625,
      "step": 1207
    },
    {
      "epoch": 0.25356842989084805,
      "grad_norm": 0.14439460635185242,
      "learning_rate": 9.299582456732933e-05,
      "loss": 0.1687,
      "step": 1208
    },
    {
      "epoch": 0.25377833753148615,
      "grad_norm": 0.12489113956689835,
      "learning_rate": 9.297711022448357e-05,
      "loss": 0.1578,
      "step": 1209
    },
    {
      "epoch": 0.25398824517212426,
      "grad_norm": 0.11665599048137665,
      "learning_rate": 9.29583728019479e-05,
      "loss": 0.1585,
      "step": 1210
    },
    {
      "epoch": 0.25419815281276237,
      "grad_norm": 0.1369842141866684,
      "learning_rate": 9.293961230978476e-05,
      "loss": 0.1685,
      "step": 1211
    },
    {
      "epoch": 0.25440806045340053,
      "grad_norm": 0.1267922818660736,
      "learning_rate": 9.292082875806895e-05,
      "loss": 0.1512,
      "step": 1212
    },
    {
      "epoch": 0.25461796809403864,
      "grad_norm": 0.13632832467556,
      "learning_rate": 9.29020221568877e-05,
      "loss": 0.1588,
      "step": 1213
    },
    {
      "epoch": 0.25482787573467675,
      "grad_norm": 0.13703875243663788,
      "learning_rate": 9.288319251634057e-05,
      "loss": 0.1653,
      "step": 1214
    },
    {
      "epoch": 0.25503778337531485,
      "grad_norm": 0.1690177023410797,
      "learning_rate": 9.28643398465395e-05,
      "loss": 0.1452,
      "step": 1215
    },
    {
      "epoch": 0.25524769101595296,
      "grad_norm": 0.13149164617061615,
      "learning_rate": 9.28454641576088e-05,
      "loss": 0.1585,
      "step": 1216
    },
    {
      "epoch": 0.2554575986565911,
      "grad_norm": 0.11899828165769577,
      "learning_rate": 9.282656545968516e-05,
      "loss": 0.1666,
      "step": 1217
    },
    {
      "epoch": 0.25566750629722923,
      "grad_norm": 0.1439366489648819,
      "learning_rate": 9.28076437629176e-05,
      "loss": 0.17,
      "step": 1218
    },
    {
      "epoch": 0.25587741393786734,
      "grad_norm": 0.1383795440196991,
      "learning_rate": 9.278869907746751e-05,
      "loss": 0.1609,
      "step": 1219
    },
    {
      "epoch": 0.25608732157850544,
      "grad_norm": 0.13621094822883606,
      "learning_rate": 9.276973141350862e-05,
      "loss": 0.1657,
      "step": 1220
    },
    {
      "epoch": 0.25629722921914355,
      "grad_norm": 0.10558905452489853,
      "learning_rate": 9.2750740781227e-05,
      "loss": 0.1548,
      "step": 1221
    },
    {
      "epoch": 0.2565071368597817,
      "grad_norm": 0.1380019336938858,
      "learning_rate": 9.273172719082105e-05,
      "loss": 0.1647,
      "step": 1222
    },
    {
      "epoch": 0.2567170445004198,
      "grad_norm": 0.1091807410120964,
      "learning_rate": 9.27126906525015e-05,
      "loss": 0.1598,
      "step": 1223
    },
    {
      "epoch": 0.25692695214105793,
      "grad_norm": 0.13246595859527588,
      "learning_rate": 9.269363117649139e-05,
      "loss": 0.1684,
      "step": 1224
    },
    {
      "epoch": 0.25713685978169604,
      "grad_norm": 0.1521276980638504,
      "learning_rate": 9.267454877302614e-05,
      "loss": 0.1745,
      "step": 1225
    },
    {
      "epoch": 0.2573467674223342,
      "grad_norm": 0.13144977390766144,
      "learning_rate": 9.26554434523534e-05,
      "loss": 0.1652,
      "step": 1226
    },
    {
      "epoch": 0.2575566750629723,
      "grad_norm": 0.14672890305519104,
      "learning_rate": 9.263631522473317e-05,
      "loss": 0.1563,
      "step": 1227
    },
    {
      "epoch": 0.2577665827036104,
      "grad_norm": 0.1289316564798355,
      "learning_rate": 9.261716410043774e-05,
      "loss": 0.1594,
      "step": 1228
    },
    {
      "epoch": 0.2579764903442485,
      "grad_norm": 0.13553071022033691,
      "learning_rate": 9.259799008975173e-05,
      "loss": 0.1457,
      "step": 1229
    },
    {
      "epoch": 0.25818639798488663,
      "grad_norm": 0.11398561298847198,
      "learning_rate": 9.257879320297198e-05,
      "loss": 0.1583,
      "step": 1230
    },
    {
      "epoch": 0.2583963056255248,
      "grad_norm": 0.13140664994716644,
      "learning_rate": 9.255957345040769e-05,
      "loss": 0.1667,
      "step": 1231
    },
    {
      "epoch": 0.2586062132661629,
      "grad_norm": 0.11999331414699554,
      "learning_rate": 9.25403308423803e-05,
      "loss": 0.1656,
      "step": 1232
    },
    {
      "epoch": 0.258816120906801,
      "grad_norm": 0.14149609208106995,
      "learning_rate": 9.25210653892235e-05,
      "loss": 0.1605,
      "step": 1233
    },
    {
      "epoch": 0.2590260285474391,
      "grad_norm": 0.1261209398508072,
      "learning_rate": 9.250177710128332e-05,
      "loss": 0.1594,
      "step": 1234
    },
    {
      "epoch": 0.2592359361880772,
      "grad_norm": 0.15169532597064972,
      "learning_rate": 9.248246598891798e-05,
      "loss": 0.1744,
      "step": 1235
    },
    {
      "epoch": 0.2594458438287154,
      "grad_norm": 0.12228883057832718,
      "learning_rate": 9.246313206249802e-05,
      "loss": 0.1753,
      "step": 1236
    },
    {
      "epoch": 0.2596557514693535,
      "grad_norm": 0.14006085693836212,
      "learning_rate": 9.244377533240618e-05,
      "loss": 0.1614,
      "step": 1237
    },
    {
      "epoch": 0.2598656591099916,
      "grad_norm": 0.11327286064624786,
      "learning_rate": 9.242439580903745e-05,
      "loss": 0.1447,
      "step": 1238
    },
    {
      "epoch": 0.2600755667506297,
      "grad_norm": 0.11881330609321594,
      "learning_rate": 9.24049935027991e-05,
      "loss": 0.153,
      "step": 1239
    },
    {
      "epoch": 0.26028547439126787,
      "grad_norm": 0.11056238412857056,
      "learning_rate": 9.238556842411061e-05,
      "loss": 0.1607,
      "step": 1240
    },
    {
      "epoch": 0.260495382031906,
      "grad_norm": 0.1527404934167862,
      "learning_rate": 9.23661205834037e-05,
      "loss": 0.182,
      "step": 1241
    },
    {
      "epoch": 0.2607052896725441,
      "grad_norm": 0.11118711531162262,
      "learning_rate": 9.234664999112229e-05,
      "loss": 0.1756,
      "step": 1242
    },
    {
      "epoch": 0.2609151973131822,
      "grad_norm": 0.1669243574142456,
      "learning_rate": 9.232715665772252e-05,
      "loss": 0.1475,
      "step": 1243
    },
    {
      "epoch": 0.2611251049538203,
      "grad_norm": 0.14391769468784332,
      "learning_rate": 9.230764059367276e-05,
      "loss": 0.1807,
      "step": 1244
    },
    {
      "epoch": 0.26133501259445846,
      "grad_norm": 0.11989136785268784,
      "learning_rate": 9.228810180945362e-05,
      "loss": 0.1689,
      "step": 1245
    },
    {
      "epoch": 0.26154492023509657,
      "grad_norm": 0.1314937323331833,
      "learning_rate": 9.226854031555783e-05,
      "loss": 0.1672,
      "step": 1246
    },
    {
      "epoch": 0.2617548278757347,
      "grad_norm": 0.13055700063705444,
      "learning_rate": 9.224895612249039e-05,
      "loss": 0.1807,
      "step": 1247
    },
    {
      "epoch": 0.2619647355163728,
      "grad_norm": 0.1459030658006668,
      "learning_rate": 9.222934924076843e-05,
      "loss": 0.1669,
      "step": 1248
    },
    {
      "epoch": 0.2621746431570109,
      "grad_norm": 0.14526896178722382,
      "learning_rate": 9.22097196809213e-05,
      "loss": 0.1546,
      "step": 1249
    },
    {
      "epoch": 0.26238455079764905,
      "grad_norm": 0.12496902793645859,
      "learning_rate": 9.219006745349053e-05,
      "loss": 0.1733,
      "step": 1250
    },
    {
      "epoch": 0.26259445843828716,
      "grad_norm": 0.12503020465373993,
      "learning_rate": 9.217039256902983e-05,
      "loss": 0.1697,
      "step": 1251
    },
    {
      "epoch": 0.26280436607892527,
      "grad_norm": 0.1265639215707779,
      "learning_rate": 9.215069503810505e-05,
      "loss": 0.1688,
      "step": 1252
    },
    {
      "epoch": 0.2630142737195634,
      "grad_norm": 0.1370651125907898,
      "learning_rate": 9.213097487129418e-05,
      "loss": 0.138,
      "step": 1253
    },
    {
      "epoch": 0.26322418136020154,
      "grad_norm": 0.1354241967201233,
      "learning_rate": 9.211123207918746e-05,
      "loss": 0.1563,
      "step": 1254
    },
    {
      "epoch": 0.26343408900083964,
      "grad_norm": 0.13176575303077698,
      "learning_rate": 9.209146667238719e-05,
      "loss": 0.1683,
      "step": 1255
    },
    {
      "epoch": 0.26364399664147775,
      "grad_norm": 0.1484147608280182,
      "learning_rate": 9.207167866150783e-05,
      "loss": 0.1617,
      "step": 1256
    },
    {
      "epoch": 0.26385390428211586,
      "grad_norm": 0.1321289986371994,
      "learning_rate": 9.205186805717603e-05,
      "loss": 0.1592,
      "step": 1257
    },
    {
      "epoch": 0.26406381192275397,
      "grad_norm": 0.13387440145015717,
      "learning_rate": 9.203203487003049e-05,
      "loss": 0.1758,
      "step": 1258
    },
    {
      "epoch": 0.26427371956339213,
      "grad_norm": 0.15501335263252258,
      "learning_rate": 9.201217911072213e-05,
      "loss": 0.1748,
      "step": 1259
    },
    {
      "epoch": 0.26448362720403024,
      "grad_norm": 0.13924849033355713,
      "learning_rate": 9.199230078991393e-05,
      "loss": 0.1526,
      "step": 1260
    },
    {
      "epoch": 0.26469353484466834,
      "grad_norm": 0.155021071434021,
      "learning_rate": 9.197239991828099e-05,
      "loss": 0.1566,
      "step": 1261
    },
    {
      "epoch": 0.26490344248530645,
      "grad_norm": 0.10661531239748001,
      "learning_rate": 9.195247650651055e-05,
      "loss": 0.159,
      "step": 1262
    },
    {
      "epoch": 0.26511335012594456,
      "grad_norm": 0.15488192439079285,
      "learning_rate": 9.193253056530194e-05,
      "loss": 0.1772,
      "step": 1263
    },
    {
      "epoch": 0.2653232577665827,
      "grad_norm": 0.11265954375267029,
      "learning_rate": 9.191256210536655e-05,
      "loss": 0.1549,
      "step": 1264
    },
    {
      "epoch": 0.26553316540722083,
      "grad_norm": 0.1617230325937271,
      "learning_rate": 9.189257113742793e-05,
      "loss": 0.1737,
      "step": 1265
    },
    {
      "epoch": 0.26574307304785894,
      "grad_norm": 0.11783073842525482,
      "learning_rate": 9.187255767222169e-05,
      "loss": 0.1615,
      "step": 1266
    },
    {
      "epoch": 0.26595298068849704,
      "grad_norm": 0.14353324472904205,
      "learning_rate": 9.185252172049553e-05,
      "loss": 0.1578,
      "step": 1267
    },
    {
      "epoch": 0.2661628883291352,
      "grad_norm": 0.1482020765542984,
      "learning_rate": 9.183246329300917e-05,
      "loss": 0.1483,
      "step": 1268
    },
    {
      "epoch": 0.2663727959697733,
      "grad_norm": 0.1308319866657257,
      "learning_rate": 9.181238240053446e-05,
      "loss": 0.1742,
      "step": 1269
    },
    {
      "epoch": 0.2665827036104114,
      "grad_norm": 0.13996942341327667,
      "learning_rate": 9.179227905385532e-05,
      "loss": 0.1649,
      "step": 1270
    },
    {
      "epoch": 0.2667926112510495,
      "grad_norm": 0.13277165591716766,
      "learning_rate": 9.177215326376769e-05,
      "loss": 0.1732,
      "step": 1271
    },
    {
      "epoch": 0.26700251889168763,
      "grad_norm": 0.12486343085765839,
      "learning_rate": 9.175200504107957e-05,
      "loss": 0.1493,
      "step": 1272
    },
    {
      "epoch": 0.2672124265323258,
      "grad_norm": 0.12765567004680634,
      "learning_rate": 9.173183439661103e-05,
      "loss": 0.1533,
      "step": 1273
    },
    {
      "epoch": 0.2674223341729639,
      "grad_norm": 0.11738407611846924,
      "learning_rate": 9.171164134119417e-05,
      "loss": 0.1551,
      "step": 1274
    },
    {
      "epoch": 0.267632241813602,
      "grad_norm": 0.12661415338516235,
      "learning_rate": 9.16914258856731e-05,
      "loss": 0.1655,
      "step": 1275
    },
    {
      "epoch": 0.2678421494542401,
      "grad_norm": 0.10266046971082687,
      "learning_rate": 9.167118804090401e-05,
      "loss": 0.1506,
      "step": 1276
    },
    {
      "epoch": 0.2680520570948782,
      "grad_norm": 0.13607458770275116,
      "learning_rate": 9.165092781775506e-05,
      "loss": 0.1621,
      "step": 1277
    },
    {
      "epoch": 0.2682619647355164,
      "grad_norm": 0.14382155239582062,
      "learning_rate": 9.163064522710649e-05,
      "loss": 0.1756,
      "step": 1278
    },
    {
      "epoch": 0.2684718723761545,
      "grad_norm": 0.14696674048900604,
      "learning_rate": 9.161034027985047e-05,
      "loss": 0.1506,
      "step": 1279
    },
    {
      "epoch": 0.2686817800167926,
      "grad_norm": 0.12927404046058655,
      "learning_rate": 9.159001298689122e-05,
      "loss": 0.1693,
      "step": 1280
    },
    {
      "epoch": 0.2688916876574307,
      "grad_norm": 0.1403432935476303,
      "learning_rate": 9.1569663359145e-05,
      "loss": 0.178,
      "step": 1281
    },
    {
      "epoch": 0.2691015952980689,
      "grad_norm": 0.12026988714933395,
      "learning_rate": 9.154929140754e-05,
      "loss": 0.1611,
      "step": 1282
    },
    {
      "epoch": 0.269311502938707,
      "grad_norm": 0.1373865157365799,
      "learning_rate": 9.152889714301645e-05,
      "loss": 0.1696,
      "step": 1283
    },
    {
      "epoch": 0.2695214105793451,
      "grad_norm": 0.13119877874851227,
      "learning_rate": 9.150848057652648e-05,
      "loss": 0.178,
      "step": 1284
    },
    {
      "epoch": 0.2697313182199832,
      "grad_norm": 0.12198714911937714,
      "learning_rate": 9.148804171903432e-05,
      "loss": 0.1579,
      "step": 1285
    },
    {
      "epoch": 0.2699412258606213,
      "grad_norm": 0.12499682605266571,
      "learning_rate": 9.146758058151603e-05,
      "loss": 0.1669,
      "step": 1286
    },
    {
      "epoch": 0.27015113350125947,
      "grad_norm": 0.12025036662817001,
      "learning_rate": 9.144709717495978e-05,
      "loss": 0.1602,
      "step": 1287
    },
    {
      "epoch": 0.2703610411418976,
      "grad_norm": 0.1325342059135437,
      "learning_rate": 9.142659151036558e-05,
      "loss": 0.1559,
      "step": 1288
    },
    {
      "epoch": 0.2705709487825357,
      "grad_norm": 0.14255130290985107,
      "learning_rate": 9.140606359874546e-05,
      "loss": 0.1441,
      "step": 1289
    },
    {
      "epoch": 0.2707808564231738,
      "grad_norm": 0.13800451159477234,
      "learning_rate": 9.138551345112337e-05,
      "loss": 0.1543,
      "step": 1290
    },
    {
      "epoch": 0.2709907640638119,
      "grad_norm": 0.10976230353116989,
      "learning_rate": 9.136494107853521e-05,
      "loss": 0.1513,
      "step": 1291
    },
    {
      "epoch": 0.27120067170445006,
      "grad_norm": 0.1330142766237259,
      "learning_rate": 9.134434649202882e-05,
      "loss": 0.1527,
      "step": 1292
    },
    {
      "epoch": 0.27141057934508817,
      "grad_norm": 0.13712665438652039,
      "learning_rate": 9.132372970266397e-05,
      "loss": 0.1621,
      "step": 1293
    },
    {
      "epoch": 0.2716204869857263,
      "grad_norm": 0.15420737862586975,
      "learning_rate": 9.130309072151231e-05,
      "loss": 0.1773,
      "step": 1294
    },
    {
      "epoch": 0.2718303946263644,
      "grad_norm": 0.14155341684818268,
      "learning_rate": 9.12824295596575e-05,
      "loss": 0.1615,
      "step": 1295
    },
    {
      "epoch": 0.27204030226700254,
      "grad_norm": 0.14036308228969574,
      "learning_rate": 9.126174622819499e-05,
      "loss": 0.1496,
      "step": 1296
    },
    {
      "epoch": 0.27225020990764065,
      "grad_norm": 0.13568180799484253,
      "learning_rate": 9.124104073823225e-05,
      "loss": 0.1562,
      "step": 1297
    },
    {
      "epoch": 0.27246011754827876,
      "grad_norm": 0.12056776136159897,
      "learning_rate": 9.122031310088861e-05,
      "loss": 0.1722,
      "step": 1298
    },
    {
      "epoch": 0.27267002518891686,
      "grad_norm": 0.139260396361351,
      "learning_rate": 9.119956332729524e-05,
      "loss": 0.1797,
      "step": 1299
    },
    {
      "epoch": 0.27287993282955497,
      "grad_norm": 0.16409984230995178,
      "learning_rate": 9.117879142859527e-05,
      "loss": 0.1661,
      "step": 1300
    },
    {
      "epoch": 0.27308984047019313,
      "grad_norm": 0.1330333650112152,
      "learning_rate": 9.115799741594368e-05,
      "loss": 0.1794,
      "step": 1301
    },
    {
      "epoch": 0.27329974811083124,
      "grad_norm": 0.12173225730657578,
      "learning_rate": 9.113718130050734e-05,
      "loss": 0.1449,
      "step": 1302
    },
    {
      "epoch": 0.27350965575146935,
      "grad_norm": 0.10996419191360474,
      "learning_rate": 9.111634309346496e-05,
      "loss": 0.1626,
      "step": 1303
    },
    {
      "epoch": 0.27371956339210746,
      "grad_norm": 0.13602428138256073,
      "learning_rate": 9.109548280600715e-05,
      "loss": 0.1769,
      "step": 1304
    },
    {
      "epoch": 0.2739294710327456,
      "grad_norm": 0.11914105713367462,
      "learning_rate": 9.107460044933635e-05,
      "loss": 0.1664,
      "step": 1305
    },
    {
      "epoch": 0.2741393786733837,
      "grad_norm": 0.12417447566986084,
      "learning_rate": 9.105369603466686e-05,
      "loss": 0.1455,
      "step": 1306
    },
    {
      "epoch": 0.27434928631402183,
      "grad_norm": 0.13093994557857513,
      "learning_rate": 9.103276957322484e-05,
      "loss": 0.1721,
      "step": 1307
    },
    {
      "epoch": 0.27455919395465994,
      "grad_norm": 0.13235576450824738,
      "learning_rate": 9.101182107624827e-05,
      "loss": 0.1614,
      "step": 1308
    },
    {
      "epoch": 0.27476910159529805,
      "grad_norm": 0.11165154725313187,
      "learning_rate": 9.099085055498696e-05,
      "loss": 0.1656,
      "step": 1309
    },
    {
      "epoch": 0.2749790092359362,
      "grad_norm": 0.10962437093257904,
      "learning_rate": 9.096985802070259e-05,
      "loss": 0.1633,
      "step": 1310
    },
    {
      "epoch": 0.2751889168765743,
      "grad_norm": 0.12218458205461502,
      "learning_rate": 9.09488434846686e-05,
      "loss": 0.1542,
      "step": 1311
    },
    {
      "epoch": 0.2753988245172124,
      "grad_norm": 0.11284459382295609,
      "learning_rate": 9.09278069581703e-05,
      "loss": 0.1591,
      "step": 1312
    },
    {
      "epoch": 0.27560873215785053,
      "grad_norm": 0.10450571775436401,
      "learning_rate": 9.090674845250475e-05,
      "loss": 0.1571,
      "step": 1313
    },
    {
      "epoch": 0.27581863979848864,
      "grad_norm": 0.12627463042736053,
      "learning_rate": 9.088566797898089e-05,
      "loss": 0.1613,
      "step": 1314
    },
    {
      "epoch": 0.2760285474391268,
      "grad_norm": 0.1578662246465683,
      "learning_rate": 9.08645655489194e-05,
      "loss": 0.1506,
      "step": 1315
    },
    {
      "epoch": 0.2762384550797649,
      "grad_norm": 0.11859230697154999,
      "learning_rate": 9.084344117365276e-05,
      "loss": 0.1502,
      "step": 1316
    },
    {
      "epoch": 0.276448362720403,
      "grad_norm": 0.11622706055641174,
      "learning_rate": 9.082229486452524e-05,
      "loss": 0.1674,
      "step": 1317
    },
    {
      "epoch": 0.2766582703610411,
      "grad_norm": 0.12626247107982635,
      "learning_rate": 9.08011266328929e-05,
      "loss": 0.1522,
      "step": 1318
    },
    {
      "epoch": 0.2768681780016793,
      "grad_norm": 0.13337282836437225,
      "learning_rate": 9.077993649012357e-05,
      "loss": 0.1647,
      "step": 1319
    },
    {
      "epoch": 0.2770780856423174,
      "grad_norm": 0.1255621761083603,
      "learning_rate": 9.075872444759683e-05,
      "loss": 0.1564,
      "step": 1320
    },
    {
      "epoch": 0.2772879932829555,
      "grad_norm": 0.12364380061626434,
      "learning_rate": 9.073749051670403e-05,
      "loss": 0.1555,
      "step": 1321
    },
    {
      "epoch": 0.2774979009235936,
      "grad_norm": 0.11966480314731598,
      "learning_rate": 9.07162347088483e-05,
      "loss": 0.163,
      "step": 1322
    },
    {
      "epoch": 0.2777078085642317,
      "grad_norm": 0.10513235628604889,
      "learning_rate": 9.069495703544443e-05,
      "loss": 0.1578,
      "step": 1323
    },
    {
      "epoch": 0.2779177162048699,
      "grad_norm": 0.10422056168317795,
      "learning_rate": 9.06736575079191e-05,
      "loss": 0.1435,
      "step": 1324
    },
    {
      "epoch": 0.278127623845508,
      "grad_norm": 0.12099135667085648,
      "learning_rate": 9.065233613771059e-05,
      "loss": 0.1606,
      "step": 1325
    },
    {
      "epoch": 0.2783375314861461,
      "grad_norm": 0.10527081042528152,
      "learning_rate": 9.063099293626898e-05,
      "loss": 0.1606,
      "step": 1326
    },
    {
      "epoch": 0.2785474391267842,
      "grad_norm": 0.15932826697826385,
      "learning_rate": 9.060962791505605e-05,
      "loss": 0.1626,
      "step": 1327
    },
    {
      "epoch": 0.2787573467674223,
      "grad_norm": 0.14808166027069092,
      "learning_rate": 9.05882410855453e-05,
      "loss": 0.1648,
      "step": 1328
    },
    {
      "epoch": 0.27896725440806047,
      "grad_norm": 0.1467159539461136,
      "learning_rate": 9.056683245922196e-05,
      "loss": 0.1533,
      "step": 1329
    },
    {
      "epoch": 0.2791771620486986,
      "grad_norm": 0.13326837122440338,
      "learning_rate": 9.054540204758295e-05,
      "loss": 0.1638,
      "step": 1330
    },
    {
      "epoch": 0.2793870696893367,
      "grad_norm": 0.14456981420516968,
      "learning_rate": 9.052394986213688e-05,
      "loss": 0.1639,
      "step": 1331
    },
    {
      "epoch": 0.2795969773299748,
      "grad_norm": 0.1200864389538765,
      "learning_rate": 9.050247591440407e-05,
      "loss": 0.1556,
      "step": 1332
    },
    {
      "epoch": 0.27980688497061296,
      "grad_norm": 0.1340358853340149,
      "learning_rate": 9.04809802159165e-05,
      "loss": 0.1627,
      "step": 1333
    },
    {
      "epoch": 0.28001679261125106,
      "grad_norm": 0.14233215153217316,
      "learning_rate": 9.045946277821791e-05,
      "loss": 0.1604,
      "step": 1334
    },
    {
      "epoch": 0.28022670025188917,
      "grad_norm": 0.15358757972717285,
      "learning_rate": 9.04379236128636e-05,
      "loss": 0.1709,
      "step": 1335
    },
    {
      "epoch": 0.2804366078925273,
      "grad_norm": 0.14173360168933868,
      "learning_rate": 9.041636273142061e-05,
      "loss": 0.1659,
      "step": 1336
    },
    {
      "epoch": 0.2806465155331654,
      "grad_norm": 0.13441811501979828,
      "learning_rate": 9.039478014546761e-05,
      "loss": 0.171,
      "step": 1337
    },
    {
      "epoch": 0.28085642317380355,
      "grad_norm": 0.12063397467136383,
      "learning_rate": 9.037317586659498e-05,
      "loss": 0.1514,
      "step": 1338
    },
    {
      "epoch": 0.28106633081444166,
      "grad_norm": 0.15047864615917206,
      "learning_rate": 9.035154990640466e-05,
      "loss": 0.179,
      "step": 1339
    },
    {
      "epoch": 0.28127623845507976,
      "grad_norm": 0.1325361728668213,
      "learning_rate": 9.032990227651034e-05,
      "loss": 0.1659,
      "step": 1340
    },
    {
      "epoch": 0.28148614609571787,
      "grad_norm": 0.14775165915489197,
      "learning_rate": 9.030823298853725e-05,
      "loss": 0.1732,
      "step": 1341
    },
    {
      "epoch": 0.281696053736356,
      "grad_norm": 0.1677199900150299,
      "learning_rate": 9.028654205412233e-05,
      "loss": 0.1696,
      "step": 1342
    },
    {
      "epoch": 0.28190596137699414,
      "grad_norm": 0.1279178410768509,
      "learning_rate": 9.026482948491407e-05,
      "loss": 0.1529,
      "step": 1343
    },
    {
      "epoch": 0.28211586901763225,
      "grad_norm": 0.12682883441448212,
      "learning_rate": 9.024309529257262e-05,
      "loss": 0.1513,
      "step": 1344
    },
    {
      "epoch": 0.28232577665827036,
      "grad_norm": 0.10392288863658905,
      "learning_rate": 9.022133948876975e-05,
      "loss": 0.1611,
      "step": 1345
    },
    {
      "epoch": 0.28253568429890846,
      "grad_norm": 0.11686202883720398,
      "learning_rate": 9.019956208518883e-05,
      "loss": 0.1672,
      "step": 1346
    },
    {
      "epoch": 0.2827455919395466,
      "grad_norm": 0.12934426963329315,
      "learning_rate": 9.017776309352481e-05,
      "loss": 0.1433,
      "step": 1347
    },
    {
      "epoch": 0.28295549958018473,
      "grad_norm": 0.11639254540205002,
      "learning_rate": 9.015594252548426e-05,
      "loss": 0.1707,
      "step": 1348
    },
    {
      "epoch": 0.28316540722082284,
      "grad_norm": 0.12937045097351074,
      "learning_rate": 9.013410039278531e-05,
      "loss": 0.1553,
      "step": 1349
    },
    {
      "epoch": 0.28337531486146095,
      "grad_norm": 0.1247095987200737,
      "learning_rate": 9.01122367071577e-05,
      "loss": 0.1679,
      "step": 1350
    },
    {
      "epoch": 0.28358522250209905,
      "grad_norm": 0.12007705122232437,
      "learning_rate": 9.00903514803427e-05,
      "loss": 0.1554,
      "step": 1351
    },
    {
      "epoch": 0.2837951301427372,
      "grad_norm": 0.17036741971969604,
      "learning_rate": 9.00684447240932e-05,
      "loss": 0.1698,
      "step": 1352
    },
    {
      "epoch": 0.2840050377833753,
      "grad_norm": 0.11295659095048904,
      "learning_rate": 9.004651645017363e-05,
      "loss": 0.1635,
      "step": 1353
    },
    {
      "epoch": 0.28421494542401343,
      "grad_norm": 0.1332911103963852,
      "learning_rate": 9.002456667035997e-05,
      "loss": 0.169,
      "step": 1354
    },
    {
      "epoch": 0.28442485306465154,
      "grad_norm": 0.1429576724767685,
      "learning_rate": 9.000259539643972e-05,
      "loss": 0.1641,
      "step": 1355
    },
    {
      "epoch": 0.28463476070528965,
      "grad_norm": 0.14206334948539734,
      "learning_rate": 8.998060264021201e-05,
      "loss": 0.1692,
      "step": 1356
    },
    {
      "epoch": 0.2848446683459278,
      "grad_norm": 0.11948953568935394,
      "learning_rate": 8.995858841348743e-05,
      "loss": 0.1594,
      "step": 1357
    },
    {
      "epoch": 0.2850545759865659,
      "grad_norm": 0.12463600933551788,
      "learning_rate": 8.993655272808811e-05,
      "loss": 0.1657,
      "step": 1358
    },
    {
      "epoch": 0.285264483627204,
      "grad_norm": 0.15918885171413422,
      "learning_rate": 8.991449559584775e-05,
      "loss": 0.1532,
      "step": 1359
    },
    {
      "epoch": 0.28547439126784213,
      "grad_norm": 0.12887489795684814,
      "learning_rate": 8.989241702861149e-05,
      "loss": 0.1724,
      "step": 1360
    },
    {
      "epoch": 0.2856842989084803,
      "grad_norm": 0.1456436812877655,
      "learning_rate": 8.987031703823606e-05,
      "loss": 0.1529,
      "step": 1361
    },
    {
      "epoch": 0.2858942065491184,
      "grad_norm": 0.16147381067276,
      "learning_rate": 8.984819563658964e-05,
      "loss": 0.1655,
      "step": 1362
    },
    {
      "epoch": 0.2861041141897565,
      "grad_norm": 0.18096451461315155,
      "learning_rate": 8.982605283555192e-05,
      "loss": 0.152,
      "step": 1363
    },
    {
      "epoch": 0.2863140218303946,
      "grad_norm": 0.12937211990356445,
      "learning_rate": 8.980388864701412e-05,
      "loss": 0.1627,
      "step": 1364
    },
    {
      "epoch": 0.2865239294710327,
      "grad_norm": 0.16723109781742096,
      "learning_rate": 8.97817030828789e-05,
      "loss": 0.1666,
      "step": 1365
    },
    {
      "epoch": 0.2867338371116709,
      "grad_norm": 0.1125638410449028,
      "learning_rate": 8.97594961550604e-05,
      "loss": 0.1514,
      "step": 1366
    },
    {
      "epoch": 0.286943744752309,
      "grad_norm": 0.12200607359409332,
      "learning_rate": 8.973726787548427e-05,
      "loss": 0.1593,
      "step": 1367
    },
    {
      "epoch": 0.2871536523929471,
      "grad_norm": 0.12197429686784744,
      "learning_rate": 8.971501825608762e-05,
      "loss": 0.1662,
      "step": 1368
    },
    {
      "epoch": 0.2873635600335852,
      "grad_norm": 0.11950329691171646,
      "learning_rate": 8.969274730881893e-05,
      "loss": 0.1619,
      "step": 1369
    },
    {
      "epoch": 0.2875734676742233,
      "grad_norm": 0.13871550559997559,
      "learning_rate": 8.96704550456383e-05,
      "loss": 0.1467,
      "step": 1370
    },
    {
      "epoch": 0.2877833753148615,
      "grad_norm": 0.11093387007713318,
      "learning_rate": 8.964814147851713e-05,
      "loss": 0.1703,
      "step": 1371
    },
    {
      "epoch": 0.2879932829554996,
      "grad_norm": 0.13531048595905304,
      "learning_rate": 8.962580661943831e-05,
      "loss": 0.1587,
      "step": 1372
    },
    {
      "epoch": 0.2882031905961377,
      "grad_norm": 0.11192765831947327,
      "learning_rate": 8.96034504803962e-05,
      "loss": 0.1537,
      "step": 1373
    },
    {
      "epoch": 0.2884130982367758,
      "grad_norm": 0.12293984740972519,
      "learning_rate": 8.958107307339653e-05,
      "loss": 0.1644,
      "step": 1374
    },
    {
      "epoch": 0.28862300587741396,
      "grad_norm": 0.12963594496250153,
      "learning_rate": 8.955867441045652e-05,
      "loss": 0.16,
      "step": 1375
    },
    {
      "epoch": 0.28883291351805207,
      "grad_norm": 0.11687605082988739,
      "learning_rate": 8.95362545036047e-05,
      "loss": 0.157,
      "step": 1376
    },
    {
      "epoch": 0.2890428211586902,
      "grad_norm": 0.11696937680244446,
      "learning_rate": 8.951381336488114e-05,
      "loss": 0.1572,
      "step": 1377
    },
    {
      "epoch": 0.2892527287993283,
      "grad_norm": 0.12374909222126007,
      "learning_rate": 8.949135100633718e-05,
      "loss": 0.1584,
      "step": 1378
    },
    {
      "epoch": 0.2894626364399664,
      "grad_norm": 0.133868008852005,
      "learning_rate": 8.946886744003569e-05,
      "loss": 0.1415,
      "step": 1379
    },
    {
      "epoch": 0.28967254408060455,
      "grad_norm": 0.11039866507053375,
      "learning_rate": 8.94463626780508e-05,
      "loss": 0.1783,
      "step": 1380
    },
    {
      "epoch": 0.28988245172124266,
      "grad_norm": 0.1570889800786972,
      "learning_rate": 8.94238367324681e-05,
      "loss": 0.1707,
      "step": 1381
    },
    {
      "epoch": 0.29009235936188077,
      "grad_norm": 0.1414021998643875,
      "learning_rate": 8.940128961538454e-05,
      "loss": 0.1339,
      "step": 1382
    },
    {
      "epoch": 0.2903022670025189,
      "grad_norm": 0.12162122875452042,
      "learning_rate": 8.937872133890843e-05,
      "loss": 0.1595,
      "step": 1383
    },
    {
      "epoch": 0.290512174643157,
      "grad_norm": 0.1278192549943924,
      "learning_rate": 8.935613191515946e-05,
      "loss": 0.1386,
      "step": 1384
    },
    {
      "epoch": 0.29072208228379515,
      "grad_norm": 0.12220022082328796,
      "learning_rate": 8.933352135626867e-05,
      "loss": 0.1663,
      "step": 1385
    },
    {
      "epoch": 0.29093198992443325,
      "grad_norm": 0.15517336130142212,
      "learning_rate": 8.931088967437844e-05,
      "loss": 0.1596,
      "step": 1386
    },
    {
      "epoch": 0.29114189756507136,
      "grad_norm": 0.13512621819972992,
      "learning_rate": 8.928823688164249e-05,
      "loss": 0.1412,
      "step": 1387
    },
    {
      "epoch": 0.29135180520570947,
      "grad_norm": 0.10588439553976059,
      "learning_rate": 8.92655629902259e-05,
      "loss": 0.1502,
      "step": 1388
    },
    {
      "epoch": 0.29156171284634763,
      "grad_norm": 0.16245464980602264,
      "learning_rate": 8.924286801230506e-05,
      "loss": 0.1669,
      "step": 1389
    },
    {
      "epoch": 0.29177162048698574,
      "grad_norm": 0.12498240917921066,
      "learning_rate": 8.92201519600677e-05,
      "loss": 0.1518,
      "step": 1390
    },
    {
      "epoch": 0.29198152812762385,
      "grad_norm": 0.13893572986125946,
      "learning_rate": 8.919741484571286e-05,
      "loss": 0.18,
      "step": 1391
    },
    {
      "epoch": 0.29219143576826195,
      "grad_norm": 0.1361815184354782,
      "learning_rate": 8.917465668145086e-05,
      "loss": 0.166,
      "step": 1392
    },
    {
      "epoch": 0.29240134340890006,
      "grad_norm": 0.12212132662534714,
      "learning_rate": 8.915187747950339e-05,
      "loss": 0.1606,
      "step": 1393
    },
    {
      "epoch": 0.2926112510495382,
      "grad_norm": 0.12465030699968338,
      "learning_rate": 8.912907725210342e-05,
      "loss": 0.1454,
      "step": 1394
    },
    {
      "epoch": 0.29282115869017633,
      "grad_norm": 0.15186457335948944,
      "learning_rate": 8.910625601149512e-05,
      "loss": 0.1581,
      "step": 1395
    },
    {
      "epoch": 0.29303106633081444,
      "grad_norm": 0.1369972676038742,
      "learning_rate": 8.908341376993409e-05,
      "loss": 0.1636,
      "step": 1396
    },
    {
      "epoch": 0.29324097397145255,
      "grad_norm": 0.12631604075431824,
      "learning_rate": 8.90605505396871e-05,
      "loss": 0.1536,
      "step": 1397
    },
    {
      "epoch": 0.29345088161209065,
      "grad_norm": 0.13023649156093597,
      "learning_rate": 8.903766633303222e-05,
      "loss": 0.1615,
      "step": 1398
    },
    {
      "epoch": 0.2936607892527288,
      "grad_norm": 0.13841018080711365,
      "learning_rate": 8.90147611622588e-05,
      "loss": 0.1756,
      "step": 1399
    },
    {
      "epoch": 0.2938706968933669,
      "grad_norm": 0.14210903644561768,
      "learning_rate": 8.899183503966746e-05,
      "loss": 0.1661,
      "step": 1400
    },
    {
      "epoch": 0.29408060453400503,
      "grad_norm": 0.13981324434280396,
      "learning_rate": 8.896888797757003e-05,
      "loss": 0.1547,
      "step": 1401
    },
    {
      "epoch": 0.29429051217464314,
      "grad_norm": 0.11732388287782669,
      "learning_rate": 8.89459199882896e-05,
      "loss": 0.1648,
      "step": 1402
    },
    {
      "epoch": 0.2945004198152813,
      "grad_norm": 0.13775314390659332,
      "learning_rate": 8.892293108416055e-05,
      "loss": 0.171,
      "step": 1403
    },
    {
      "epoch": 0.2947103274559194,
      "grad_norm": 0.11232218891382217,
      "learning_rate": 8.889992127752839e-05,
      "loss": 0.1813,
      "step": 1404
    },
    {
      "epoch": 0.2949202350965575,
      "grad_norm": 0.12403729557991028,
      "learning_rate": 8.887689058074994e-05,
      "loss": 0.1694,
      "step": 1405
    },
    {
      "epoch": 0.2951301427371956,
      "grad_norm": 0.12742987275123596,
      "learning_rate": 8.885383900619321e-05,
      "loss": 0.1584,
      "step": 1406
    },
    {
      "epoch": 0.29534005037783373,
      "grad_norm": 0.12318763881921768,
      "learning_rate": 8.883076656623741e-05,
      "loss": 0.1484,
      "step": 1407
    },
    {
      "epoch": 0.2955499580184719,
      "grad_norm": 0.1265798807144165,
      "learning_rate": 8.880767327327297e-05,
      "loss": 0.1708,
      "step": 1408
    },
    {
      "epoch": 0.29575986565911,
      "grad_norm": 0.12394752353429794,
      "learning_rate": 8.878455913970154e-05,
      "loss": 0.1599,
      "step": 1409
    },
    {
      "epoch": 0.2959697732997481,
      "grad_norm": 0.1458372324705124,
      "learning_rate": 8.876142417793591e-05,
      "loss": 0.1751,
      "step": 1410
    },
    {
      "epoch": 0.2961796809403862,
      "grad_norm": 0.14673961699008942,
      "learning_rate": 8.87382684004001e-05,
      "loss": 0.1651,
      "step": 1411
    },
    {
      "epoch": 0.2963895885810244,
      "grad_norm": 0.15783408284187317,
      "learning_rate": 8.871509181952925e-05,
      "loss": 0.1788,
      "step": 1412
    },
    {
      "epoch": 0.2965994962216625,
      "grad_norm": 0.1385459154844284,
      "learning_rate": 8.869189444776979e-05,
      "loss": 0.158,
      "step": 1413
    },
    {
      "epoch": 0.2968094038623006,
      "grad_norm": 0.12669824063777924,
      "learning_rate": 8.866867629757916e-05,
      "loss": 0.1556,
      "step": 1414
    },
    {
      "epoch": 0.2970193115029387,
      "grad_norm": 0.12607593834400177,
      "learning_rate": 8.86454373814261e-05,
      "loss": 0.145,
      "step": 1415
    },
    {
      "epoch": 0.2972292191435768,
      "grad_norm": 0.15276843309402466,
      "learning_rate": 8.86221777117904e-05,
      "loss": 0.1644,
      "step": 1416
    },
    {
      "epoch": 0.29743912678421497,
      "grad_norm": 0.154046893119812,
      "learning_rate": 8.859889730116304e-05,
      "loss": 0.162,
      "step": 1417
    },
    {
      "epoch": 0.2976490344248531,
      "grad_norm": 0.13313519954681396,
      "learning_rate": 8.857559616204613e-05,
      "loss": 0.1659,
      "step": 1418
    },
    {
      "epoch": 0.2978589420654912,
      "grad_norm": 0.11869316548109055,
      "learning_rate": 8.855227430695294e-05,
      "loss": 0.149,
      "step": 1419
    },
    {
      "epoch": 0.2980688497061293,
      "grad_norm": 0.13355213403701782,
      "learning_rate": 8.852893174840782e-05,
      "loss": 0.1674,
      "step": 1420
    },
    {
      "epoch": 0.2982787573467674,
      "grad_norm": 0.11961048096418381,
      "learning_rate": 8.850556849894625e-05,
      "loss": 0.1596,
      "step": 1421
    },
    {
      "epoch": 0.29848866498740556,
      "grad_norm": 0.15859396755695343,
      "learning_rate": 8.848218457111483e-05,
      "loss": 0.1688,
      "step": 1422
    },
    {
      "epoch": 0.29869857262804367,
      "grad_norm": 0.12413342297077179,
      "learning_rate": 8.845877997747127e-05,
      "loss": 0.1764,
      "step": 1423
    },
    {
      "epoch": 0.2989084802686818,
      "grad_norm": 0.1361013799905777,
      "learning_rate": 8.843535473058437e-05,
      "loss": 0.166,
      "step": 1424
    },
    {
      "epoch": 0.2991183879093199,
      "grad_norm": 0.17420732975006104,
      "learning_rate": 8.841190884303402e-05,
      "loss": 0.1699,
      "step": 1425
    },
    {
      "epoch": 0.29932829554995805,
      "grad_norm": 0.13347145915031433,
      "learning_rate": 8.838844232741122e-05,
      "loss": 0.1642,
      "step": 1426
    },
    {
      "epoch": 0.29953820319059615,
      "grad_norm": 0.1271234154701233,
      "learning_rate": 8.8364955196318e-05,
      "loss": 0.1425,
      "step": 1427
    },
    {
      "epoch": 0.29974811083123426,
      "grad_norm": 0.17551806569099426,
      "learning_rate": 8.834144746236748e-05,
      "loss": 0.1771,
      "step": 1428
    },
    {
      "epoch": 0.29995801847187237,
      "grad_norm": 0.16244301199913025,
      "learning_rate": 8.831791913818387e-05,
      "loss": 0.1598,
      "step": 1429
    },
    {
      "epoch": 0.3001679261125105,
      "grad_norm": 0.1273384541273117,
      "learning_rate": 8.829437023640239e-05,
      "loss": 0.1448,
      "step": 1430
    },
    {
      "epoch": 0.30037783375314864,
      "grad_norm": 0.10530916601419449,
      "learning_rate": 8.827080076966938e-05,
      "loss": 0.1502,
      "step": 1431
    },
    {
      "epoch": 0.30058774139378674,
      "grad_norm": 0.10858159512281418,
      "learning_rate": 8.824721075064215e-05,
      "loss": 0.1674,
      "step": 1432
    },
    {
      "epoch": 0.30079764903442485,
      "grad_norm": 0.12510398030281067,
      "learning_rate": 8.82236001919891e-05,
      "loss": 0.1739,
      "step": 1433
    },
    {
      "epoch": 0.30100755667506296,
      "grad_norm": 0.12490396946668625,
      "learning_rate": 8.819996910638962e-05,
      "loss": 0.1441,
      "step": 1434
    },
    {
      "epoch": 0.30121746431570107,
      "grad_norm": 0.13503123819828033,
      "learning_rate": 8.817631750653412e-05,
      "loss": 0.1545,
      "step": 1435
    },
    {
      "epoch": 0.30142737195633923,
      "grad_norm": 0.12778516113758087,
      "learning_rate": 8.815264540512411e-05,
      "loss": 0.1618,
      "step": 1436
    },
    {
      "epoch": 0.30163727959697734,
      "grad_norm": 0.12481486797332764,
      "learning_rate": 8.812895281487201e-05,
      "loss": 0.1541,
      "step": 1437
    },
    {
      "epoch": 0.30184718723761544,
      "grad_norm": 0.15225140750408173,
      "learning_rate": 8.810523974850131e-05,
      "loss": 0.1737,
      "step": 1438
    },
    {
      "epoch": 0.30205709487825355,
      "grad_norm": 0.12728790938854218,
      "learning_rate": 8.808150621874643e-05,
      "loss": 0.1627,
      "step": 1439
    },
    {
      "epoch": 0.3022670025188917,
      "grad_norm": 0.1369866281747818,
      "learning_rate": 8.805775223835284e-05,
      "loss": 0.1685,
      "step": 1440
    },
    {
      "epoch": 0.3024769101595298,
      "grad_norm": 0.13604697585105896,
      "learning_rate": 8.803397782007697e-05,
      "loss": 0.1623,
      "step": 1441
    },
    {
      "epoch": 0.30268681780016793,
      "grad_norm": 0.12286148965358734,
      "learning_rate": 8.801018297668622e-05,
      "loss": 0.1544,
      "step": 1442
    },
    {
      "epoch": 0.30289672544080604,
      "grad_norm": 0.12319783121347427,
      "learning_rate": 8.798636772095896e-05,
      "loss": 0.1559,
      "step": 1443
    },
    {
      "epoch": 0.30310663308144414,
      "grad_norm": 0.13532286882400513,
      "learning_rate": 8.796253206568454e-05,
      "loss": 0.1462,
      "step": 1444
    },
    {
      "epoch": 0.3033165407220823,
      "grad_norm": 0.14293217658996582,
      "learning_rate": 8.793867602366326e-05,
      "loss": 0.167,
      "step": 1445
    },
    {
      "epoch": 0.3035264483627204,
      "grad_norm": 0.1477852612733841,
      "learning_rate": 8.791479960770633e-05,
      "loss": 0.166,
      "step": 1446
    },
    {
      "epoch": 0.3037363560033585,
      "grad_norm": 0.1326388120651245,
      "learning_rate": 8.789090283063595e-05,
      "loss": 0.1737,
      "step": 1447
    },
    {
      "epoch": 0.30394626364399663,
      "grad_norm": 0.14401552081108093,
      "learning_rate": 8.786698570528522e-05,
      "loss": 0.1645,
      "step": 1448
    },
    {
      "epoch": 0.30415617128463474,
      "grad_norm": 0.13078097999095917,
      "learning_rate": 8.784304824449819e-05,
      "loss": 0.1712,
      "step": 1449
    },
    {
      "epoch": 0.3043660789252729,
      "grad_norm": 0.12756723165512085,
      "learning_rate": 8.781909046112983e-05,
      "loss": 0.1606,
      "step": 1450
    },
    {
      "epoch": 0.304575986565911,
      "grad_norm": 0.13873012363910675,
      "learning_rate": 8.779511236804603e-05,
      "loss": 0.1682,
      "step": 1451
    },
    {
      "epoch": 0.3047858942065491,
      "grad_norm": 0.12623098492622375,
      "learning_rate": 8.777111397812354e-05,
      "loss": 0.1617,
      "step": 1452
    },
    {
      "epoch": 0.3049958018471872,
      "grad_norm": 0.1419798731803894,
      "learning_rate": 8.774709530425006e-05,
      "loss": 0.1446,
      "step": 1453
    },
    {
      "epoch": 0.3052057094878254,
      "grad_norm": 0.1524171084165573,
      "learning_rate": 8.772305635932416e-05,
      "loss": 0.1606,
      "step": 1454
    },
    {
      "epoch": 0.3054156171284635,
      "grad_norm": 0.1456252783536911,
      "learning_rate": 8.769899715625533e-05,
      "loss": 0.1491,
      "step": 1455
    },
    {
      "epoch": 0.3056255247691016,
      "grad_norm": 0.16533705592155457,
      "learning_rate": 8.767491770796388e-05,
      "loss": 0.1455,
      "step": 1456
    },
    {
      "epoch": 0.3058354324097397,
      "grad_norm": 0.1010533794760704,
      "learning_rate": 8.765081802738104e-05,
      "loss": 0.1599,
      "step": 1457
    },
    {
      "epoch": 0.3060453400503778,
      "grad_norm": 0.1445004791021347,
      "learning_rate": 8.762669812744887e-05,
      "loss": 0.1788,
      "step": 1458
    },
    {
      "epoch": 0.306255247691016,
      "grad_norm": 0.1463691145181656,
      "learning_rate": 8.760255802112032e-05,
      "loss": 0.1554,
      "step": 1459
    },
    {
      "epoch": 0.3064651553316541,
      "grad_norm": 0.14523173868656158,
      "learning_rate": 8.757839772135919e-05,
      "loss": 0.1768,
      "step": 1460
    },
    {
      "epoch": 0.3066750629722922,
      "grad_norm": 0.12887312471866608,
      "learning_rate": 8.75542172411401e-05,
      "loss": 0.1385,
      "step": 1461
    },
    {
      "epoch": 0.3068849706129303,
      "grad_norm": 0.12244197726249695,
      "learning_rate": 8.753001659344852e-05,
      "loss": 0.177,
      "step": 1462
    },
    {
      "epoch": 0.3070948782535684,
      "grad_norm": 0.1390761286020279,
      "learning_rate": 8.750579579128077e-05,
      "loss": 0.152,
      "step": 1463
    },
    {
      "epoch": 0.30730478589420657,
      "grad_norm": 0.1180199533700943,
      "learning_rate": 8.748155484764393e-05,
      "loss": 0.1822,
      "step": 1464
    },
    {
      "epoch": 0.3075146935348447,
      "grad_norm": 0.12501199543476105,
      "learning_rate": 8.745729377555598e-05,
      "loss": 0.1585,
      "step": 1465
    },
    {
      "epoch": 0.3077246011754828,
      "grad_norm": 0.1500568687915802,
      "learning_rate": 8.743301258804567e-05,
      "loss": 0.1571,
      "step": 1466
    },
    {
      "epoch": 0.3079345088161209,
      "grad_norm": 0.10966819524765015,
      "learning_rate": 8.740871129815253e-05,
      "loss": 0.1654,
      "step": 1467
    },
    {
      "epoch": 0.30814441645675905,
      "grad_norm": 0.14994241297245026,
      "learning_rate": 8.73843899189269e-05,
      "loss": 0.156,
      "step": 1468
    },
    {
      "epoch": 0.30835432409739716,
      "grad_norm": 0.13417619466781616,
      "learning_rate": 8.736004846342996e-05,
      "loss": 0.16,
      "step": 1469
    },
    {
      "epoch": 0.30856423173803527,
      "grad_norm": 0.12869034707546234,
      "learning_rate": 8.733568694473359e-05,
      "loss": 0.1526,
      "step": 1470
    },
    {
      "epoch": 0.3087741393786734,
      "grad_norm": 0.14110779762268066,
      "learning_rate": 8.731130537592048e-05,
      "loss": 0.1677,
      "step": 1471
    },
    {
      "epoch": 0.3089840470193115,
      "grad_norm": 0.11759759485721588,
      "learning_rate": 8.728690377008412e-05,
      "loss": 0.1583,
      "step": 1472
    },
    {
      "epoch": 0.30919395465994964,
      "grad_norm": 0.13631226122379303,
      "learning_rate": 8.726248214032871e-05,
      "loss": 0.151,
      "step": 1473
    },
    {
      "epoch": 0.30940386230058775,
      "grad_norm": 0.13227972388267517,
      "learning_rate": 8.723804049976922e-05,
      "loss": 0.1616,
      "step": 1474
    },
    {
      "epoch": 0.30961376994122586,
      "grad_norm": 0.15408119559288025,
      "learning_rate": 8.721357886153137e-05,
      "loss": 0.1731,
      "step": 1475
    },
    {
      "epoch": 0.30982367758186397,
      "grad_norm": 0.14567722380161285,
      "learning_rate": 8.718909723875163e-05,
      "loss": 0.1555,
      "step": 1476
    },
    {
      "epoch": 0.3100335852225021,
      "grad_norm": 0.1292102187871933,
      "learning_rate": 8.71645956445772e-05,
      "loss": 0.1566,
      "step": 1477
    },
    {
      "epoch": 0.31024349286314024,
      "grad_norm": 0.12155589461326599,
      "learning_rate": 8.714007409216595e-05,
      "loss": 0.168,
      "step": 1478
    },
    {
      "epoch": 0.31045340050377834,
      "grad_norm": 0.1343718022108078,
      "learning_rate": 8.711553259468657e-05,
      "loss": 0.1673,
      "step": 1479
    },
    {
      "epoch": 0.31066330814441645,
      "grad_norm": 0.11374703049659729,
      "learning_rate": 8.709097116531835e-05,
      "loss": 0.1655,
      "step": 1480
    },
    {
      "epoch": 0.31087321578505456,
      "grad_norm": 0.13738465309143066,
      "learning_rate": 8.706638981725139e-05,
      "loss": 0.1573,
      "step": 1481
    },
    {
      "epoch": 0.3110831234256927,
      "grad_norm": 0.13506565988063812,
      "learning_rate": 8.70417885636864e-05,
      "loss": 0.1649,
      "step": 1482
    },
    {
      "epoch": 0.3112930310663308,
      "grad_norm": 0.1244024857878685,
      "learning_rate": 8.701716741783483e-05,
      "loss": 0.1605,
      "step": 1483
    },
    {
      "epoch": 0.31150293870696893,
      "grad_norm": 0.15039315819740295,
      "learning_rate": 8.69925263929188e-05,
      "loss": 0.1634,
      "step": 1484
    },
    {
      "epoch": 0.31171284634760704,
      "grad_norm": 0.12742501497268677,
      "learning_rate": 8.69678655021711e-05,
      "loss": 0.1649,
      "step": 1485
    },
    {
      "epoch": 0.31192275398824515,
      "grad_norm": 0.1437070369720459,
      "learning_rate": 8.694318475883518e-05,
      "loss": 0.1565,
      "step": 1486
    },
    {
      "epoch": 0.3121326616288833,
      "grad_norm": 0.13695649802684784,
      "learning_rate": 8.691848417616517e-05,
      "loss": 0.1628,
      "step": 1487
    },
    {
      "epoch": 0.3123425692695214,
      "grad_norm": 0.12154117226600647,
      "learning_rate": 8.689376376742586e-05,
      "loss": 0.159,
      "step": 1488
    },
    {
      "epoch": 0.3125524769101595,
      "grad_norm": 0.12522564828395844,
      "learning_rate": 8.686902354589266e-05,
      "loss": 0.1567,
      "step": 1489
    },
    {
      "epoch": 0.31276238455079763,
      "grad_norm": 0.14949142932891846,
      "learning_rate": 8.684426352485165e-05,
      "loss": 0.1602,
      "step": 1490
    },
    {
      "epoch": 0.31297229219143574,
      "grad_norm": 0.13096019625663757,
      "learning_rate": 8.68194837175995e-05,
      "loss": 0.157,
      "step": 1491
    },
    {
      "epoch": 0.3131821998320739,
      "grad_norm": 0.12589357793331146,
      "learning_rate": 8.679468413744356e-05,
      "loss": 0.1711,
      "step": 1492
    },
    {
      "epoch": 0.313392107472712,
      "grad_norm": 0.1338101178407669,
      "learning_rate": 8.676986479770175e-05,
      "loss": 0.1628,
      "step": 1493
    },
    {
      "epoch": 0.3136020151133501,
      "grad_norm": 0.13691024482250214,
      "learning_rate": 8.674502571170262e-05,
      "loss": 0.1604,
      "step": 1494
    },
    {
      "epoch": 0.3138119227539882,
      "grad_norm": 0.14150124788284302,
      "learning_rate": 8.672016689278535e-05,
      "loss": 0.1707,
      "step": 1495
    },
    {
      "epoch": 0.3140218303946264,
      "grad_norm": 0.13233508169651031,
      "learning_rate": 8.669528835429969e-05,
      "loss": 0.1421,
      "step": 1496
    },
    {
      "epoch": 0.3142317380352645,
      "grad_norm": 0.11874301731586456,
      "learning_rate": 8.667039010960596e-05,
      "loss": 0.1576,
      "step": 1497
    },
    {
      "epoch": 0.3144416456759026,
      "grad_norm": 0.10090667009353638,
      "learning_rate": 8.66454721720751e-05,
      "loss": 0.158,
      "step": 1498
    },
    {
      "epoch": 0.3146515533165407,
      "grad_norm": 0.126051664352417,
      "learning_rate": 8.662053455508862e-05,
      "loss": 0.1725,
      "step": 1499
    },
    {
      "epoch": 0.3148614609571788,
      "grad_norm": 0.11627226322889328,
      "learning_rate": 8.659557727203858e-05,
      "loss": 0.1572,
      "step": 1500
    },
    {
      "epoch": 0.315071368597817,
      "grad_norm": 0.12283121049404144,
      "learning_rate": 8.657060033632763e-05,
      "loss": 0.1633,
      "step": 1501
    },
    {
      "epoch": 0.3152812762384551,
      "grad_norm": 0.11624766141176224,
      "learning_rate": 8.654560376136891e-05,
      "loss": 0.1762,
      "step": 1502
    },
    {
      "epoch": 0.3154911838790932,
      "grad_norm": 0.13653838634490967,
      "learning_rate": 8.65205875605862e-05,
      "loss": 0.1784,
      "step": 1503
    },
    {
      "epoch": 0.3157010915197313,
      "grad_norm": 0.12123393267393112,
      "learning_rate": 8.649555174741374e-05,
      "loss": 0.1551,
      "step": 1504
    },
    {
      "epoch": 0.3159109991603694,
      "grad_norm": 0.12023040652275085,
      "learning_rate": 8.647049633529635e-05,
      "loss": 0.174,
      "step": 1505
    },
    {
      "epoch": 0.3161209068010076,
      "grad_norm": 0.12046807259321213,
      "learning_rate": 8.644542133768936e-05,
      "loss": 0.1703,
      "step": 1506
    },
    {
      "epoch": 0.3163308144416457,
      "grad_norm": 0.11224795132875443,
      "learning_rate": 8.64203267680586e-05,
      "loss": 0.1435,
      "step": 1507
    },
    {
      "epoch": 0.3165407220822838,
      "grad_norm": 0.1572083830833435,
      "learning_rate": 8.639521263988044e-05,
      "loss": 0.153,
      "step": 1508
    },
    {
      "epoch": 0.3167506297229219,
      "grad_norm": 0.1247357502579689,
      "learning_rate": 8.637007896664173e-05,
      "loss": 0.1566,
      "step": 1509
    },
    {
      "epoch": 0.31696053736356006,
      "grad_norm": 0.12798605859279633,
      "learning_rate": 8.634492576183984e-05,
      "loss": 0.1526,
      "step": 1510
    },
    {
      "epoch": 0.31717044500419816,
      "grad_norm": 0.1474895477294922,
      "learning_rate": 8.631975303898261e-05,
      "loss": 0.169,
      "step": 1511
    },
    {
      "epoch": 0.31738035264483627,
      "grad_norm": 0.11636695265769958,
      "learning_rate": 8.629456081158838e-05,
      "loss": 0.1613,
      "step": 1512
    },
    {
      "epoch": 0.3175902602854744,
      "grad_norm": 0.11031098663806915,
      "learning_rate": 8.626934909318591e-05,
      "loss": 0.1576,
      "step": 1513
    },
    {
      "epoch": 0.3178001679261125,
      "grad_norm": 0.12879802286624908,
      "learning_rate": 8.624411789731452e-05,
      "loss": 0.1556,
      "step": 1514
    },
    {
      "epoch": 0.31801007556675065,
      "grad_norm": 0.1240730881690979,
      "learning_rate": 8.621886723752392e-05,
      "loss": 0.1633,
      "step": 1515
    },
    {
      "epoch": 0.31821998320738876,
      "grad_norm": 0.10870420932769775,
      "learning_rate": 8.619359712737427e-05,
      "loss": 0.1385,
      "step": 1516
    },
    {
      "epoch": 0.31842989084802686,
      "grad_norm": 0.11011097580194473,
      "learning_rate": 8.616830758043622e-05,
      "loss": 0.1655,
      "step": 1517
    },
    {
      "epoch": 0.31863979848866497,
      "grad_norm": 0.17963182926177979,
      "learning_rate": 8.614299861029083e-05,
      "loss": 0.1651,
      "step": 1518
    },
    {
      "epoch": 0.31884970612930313,
      "grad_norm": 0.12011148035526276,
      "learning_rate": 8.61176702305296e-05,
      "loss": 0.1657,
      "step": 1519
    },
    {
      "epoch": 0.31905961376994124,
      "grad_norm": 0.1365176886320114,
      "learning_rate": 8.609232245475443e-05,
      "loss": 0.1839,
      "step": 1520
    },
    {
      "epoch": 0.31926952141057935,
      "grad_norm": 0.1179775819182396,
      "learning_rate": 8.606695529657767e-05,
      "loss": 0.1581,
      "step": 1521
    },
    {
      "epoch": 0.31947942905121746,
      "grad_norm": 0.15118494629859924,
      "learning_rate": 8.604156876962206e-05,
      "loss": 0.1831,
      "step": 1522
    },
    {
      "epoch": 0.31968933669185556,
      "grad_norm": 0.14029067754745483,
      "learning_rate": 8.601616288752073e-05,
      "loss": 0.1708,
      "step": 1523
    },
    {
      "epoch": 0.3198992443324937,
      "grad_norm": 0.14659684896469116,
      "learning_rate": 8.599073766391725e-05,
      "loss": 0.1599,
      "step": 1524
    },
    {
      "epoch": 0.32010915197313183,
      "grad_norm": 0.1260727345943451,
      "learning_rate": 8.596529311246551e-05,
      "loss": 0.1623,
      "step": 1525
    },
    {
      "epoch": 0.32031905961376994,
      "grad_norm": 0.12665653228759766,
      "learning_rate": 8.593982924682986e-05,
      "loss": 0.1672,
      "step": 1526
    },
    {
      "epoch": 0.32052896725440805,
      "grad_norm": 0.12388867139816284,
      "learning_rate": 8.591434608068493e-05,
      "loss": 0.1738,
      "step": 1527
    },
    {
      "epoch": 0.32073887489504616,
      "grad_norm": 0.12167280912399292,
      "learning_rate": 8.58888436277158e-05,
      "loss": 0.1627,
      "step": 1528
    },
    {
      "epoch": 0.3209487825356843,
      "grad_norm": 0.12876589596271515,
      "learning_rate": 8.586332190161786e-05,
      "loss": 0.1646,
      "step": 1529
    },
    {
      "epoch": 0.3211586901763224,
      "grad_norm": 0.12370327860116959,
      "learning_rate": 8.583778091609683e-05,
      "loss": 0.156,
      "step": 1530
    },
    {
      "epoch": 0.32136859781696053,
      "grad_norm": 0.11784835159778595,
      "learning_rate": 8.581222068486884e-05,
      "loss": 0.1664,
      "step": 1531
    },
    {
      "epoch": 0.32157850545759864,
      "grad_norm": 0.14363597333431244,
      "learning_rate": 8.578664122166029e-05,
      "loss": 0.1724,
      "step": 1532
    },
    {
      "epoch": 0.3217884130982368,
      "grad_norm": 0.11469676345586777,
      "learning_rate": 8.576104254020796e-05,
      "loss": 0.1469,
      "step": 1533
    },
    {
      "epoch": 0.3219983207388749,
      "grad_norm": 0.12534739077091217,
      "learning_rate": 8.573542465425892e-05,
      "loss": 0.1687,
      "step": 1534
    },
    {
      "epoch": 0.322208228379513,
      "grad_norm": 0.12772725522518158,
      "learning_rate": 8.570978757757057e-05,
      "loss": 0.159,
      "step": 1535
    },
    {
      "epoch": 0.3224181360201511,
      "grad_norm": 0.14892329275608063,
      "learning_rate": 8.568413132391058e-05,
      "loss": 0.1586,
      "step": 1536
    },
    {
      "epoch": 0.32262804366078923,
      "grad_norm": 0.1334220916032791,
      "learning_rate": 8.565845590705695e-05,
      "loss": 0.1692,
      "step": 1537
    },
    {
      "epoch": 0.3228379513014274,
      "grad_norm": 0.1165885403752327,
      "learning_rate": 8.563276134079798e-05,
      "loss": 0.1673,
      "step": 1538
    },
    {
      "epoch": 0.3230478589420655,
      "grad_norm": 0.11613230407238007,
      "learning_rate": 8.560704763893224e-05,
      "loss": 0.1512,
      "step": 1539
    },
    {
      "epoch": 0.3232577665827036,
      "grad_norm": 0.11873317509889603,
      "learning_rate": 8.558131481526856e-05,
      "loss": 0.1556,
      "step": 1540
    },
    {
      "epoch": 0.3234676742233417,
      "grad_norm": 0.12491460889577866,
      "learning_rate": 8.555556288362605e-05,
      "loss": 0.1728,
      "step": 1541
    },
    {
      "epoch": 0.3236775818639798,
      "grad_norm": 0.1342668980360031,
      "learning_rate": 8.552979185783412e-05,
      "loss": 0.1746,
      "step": 1542
    },
    {
      "epoch": 0.323887489504618,
      "grad_norm": 0.1082087978720665,
      "learning_rate": 8.550400175173236e-05,
      "loss": 0.1668,
      "step": 1543
    },
    {
      "epoch": 0.3240973971452561,
      "grad_norm": 0.12464690953493118,
      "learning_rate": 8.547819257917065e-05,
      "loss": 0.1642,
      "step": 1544
    },
    {
      "epoch": 0.3243073047858942,
      "grad_norm": 0.11872904002666473,
      "learning_rate": 8.545236435400914e-05,
      "loss": 0.1586,
      "step": 1545
    },
    {
      "epoch": 0.3245172124265323,
      "grad_norm": 0.14073589444160461,
      "learning_rate": 8.542651709011814e-05,
      "loss": 0.1733,
      "step": 1546
    },
    {
      "epoch": 0.32472712006717047,
      "grad_norm": 0.11652261763811111,
      "learning_rate": 8.540065080137824e-05,
      "loss": 0.1689,
      "step": 1547
    },
    {
      "epoch": 0.3249370277078086,
      "grad_norm": 0.14177857339382172,
      "learning_rate": 8.537476550168022e-05,
      "loss": 0.1504,
      "step": 1548
    },
    {
      "epoch": 0.3251469353484467,
      "grad_norm": 0.11179566383361816,
      "learning_rate": 8.534886120492509e-05,
      "loss": 0.1504,
      "step": 1549
    },
    {
      "epoch": 0.3253568429890848,
      "grad_norm": 0.141026109457016,
      "learning_rate": 8.532293792502403e-05,
      "loss": 0.1642,
      "step": 1550
    },
    {
      "epoch": 0.3255667506297229,
      "grad_norm": 0.11241170763969421,
      "learning_rate": 8.529699567589844e-05,
      "loss": 0.1662,
      "step": 1551
    },
    {
      "epoch": 0.32577665827036106,
      "grad_norm": 0.11532200872898102,
      "learning_rate": 8.52710344714799e-05,
      "loss": 0.1588,
      "step": 1552
    },
    {
      "epoch": 0.32598656591099917,
      "grad_norm": 0.13776087760925293,
      "learning_rate": 8.524505432571019e-05,
      "loss": 0.1654,
      "step": 1553
    },
    {
      "epoch": 0.3261964735516373,
      "grad_norm": 0.11376765370368958,
      "learning_rate": 8.521905525254123e-05,
      "loss": 0.1536,
      "step": 1554
    },
    {
      "epoch": 0.3264063811922754,
      "grad_norm": 0.11074686050415039,
      "learning_rate": 8.519303726593508e-05,
      "loss": 0.1589,
      "step": 1555
    },
    {
      "epoch": 0.3266162888329135,
      "grad_norm": 0.11213414371013641,
      "learning_rate": 8.516700037986406e-05,
      "loss": 0.1548,
      "step": 1556
    },
    {
      "epoch": 0.32682619647355166,
      "grad_norm": 0.14128370583057404,
      "learning_rate": 8.514094460831052e-05,
      "loss": 0.1556,
      "step": 1557
    },
    {
      "epoch": 0.32703610411418976,
      "grad_norm": 0.1197216585278511,
      "learning_rate": 8.5114869965267e-05,
      "loss": 0.146,
      "step": 1558
    },
    {
      "epoch": 0.32724601175482787,
      "grad_norm": 0.12456157803535461,
      "learning_rate": 8.508877646473623e-05,
      "loss": 0.1586,
      "step": 1559
    },
    {
      "epoch": 0.327455919395466,
      "grad_norm": 0.11759025603532791,
      "learning_rate": 8.506266412073099e-05,
      "loss": 0.1564,
      "step": 1560
    },
    {
      "epoch": 0.32766582703610414,
      "grad_norm": 0.1229950562119484,
      "learning_rate": 8.503653294727418e-05,
      "loss": 0.159,
      "step": 1561
    },
    {
      "epoch": 0.32787573467674225,
      "grad_norm": 0.11062034219503403,
      "learning_rate": 8.501038295839887e-05,
      "loss": 0.1536,
      "step": 1562
    },
    {
      "epoch": 0.32808564231738035,
      "grad_norm": 0.1223401203751564,
      "learning_rate": 8.498421416814818e-05,
      "loss": 0.1473,
      "step": 1563
    },
    {
      "epoch": 0.32829554995801846,
      "grad_norm": 0.13364095985889435,
      "learning_rate": 8.495802659057536e-05,
      "loss": 0.169,
      "step": 1564
    },
    {
      "epoch": 0.32850545759865657,
      "grad_norm": 0.1344279795885086,
      "learning_rate": 8.493182023974372e-05,
      "loss": 0.1537,
      "step": 1565
    },
    {
      "epoch": 0.32871536523929473,
      "grad_norm": 0.17328642308712006,
      "learning_rate": 8.490559512972671e-05,
      "loss": 0.1476,
      "step": 1566
    },
    {
      "epoch": 0.32892527287993284,
      "grad_norm": 0.1484242081642151,
      "learning_rate": 8.487935127460779e-05,
      "loss": 0.1666,
      "step": 1567
    },
    {
      "epoch": 0.32913518052057095,
      "grad_norm": 0.17023669183254242,
      "learning_rate": 8.485308868848049e-05,
      "loss": 0.1653,
      "step": 1568
    },
    {
      "epoch": 0.32934508816120905,
      "grad_norm": 0.14054429531097412,
      "learning_rate": 8.482680738544843e-05,
      "loss": 0.1675,
      "step": 1569
    },
    {
      "epoch": 0.32955499580184716,
      "grad_norm": 0.1140383705496788,
      "learning_rate": 8.480050737962531e-05,
      "loss": 0.1612,
      "step": 1570
    },
    {
      "epoch": 0.3297649034424853,
      "grad_norm": 0.13256023824214935,
      "learning_rate": 8.477418868513476e-05,
      "loss": 0.1595,
      "step": 1571
    },
    {
      "epoch": 0.32997481108312343,
      "grad_norm": 0.13051024079322815,
      "learning_rate": 8.474785131611058e-05,
      "loss": 0.1621,
      "step": 1572
    },
    {
      "epoch": 0.33018471872376154,
      "grad_norm": 0.11545020341873169,
      "learning_rate": 8.472149528669651e-05,
      "loss": 0.1683,
      "step": 1573
    },
    {
      "epoch": 0.33039462636439965,
      "grad_norm": 0.13742703199386597,
      "learning_rate": 8.469512061104635e-05,
      "loss": 0.1611,
      "step": 1574
    },
    {
      "epoch": 0.3306045340050378,
      "grad_norm": 0.13308970630168915,
      "learning_rate": 8.46687273033239e-05,
      "loss": 0.148,
      "step": 1575
    },
    {
      "epoch": 0.3308144416456759,
      "grad_norm": 0.13026942312717438,
      "learning_rate": 8.464231537770298e-05,
      "loss": 0.1772,
      "step": 1576
    },
    {
      "epoch": 0.331024349286314,
      "grad_norm": 0.11242230236530304,
      "learning_rate": 8.461588484836738e-05,
      "loss": 0.1632,
      "step": 1577
    },
    {
      "epoch": 0.33123425692695213,
      "grad_norm": 0.13627904653549194,
      "learning_rate": 8.45894357295109e-05,
      "loss": 0.1644,
      "step": 1578
    },
    {
      "epoch": 0.33144416456759024,
      "grad_norm": 0.11995008587837219,
      "learning_rate": 8.456296803533734e-05,
      "loss": 0.1532,
      "step": 1579
    },
    {
      "epoch": 0.3316540722082284,
      "grad_norm": 0.12165706604719162,
      "learning_rate": 8.453648178006044e-05,
      "loss": 0.1596,
      "step": 1580
    },
    {
      "epoch": 0.3318639798488665,
      "grad_norm": 0.1462729126214981,
      "learning_rate": 8.450997697790393e-05,
      "loss": 0.1572,
      "step": 1581
    },
    {
      "epoch": 0.3320738874895046,
      "grad_norm": 0.11989284306764603,
      "learning_rate": 8.448345364310149e-05,
      "loss": 0.1631,
      "step": 1582
    },
    {
      "epoch": 0.3322837951301427,
      "grad_norm": 0.1307172328233719,
      "learning_rate": 8.445691178989674e-05,
      "loss": 0.1639,
      "step": 1583
    },
    {
      "epoch": 0.33249370277078083,
      "grad_norm": 0.13907490670681,
      "learning_rate": 8.44303514325433e-05,
      "loss": 0.1466,
      "step": 1584
    },
    {
      "epoch": 0.332703610411419,
      "grad_norm": 0.21999351680278778,
      "learning_rate": 8.440377258530465e-05,
      "loss": 0.1596,
      "step": 1585
    },
    {
      "epoch": 0.3329135180520571,
      "grad_norm": 0.11224652826786041,
      "learning_rate": 8.437717526245428e-05,
      "loss": 0.1543,
      "step": 1586
    },
    {
      "epoch": 0.3331234256926952,
      "grad_norm": 0.12237223982810974,
      "learning_rate": 8.435055947827552e-05,
      "loss": 0.149,
      "step": 1587
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.1236414983868599,
      "learning_rate": 8.432392524706168e-05,
      "loss": 0.1421,
      "step": 1588
    },
    {
      "epoch": 0.3335432409739715,
      "grad_norm": 0.12933699786663055,
      "learning_rate": 8.429727258311593e-05,
      "loss": 0.152,
      "step": 1589
    },
    {
      "epoch": 0.3337531486146096,
      "grad_norm": 0.14159703254699707,
      "learning_rate": 8.427060150075137e-05,
      "loss": 0.1705,
      "step": 1590
    },
    {
      "epoch": 0.3339630562552477,
      "grad_norm": 0.1670205146074295,
      "learning_rate": 8.424391201429099e-05,
      "loss": 0.1544,
      "step": 1591
    },
    {
      "epoch": 0.3341729638958858,
      "grad_norm": 0.11887665838003159,
      "learning_rate": 8.421720413806764e-05,
      "loss": 0.147,
      "step": 1592
    },
    {
      "epoch": 0.3343828715365239,
      "grad_norm": 0.11611668765544891,
      "learning_rate": 8.419047788642407e-05,
      "loss": 0.1758,
      "step": 1593
    },
    {
      "epoch": 0.33459277917716207,
      "grad_norm": 0.15789294242858887,
      "learning_rate": 8.416373327371287e-05,
      "loss": 0.1533,
      "step": 1594
    },
    {
      "epoch": 0.3348026868178002,
      "grad_norm": 0.13557223975658417,
      "learning_rate": 8.413697031429653e-05,
      "loss": 0.1588,
      "step": 1595
    },
    {
      "epoch": 0.3350125944584383,
      "grad_norm": 0.29409486055374146,
      "learning_rate": 8.411018902254736e-05,
      "loss": 0.1686,
      "step": 1596
    },
    {
      "epoch": 0.3352225020990764,
      "grad_norm": 0.13911570608615875,
      "learning_rate": 8.408338941284752e-05,
      "loss": 0.168,
      "step": 1597
    },
    {
      "epoch": 0.3354324097397145,
      "grad_norm": 0.1477447748184204,
      "learning_rate": 8.405657149958902e-05,
      "loss": 0.1611,
      "step": 1598
    },
    {
      "epoch": 0.33564231738035266,
      "grad_norm": 0.13031038641929626,
      "learning_rate": 8.402973529717368e-05,
      "loss": 0.1676,
      "step": 1599
    },
    {
      "epoch": 0.33585222502099077,
      "grad_norm": 0.12024478614330292,
      "learning_rate": 8.400288082001319e-05,
      "loss": 0.1567,
      "step": 1600
    },
    {
      "epoch": 0.3360621326616289,
      "grad_norm": 0.12338093668222427,
      "learning_rate": 8.397600808252897e-05,
      "loss": 0.1585,
      "step": 1601
    },
    {
      "epoch": 0.336272040302267,
      "grad_norm": 0.12878558039665222,
      "learning_rate": 8.394911709915232e-05,
      "loss": 0.1624,
      "step": 1602
    },
    {
      "epoch": 0.33648194794290515,
      "grad_norm": 0.12042135745286942,
      "learning_rate": 8.392220788432431e-05,
      "loss": 0.1719,
      "step": 1603
    },
    {
      "epoch": 0.33669185558354325,
      "grad_norm": 0.14576420187950134,
      "learning_rate": 8.389528045249579e-05,
      "loss": 0.1669,
      "step": 1604
    },
    {
      "epoch": 0.33690176322418136,
      "grad_norm": 0.13084185123443604,
      "learning_rate": 8.386833481812744e-05,
      "loss": 0.1556,
      "step": 1605
    },
    {
      "epoch": 0.33711167086481947,
      "grad_norm": 0.15233568847179413,
      "learning_rate": 8.384137099568965e-05,
      "loss": 0.1674,
      "step": 1606
    },
    {
      "epoch": 0.3373215785054576,
      "grad_norm": 0.15160681307315826,
      "learning_rate": 8.381438899966261e-05,
      "loss": 0.1727,
      "step": 1607
    },
    {
      "epoch": 0.33753148614609574,
      "grad_norm": 0.13782384991645813,
      "learning_rate": 8.378738884453627e-05,
      "loss": 0.1754,
      "step": 1608
    },
    {
      "epoch": 0.33774139378673385,
      "grad_norm": 0.12946204841136932,
      "learning_rate": 8.376037054481034e-05,
      "loss": 0.176,
      "step": 1609
    },
    {
      "epoch": 0.33795130142737195,
      "grad_norm": 0.11807133257389069,
      "learning_rate": 8.373333411499426e-05,
      "loss": 0.1465,
      "step": 1610
    },
    {
      "epoch": 0.33816120906801006,
      "grad_norm": 0.1325267106294632,
      "learning_rate": 8.370627956960721e-05,
      "loss": 0.1475,
      "step": 1611
    },
    {
      "epoch": 0.33837111670864817,
      "grad_norm": 0.12463680654764175,
      "learning_rate": 8.36792069231781e-05,
      "loss": 0.1691,
      "step": 1612
    },
    {
      "epoch": 0.33858102434928633,
      "grad_norm": 0.1311996728181839,
      "learning_rate": 8.365211619024555e-05,
      "loss": 0.1562,
      "step": 1613
    },
    {
      "epoch": 0.33879093198992444,
      "grad_norm": 0.13213030993938446,
      "learning_rate": 8.362500738535792e-05,
      "loss": 0.1611,
      "step": 1614
    },
    {
      "epoch": 0.33900083963056254,
      "grad_norm": 0.1333753913640976,
      "learning_rate": 8.359788052307324e-05,
      "loss": 0.1612,
      "step": 1615
    },
    {
      "epoch": 0.33921074727120065,
      "grad_norm": 0.11926085501909256,
      "learning_rate": 8.357073561795928e-05,
      "loss": 0.1621,
      "step": 1616
    },
    {
      "epoch": 0.3394206549118388,
      "grad_norm": 0.14635923504829407,
      "learning_rate": 8.354357268459343e-05,
      "loss": 0.1716,
      "step": 1617
    },
    {
      "epoch": 0.3396305625524769,
      "grad_norm": 0.1318664848804474,
      "learning_rate": 8.351639173756284e-05,
      "loss": 0.171,
      "step": 1618
    },
    {
      "epoch": 0.33984047019311503,
      "grad_norm": 0.13774079084396362,
      "learning_rate": 8.34891927914643e-05,
      "loss": 0.1527,
      "step": 1619
    },
    {
      "epoch": 0.34005037783375314,
      "grad_norm": 0.1164809837937355,
      "learning_rate": 8.346197586090426e-05,
      "loss": 0.161,
      "step": 1620
    },
    {
      "epoch": 0.34026028547439124,
      "grad_norm": 0.13626568019390106,
      "learning_rate": 8.343474096049881e-05,
      "loss": 0.1573,
      "step": 1621
    },
    {
      "epoch": 0.3404701931150294,
      "grad_norm": 0.10750927031040192,
      "learning_rate": 8.340748810487375e-05,
      "loss": 0.1516,
      "step": 1622
    },
    {
      "epoch": 0.3406801007556675,
      "grad_norm": 0.1209016814827919,
      "learning_rate": 8.338021730866448e-05,
      "loss": 0.1604,
      "step": 1623
    },
    {
      "epoch": 0.3408900083963056,
      "grad_norm": 0.12268110364675522,
      "learning_rate": 8.335292858651601e-05,
      "loss": 0.1723,
      "step": 1624
    },
    {
      "epoch": 0.34109991603694373,
      "grad_norm": 0.1716935932636261,
      "learning_rate": 8.332562195308305e-05,
      "loss": 0.1617,
      "step": 1625
    },
    {
      "epoch": 0.3413098236775819,
      "grad_norm": 0.12799715995788574,
      "learning_rate": 8.329829742302984e-05,
      "loss": 0.1677,
      "step": 1626
    },
    {
      "epoch": 0.34151973131822,
      "grad_norm": 0.12748028337955475,
      "learning_rate": 8.327095501103029e-05,
      "loss": 0.1587,
      "step": 1627
    },
    {
      "epoch": 0.3417296389588581,
      "grad_norm": 0.15869416296482086,
      "learning_rate": 8.324359473176793e-05,
      "loss": 0.1644,
      "step": 1628
    },
    {
      "epoch": 0.3419395465994962,
      "grad_norm": 0.14217017590999603,
      "learning_rate": 8.321621659993583e-05,
      "loss": 0.1688,
      "step": 1629
    },
    {
      "epoch": 0.3421494542401343,
      "grad_norm": 0.11858684569597244,
      "learning_rate": 8.318882063023669e-05,
      "loss": 0.1513,
      "step": 1630
    },
    {
      "epoch": 0.3423593618807725,
      "grad_norm": 0.11514747142791748,
      "learning_rate": 8.316140683738274e-05,
      "loss": 0.146,
      "step": 1631
    },
    {
      "epoch": 0.3425692695214106,
      "grad_norm": 0.13995614647865295,
      "learning_rate": 8.313397523609585e-05,
      "loss": 0.1567,
      "step": 1632
    },
    {
      "epoch": 0.3427791771620487,
      "grad_norm": 0.10559704899787903,
      "learning_rate": 8.310652584110738e-05,
      "loss": 0.1577,
      "step": 1633
    },
    {
      "epoch": 0.3429890848026868,
      "grad_norm": 0.11493990570306778,
      "learning_rate": 8.30790586671583e-05,
      "loss": 0.1431,
      "step": 1634
    },
    {
      "epoch": 0.3431989924433249,
      "grad_norm": 0.14548490941524506,
      "learning_rate": 8.305157372899913e-05,
      "loss": 0.1791,
      "step": 1635
    },
    {
      "epoch": 0.3434089000839631,
      "grad_norm": 0.13234320282936096,
      "learning_rate": 8.302407104138988e-05,
      "loss": 0.1568,
      "step": 1636
    },
    {
      "epoch": 0.3436188077246012,
      "grad_norm": 0.1269916445016861,
      "learning_rate": 8.299655061910012e-05,
      "loss": 0.1494,
      "step": 1637
    },
    {
      "epoch": 0.3438287153652393,
      "grad_norm": 0.12883053719997406,
      "learning_rate": 8.296901247690896e-05,
      "loss": 0.1633,
      "step": 1638
    },
    {
      "epoch": 0.3440386230058774,
      "grad_norm": 0.12479805201292038,
      "learning_rate": 8.294145662960502e-05,
      "loss": 0.1744,
      "step": 1639
    },
    {
      "epoch": 0.34424853064651556,
      "grad_norm": 0.13455580174922943,
      "learning_rate": 8.29138830919864e-05,
      "loss": 0.1787,
      "step": 1640
    },
    {
      "epoch": 0.34445843828715367,
      "grad_norm": 0.11923965066671371,
      "learning_rate": 8.288629187886073e-05,
      "loss": 0.1703,
      "step": 1641
    },
    {
      "epoch": 0.3446683459277918,
      "grad_norm": 0.1303246170282364,
      "learning_rate": 8.285868300504511e-05,
      "loss": 0.1669,
      "step": 1642
    },
    {
      "epoch": 0.3448782535684299,
      "grad_norm": 0.09619605541229248,
      "learning_rate": 8.283105648536616e-05,
      "loss": 0.163,
      "step": 1643
    },
    {
      "epoch": 0.345088161209068,
      "grad_norm": 0.11518269032239914,
      "learning_rate": 8.280341233465992e-05,
      "loss": 0.1511,
      "step": 1644
    },
    {
      "epoch": 0.34529806884970615,
      "grad_norm": 0.10729756206274033,
      "learning_rate": 8.277575056777195e-05,
      "loss": 0.1671,
      "step": 1645
    },
    {
      "epoch": 0.34550797649034426,
      "grad_norm": 0.13708291947841644,
      "learning_rate": 8.274807119955725e-05,
      "loss": 0.1719,
      "step": 1646
    },
    {
      "epoch": 0.34571788413098237,
      "grad_norm": 0.11630459129810333,
      "learning_rate": 8.272037424488028e-05,
      "loss": 0.1505,
      "step": 1647
    },
    {
      "epoch": 0.3459277917716205,
      "grad_norm": 0.15268699824810028,
      "learning_rate": 8.26926597186149e-05,
      "loss": 0.1714,
      "step": 1648
    },
    {
      "epoch": 0.3461376994122586,
      "grad_norm": 0.1177772805094719,
      "learning_rate": 8.266492763564451e-05,
      "loss": 0.1651,
      "step": 1649
    },
    {
      "epoch": 0.34634760705289674,
      "grad_norm": 0.12576864659786224,
      "learning_rate": 8.263717801086182e-05,
      "loss": 0.1542,
      "step": 1650
    },
    {
      "epoch": 0.34655751469353485,
      "grad_norm": 0.10988842695951462,
      "learning_rate": 8.260941085916901e-05,
      "loss": 0.1436,
      "step": 1651
    },
    {
      "epoch": 0.34676742233417296,
      "grad_norm": 0.12295406311750412,
      "learning_rate": 8.258162619547771e-05,
      "loss": 0.1611,
      "step": 1652
    },
    {
      "epoch": 0.34697732997481107,
      "grad_norm": 0.12765416502952576,
      "learning_rate": 8.25538240347089e-05,
      "loss": 0.1707,
      "step": 1653
    },
    {
      "epoch": 0.34718723761544923,
      "grad_norm": 0.128097802400589,
      "learning_rate": 8.252600439179295e-05,
      "loss": 0.1777,
      "step": 1654
    },
    {
      "epoch": 0.34739714525608734,
      "grad_norm": 0.12253068387508392,
      "learning_rate": 8.24981672816697e-05,
      "loss": 0.1625,
      "step": 1655
    },
    {
      "epoch": 0.34760705289672544,
      "grad_norm": 0.12777206301689148,
      "learning_rate": 8.247031271928826e-05,
      "loss": 0.1822,
      "step": 1656
    },
    {
      "epoch": 0.34781696053736355,
      "grad_norm": 0.10205968469381332,
      "learning_rate": 8.24424407196072e-05,
      "loss": 0.1459,
      "step": 1657
    },
    {
      "epoch": 0.34802686817800166,
      "grad_norm": 0.11570538580417633,
      "learning_rate": 8.241455129759441e-05,
      "loss": 0.1719,
      "step": 1658
    },
    {
      "epoch": 0.3482367758186398,
      "grad_norm": 0.12801894545555115,
      "learning_rate": 8.238664446822715e-05,
      "loss": 0.1628,
      "step": 1659
    },
    {
      "epoch": 0.34844668345927793,
      "grad_norm": 0.13114169239997864,
      "learning_rate": 8.2358720246492e-05,
      "loss": 0.1768,
      "step": 1660
    },
    {
      "epoch": 0.34865659109991604,
      "grad_norm": 0.14204199612140656,
      "learning_rate": 8.233077864738495e-05,
      "loss": 0.1359,
      "step": 1661
    },
    {
      "epoch": 0.34886649874055414,
      "grad_norm": 0.13539434969425201,
      "learning_rate": 8.230281968591125e-05,
      "loss": 0.1696,
      "step": 1662
    },
    {
      "epoch": 0.34907640638119225,
      "grad_norm": 0.10501176118850708,
      "learning_rate": 8.22748433770855e-05,
      "loss": 0.1593,
      "step": 1663
    },
    {
      "epoch": 0.3492863140218304,
      "grad_norm": 0.10992568731307983,
      "learning_rate": 8.224684973593161e-05,
      "loss": 0.167,
      "step": 1664
    },
    {
      "epoch": 0.3494962216624685,
      "grad_norm": 0.10650011897087097,
      "learning_rate": 8.221883877748285e-05,
      "loss": 0.1533,
      "step": 1665
    },
    {
      "epoch": 0.3497061293031066,
      "grad_norm": 0.10955175757408142,
      "learning_rate": 8.219081051678167e-05,
      "loss": 0.1684,
      "step": 1666
    },
    {
      "epoch": 0.34991603694374473,
      "grad_norm": 0.11864706128835678,
      "learning_rate": 8.216276496887995e-05,
      "loss": 0.1662,
      "step": 1667
    },
    {
      "epoch": 0.3501259445843829,
      "grad_norm": 0.11310485005378723,
      "learning_rate": 8.213470214883876e-05,
      "loss": 0.1551,
      "step": 1668
    },
    {
      "epoch": 0.350335852225021,
      "grad_norm": 0.11496865004301071,
      "learning_rate": 8.210662207172848e-05,
      "loss": 0.1599,
      "step": 1669
    },
    {
      "epoch": 0.3505457598656591,
      "grad_norm": 0.16101498901844025,
      "learning_rate": 8.207852475262876e-05,
      "loss": 0.1599,
      "step": 1670
    },
    {
      "epoch": 0.3507556675062972,
      "grad_norm": 0.12098553776741028,
      "learning_rate": 8.205041020662849e-05,
      "loss": 0.1541,
      "step": 1671
    },
    {
      "epoch": 0.3509655751469353,
      "grad_norm": 0.14668092131614685,
      "learning_rate": 8.202227844882583e-05,
      "loss": 0.1625,
      "step": 1672
    },
    {
      "epoch": 0.3511754827875735,
      "grad_norm": 0.11160324513912201,
      "learning_rate": 8.199412949432817e-05,
      "loss": 0.1534,
      "step": 1673
    },
    {
      "epoch": 0.3513853904282116,
      "grad_norm": 0.1357201784849167,
      "learning_rate": 8.196596335825217e-05,
      "loss": 0.1702,
      "step": 1674
    },
    {
      "epoch": 0.3515952980688497,
      "grad_norm": 0.11901015788316727,
      "learning_rate": 8.193778005572365e-05,
      "loss": 0.1581,
      "step": 1675
    },
    {
      "epoch": 0.3518052057094878,
      "grad_norm": 0.12563678622245789,
      "learning_rate": 8.19095796018777e-05,
      "loss": 0.1681,
      "step": 1676
    },
    {
      "epoch": 0.3520151133501259,
      "grad_norm": 0.12322632968425751,
      "learning_rate": 8.188136201185863e-05,
      "loss": 0.1503,
      "step": 1677
    },
    {
      "epoch": 0.3522250209907641,
      "grad_norm": 0.13357025384902954,
      "learning_rate": 8.185312730081991e-05,
      "loss": 0.1704,
      "step": 1678
    },
    {
      "epoch": 0.3524349286314022,
      "grad_norm": 0.12442915886640549,
      "learning_rate": 8.182487548392422e-05,
      "loss": 0.163,
      "step": 1679
    },
    {
      "epoch": 0.3526448362720403,
      "grad_norm": 0.12422217428684235,
      "learning_rate": 8.179660657634342e-05,
      "loss": 0.1492,
      "step": 1680
    },
    {
      "epoch": 0.3528547439126784,
      "grad_norm": 0.12983065843582153,
      "learning_rate": 8.176832059325861e-05,
      "loss": 0.1509,
      "step": 1681
    },
    {
      "epoch": 0.35306465155331657,
      "grad_norm": 0.10647305846214294,
      "learning_rate": 8.174001754985996e-05,
      "loss": 0.1775,
      "step": 1682
    },
    {
      "epoch": 0.3532745591939547,
      "grad_norm": 0.1141040176153183,
      "learning_rate": 8.171169746134686e-05,
      "loss": 0.1617,
      "step": 1683
    },
    {
      "epoch": 0.3534844668345928,
      "grad_norm": 0.13015101850032806,
      "learning_rate": 8.168336034292786e-05,
      "loss": 0.16,
      "step": 1684
    },
    {
      "epoch": 0.3536943744752309,
      "grad_norm": 0.1377038061618805,
      "learning_rate": 8.165500620982062e-05,
      "loss": 0.1524,
      "step": 1685
    },
    {
      "epoch": 0.353904282115869,
      "grad_norm": 0.1097029447555542,
      "learning_rate": 8.162663507725198e-05,
      "loss": 0.1697,
      "step": 1686
    },
    {
      "epoch": 0.35411418975650716,
      "grad_norm": 0.1175478920340538,
      "learning_rate": 8.159824696045788e-05,
      "loss": 0.1826,
      "step": 1687
    },
    {
      "epoch": 0.35432409739714527,
      "grad_norm": 0.10951685905456543,
      "learning_rate": 8.156984187468337e-05,
      "loss": 0.156,
      "step": 1688
    },
    {
      "epoch": 0.3545340050377834,
      "grad_norm": 0.11676156520843506,
      "learning_rate": 8.154141983518265e-05,
      "loss": 0.1443,
      "step": 1689
    },
    {
      "epoch": 0.3547439126784215,
      "grad_norm": 0.10480359196662903,
      "learning_rate": 8.151298085721899e-05,
      "loss": 0.1675,
      "step": 1690
    },
    {
      "epoch": 0.3549538203190596,
      "grad_norm": 0.11646043509244919,
      "learning_rate": 8.148452495606478e-05,
      "loss": 0.1576,
      "step": 1691
    },
    {
      "epoch": 0.35516372795969775,
      "grad_norm": 0.1242692843079567,
      "learning_rate": 8.14560521470015e-05,
      "loss": 0.1486,
      "step": 1692
    },
    {
      "epoch": 0.35537363560033586,
      "grad_norm": 0.09616512805223465,
      "learning_rate": 8.142756244531967e-05,
      "loss": 0.1522,
      "step": 1693
    },
    {
      "epoch": 0.35558354324097396,
      "grad_norm": 0.139311745762825,
      "learning_rate": 8.13990558663189e-05,
      "loss": 0.1613,
      "step": 1694
    },
    {
      "epoch": 0.35579345088161207,
      "grad_norm": 0.11522821336984634,
      "learning_rate": 8.137053242530795e-05,
      "loss": 0.1603,
      "step": 1695
    },
    {
      "epoch": 0.35600335852225024,
      "grad_norm": 0.11472437530755997,
      "learning_rate": 8.134199213760447e-05,
      "loss": 0.1537,
      "step": 1696
    },
    {
      "epoch": 0.35621326616288834,
      "grad_norm": 0.14459721744060516,
      "learning_rate": 8.13134350185353e-05,
      "loss": 0.151,
      "step": 1697
    },
    {
      "epoch": 0.35642317380352645,
      "grad_norm": 0.12816378474235535,
      "learning_rate": 8.128486108343625e-05,
      "loss": 0.1677,
      "step": 1698
    },
    {
      "epoch": 0.35663308144416456,
      "grad_norm": 0.11897154897451401,
      "learning_rate": 8.125627034765218e-05,
      "loss": 0.1802,
      "step": 1699
    },
    {
      "epoch": 0.35684298908480266,
      "grad_norm": 0.12595590949058533,
      "learning_rate": 8.122766282653696e-05,
      "loss": 0.1668,
      "step": 1700
    },
    {
      "epoch": 0.3570528967254408,
      "grad_norm": 0.1388867050409317,
      "learning_rate": 8.119903853545348e-05,
      "loss": 0.1826,
      "step": 1701
    },
    {
      "epoch": 0.35726280436607893,
      "grad_norm": 0.1385848969221115,
      "learning_rate": 8.117039748977366e-05,
      "loss": 0.1581,
      "step": 1702
    },
    {
      "epoch": 0.35747271200671704,
      "grad_norm": 0.14706510305404663,
      "learning_rate": 8.114173970487838e-05,
      "loss": 0.1618,
      "step": 1703
    },
    {
      "epoch": 0.35768261964735515,
      "grad_norm": 0.15850181877613068,
      "learning_rate": 8.111306519615751e-05,
      "loss": 0.1527,
      "step": 1704
    },
    {
      "epoch": 0.35789252728799326,
      "grad_norm": 0.11479350924491882,
      "learning_rate": 8.108437397900995e-05,
      "loss": 0.166,
      "step": 1705
    },
    {
      "epoch": 0.3581024349286314,
      "grad_norm": 0.14089131355285645,
      "learning_rate": 8.10556660688435e-05,
      "loss": 0.1599,
      "step": 1706
    },
    {
      "epoch": 0.3583123425692695,
      "grad_norm": 0.11980936676263809,
      "learning_rate": 8.102694148107499e-05,
      "loss": 0.1685,
      "step": 1707
    },
    {
      "epoch": 0.35852225020990763,
      "grad_norm": 0.12391865998506546,
      "learning_rate": 8.099820023113018e-05,
      "loss": 0.164,
      "step": 1708
    },
    {
      "epoch": 0.35873215785054574,
      "grad_norm": 0.11180952191352844,
      "learning_rate": 8.096944233444379e-05,
      "loss": 0.1498,
      "step": 1709
    },
    {
      "epoch": 0.3589420654911839,
      "grad_norm": 0.12841565907001495,
      "learning_rate": 8.094066780645943e-05,
      "loss": 0.1606,
      "step": 1710
    },
    {
      "epoch": 0.359151973131822,
      "grad_norm": 0.11051752418279648,
      "learning_rate": 8.09118766626297e-05,
      "loss": 0.1519,
      "step": 1711
    },
    {
      "epoch": 0.3593618807724601,
      "grad_norm": 0.12816421687602997,
      "learning_rate": 8.088306891841609e-05,
      "loss": 0.1657,
      "step": 1712
    },
    {
      "epoch": 0.3595717884130982,
      "grad_norm": 0.10819518566131592,
      "learning_rate": 8.085424458928904e-05,
      "loss": 0.1805,
      "step": 1713
    },
    {
      "epoch": 0.35978169605373633,
      "grad_norm": 0.10757758468389511,
      "learning_rate": 8.082540369072786e-05,
      "loss": 0.1524,
      "step": 1714
    },
    {
      "epoch": 0.3599916036943745,
      "grad_norm": 0.13418933749198914,
      "learning_rate": 8.079654623822077e-05,
      "loss": 0.1623,
      "step": 1715
    },
    {
      "epoch": 0.3602015113350126,
      "grad_norm": 0.11752188950777054,
      "learning_rate": 8.076767224726487e-05,
      "loss": 0.1736,
      "step": 1716
    },
    {
      "epoch": 0.3604114189756507,
      "grad_norm": 0.1459878832101822,
      "learning_rate": 8.073878173336618e-05,
      "loss": 0.1502,
      "step": 1717
    },
    {
      "epoch": 0.3606213266162888,
      "grad_norm": 0.10812754929065704,
      "learning_rate": 8.070987471203955e-05,
      "loss": 0.1484,
      "step": 1718
    },
    {
      "epoch": 0.3608312342569269,
      "grad_norm": 0.1168069913983345,
      "learning_rate": 8.06809511988087e-05,
      "loss": 0.1355,
      "step": 1719
    },
    {
      "epoch": 0.3610411418975651,
      "grad_norm": 0.133521169424057,
      "learning_rate": 8.065201120920625e-05,
      "loss": 0.1707,
      "step": 1720
    },
    {
      "epoch": 0.3612510495382032,
      "grad_norm": 0.12936338782310486,
      "learning_rate": 8.062305475877358e-05,
      "loss": 0.1518,
      "step": 1721
    },
    {
      "epoch": 0.3614609571788413,
      "grad_norm": 0.11167915165424347,
      "learning_rate": 8.059408186306104e-05,
      "loss": 0.1365,
      "step": 1722
    },
    {
      "epoch": 0.3616708648194794,
      "grad_norm": 0.1492997705936432,
      "learning_rate": 8.056509253762768e-05,
      "loss": 0.1719,
      "step": 1723
    },
    {
      "epoch": 0.3618807724601176,
      "grad_norm": 0.1078883707523346,
      "learning_rate": 8.053608679804144e-05,
      "loss": 0.1466,
      "step": 1724
    },
    {
      "epoch": 0.3620906801007557,
      "grad_norm": 0.11772193014621735,
      "learning_rate": 8.050706465987907e-05,
      "loss": 0.1607,
      "step": 1725
    },
    {
      "epoch": 0.3623005877413938,
      "grad_norm": 0.11475538462400436,
      "learning_rate": 8.047802613872615e-05,
      "loss": 0.1727,
      "step": 1726
    },
    {
      "epoch": 0.3625104953820319,
      "grad_norm": 0.10466550290584564,
      "learning_rate": 8.044897125017696e-05,
      "loss": 0.1586,
      "step": 1727
    },
    {
      "epoch": 0.36272040302267,
      "grad_norm": 0.13003644347190857,
      "learning_rate": 8.041990000983472e-05,
      "loss": 0.168,
      "step": 1728
    },
    {
      "epoch": 0.36293031066330816,
      "grad_norm": 0.11580996215343475,
      "learning_rate": 8.039081243331128e-05,
      "loss": 0.1532,
      "step": 1729
    },
    {
      "epoch": 0.36314021830394627,
      "grad_norm": 0.11512108892202377,
      "learning_rate": 8.036170853622738e-05,
      "loss": 0.153,
      "step": 1730
    },
    {
      "epoch": 0.3633501259445844,
      "grad_norm": 0.14258354902267456,
      "learning_rate": 8.033258833421246e-05,
      "loss": 0.1542,
      "step": 1731
    },
    {
      "epoch": 0.3635600335852225,
      "grad_norm": 0.11443816125392914,
      "learning_rate": 8.030345184290473e-05,
      "loss": 0.1523,
      "step": 1732
    },
    {
      "epoch": 0.36376994122586065,
      "grad_norm": 0.098850317299366,
      "learning_rate": 8.027429907795116e-05,
      "loss": 0.1413,
      "step": 1733
    },
    {
      "epoch": 0.36397984886649876,
      "grad_norm": 0.09741919487714767,
      "learning_rate": 8.024513005500744e-05,
      "loss": 0.1482,
      "step": 1734
    },
    {
      "epoch": 0.36418975650713686,
      "grad_norm": 0.12056808918714523,
      "learning_rate": 8.021594478973801e-05,
      "loss": 0.1621,
      "step": 1735
    },
    {
      "epoch": 0.36439966414777497,
      "grad_norm": 0.10978755354881287,
      "learning_rate": 8.018674329781603e-05,
      "loss": 0.1578,
      "step": 1736
    },
    {
      "epoch": 0.3646095717884131,
      "grad_norm": 0.12888531386852264,
      "learning_rate": 8.015752559492335e-05,
      "loss": 0.1552,
      "step": 1737
    },
    {
      "epoch": 0.36481947942905124,
      "grad_norm": 0.12127751111984253,
      "learning_rate": 8.012829169675056e-05,
      "loss": 0.165,
      "step": 1738
    },
    {
      "epoch": 0.36502938706968935,
      "grad_norm": 0.10316291451454163,
      "learning_rate": 8.009904161899692e-05,
      "loss": 0.1306,
      "step": 1739
    },
    {
      "epoch": 0.36523929471032746,
      "grad_norm": 0.12621663510799408,
      "learning_rate": 8.006977537737043e-05,
      "loss": 0.1586,
      "step": 1740
    },
    {
      "epoch": 0.36544920235096556,
      "grad_norm": 0.13226893544197083,
      "learning_rate": 8.004049298758767e-05,
      "loss": 0.1432,
      "step": 1741
    },
    {
      "epoch": 0.36565910999160367,
      "grad_norm": 0.11602537333965302,
      "learning_rate": 8.001119446537397e-05,
      "loss": 0.152,
      "step": 1742
    },
    {
      "epoch": 0.36586901763224183,
      "grad_norm": 0.10899031907320023,
      "learning_rate": 7.998187982646332e-05,
      "loss": 0.1489,
      "step": 1743
    },
    {
      "epoch": 0.36607892527287994,
      "grad_norm": 0.11780837923288345,
      "learning_rate": 7.995254908659836e-05,
      "loss": 0.166,
      "step": 1744
    },
    {
      "epoch": 0.36628883291351805,
      "grad_norm": 0.13604290783405304,
      "learning_rate": 7.992320226153032e-05,
      "loss": 0.1731,
      "step": 1745
    },
    {
      "epoch": 0.36649874055415615,
      "grad_norm": 0.1298649162054062,
      "learning_rate": 7.989383936701916e-05,
      "loss": 0.151,
      "step": 1746
    },
    {
      "epoch": 0.3667086481947943,
      "grad_norm": 0.12146075814962387,
      "learning_rate": 7.986446041883342e-05,
      "loss": 0.1674,
      "step": 1747
    },
    {
      "epoch": 0.3669185558354324,
      "grad_norm": 0.13171765208244324,
      "learning_rate": 7.983506543275025e-05,
      "loss": 0.147,
      "step": 1748
    },
    {
      "epoch": 0.36712846347607053,
      "grad_norm": 0.13356606662273407,
      "learning_rate": 7.980565442455545e-05,
      "loss": 0.1724,
      "step": 1749
    },
    {
      "epoch": 0.36733837111670864,
      "grad_norm": 0.13275739550590515,
      "learning_rate": 7.977622741004338e-05,
      "loss": 0.1586,
      "step": 1750
    },
    {
      "epoch": 0.36754827875734675,
      "grad_norm": 0.1208769828081131,
      "learning_rate": 7.974678440501703e-05,
      "loss": 0.1558,
      "step": 1751
    },
    {
      "epoch": 0.3677581863979849,
      "grad_norm": 0.15097500383853912,
      "learning_rate": 7.971732542528798e-05,
      "loss": 0.157,
      "step": 1752
    },
    {
      "epoch": 0.367968094038623,
      "grad_norm": 0.13939009606838226,
      "learning_rate": 7.968785048667634e-05,
      "loss": 0.1483,
      "step": 1753
    },
    {
      "epoch": 0.3681780016792611,
      "grad_norm": 0.1147853285074234,
      "learning_rate": 7.965835960501086e-05,
      "loss": 0.1638,
      "step": 1754
    },
    {
      "epoch": 0.36838790931989923,
      "grad_norm": 0.10615186393260956,
      "learning_rate": 7.96288527961288e-05,
      "loss": 0.1674,
      "step": 1755
    },
    {
      "epoch": 0.36859781696053734,
      "grad_norm": 0.10610269010066986,
      "learning_rate": 7.959933007587598e-05,
      "loss": 0.1598,
      "step": 1756
    },
    {
      "epoch": 0.3688077246011755,
      "grad_norm": 0.11840075999498367,
      "learning_rate": 7.956979146010683e-05,
      "loss": 0.1673,
      "step": 1757
    },
    {
      "epoch": 0.3690176322418136,
      "grad_norm": 0.1263597160577774,
      "learning_rate": 7.954023696468418e-05,
      "loss": 0.1332,
      "step": 1758
    },
    {
      "epoch": 0.3692275398824517,
      "grad_norm": 0.13218241930007935,
      "learning_rate": 7.951066660547952e-05,
      "loss": 0.165,
      "step": 1759
    },
    {
      "epoch": 0.3694374475230898,
      "grad_norm": 0.11182461678981781,
      "learning_rate": 7.948108039837281e-05,
      "loss": 0.175,
      "step": 1760
    },
    {
      "epoch": 0.369647355163728,
      "grad_norm": 0.12093862891197205,
      "learning_rate": 7.94514783592525e-05,
      "loss": 0.17,
      "step": 1761
    },
    {
      "epoch": 0.3698572628043661,
      "grad_norm": 0.1129969134926796,
      "learning_rate": 7.942186050401556e-05,
      "loss": 0.1605,
      "step": 1762
    },
    {
      "epoch": 0.3700671704450042,
      "grad_norm": 0.11315370351076126,
      "learning_rate": 7.939222684856747e-05,
      "loss": 0.1724,
      "step": 1763
    },
    {
      "epoch": 0.3702770780856423,
      "grad_norm": 0.12104691565036774,
      "learning_rate": 7.936257740882218e-05,
      "loss": 0.1693,
      "step": 1764
    },
    {
      "epoch": 0.3704869857262804,
      "grad_norm": 0.12000355124473572,
      "learning_rate": 7.933291220070209e-05,
      "loss": 0.1855,
      "step": 1765
    },
    {
      "epoch": 0.3706968933669186,
      "grad_norm": 0.11460446566343307,
      "learning_rate": 7.930323124013812e-05,
      "loss": 0.1531,
      "step": 1766
    },
    {
      "epoch": 0.3709068010075567,
      "grad_norm": 0.12112744152545929,
      "learning_rate": 7.927353454306962e-05,
      "loss": 0.166,
      "step": 1767
    },
    {
      "epoch": 0.3711167086481948,
      "grad_norm": 0.11343761533498764,
      "learning_rate": 7.92438221254444e-05,
      "loss": 0.1718,
      "step": 1768
    },
    {
      "epoch": 0.3713266162888329,
      "grad_norm": 0.09574848413467407,
      "learning_rate": 7.921409400321869e-05,
      "loss": 0.1552,
      "step": 1769
    },
    {
      "epoch": 0.371536523929471,
      "grad_norm": 0.12521985173225403,
      "learning_rate": 7.918435019235719e-05,
      "loss": 0.173,
      "step": 1770
    },
    {
      "epoch": 0.37174643157010917,
      "grad_norm": 0.13208284974098206,
      "learning_rate": 7.915459070883297e-05,
      "loss": 0.1461,
      "step": 1771
    },
    {
      "epoch": 0.3719563392107473,
      "grad_norm": 0.13827428221702576,
      "learning_rate": 7.91248155686276e-05,
      "loss": 0.1763,
      "step": 1772
    },
    {
      "epoch": 0.3721662468513854,
      "grad_norm": 0.11231647431850433,
      "learning_rate": 7.909502478773099e-05,
      "loss": 0.1507,
      "step": 1773
    },
    {
      "epoch": 0.3723761544920235,
      "grad_norm": 0.13902410864830017,
      "learning_rate": 7.906521838214147e-05,
      "loss": 0.1658,
      "step": 1774
    },
    {
      "epoch": 0.37258606213266166,
      "grad_norm": 0.1284346878528595,
      "learning_rate": 7.903539636786574e-05,
      "loss": 0.1473,
      "step": 1775
    },
    {
      "epoch": 0.37279596977329976,
      "grad_norm": 0.09582658857107162,
      "learning_rate": 7.900555876091894e-05,
      "loss": 0.1578,
      "step": 1776
    },
    {
      "epoch": 0.37300587741393787,
      "grad_norm": 0.11832595616579056,
      "learning_rate": 7.897570557732452e-05,
      "loss": 0.1528,
      "step": 1777
    },
    {
      "epoch": 0.373215785054576,
      "grad_norm": 0.12966689467430115,
      "learning_rate": 7.894583683311434e-05,
      "loss": 0.1861,
      "step": 1778
    },
    {
      "epoch": 0.3734256926952141,
      "grad_norm": 0.11145227402448654,
      "learning_rate": 7.891595254432857e-05,
      "loss": 0.1555,
      "step": 1779
    },
    {
      "epoch": 0.37363560033585225,
      "grad_norm": 0.13669519126415253,
      "learning_rate": 7.888605272701578e-05,
      "loss": 0.1618,
      "step": 1780
    },
    {
      "epoch": 0.37384550797649035,
      "grad_norm": 0.12473908811807632,
      "learning_rate": 7.885613739723285e-05,
      "loss": 0.1604,
      "step": 1781
    },
    {
      "epoch": 0.37405541561712846,
      "grad_norm": 0.10881557315587997,
      "learning_rate": 7.8826206571045e-05,
      "loss": 0.1701,
      "step": 1782
    },
    {
      "epoch": 0.37426532325776657,
      "grad_norm": 0.11604516953229904,
      "learning_rate": 7.879626026452576e-05,
      "loss": 0.1467,
      "step": 1783
    },
    {
      "epoch": 0.3744752308984047,
      "grad_norm": 0.11594154685735703,
      "learning_rate": 7.876629849375696e-05,
      "loss": 0.181,
      "step": 1784
    },
    {
      "epoch": 0.37468513853904284,
      "grad_norm": 0.11912267655134201,
      "learning_rate": 7.87363212748288e-05,
      "loss": 0.1451,
      "step": 1785
    },
    {
      "epoch": 0.37489504617968095,
      "grad_norm": 0.11206775903701782,
      "learning_rate": 7.87063286238397e-05,
      "loss": 0.1706,
      "step": 1786
    },
    {
      "epoch": 0.37510495382031905,
      "grad_norm": 0.12397405505180359,
      "learning_rate": 7.86763205568964e-05,
      "loss": 0.1599,
      "step": 1787
    },
    {
      "epoch": 0.37531486146095716,
      "grad_norm": 0.1209942027926445,
      "learning_rate": 7.864629709011395e-05,
      "loss": 0.139,
      "step": 1788
    },
    {
      "epoch": 0.3755247691015953,
      "grad_norm": 0.10950622707605362,
      "learning_rate": 7.86162582396156e-05,
      "loss": 0.1625,
      "step": 1789
    },
    {
      "epoch": 0.37573467674223343,
      "grad_norm": 0.12611068785190582,
      "learning_rate": 7.85862040215329e-05,
      "loss": 0.1719,
      "step": 1790
    },
    {
      "epoch": 0.37594458438287154,
      "grad_norm": 0.1125999391078949,
      "learning_rate": 7.855613445200567e-05,
      "loss": 0.1732,
      "step": 1791
    },
    {
      "epoch": 0.37615449202350965,
      "grad_norm": 0.12845373153686523,
      "learning_rate": 7.852604954718195e-05,
      "loss": 0.1477,
      "step": 1792
    },
    {
      "epoch": 0.37636439966414775,
      "grad_norm": 0.17373661696910858,
      "learning_rate": 7.8495949323218e-05,
      "loss": 0.158,
      "step": 1793
    },
    {
      "epoch": 0.3765743073047859,
      "grad_norm": 0.10421296954154968,
      "learning_rate": 7.846583379627836e-05,
      "loss": 0.1683,
      "step": 1794
    },
    {
      "epoch": 0.376784214945424,
      "grad_norm": 0.1311960518360138,
      "learning_rate": 7.843570298253577e-05,
      "loss": 0.1361,
      "step": 1795
    },
    {
      "epoch": 0.37699412258606213,
      "grad_norm": 0.17410288751125336,
      "learning_rate": 7.840555689817111e-05,
      "loss": 0.1678,
      "step": 1796
    },
    {
      "epoch": 0.37720403022670024,
      "grad_norm": 0.11635412275791168,
      "learning_rate": 7.837539555937354e-05,
      "loss": 0.1582,
      "step": 1797
    },
    {
      "epoch": 0.37741393786733834,
      "grad_norm": 0.1161051094532013,
      "learning_rate": 7.83452189823404e-05,
      "loss": 0.1686,
      "step": 1798
    },
    {
      "epoch": 0.3776238455079765,
      "grad_norm": 0.11930254846811295,
      "learning_rate": 7.831502718327719e-05,
      "loss": 0.1717,
      "step": 1799
    },
    {
      "epoch": 0.3778337531486146,
      "grad_norm": 0.12649065256118774,
      "learning_rate": 7.82848201783976e-05,
      "loss": 0.1418,
      "step": 1800
    },
    {
      "epoch": 0.3780436607892527,
      "grad_norm": 0.1233733594417572,
      "learning_rate": 7.825459798392349e-05,
      "loss": 0.1412,
      "step": 1801
    },
    {
      "epoch": 0.37825356842989083,
      "grad_norm": 0.12751632928848267,
      "learning_rate": 7.822436061608485e-05,
      "loss": 0.1591,
      "step": 1802
    },
    {
      "epoch": 0.378463476070529,
      "grad_norm": 0.11433545500040054,
      "learning_rate": 7.819410809111984e-05,
      "loss": 0.1608,
      "step": 1803
    },
    {
      "epoch": 0.3786733837111671,
      "grad_norm": 0.12332860380411148,
      "learning_rate": 7.816384042527479e-05,
      "loss": 0.1544,
      "step": 1804
    },
    {
      "epoch": 0.3788832913518052,
      "grad_norm": 0.14029990136623383,
      "learning_rate": 7.813355763480407e-05,
      "loss": 0.1652,
      "step": 1805
    },
    {
      "epoch": 0.3790931989924433,
      "grad_norm": 0.12804871797561646,
      "learning_rate": 7.810325973597027e-05,
      "loss": 0.1464,
      "step": 1806
    },
    {
      "epoch": 0.3793031066330814,
      "grad_norm": 0.10844779014587402,
      "learning_rate": 7.807294674504405e-05,
      "loss": 0.1517,
      "step": 1807
    },
    {
      "epoch": 0.3795130142737196,
      "grad_norm": 0.1057022362947464,
      "learning_rate": 7.804261867830418e-05,
      "loss": 0.1558,
      "step": 1808
    },
    {
      "epoch": 0.3797229219143577,
      "grad_norm": 0.1034378632903099,
      "learning_rate": 7.801227555203751e-05,
      "loss": 0.1519,
      "step": 1809
    },
    {
      "epoch": 0.3799328295549958,
      "grad_norm": 0.13120600581169128,
      "learning_rate": 7.7981917382539e-05,
      "loss": 0.1689,
      "step": 1810
    },
    {
      "epoch": 0.3801427371956339,
      "grad_norm": 0.14298629760742188,
      "learning_rate": 7.795154418611169e-05,
      "loss": 0.1609,
      "step": 1811
    },
    {
      "epoch": 0.380352644836272,
      "grad_norm": 0.1206570416688919,
      "learning_rate": 7.792115597906666e-05,
      "loss": 0.1481,
      "step": 1812
    },
    {
      "epoch": 0.3805625524769102,
      "grad_norm": 0.12700314819812775,
      "learning_rate": 7.789075277772309e-05,
      "loss": 0.1526,
      "step": 1813
    },
    {
      "epoch": 0.3807724601175483,
      "grad_norm": 0.13419455289840698,
      "learning_rate": 7.786033459840817e-05,
      "loss": 0.1665,
      "step": 1814
    },
    {
      "epoch": 0.3809823677581864,
      "grad_norm": 0.11350502818822861,
      "learning_rate": 7.78299014574572e-05,
      "loss": 0.1293,
      "step": 1815
    },
    {
      "epoch": 0.3811922753988245,
      "grad_norm": 0.12667128443717957,
      "learning_rate": 7.779945337121343e-05,
      "loss": 0.1474,
      "step": 1816
    },
    {
      "epoch": 0.38140218303946266,
      "grad_norm": 0.14447607100009918,
      "learning_rate": 7.77689903560282e-05,
      "loss": 0.1642,
      "step": 1817
    },
    {
      "epoch": 0.38161209068010077,
      "grad_norm": 0.1254795789718628,
      "learning_rate": 7.773851242826083e-05,
      "loss": 0.1565,
      "step": 1818
    },
    {
      "epoch": 0.3818219983207389,
      "grad_norm": 0.1166321262717247,
      "learning_rate": 7.770801960427869e-05,
      "loss": 0.1583,
      "step": 1819
    },
    {
      "epoch": 0.382031905961377,
      "grad_norm": 0.11959258466959,
      "learning_rate": 7.767751190045706e-05,
      "loss": 0.1623,
      "step": 1820
    },
    {
      "epoch": 0.3822418136020151,
      "grad_norm": 0.09385194629430771,
      "learning_rate": 7.764698933317935e-05,
      "loss": 0.1525,
      "step": 1821
    },
    {
      "epoch": 0.38245172124265325,
      "grad_norm": 0.1065850630402565,
      "learning_rate": 7.761645191883684e-05,
      "loss": 0.1538,
      "step": 1822
    },
    {
      "epoch": 0.38266162888329136,
      "grad_norm": 0.13342434167861938,
      "learning_rate": 7.758589967382883e-05,
      "loss": 0.1488,
      "step": 1823
    },
    {
      "epoch": 0.38287153652392947,
      "grad_norm": 0.13576741516590118,
      "learning_rate": 7.755533261456256e-05,
      "loss": 0.1693,
      "step": 1824
    },
    {
      "epoch": 0.3830814441645676,
      "grad_norm": 0.12756936252117157,
      "learning_rate": 7.752475075745325e-05,
      "loss": 0.1573,
      "step": 1825
    },
    {
      "epoch": 0.3832913518052057,
      "grad_norm": 0.1328425407409668,
      "learning_rate": 7.749415411892404e-05,
      "loss": 0.142,
      "step": 1826
    },
    {
      "epoch": 0.38350125944584385,
      "grad_norm": 0.1451667994260788,
      "learning_rate": 7.746354271540605e-05,
      "loss": 0.1544,
      "step": 1827
    },
    {
      "epoch": 0.38371116708648195,
      "grad_norm": 0.13675998151302338,
      "learning_rate": 7.743291656333826e-05,
      "loss": 0.1576,
      "step": 1828
    },
    {
      "epoch": 0.38392107472712006,
      "grad_norm": 0.11071248352527618,
      "learning_rate": 7.740227567916767e-05,
      "loss": 0.1636,
      "step": 1829
    },
    {
      "epoch": 0.38413098236775817,
      "grad_norm": 0.13009920716285706,
      "learning_rate": 7.73716200793491e-05,
      "loss": 0.1705,
      "step": 1830
    },
    {
      "epoch": 0.38434089000839633,
      "grad_norm": 0.13077101111412048,
      "learning_rate": 7.734094978034532e-05,
      "loss": 0.1715,
      "step": 1831
    },
    {
      "epoch": 0.38455079764903444,
      "grad_norm": 0.11513860523700714,
      "learning_rate": 7.731026479862696e-05,
      "loss": 0.1584,
      "step": 1832
    },
    {
      "epoch": 0.38476070528967254,
      "grad_norm": 0.10579150915145874,
      "learning_rate": 7.727956515067259e-05,
      "loss": 0.1618,
      "step": 1833
    },
    {
      "epoch": 0.38497061293031065,
      "grad_norm": 0.1078706756234169,
      "learning_rate": 7.724885085296858e-05,
      "loss": 0.167,
      "step": 1834
    },
    {
      "epoch": 0.38518052057094876,
      "grad_norm": 0.131106436252594,
      "learning_rate": 7.721812192200924e-05,
      "loss": 0.1525,
      "step": 1835
    },
    {
      "epoch": 0.3853904282115869,
      "grad_norm": 0.15161815285682678,
      "learning_rate": 7.718737837429668e-05,
      "loss": 0.1523,
      "step": 1836
    },
    {
      "epoch": 0.38560033585222503,
      "grad_norm": 0.11949379742145538,
      "learning_rate": 7.715662022634094e-05,
      "loss": 0.1543,
      "step": 1837
    },
    {
      "epoch": 0.38581024349286314,
      "grad_norm": 0.16841568052768707,
      "learning_rate": 7.712584749465981e-05,
      "loss": 0.1754,
      "step": 1838
    },
    {
      "epoch": 0.38602015113350124,
      "grad_norm": 0.1119508296251297,
      "learning_rate": 7.709506019577895e-05,
      "loss": 0.16,
      "step": 1839
    },
    {
      "epoch": 0.38623005877413935,
      "grad_norm": 0.12148314714431763,
      "learning_rate": 7.706425834623184e-05,
      "loss": 0.1545,
      "step": 1840
    },
    {
      "epoch": 0.3864399664147775,
      "grad_norm": 0.13948237895965576,
      "learning_rate": 7.703344196255979e-05,
      "loss": 0.156,
      "step": 1841
    },
    {
      "epoch": 0.3866498740554156,
      "grad_norm": 0.11464816331863403,
      "learning_rate": 7.700261106131191e-05,
      "loss": 0.172,
      "step": 1842
    },
    {
      "epoch": 0.38685978169605373,
      "grad_norm": 0.11518428474664688,
      "learning_rate": 7.697176565904507e-05,
      "loss": 0.171,
      "step": 1843
    },
    {
      "epoch": 0.38706968933669184,
      "grad_norm": 0.11466710269451141,
      "learning_rate": 7.694090577232398e-05,
      "loss": 0.1545,
      "step": 1844
    },
    {
      "epoch": 0.38727959697733,
      "grad_norm": 0.13466320931911469,
      "learning_rate": 7.691003141772108e-05,
      "loss": 0.1487,
      "step": 1845
    },
    {
      "epoch": 0.3874895046179681,
      "grad_norm": 0.1101074293255806,
      "learning_rate": 7.687914261181664e-05,
      "loss": 0.1659,
      "step": 1846
    },
    {
      "epoch": 0.3876994122586062,
      "grad_norm": 0.12081106752157211,
      "learning_rate": 7.684823937119862e-05,
      "loss": 0.1558,
      "step": 1847
    },
    {
      "epoch": 0.3879093198992443,
      "grad_norm": 0.12375544011592865,
      "learning_rate": 7.681732171246276e-05,
      "loss": 0.1626,
      "step": 1848
    },
    {
      "epoch": 0.3881192275398824,
      "grad_norm": 0.13245245814323425,
      "learning_rate": 7.678638965221262e-05,
      "loss": 0.1445,
      "step": 1849
    },
    {
      "epoch": 0.3883291351805206,
      "grad_norm": 0.12885937094688416,
      "learning_rate": 7.675544320705932e-05,
      "loss": 0.151,
      "step": 1850
    },
    {
      "epoch": 0.3885390428211587,
      "grad_norm": 0.15650451183319092,
      "learning_rate": 7.67244823936219e-05,
      "loss": 0.1456,
      "step": 1851
    },
    {
      "epoch": 0.3887489504617968,
      "grad_norm": 0.13623982667922974,
      "learning_rate": 7.669350722852693e-05,
      "loss": 0.1726,
      "step": 1852
    },
    {
      "epoch": 0.3889588581024349,
      "grad_norm": 0.1482025533914566,
      "learning_rate": 7.666251772840887e-05,
      "loss": 0.1594,
      "step": 1853
    },
    {
      "epoch": 0.3891687657430731,
      "grad_norm": 0.12205847352743149,
      "learning_rate": 7.663151390990975e-05,
      "loss": 0.1656,
      "step": 1854
    },
    {
      "epoch": 0.3893786733837112,
      "grad_norm": 0.11122436821460724,
      "learning_rate": 7.660049578967934e-05,
      "loss": 0.1674,
      "step": 1855
    },
    {
      "epoch": 0.3895885810243493,
      "grad_norm": 0.13141264021396637,
      "learning_rate": 7.656946338437505e-05,
      "loss": 0.1543,
      "step": 1856
    },
    {
      "epoch": 0.3897984886649874,
      "grad_norm": 0.12650364637374878,
      "learning_rate": 7.653841671066203e-05,
      "loss": 0.1469,
      "step": 1857
    },
    {
      "epoch": 0.3900083963056255,
      "grad_norm": 0.12014992535114288,
      "learning_rate": 7.650735578521302e-05,
      "loss": 0.1541,
      "step": 1858
    },
    {
      "epoch": 0.39021830394626367,
      "grad_norm": 0.1951693892478943,
      "learning_rate": 7.647628062470848e-05,
      "loss": 0.1653,
      "step": 1859
    },
    {
      "epoch": 0.3904282115869018,
      "grad_norm": 0.1394760012626648,
      "learning_rate": 7.644519124583643e-05,
      "loss": 0.1573,
      "step": 1860
    },
    {
      "epoch": 0.3906381192275399,
      "grad_norm": 0.14827001094818115,
      "learning_rate": 7.641408766529266e-05,
      "loss": 0.1585,
      "step": 1861
    },
    {
      "epoch": 0.390848026868178,
      "grad_norm": 0.1227707639336586,
      "learning_rate": 7.63829698997804e-05,
      "loss": 0.1548,
      "step": 1862
    },
    {
      "epoch": 0.3910579345088161,
      "grad_norm": 0.13828247785568237,
      "learning_rate": 7.635183796601068e-05,
      "loss": 0.1789,
      "step": 1863
    },
    {
      "epoch": 0.39126784214945426,
      "grad_norm": 0.13925831019878387,
      "learning_rate": 7.632069188070204e-05,
      "loss": 0.1699,
      "step": 1864
    },
    {
      "epoch": 0.39147774979009237,
      "grad_norm": 0.13822948932647705,
      "learning_rate": 7.628953166058062e-05,
      "loss": 0.1674,
      "step": 1865
    },
    {
      "epoch": 0.3916876574307305,
      "grad_norm": 0.14011485874652863,
      "learning_rate": 7.625835732238018e-05,
      "loss": 0.1778,
      "step": 1866
    },
    {
      "epoch": 0.3918975650713686,
      "grad_norm": 0.1188986524939537,
      "learning_rate": 7.622716888284205e-05,
      "loss": 0.1563,
      "step": 1867
    },
    {
      "epoch": 0.39210747271200674,
      "grad_norm": 0.10080984979867935,
      "learning_rate": 7.619596635871513e-05,
      "loss": 0.169,
      "step": 1868
    },
    {
      "epoch": 0.39231738035264485,
      "grad_norm": 0.11640693992376328,
      "learning_rate": 7.616474976675592e-05,
      "loss": 0.1699,
      "step": 1869
    },
    {
      "epoch": 0.39252728799328296,
      "grad_norm": 0.1325574517250061,
      "learning_rate": 7.613351912372842e-05,
      "loss": 0.1727,
      "step": 1870
    },
    {
      "epoch": 0.39273719563392107,
      "grad_norm": 0.12673214077949524,
      "learning_rate": 7.610227444640417e-05,
      "loss": 0.1579,
      "step": 1871
    },
    {
      "epoch": 0.3929471032745592,
      "grad_norm": 0.10220383107662201,
      "learning_rate": 7.607101575156232e-05,
      "loss": 0.1559,
      "step": 1872
    },
    {
      "epoch": 0.39315701091519734,
      "grad_norm": 0.11445717513561249,
      "learning_rate": 7.603974305598948e-05,
      "loss": 0.1609,
      "step": 1873
    },
    {
      "epoch": 0.39336691855583544,
      "grad_norm": 0.10287608951330185,
      "learning_rate": 7.600845637647981e-05,
      "loss": 0.1442,
      "step": 1874
    },
    {
      "epoch": 0.39357682619647355,
      "grad_norm": 0.11501127481460571,
      "learning_rate": 7.597715572983499e-05,
      "loss": 0.162,
      "step": 1875
    },
    {
      "epoch": 0.39378673383711166,
      "grad_norm": 0.18507914245128632,
      "learning_rate": 7.594584113286413e-05,
      "loss": 0.1611,
      "step": 1876
    },
    {
      "epoch": 0.39399664147774976,
      "grad_norm": 0.11232730746269226,
      "learning_rate": 7.591451260238394e-05,
      "loss": 0.1555,
      "step": 1877
    },
    {
      "epoch": 0.39420654911838793,
      "grad_norm": 0.0994970053434372,
      "learning_rate": 7.588317015521853e-05,
      "loss": 0.1331,
      "step": 1878
    },
    {
      "epoch": 0.39441645675902604,
      "grad_norm": 0.11033397912979126,
      "learning_rate": 7.58518138081995e-05,
      "loss": 0.1611,
      "step": 1879
    },
    {
      "epoch": 0.39462636439966414,
      "grad_norm": 0.12892796099185944,
      "learning_rate": 7.582044357816593e-05,
      "loss": 0.1597,
      "step": 1880
    },
    {
      "epoch": 0.39483627204030225,
      "grad_norm": 0.13372941315174103,
      "learning_rate": 7.578905948196437e-05,
      "loss": 0.1605,
      "step": 1881
    },
    {
      "epoch": 0.3950461796809404,
      "grad_norm": 0.11890514194965363,
      "learning_rate": 7.57576615364488e-05,
      "loss": 0.1598,
      "step": 1882
    },
    {
      "epoch": 0.3952560873215785,
      "grad_norm": 0.14822334051132202,
      "learning_rate": 7.57262497584806e-05,
      "loss": 0.1801,
      "step": 1883
    },
    {
      "epoch": 0.3954659949622166,
      "grad_norm": 0.1431664526462555,
      "learning_rate": 7.569482416492863e-05,
      "loss": 0.1632,
      "step": 1884
    },
    {
      "epoch": 0.39567590260285473,
      "grad_norm": 0.11748656630516052,
      "learning_rate": 7.566338477266914e-05,
      "loss": 0.1479,
      "step": 1885
    },
    {
      "epoch": 0.39588581024349284,
      "grad_norm": 0.11131531000137329,
      "learning_rate": 7.563193159858584e-05,
      "loss": 0.1545,
      "step": 1886
    },
    {
      "epoch": 0.396095717884131,
      "grad_norm": 0.11406468600034714,
      "learning_rate": 7.560046465956975e-05,
      "loss": 0.1505,
      "step": 1887
    },
    {
      "epoch": 0.3963056255247691,
      "grad_norm": 0.12816153466701508,
      "learning_rate": 7.556898397251938e-05,
      "loss": 0.1709,
      "step": 1888
    },
    {
      "epoch": 0.3965155331654072,
      "grad_norm": 0.11210453510284424,
      "learning_rate": 7.553748955434054e-05,
      "loss": 0.1575,
      "step": 1889
    },
    {
      "epoch": 0.3967254408060453,
      "grad_norm": 0.11992809921503067,
      "learning_rate": 7.55059814219465e-05,
      "loss": 0.1396,
      "step": 1890
    },
    {
      "epoch": 0.39693534844668343,
      "grad_norm": 0.11076368391513824,
      "learning_rate": 7.54744595922578e-05,
      "loss": 0.1574,
      "step": 1891
    },
    {
      "epoch": 0.3971452560873216,
      "grad_norm": 0.12722079455852509,
      "learning_rate": 7.544292408220241e-05,
      "loss": 0.1595,
      "step": 1892
    },
    {
      "epoch": 0.3973551637279597,
      "grad_norm": 0.12932024896144867,
      "learning_rate": 7.541137490871559e-05,
      "loss": 0.1822,
      "step": 1893
    },
    {
      "epoch": 0.3975650713685978,
      "grad_norm": 0.1031857281923294,
      "learning_rate": 7.537981208874002e-05,
      "loss": 0.1624,
      "step": 1894
    },
    {
      "epoch": 0.3977749790092359,
      "grad_norm": 0.10605569183826447,
      "learning_rate": 7.534823563922558e-05,
      "loss": 0.1553,
      "step": 1895
    },
    {
      "epoch": 0.3979848866498741,
      "grad_norm": 0.11709260195493698,
      "learning_rate": 7.531664557712962e-05,
      "loss": 0.1636,
      "step": 1896
    },
    {
      "epoch": 0.3981947942905122,
      "grad_norm": 0.12204380333423615,
      "learning_rate": 7.528504191941668e-05,
      "loss": 0.1446,
      "step": 1897
    },
    {
      "epoch": 0.3984047019311503,
      "grad_norm": 0.1183619275689125,
      "learning_rate": 7.525342468305865e-05,
      "loss": 0.1715,
      "step": 1898
    },
    {
      "epoch": 0.3986146095717884,
      "grad_norm": 0.15471158921718597,
      "learning_rate": 7.522179388503471e-05,
      "loss": 0.1648,
      "step": 1899
    },
    {
      "epoch": 0.3988245172124265,
      "grad_norm": 0.11495169252157211,
      "learning_rate": 7.519014954233136e-05,
      "loss": 0.1752,
      "step": 1900
    },
    {
      "epoch": 0.3990344248530647,
      "grad_norm": 0.1107286810874939,
      "learning_rate": 7.515849167194227e-05,
      "loss": 0.1535,
      "step": 1901
    },
    {
      "epoch": 0.3992443324937028,
      "grad_norm": 0.13978683948516846,
      "learning_rate": 7.51268202908685e-05,
      "loss": 0.1535,
      "step": 1902
    },
    {
      "epoch": 0.3994542401343409,
      "grad_norm": 0.13453927636146545,
      "learning_rate": 7.509513541611826e-05,
      "loss": 0.154,
      "step": 1903
    },
    {
      "epoch": 0.399664147774979,
      "grad_norm": 0.14580851793289185,
      "learning_rate": 7.506343706470707e-05,
      "loss": 0.1758,
      "step": 1904
    },
    {
      "epoch": 0.3998740554156171,
      "grad_norm": 0.14356370270252228,
      "learning_rate": 7.503172525365766e-05,
      "loss": 0.1781,
      "step": 1905
    },
    {
      "epoch": 0.40008396305625527,
      "grad_norm": 0.1313529908657074,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.1627,
      "step": 1906
    },
    {
      "epoch": 0.4002938706968934,
      "grad_norm": 0.11975359171628952,
      "learning_rate": 7.49682613207713e-05,
      "loss": 0.1542,
      "step": 1907
    },
    {
      "epoch": 0.4005037783375315,
      "grad_norm": 0.1087488904595375,
      "learning_rate": 7.493650923301592e-05,
      "loss": 0.1641,
      "step": 1908
    },
    {
      "epoch": 0.4007136859781696,
      "grad_norm": 0.1269085705280304,
      "learning_rate": 7.490474375378546e-05,
      "loss": 0.1627,
      "step": 1909
    },
    {
      "epoch": 0.40092359361880775,
      "grad_norm": 0.1457659900188446,
      "learning_rate": 7.487296490013875e-05,
      "loss": 0.1808,
      "step": 1910
    },
    {
      "epoch": 0.40113350125944586,
      "grad_norm": 0.1324755698442459,
      "learning_rate": 7.484117268914171e-05,
      "loss": 0.1625,
      "step": 1911
    },
    {
      "epoch": 0.40134340890008396,
      "grad_norm": 0.14338278770446777,
      "learning_rate": 7.48093671378675e-05,
      "loss": 0.1551,
      "step": 1912
    },
    {
      "epoch": 0.40155331654072207,
      "grad_norm": 0.13183006644248962,
      "learning_rate": 7.477754826339645e-05,
      "loss": 0.1611,
      "step": 1913
    },
    {
      "epoch": 0.4017632241813602,
      "grad_norm": 0.11733553558588028,
      "learning_rate": 7.474571608281599e-05,
      "loss": 0.1514,
      "step": 1914
    },
    {
      "epoch": 0.40197313182199834,
      "grad_norm": 0.12164483219385147,
      "learning_rate": 7.471387061322074e-05,
      "loss": 0.1609,
      "step": 1915
    },
    {
      "epoch": 0.40218303946263645,
      "grad_norm": 0.1412167251110077,
      "learning_rate": 7.468201187171245e-05,
      "loss": 0.1594,
      "step": 1916
    },
    {
      "epoch": 0.40239294710327456,
      "grad_norm": 0.14494703710079193,
      "learning_rate": 7.46501398754e-05,
      "loss": 0.1702,
      "step": 1917
    },
    {
      "epoch": 0.40260285474391266,
      "grad_norm": 0.1307775378227234,
      "learning_rate": 7.461825464139934e-05,
      "loss": 0.1626,
      "step": 1918
    },
    {
      "epoch": 0.40281276238455077,
      "grad_norm": 0.11550642549991608,
      "learning_rate": 7.458635618683362e-05,
      "loss": 0.1708,
      "step": 1919
    },
    {
      "epoch": 0.40302267002518893,
      "grad_norm": 0.11663535982370377,
      "learning_rate": 7.4554444528833e-05,
      "loss": 0.1467,
      "step": 1920
    },
    {
      "epoch": 0.40323257766582704,
      "grad_norm": 0.11326033622026443,
      "learning_rate": 7.452251968453481e-05,
      "loss": 0.1555,
      "step": 1921
    },
    {
      "epoch": 0.40344248530646515,
      "grad_norm": 0.10349076241254807,
      "learning_rate": 7.449058167108337e-05,
      "loss": 0.1725,
      "step": 1922
    },
    {
      "epoch": 0.40365239294710326,
      "grad_norm": 0.12648595869541168,
      "learning_rate": 7.445863050563017e-05,
      "loss": 0.1535,
      "step": 1923
    },
    {
      "epoch": 0.4038623005877414,
      "grad_norm": 0.12781694531440735,
      "learning_rate": 7.442666620533368e-05,
      "loss": 0.1718,
      "step": 1924
    },
    {
      "epoch": 0.4040722082283795,
      "grad_norm": 0.14075273275375366,
      "learning_rate": 7.439468878735947e-05,
      "loss": 0.1496,
      "step": 1925
    },
    {
      "epoch": 0.40428211586901763,
      "grad_norm": 0.10260710120201111,
      "learning_rate": 7.436269826888015e-05,
      "loss": 0.1569,
      "step": 1926
    },
    {
      "epoch": 0.40449202350965574,
      "grad_norm": 0.15809062123298645,
      "learning_rate": 7.433069466707537e-05,
      "loss": 0.1714,
      "step": 1927
    },
    {
      "epoch": 0.40470193115029385,
      "grad_norm": 0.12049717456102371,
      "learning_rate": 7.429867799913177e-05,
      "loss": 0.1624,
      "step": 1928
    },
    {
      "epoch": 0.404911838790932,
      "grad_norm": 0.11036177724599838,
      "learning_rate": 7.426664828224304e-05,
      "loss": 0.1672,
      "step": 1929
    },
    {
      "epoch": 0.4051217464315701,
      "grad_norm": 0.12535598874092102,
      "learning_rate": 7.423460553360988e-05,
      "loss": 0.1677,
      "step": 1930
    },
    {
      "epoch": 0.4053316540722082,
      "grad_norm": 0.11823215335607529,
      "learning_rate": 7.420254977043998e-05,
      "loss": 0.159,
      "step": 1931
    },
    {
      "epoch": 0.40554156171284633,
      "grad_norm": 0.13293218612670898,
      "learning_rate": 7.4170481009948e-05,
      "loss": 0.1688,
      "step": 1932
    },
    {
      "epoch": 0.40575146935348444,
      "grad_norm": 0.11967580020427704,
      "learning_rate": 7.41383992693556e-05,
      "loss": 0.1649,
      "step": 1933
    },
    {
      "epoch": 0.4059613769941226,
      "grad_norm": 0.12903308868408203,
      "learning_rate": 7.410630456589141e-05,
      "loss": 0.1527,
      "step": 1934
    },
    {
      "epoch": 0.4061712846347607,
      "grad_norm": 0.13130050897598267,
      "learning_rate": 7.407419691679104e-05,
      "loss": 0.1607,
      "step": 1935
    },
    {
      "epoch": 0.4063811922753988,
      "grad_norm": 0.12664926052093506,
      "learning_rate": 7.404207633929699e-05,
      "loss": 0.151,
      "step": 1936
    },
    {
      "epoch": 0.4065910999160369,
      "grad_norm": 0.11870288848876953,
      "learning_rate": 7.400994285065878e-05,
      "loss": 0.151,
      "step": 1937
    },
    {
      "epoch": 0.4068010075566751,
      "grad_norm": 0.10910684615373611,
      "learning_rate": 7.397779646813282e-05,
      "loss": 0.1393,
      "step": 1938
    },
    {
      "epoch": 0.4070109151973132,
      "grad_norm": 0.13992692530155182,
      "learning_rate": 7.394563720898245e-05,
      "loss": 0.18,
      "step": 1939
    },
    {
      "epoch": 0.4072208228379513,
      "grad_norm": 0.1333228498697281,
      "learning_rate": 7.39134650904779e-05,
      "loss": 0.1536,
      "step": 1940
    },
    {
      "epoch": 0.4074307304785894,
      "grad_norm": 0.11630193144083023,
      "learning_rate": 7.388128012989636e-05,
      "loss": 0.1603,
      "step": 1941
    },
    {
      "epoch": 0.4076406381192275,
      "grad_norm": 0.12996290624141693,
      "learning_rate": 7.38490823445219e-05,
      "loss": 0.1532,
      "step": 1942
    },
    {
      "epoch": 0.4078505457598657,
      "grad_norm": 0.12526121735572815,
      "learning_rate": 7.381687175164544e-05,
      "loss": 0.1684,
      "step": 1943
    },
    {
      "epoch": 0.4080604534005038,
      "grad_norm": 0.13063152134418488,
      "learning_rate": 7.378464836856481e-05,
      "loss": 0.1637,
      "step": 1944
    },
    {
      "epoch": 0.4082703610411419,
      "grad_norm": 0.11410169303417206,
      "learning_rate": 7.375241221258471e-05,
      "loss": 0.1621,
      "step": 1945
    },
    {
      "epoch": 0.40848026868178,
      "grad_norm": 0.1353740096092224,
      "learning_rate": 7.37201633010167e-05,
      "loss": 0.1274,
      "step": 1946
    },
    {
      "epoch": 0.4086901763224181,
      "grad_norm": 0.1723635345697403,
      "learning_rate": 7.368790165117916e-05,
      "loss": 0.1589,
      "step": 1947
    },
    {
      "epoch": 0.40890008396305627,
      "grad_norm": 0.111173614859581,
      "learning_rate": 7.365562728039734e-05,
      "loss": 0.1626,
      "step": 1948
    },
    {
      "epoch": 0.4091099916036944,
      "grad_norm": 0.10644244402647018,
      "learning_rate": 7.362334020600334e-05,
      "loss": 0.1709,
      "step": 1949
    },
    {
      "epoch": 0.4093198992443325,
      "grad_norm": 0.12528212368488312,
      "learning_rate": 7.359104044533601e-05,
      "loss": 0.1688,
      "step": 1950
    },
    {
      "epoch": 0.4095298068849706,
      "grad_norm": 0.11388932913541794,
      "learning_rate": 7.35587280157411e-05,
      "loss": 0.1624,
      "step": 1951
    },
    {
      "epoch": 0.40973971452560876,
      "grad_norm": 0.11479576677083969,
      "learning_rate": 7.35264029345711e-05,
      "loss": 0.1728,
      "step": 1952
    },
    {
      "epoch": 0.40994962216624686,
      "grad_norm": 0.11558470875024796,
      "learning_rate": 7.349406521918533e-05,
      "loss": 0.1286,
      "step": 1953
    },
    {
      "epoch": 0.41015952980688497,
      "grad_norm": 0.12116646766662598,
      "learning_rate": 7.346171488694988e-05,
      "loss": 0.1704,
      "step": 1954
    },
    {
      "epoch": 0.4103694374475231,
      "grad_norm": 0.12693926692008972,
      "learning_rate": 7.34293519552376e-05,
      "loss": 0.165,
      "step": 1955
    },
    {
      "epoch": 0.4105793450881612,
      "grad_norm": 0.11315865814685822,
      "learning_rate": 7.339697644142815e-05,
      "loss": 0.1562,
      "step": 1956
    },
    {
      "epoch": 0.41078925272879935,
      "grad_norm": 0.13625770807266235,
      "learning_rate": 7.33645883629079e-05,
      "loss": 0.1792,
      "step": 1957
    },
    {
      "epoch": 0.41099916036943746,
      "grad_norm": 0.11685556173324585,
      "learning_rate": 7.333218773707e-05,
      "loss": 0.1626,
      "step": 1958
    },
    {
      "epoch": 0.41120906801007556,
      "grad_norm": 0.13109751045703888,
      "learning_rate": 7.329977458131431e-05,
      "loss": 0.1563,
      "step": 1959
    },
    {
      "epoch": 0.41141897565071367,
      "grad_norm": 0.1340368390083313,
      "learning_rate": 7.326734891304745e-05,
      "loss": 0.1559,
      "step": 1960
    },
    {
      "epoch": 0.41162888329135183,
      "grad_norm": 0.12216036021709442,
      "learning_rate": 7.323491074968275e-05,
      "loss": 0.1604,
      "step": 1961
    },
    {
      "epoch": 0.41183879093198994,
      "grad_norm": 0.11884403228759766,
      "learning_rate": 7.320246010864023e-05,
      "loss": 0.1572,
      "step": 1962
    },
    {
      "epoch": 0.41204869857262805,
      "grad_norm": 0.11110524088144302,
      "learning_rate": 7.316999700734663e-05,
      "loss": 0.1606,
      "step": 1963
    },
    {
      "epoch": 0.41225860621326615,
      "grad_norm": 0.12798446416854858,
      "learning_rate": 7.313752146323539e-05,
      "loss": 0.1625,
      "step": 1964
    },
    {
      "epoch": 0.41246851385390426,
      "grad_norm": 0.11968263983726501,
      "learning_rate": 7.31050334937466e-05,
      "loss": 0.1749,
      "step": 1965
    },
    {
      "epoch": 0.4126784214945424,
      "grad_norm": 0.10930939763784409,
      "learning_rate": 7.307253311632704e-05,
      "loss": 0.1722,
      "step": 1966
    },
    {
      "epoch": 0.41288832913518053,
      "grad_norm": 0.12315979599952698,
      "learning_rate": 7.304002034843015e-05,
      "loss": 0.1776,
      "step": 1967
    },
    {
      "epoch": 0.41309823677581864,
      "grad_norm": 0.1320822834968567,
      "learning_rate": 7.300749520751608e-05,
      "loss": 0.1539,
      "step": 1968
    },
    {
      "epoch": 0.41330814441645675,
      "grad_norm": 0.12075715512037277,
      "learning_rate": 7.297495771105152e-05,
      "loss": 0.1414,
      "step": 1969
    },
    {
      "epoch": 0.41351805205709485,
      "grad_norm": 0.1056932806968689,
      "learning_rate": 7.294240787650986e-05,
      "loss": 0.1485,
      "step": 1970
    },
    {
      "epoch": 0.413727959697733,
      "grad_norm": 0.10487929731607437,
      "learning_rate": 7.290984572137111e-05,
      "loss": 0.1728,
      "step": 1971
    },
    {
      "epoch": 0.4139378673383711,
      "grad_norm": 0.10982023179531097,
      "learning_rate": 7.287727126312191e-05,
      "loss": 0.1608,
      "step": 1972
    },
    {
      "epoch": 0.41414777497900923,
      "grad_norm": 0.09598106890916824,
      "learning_rate": 7.284468451925546e-05,
      "loss": 0.1503,
      "step": 1973
    },
    {
      "epoch": 0.41435768261964734,
      "grad_norm": 0.14356835186481476,
      "learning_rate": 7.281208550727159e-05,
      "loss": 0.1495,
      "step": 1974
    },
    {
      "epoch": 0.4145675902602855,
      "grad_norm": 0.10420846194028854,
      "learning_rate": 7.277947424467672e-05,
      "loss": 0.1586,
      "step": 1975
    },
    {
      "epoch": 0.4147774979009236,
      "grad_norm": 0.12487095594406128,
      "learning_rate": 7.274685074898386e-05,
      "loss": 0.175,
      "step": 1976
    },
    {
      "epoch": 0.4149874055415617,
      "grad_norm": 0.11194780468940735,
      "learning_rate": 7.271421503771251e-05,
      "loss": 0.1574,
      "step": 1977
    },
    {
      "epoch": 0.4151973131821998,
      "grad_norm": 0.12557700276374817,
      "learning_rate": 7.268156712838888e-05,
      "loss": 0.1529,
      "step": 1978
    },
    {
      "epoch": 0.41540722082283793,
      "grad_norm": 0.10330946743488312,
      "learning_rate": 7.264890703854559e-05,
      "loss": 0.1311,
      "step": 1979
    },
    {
      "epoch": 0.4156171284634761,
      "grad_norm": 0.11576497554779053,
      "learning_rate": 7.261623478572185e-05,
      "loss": 0.1518,
      "step": 1980
    },
    {
      "epoch": 0.4158270361041142,
      "grad_norm": 0.1512029618024826,
      "learning_rate": 7.258355038746345e-05,
      "loss": 0.1629,
      "step": 1981
    },
    {
      "epoch": 0.4160369437447523,
      "grad_norm": 0.142082080245018,
      "learning_rate": 7.25508538613226e-05,
      "loss": 0.1607,
      "step": 1982
    },
    {
      "epoch": 0.4162468513853904,
      "grad_norm": 0.13623762130737305,
      "learning_rate": 7.251814522485813e-05,
      "loss": 0.1695,
      "step": 1983
    },
    {
      "epoch": 0.4164567590260285,
      "grad_norm": 0.1192348450422287,
      "learning_rate": 7.248542449563529e-05,
      "loss": 0.1668,
      "step": 1984
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 0.12478368729352951,
      "learning_rate": 7.245269169122588e-05,
      "loss": 0.1427,
      "step": 1985
    },
    {
      "epoch": 0.4168765743073048,
      "grad_norm": 0.1585802137851715,
      "learning_rate": 7.241994682920816e-05,
      "loss": 0.1778,
      "step": 1986
    },
    {
      "epoch": 0.4170864819479429,
      "grad_norm": 0.10179055482149124,
      "learning_rate": 7.238718992716687e-05,
      "loss": 0.1358,
      "step": 1987
    },
    {
      "epoch": 0.417296389588581,
      "grad_norm": 0.11688593029975891,
      "learning_rate": 7.23544210026932e-05,
      "loss": 0.1603,
      "step": 1988
    },
    {
      "epoch": 0.41750629722921917,
      "grad_norm": 0.1158967837691307,
      "learning_rate": 7.232164007338484e-05,
      "loss": 0.1408,
      "step": 1989
    },
    {
      "epoch": 0.4177162048698573,
      "grad_norm": 0.1265357881784439,
      "learning_rate": 7.228884715684588e-05,
      "loss": 0.1402,
      "step": 1990
    },
    {
      "epoch": 0.4179261125104954,
      "grad_norm": 0.10598758608102798,
      "learning_rate": 7.225604227068684e-05,
      "loss": 0.1431,
      "step": 1991
    },
    {
      "epoch": 0.4181360201511335,
      "grad_norm": 0.12342094630002975,
      "learning_rate": 7.222322543252474e-05,
      "loss": 0.1539,
      "step": 1992
    },
    {
      "epoch": 0.4183459277917716,
      "grad_norm": 0.12338914722204208,
      "learning_rate": 7.219039665998295e-05,
      "loss": 0.1719,
      "step": 1993
    },
    {
      "epoch": 0.41855583543240976,
      "grad_norm": 0.11439438164234161,
      "learning_rate": 7.215755597069126e-05,
      "loss": 0.1639,
      "step": 1994
    },
    {
      "epoch": 0.41876574307304787,
      "grad_norm": 0.12324292212724686,
      "learning_rate": 7.212470338228589e-05,
      "loss": 0.1759,
      "step": 1995
    },
    {
      "epoch": 0.418975650713686,
      "grad_norm": 0.10522071272134781,
      "learning_rate": 7.209183891240941e-05,
      "loss": 0.1475,
      "step": 1996
    },
    {
      "epoch": 0.4191855583543241,
      "grad_norm": 0.09488867968320847,
      "learning_rate": 7.205896257871082e-05,
      "loss": 0.1564,
      "step": 1997
    },
    {
      "epoch": 0.4193954659949622,
      "grad_norm": 0.12088136374950409,
      "learning_rate": 7.202607439884543e-05,
      "loss": 0.1671,
      "step": 1998
    },
    {
      "epoch": 0.41960537363560035,
      "grad_norm": 0.11763715744018555,
      "learning_rate": 7.199317439047499e-05,
      "loss": 0.1463,
      "step": 1999
    },
    {
      "epoch": 0.41981528127623846,
      "grad_norm": 0.10898829251527786,
      "learning_rate": 7.196026257126749e-05,
      "loss": 0.1697,
      "step": 2000
    },
    {
      "epoch": 0.42002518891687657,
      "grad_norm": 0.114218570291996,
      "learning_rate": 7.192733895889737e-05,
      "loss": 0.1693,
      "step": 2001
    },
    {
      "epoch": 0.4202350965575147,
      "grad_norm": 0.11072574555873871,
      "learning_rate": 7.189440357104536e-05,
      "loss": 0.1599,
      "step": 2002
    },
    {
      "epoch": 0.42044500419815284,
      "grad_norm": 0.118053138256073,
      "learning_rate": 7.186145642539852e-05,
      "loss": 0.1697,
      "step": 2003
    },
    {
      "epoch": 0.42065491183879095,
      "grad_norm": 0.11998876184225082,
      "learning_rate": 7.18284975396502e-05,
      "loss": 0.1666,
      "step": 2004
    },
    {
      "epoch": 0.42086481947942905,
      "grad_norm": 0.12326207011938095,
      "learning_rate": 7.179552693150009e-05,
      "loss": 0.137,
      "step": 2005
    },
    {
      "epoch": 0.42107472712006716,
      "grad_norm": 0.11958475410938263,
      "learning_rate": 7.176254461865415e-05,
      "loss": 0.1571,
      "step": 2006
    },
    {
      "epoch": 0.42128463476070527,
      "grad_norm": 0.1109607145190239,
      "learning_rate": 7.172955061882465e-05,
      "loss": 0.1359,
      "step": 2007
    },
    {
      "epoch": 0.42149454240134343,
      "grad_norm": 0.1003115177154541,
      "learning_rate": 7.16965449497301e-05,
      "loss": 0.1643,
      "step": 2008
    },
    {
      "epoch": 0.42170445004198154,
      "grad_norm": 0.10504936426877975,
      "learning_rate": 7.166352762909532e-05,
      "loss": 0.1469,
      "step": 2009
    },
    {
      "epoch": 0.42191435768261965,
      "grad_norm": 0.11309293657541275,
      "learning_rate": 7.163049867465135e-05,
      "loss": 0.1474,
      "step": 2010
    },
    {
      "epoch": 0.42212426532325775,
      "grad_norm": 0.10572872310876846,
      "learning_rate": 7.15974581041355e-05,
      "loss": 0.1622,
      "step": 2011
    },
    {
      "epoch": 0.42233417296389586,
      "grad_norm": 0.10934752225875854,
      "learning_rate": 7.156440593529132e-05,
      "loss": 0.162,
      "step": 2012
    },
    {
      "epoch": 0.422544080604534,
      "grad_norm": 0.12137974798679352,
      "learning_rate": 7.153134218586857e-05,
      "loss": 0.1438,
      "step": 2013
    },
    {
      "epoch": 0.42275398824517213,
      "grad_norm": 0.12691549956798553,
      "learning_rate": 7.149826687362322e-05,
      "loss": 0.1494,
      "step": 2014
    },
    {
      "epoch": 0.42296389588581024,
      "grad_norm": 0.12283293902873993,
      "learning_rate": 7.146518001631751e-05,
      "loss": 0.1578,
      "step": 2015
    },
    {
      "epoch": 0.42317380352644834,
      "grad_norm": 0.13865120708942413,
      "learning_rate": 7.143208163171979e-05,
      "loss": 0.1652,
      "step": 2016
    },
    {
      "epoch": 0.4233837111670865,
      "grad_norm": 0.11391118913888931,
      "learning_rate": 7.139897173760469e-05,
      "loss": 0.148,
      "step": 2017
    },
    {
      "epoch": 0.4235936188077246,
      "grad_norm": 0.12447668612003326,
      "learning_rate": 7.136585035175294e-05,
      "loss": 0.1672,
      "step": 2018
    },
    {
      "epoch": 0.4238035264483627,
      "grad_norm": 0.1198267936706543,
      "learning_rate": 7.133271749195153e-05,
      "loss": 0.1572,
      "step": 2019
    },
    {
      "epoch": 0.42401343408900083,
      "grad_norm": 0.10948833078145981,
      "learning_rate": 7.129957317599352e-05,
      "loss": 0.1612,
      "step": 2020
    },
    {
      "epoch": 0.42422334172963894,
      "grad_norm": 0.12680894136428833,
      "learning_rate": 7.126641742167819e-05,
      "loss": 0.1648,
      "step": 2021
    },
    {
      "epoch": 0.4244332493702771,
      "grad_norm": 0.14302381873130798,
      "learning_rate": 7.123325024681093e-05,
      "loss": 0.1617,
      "step": 2022
    },
    {
      "epoch": 0.4246431570109152,
      "grad_norm": 0.13583184778690338,
      "learning_rate": 7.120007166920326e-05,
      "loss": 0.1666,
      "step": 2023
    },
    {
      "epoch": 0.4248530646515533,
      "grad_norm": 0.11573361605405807,
      "learning_rate": 7.116688170667284e-05,
      "loss": 0.1523,
      "step": 2024
    },
    {
      "epoch": 0.4250629722921914,
      "grad_norm": 0.13895514607429504,
      "learning_rate": 7.113368037704345e-05,
      "loss": 0.158,
      "step": 2025
    },
    {
      "epoch": 0.42527287993282953,
      "grad_norm": 0.1131201833486557,
      "learning_rate": 7.110046769814496e-05,
      "loss": 0.1764,
      "step": 2026
    },
    {
      "epoch": 0.4254827875734677,
      "grad_norm": 0.13642175495624542,
      "learning_rate": 7.106724368781331e-05,
      "loss": 0.1797,
      "step": 2027
    },
    {
      "epoch": 0.4256926952141058,
      "grad_norm": 0.11173267662525177,
      "learning_rate": 7.103400836389059e-05,
      "loss": 0.1685,
      "step": 2028
    },
    {
      "epoch": 0.4259026028547439,
      "grad_norm": 0.11620894074440002,
      "learning_rate": 7.10007617442249e-05,
      "loss": 0.1552,
      "step": 2029
    },
    {
      "epoch": 0.426112510495382,
      "grad_norm": 0.11680643260478973,
      "learning_rate": 7.096750384667044e-05,
      "loss": 0.1576,
      "step": 2030
    },
    {
      "epoch": 0.4263224181360202,
      "grad_norm": 0.13483844697475433,
      "learning_rate": 7.093423468908748e-05,
      "loss": 0.1609,
      "step": 2031
    },
    {
      "epoch": 0.4265323257766583,
      "grad_norm": 0.11245349794626236,
      "learning_rate": 7.09009542893423e-05,
      "loss": 0.165,
      "step": 2032
    },
    {
      "epoch": 0.4267422334172964,
      "grad_norm": 0.1436799019575119,
      "learning_rate": 7.086766266530723e-05,
      "loss": 0.1439,
      "step": 2033
    },
    {
      "epoch": 0.4269521410579345,
      "grad_norm": 0.11214408278465271,
      "learning_rate": 7.083435983486066e-05,
      "loss": 0.1678,
      "step": 2034
    },
    {
      "epoch": 0.4271620486985726,
      "grad_norm": 0.11312124133110046,
      "learning_rate": 7.080104581588693e-05,
      "loss": 0.1683,
      "step": 2035
    },
    {
      "epoch": 0.42737195633921077,
      "grad_norm": 0.12795935571193695,
      "learning_rate": 7.076772062627647e-05,
      "loss": 0.1657,
      "step": 2036
    },
    {
      "epoch": 0.4275818639798489,
      "grad_norm": 0.10281221568584442,
      "learning_rate": 7.073438428392562e-05,
      "loss": 0.1394,
      "step": 2037
    },
    {
      "epoch": 0.427791771620487,
      "grad_norm": 0.10964559763669968,
      "learning_rate": 7.070103680673683e-05,
      "loss": 0.1598,
      "step": 2038
    },
    {
      "epoch": 0.4280016792611251,
      "grad_norm": 0.10823410749435425,
      "learning_rate": 7.066767821261837e-05,
      "loss": 0.1601,
      "step": 2039
    },
    {
      "epoch": 0.4282115869017632,
      "grad_norm": 0.11483200639486313,
      "learning_rate": 7.063430851948463e-05,
      "loss": 0.14,
      "step": 2040
    },
    {
      "epoch": 0.42842149454240136,
      "grad_norm": 0.1289099007844925,
      "learning_rate": 7.060092774525588e-05,
      "loss": 0.1567,
      "step": 2041
    },
    {
      "epoch": 0.42863140218303947,
      "grad_norm": 0.11432711780071259,
      "learning_rate": 7.056753590785835e-05,
      "loss": 0.1559,
      "step": 2042
    },
    {
      "epoch": 0.4288413098236776,
      "grad_norm": 0.13700692355632782,
      "learning_rate": 7.053413302522423e-05,
      "loss": 0.1716,
      "step": 2043
    },
    {
      "epoch": 0.4290512174643157,
      "grad_norm": 0.12998148798942566,
      "learning_rate": 7.050071911529163e-05,
      "loss": 0.1539,
      "step": 2044
    },
    {
      "epoch": 0.42926112510495384,
      "grad_norm": 0.1319163739681244,
      "learning_rate": 7.046729419600457e-05,
      "loss": 0.1638,
      "step": 2045
    },
    {
      "epoch": 0.42947103274559195,
      "grad_norm": 0.1296456903219223,
      "learning_rate": 7.043385828531299e-05,
      "loss": 0.1765,
      "step": 2046
    },
    {
      "epoch": 0.42968094038623006,
      "grad_norm": 0.133979931473732,
      "learning_rate": 7.040041140117278e-05,
      "loss": 0.1721,
      "step": 2047
    },
    {
      "epoch": 0.42989084802686817,
      "grad_norm": 0.1405065357685089,
      "learning_rate": 7.036695356154564e-05,
      "loss": 0.1552,
      "step": 2048
    },
    {
      "epoch": 0.4301007556675063,
      "grad_norm": 0.13646145164966583,
      "learning_rate": 7.033348478439921e-05,
      "loss": 0.1515,
      "step": 2049
    },
    {
      "epoch": 0.43031066330814444,
      "grad_norm": 0.1469520926475525,
      "learning_rate": 7.030000508770698e-05,
      "loss": 0.1447,
      "step": 2050
    },
    {
      "epoch": 0.43052057094878254,
      "grad_norm": 0.13606610894203186,
      "learning_rate": 7.026651448944833e-05,
      "loss": 0.1728,
      "step": 2051
    },
    {
      "epoch": 0.43073047858942065,
      "grad_norm": 0.14016586542129517,
      "learning_rate": 7.023301300760843e-05,
      "loss": 0.161,
      "step": 2052
    },
    {
      "epoch": 0.43094038623005876,
      "grad_norm": 0.12709060311317444,
      "learning_rate": 7.019950066017839e-05,
      "loss": 0.1502,
      "step": 2053
    },
    {
      "epoch": 0.43115029387069687,
      "grad_norm": 0.11851731687784195,
      "learning_rate": 7.016597746515511e-05,
      "loss": 0.1493,
      "step": 2054
    },
    {
      "epoch": 0.43136020151133503,
      "grad_norm": 0.16392657160758972,
      "learning_rate": 7.013244344054127e-05,
      "loss": 0.1776,
      "step": 2055
    },
    {
      "epoch": 0.43157010915197314,
      "grad_norm": 0.122653529047966,
      "learning_rate": 7.009889860434544e-05,
      "loss": 0.1408,
      "step": 2056
    },
    {
      "epoch": 0.43178001679261124,
      "grad_norm": 0.12323731184005737,
      "learning_rate": 7.006534297458195e-05,
      "loss": 0.1467,
      "step": 2057
    },
    {
      "epoch": 0.43198992443324935,
      "grad_norm": 0.12453500181436539,
      "learning_rate": 7.003177656927095e-05,
      "loss": 0.1561,
      "step": 2058
    },
    {
      "epoch": 0.4321998320738875,
      "grad_norm": 0.12548905611038208,
      "learning_rate": 6.999819940643833e-05,
      "loss": 0.1551,
      "step": 2059
    },
    {
      "epoch": 0.4324097397145256,
      "grad_norm": 0.10663434118032455,
      "learning_rate": 6.996461150411584e-05,
      "loss": 0.1586,
      "step": 2060
    },
    {
      "epoch": 0.43261964735516373,
      "grad_norm": 0.11009252071380615,
      "learning_rate": 6.993101288034094e-05,
      "loss": 0.1428,
      "step": 2061
    },
    {
      "epoch": 0.43282955499580184,
      "grad_norm": 0.11233685165643692,
      "learning_rate": 6.989740355315683e-05,
      "loss": 0.1484,
      "step": 2062
    },
    {
      "epoch": 0.43303946263643994,
      "grad_norm": 0.12727439403533936,
      "learning_rate": 6.986378354061254e-05,
      "loss": 0.161,
      "step": 2063
    },
    {
      "epoch": 0.4332493702770781,
      "grad_norm": 0.13386842608451843,
      "learning_rate": 6.983015286076272e-05,
      "loss": 0.1355,
      "step": 2064
    },
    {
      "epoch": 0.4334592779177162,
      "grad_norm": 0.13135772943496704,
      "learning_rate": 6.979651153166786e-05,
      "loss": 0.1719,
      "step": 2065
    },
    {
      "epoch": 0.4336691855583543,
      "grad_norm": 0.12024439871311188,
      "learning_rate": 6.97628595713941e-05,
      "loss": 0.1664,
      "step": 2066
    },
    {
      "epoch": 0.4338790931989924,
      "grad_norm": 0.10455571860074997,
      "learning_rate": 6.972919699801333e-05,
      "loss": 0.1484,
      "step": 2067
    },
    {
      "epoch": 0.4340890008396306,
      "grad_norm": 0.15053103864192963,
      "learning_rate": 6.96955238296031e-05,
      "loss": 0.168,
      "step": 2068
    },
    {
      "epoch": 0.4342989084802687,
      "grad_norm": 0.11767338216304779,
      "learning_rate": 6.966184008424668e-05,
      "loss": 0.1642,
      "step": 2069
    },
    {
      "epoch": 0.4345088161209068,
      "grad_norm": 0.11430288106203079,
      "learning_rate": 6.9628145780033e-05,
      "loss": 0.1595,
      "step": 2070
    },
    {
      "epoch": 0.4347187237615449,
      "grad_norm": 0.1504775881767273,
      "learning_rate": 6.959444093505671e-05,
      "loss": 0.1553,
      "step": 2071
    },
    {
      "epoch": 0.434928631402183,
      "grad_norm": 0.12450301647186279,
      "learning_rate": 6.956072556741804e-05,
      "loss": 0.1516,
      "step": 2072
    },
    {
      "epoch": 0.4351385390428212,
      "grad_norm": 0.15835943818092346,
      "learning_rate": 6.952699969522292e-05,
      "loss": 0.1733,
      "step": 2073
    },
    {
      "epoch": 0.4353484466834593,
      "grad_norm": 0.1206754669547081,
      "learning_rate": 6.949326333658292e-05,
      "loss": 0.1613,
      "step": 2074
    },
    {
      "epoch": 0.4355583543240974,
      "grad_norm": 0.10483954846858978,
      "learning_rate": 6.945951650961524e-05,
      "loss": 0.1635,
      "step": 2075
    },
    {
      "epoch": 0.4357682619647355,
      "grad_norm": 0.11579787731170654,
      "learning_rate": 6.942575923244268e-05,
      "loss": 0.1521,
      "step": 2076
    },
    {
      "epoch": 0.4359781696053736,
      "grad_norm": 0.16092592477798462,
      "learning_rate": 6.93919915231937e-05,
      "loss": 0.1611,
      "step": 2077
    },
    {
      "epoch": 0.4361880772460118,
      "grad_norm": 0.11030247807502747,
      "learning_rate": 6.935821340000231e-05,
      "loss": 0.1465,
      "step": 2078
    },
    {
      "epoch": 0.4363979848866499,
      "grad_norm": 0.11840491741895676,
      "learning_rate": 6.932442488100813e-05,
      "loss": 0.1565,
      "step": 2079
    },
    {
      "epoch": 0.436607892527288,
      "grad_norm": 0.14345033466815948,
      "learning_rate": 6.929062598435636e-05,
      "loss": 0.1655,
      "step": 2080
    },
    {
      "epoch": 0.4368178001679261,
      "grad_norm": 0.12973609566688538,
      "learning_rate": 6.925681672819781e-05,
      "loss": 0.131,
      "step": 2081
    },
    {
      "epoch": 0.43702770780856426,
      "grad_norm": 0.11152931302785873,
      "learning_rate": 6.922299713068878e-05,
      "loss": 0.1545,
      "step": 2082
    },
    {
      "epoch": 0.43723761544920237,
      "grad_norm": 0.11546137928962708,
      "learning_rate": 6.918916720999123e-05,
      "loss": 0.1683,
      "step": 2083
    },
    {
      "epoch": 0.4374475230898405,
      "grad_norm": 0.11870436370372772,
      "learning_rate": 6.915532698427254e-05,
      "loss": 0.1493,
      "step": 2084
    },
    {
      "epoch": 0.4376574307304786,
      "grad_norm": 0.09569094330072403,
      "learning_rate": 6.912147647170571e-05,
      "loss": 0.1415,
      "step": 2085
    },
    {
      "epoch": 0.4378673383711167,
      "grad_norm": 0.1221393570303917,
      "learning_rate": 6.908761569046923e-05,
      "loss": 0.1463,
      "step": 2086
    },
    {
      "epoch": 0.43807724601175485,
      "grad_norm": 0.12907950580120087,
      "learning_rate": 6.90537446587471e-05,
      "loss": 0.1615,
      "step": 2087
    },
    {
      "epoch": 0.43828715365239296,
      "grad_norm": 0.12217225879430771,
      "learning_rate": 6.901986339472888e-05,
      "loss": 0.1592,
      "step": 2088
    },
    {
      "epoch": 0.43849706129303107,
      "grad_norm": 0.12174089252948761,
      "learning_rate": 6.898597191660956e-05,
      "loss": 0.1526,
      "step": 2089
    },
    {
      "epoch": 0.4387069689336692,
      "grad_norm": 0.11348958313465118,
      "learning_rate": 6.89520702425896e-05,
      "loss": 0.157,
      "step": 2090
    },
    {
      "epoch": 0.4389168765743073,
      "grad_norm": 0.11617136001586914,
      "learning_rate": 6.891815839087505e-05,
      "loss": 0.1511,
      "step": 2091
    },
    {
      "epoch": 0.43912678421494544,
      "grad_norm": 0.11774633824825287,
      "learning_rate": 6.888423637967728e-05,
      "loss": 0.1561,
      "step": 2092
    },
    {
      "epoch": 0.43933669185558355,
      "grad_norm": 0.11033912003040314,
      "learning_rate": 6.885030422721324e-05,
      "loss": 0.1508,
      "step": 2093
    },
    {
      "epoch": 0.43954659949622166,
      "grad_norm": 0.12567701935768127,
      "learning_rate": 6.881636195170521e-05,
      "loss": 0.1489,
      "step": 2094
    },
    {
      "epoch": 0.43975650713685976,
      "grad_norm": 0.1044977456331253,
      "learning_rate": 6.878240957138102e-05,
      "loss": 0.1499,
      "step": 2095
    },
    {
      "epoch": 0.4399664147774979,
      "grad_norm": 0.1258642077445984,
      "learning_rate": 6.874844710447386e-05,
      "loss": 0.1473,
      "step": 2096
    },
    {
      "epoch": 0.44017632241813603,
      "grad_norm": 0.11813251674175262,
      "learning_rate": 6.871447456922233e-05,
      "loss": 0.1404,
      "step": 2097
    },
    {
      "epoch": 0.44038623005877414,
      "grad_norm": 0.11908237636089325,
      "learning_rate": 6.868049198387047e-05,
      "loss": 0.1612,
      "step": 2098
    },
    {
      "epoch": 0.44059613769941225,
      "grad_norm": 0.1303856521844864,
      "learning_rate": 6.864649936666772e-05,
      "loss": 0.1722,
      "step": 2099
    },
    {
      "epoch": 0.44080604534005036,
      "grad_norm": 0.1194954439997673,
      "learning_rate": 6.861249673586887e-05,
      "loss": 0.1446,
      "step": 2100
    },
    {
      "epoch": 0.4410159529806885,
      "grad_norm": 0.1111513301730156,
      "learning_rate": 6.857848410973413e-05,
      "loss": 0.1462,
      "step": 2101
    },
    {
      "epoch": 0.4412258606213266,
      "grad_norm": 0.11029879748821259,
      "learning_rate": 6.854446150652905e-05,
      "loss": 0.1533,
      "step": 2102
    },
    {
      "epoch": 0.44143576826196473,
      "grad_norm": 0.1370946615934372,
      "learning_rate": 6.851042894452452e-05,
      "loss": 0.1477,
      "step": 2103
    },
    {
      "epoch": 0.44164567590260284,
      "grad_norm": 0.09855300188064575,
      "learning_rate": 6.847638644199684e-05,
      "loss": 0.1449,
      "step": 2104
    },
    {
      "epoch": 0.44185558354324095,
      "grad_norm": 0.11461108922958374,
      "learning_rate": 6.844233401722757e-05,
      "loss": 0.1602,
      "step": 2105
    },
    {
      "epoch": 0.4420654911838791,
      "grad_norm": 0.11520881950855255,
      "learning_rate": 6.840827168850368e-05,
      "loss": 0.1571,
      "step": 2106
    },
    {
      "epoch": 0.4422753988245172,
      "grad_norm": 0.11658153682947159,
      "learning_rate": 6.837419947411742e-05,
      "loss": 0.1533,
      "step": 2107
    },
    {
      "epoch": 0.4424853064651553,
      "grad_norm": 0.1229148879647255,
      "learning_rate": 6.834011739236632e-05,
      "loss": 0.1479,
      "step": 2108
    },
    {
      "epoch": 0.44269521410579343,
      "grad_norm": 0.12906445562839508,
      "learning_rate": 6.830602546155324e-05,
      "loss": 0.1719,
      "step": 2109
    },
    {
      "epoch": 0.4429051217464316,
      "grad_norm": 0.10934906452894211,
      "learning_rate": 6.827192369998636e-05,
      "loss": 0.1503,
      "step": 2110
    },
    {
      "epoch": 0.4431150293870697,
      "grad_norm": 0.12860219180583954,
      "learning_rate": 6.823781212597903e-05,
      "loss": 0.1578,
      "step": 2111
    },
    {
      "epoch": 0.4433249370277078,
      "grad_norm": 0.1294265240430832,
      "learning_rate": 6.820369075785003e-05,
      "loss": 0.1538,
      "step": 2112
    },
    {
      "epoch": 0.4435348446683459,
      "grad_norm": 0.10323406755924225,
      "learning_rate": 6.816955961392327e-05,
      "loss": 0.1406,
      "step": 2113
    },
    {
      "epoch": 0.443744752308984,
      "grad_norm": 0.10694414377212524,
      "learning_rate": 6.813541871252797e-05,
      "loss": 0.1636,
      "step": 2114
    },
    {
      "epoch": 0.4439546599496222,
      "grad_norm": 0.13600675761699677,
      "learning_rate": 6.810126807199854e-05,
      "loss": 0.1454,
      "step": 2115
    },
    {
      "epoch": 0.4441645675902603,
      "grad_norm": 0.12228574603796005,
      "learning_rate": 6.806710771067469e-05,
      "loss": 0.1637,
      "step": 2116
    },
    {
      "epoch": 0.4443744752308984,
      "grad_norm": 0.12474840879440308,
      "learning_rate": 6.803293764690131e-05,
      "loss": 0.1626,
      "step": 2117
    },
    {
      "epoch": 0.4445843828715365,
      "grad_norm": 0.1158130019903183,
      "learning_rate": 6.799875789902848e-05,
      "loss": 0.1538,
      "step": 2118
    },
    {
      "epoch": 0.4447942905121746,
      "grad_norm": 0.11266588419675827,
      "learning_rate": 6.796456848541152e-05,
      "loss": 0.1459,
      "step": 2119
    },
    {
      "epoch": 0.4450041981528128,
      "grad_norm": 0.12051046639680862,
      "learning_rate": 6.793036942441095e-05,
      "loss": 0.1488,
      "step": 2120
    },
    {
      "epoch": 0.4452141057934509,
      "grad_norm": 0.12589138746261597,
      "learning_rate": 6.789616073439243e-05,
      "loss": 0.1652,
      "step": 2121
    },
    {
      "epoch": 0.445424013434089,
      "grad_norm": 0.1199890673160553,
      "learning_rate": 6.78619424337268e-05,
      "loss": 0.1563,
      "step": 2122
    },
    {
      "epoch": 0.4456339210747271,
      "grad_norm": 0.11333800107240677,
      "learning_rate": 6.782771454079006e-05,
      "loss": 0.1578,
      "step": 2123
    },
    {
      "epoch": 0.44584382871536526,
      "grad_norm": 0.1303377002477646,
      "learning_rate": 6.77934770739634e-05,
      "loss": 0.1461,
      "step": 2124
    },
    {
      "epoch": 0.44605373635600337,
      "grad_norm": 0.15686602890491486,
      "learning_rate": 6.775923005163308e-05,
      "loss": 0.148,
      "step": 2125
    },
    {
      "epoch": 0.4462636439966415,
      "grad_norm": 0.11671863496303558,
      "learning_rate": 6.772497349219058e-05,
      "loss": 0.1668,
      "step": 2126
    },
    {
      "epoch": 0.4464735516372796,
      "grad_norm": 0.11080017685890198,
      "learning_rate": 6.769070741403242e-05,
      "loss": 0.1674,
      "step": 2127
    },
    {
      "epoch": 0.4466834592779177,
      "grad_norm": 0.10152662545442581,
      "learning_rate": 6.76564318355603e-05,
      "loss": 0.1476,
      "step": 2128
    },
    {
      "epoch": 0.44689336691855586,
      "grad_norm": 0.11543727666139603,
      "learning_rate": 6.762214677518096e-05,
      "loss": 0.154,
      "step": 2129
    },
    {
      "epoch": 0.44710327455919396,
      "grad_norm": 0.1334647685289383,
      "learning_rate": 6.758785225130626e-05,
      "loss": 0.1769,
      "step": 2130
    },
    {
      "epoch": 0.44731318219983207,
      "grad_norm": 0.1134229451417923,
      "learning_rate": 6.755354828235315e-05,
      "loss": 0.1508,
      "step": 2131
    },
    {
      "epoch": 0.4475230898404702,
      "grad_norm": 0.11326369643211365,
      "learning_rate": 6.751923488674366e-05,
      "loss": 0.1585,
      "step": 2132
    },
    {
      "epoch": 0.4477329974811083,
      "grad_norm": 0.12377768754959106,
      "learning_rate": 6.748491208290484e-05,
      "loss": 0.1558,
      "step": 2133
    },
    {
      "epoch": 0.44794290512174645,
      "grad_norm": 0.1556466668844223,
      "learning_rate": 6.745057988926886e-05,
      "loss": 0.1517,
      "step": 2134
    },
    {
      "epoch": 0.44815281276238456,
      "grad_norm": 0.11329285055398941,
      "learning_rate": 6.741623832427284e-05,
      "loss": 0.1695,
      "step": 2135
    },
    {
      "epoch": 0.44836272040302266,
      "grad_norm": 0.12504523992538452,
      "learning_rate": 6.738188740635902e-05,
      "loss": 0.1459,
      "step": 2136
    },
    {
      "epoch": 0.44857262804366077,
      "grad_norm": 0.10221695154905319,
      "learning_rate": 6.734752715397461e-05,
      "loss": 0.1462,
      "step": 2137
    },
    {
      "epoch": 0.44878253568429893,
      "grad_norm": 0.1422422230243683,
      "learning_rate": 6.731315758557187e-05,
      "loss": 0.1565,
      "step": 2138
    },
    {
      "epoch": 0.44899244332493704,
      "grad_norm": 0.12211299687623978,
      "learning_rate": 6.727877871960803e-05,
      "loss": 0.156,
      "step": 2139
    },
    {
      "epoch": 0.44920235096557515,
      "grad_norm": 0.11609075218439102,
      "learning_rate": 6.724439057454533e-05,
      "loss": 0.1679,
      "step": 2140
    },
    {
      "epoch": 0.44941225860621326,
      "grad_norm": 0.1233009323477745,
      "learning_rate": 6.720999316885098e-05,
      "loss": 0.1694,
      "step": 2141
    },
    {
      "epoch": 0.44962216624685136,
      "grad_norm": 0.1292707473039627,
      "learning_rate": 6.717558652099718e-05,
      "loss": 0.1669,
      "step": 2142
    },
    {
      "epoch": 0.4498320738874895,
      "grad_norm": 0.1204388439655304,
      "learning_rate": 6.71411706494611e-05,
      "loss": 0.1632,
      "step": 2143
    },
    {
      "epoch": 0.45004198152812763,
      "grad_norm": 0.12831495702266693,
      "learning_rate": 6.710674557272482e-05,
      "loss": 0.1447,
      "step": 2144
    },
    {
      "epoch": 0.45025188916876574,
      "grad_norm": 0.14179843664169312,
      "learning_rate": 6.707231130927542e-05,
      "loss": 0.159,
      "step": 2145
    },
    {
      "epoch": 0.45046179680940385,
      "grad_norm": 0.11582005023956299,
      "learning_rate": 6.703786787760487e-05,
      "loss": 0.1608,
      "step": 2146
    },
    {
      "epoch": 0.45067170445004195,
      "grad_norm": 0.11273009330034256,
      "learning_rate": 6.700341529621008e-05,
      "loss": 0.1467,
      "step": 2147
    },
    {
      "epoch": 0.4508816120906801,
      "grad_norm": 0.14487546682357788,
      "learning_rate": 6.696895358359288e-05,
      "loss": 0.1618,
      "step": 2148
    },
    {
      "epoch": 0.4510915197313182,
      "grad_norm": 0.11221778392791748,
      "learning_rate": 6.693448275826e-05,
      "loss": 0.155,
      "step": 2149
    },
    {
      "epoch": 0.45130142737195633,
      "grad_norm": 0.11724277585744858,
      "learning_rate": 6.690000283872304e-05,
      "loss": 0.1535,
      "step": 2150
    },
    {
      "epoch": 0.45151133501259444,
      "grad_norm": 0.13033629953861237,
      "learning_rate": 6.686551384349853e-05,
      "loss": 0.1818,
      "step": 2151
    },
    {
      "epoch": 0.4517212426532326,
      "grad_norm": 0.10972466319799423,
      "learning_rate": 6.683101579110783e-05,
      "loss": 0.1362,
      "step": 2152
    },
    {
      "epoch": 0.4519311502938707,
      "grad_norm": 0.12394952774047852,
      "learning_rate": 6.679650870007718e-05,
      "loss": 0.1563,
      "step": 2153
    },
    {
      "epoch": 0.4521410579345088,
      "grad_norm": 0.13337190449237823,
      "learning_rate": 6.67619925889377e-05,
      "loss": 0.1643,
      "step": 2154
    },
    {
      "epoch": 0.4523509655751469,
      "grad_norm": 0.10738395899534225,
      "learning_rate": 6.672746747622532e-05,
      "loss": 0.1372,
      "step": 2155
    },
    {
      "epoch": 0.45256087321578503,
      "grad_norm": 0.1167917475104332,
      "learning_rate": 6.669293338048079e-05,
      "loss": 0.1618,
      "step": 2156
    },
    {
      "epoch": 0.4527707808564232,
      "grad_norm": 0.12487170845270157,
      "learning_rate": 6.665839032024975e-05,
      "loss": 0.1601,
      "step": 2157
    },
    {
      "epoch": 0.4529806884970613,
      "grad_norm": 0.12744808197021484,
      "learning_rate": 6.662383831408258e-05,
      "loss": 0.1533,
      "step": 2158
    },
    {
      "epoch": 0.4531905961376994,
      "grad_norm": 0.113549143075943,
      "learning_rate": 6.658927738053451e-05,
      "loss": 0.1674,
      "step": 2159
    },
    {
      "epoch": 0.4534005037783375,
      "grad_norm": 0.12157785147428513,
      "learning_rate": 6.655470753816553e-05,
      "loss": 0.1648,
      "step": 2160
    },
    {
      "epoch": 0.4536104114189756,
      "grad_norm": 0.11910437792539597,
      "learning_rate": 6.652012880554047e-05,
      "loss": 0.1561,
      "step": 2161
    },
    {
      "epoch": 0.4538203190596138,
      "grad_norm": 0.12395650148391724,
      "learning_rate": 6.648554120122889e-05,
      "loss": 0.1508,
      "step": 2162
    },
    {
      "epoch": 0.4540302267002519,
      "grad_norm": 0.11354140937328339,
      "learning_rate": 6.64509447438051e-05,
      "loss": 0.1588,
      "step": 2163
    },
    {
      "epoch": 0.45424013434089,
      "grad_norm": 0.12797436118125916,
      "learning_rate": 6.64163394518482e-05,
      "loss": 0.166,
      "step": 2164
    },
    {
      "epoch": 0.4544500419815281,
      "grad_norm": 0.14938712120056152,
      "learning_rate": 6.638172534394204e-05,
      "loss": 0.1466,
      "step": 2165
    },
    {
      "epoch": 0.45465994962216627,
      "grad_norm": 0.12750832736492157,
      "learning_rate": 6.634710243867516e-05,
      "loss": 0.1657,
      "step": 2166
    },
    {
      "epoch": 0.4548698572628044,
      "grad_norm": 0.10431057959794998,
      "learning_rate": 6.631247075464086e-05,
      "loss": 0.1601,
      "step": 2167
    },
    {
      "epoch": 0.4550797649034425,
      "grad_norm": 0.12698952853679657,
      "learning_rate": 6.627783031043715e-05,
      "loss": 0.1571,
      "step": 2168
    },
    {
      "epoch": 0.4552896725440806,
      "grad_norm": 0.1288319081068039,
      "learning_rate": 6.624318112466676e-05,
      "loss": 0.1589,
      "step": 2169
    },
    {
      "epoch": 0.4554995801847187,
      "grad_norm": 0.12063757330179214,
      "learning_rate": 6.620852321593705e-05,
      "loss": 0.1415,
      "step": 2170
    },
    {
      "epoch": 0.45570948782535686,
      "grad_norm": 0.12932997941970825,
      "learning_rate": 6.617385660286014e-05,
      "loss": 0.1685,
      "step": 2171
    },
    {
      "epoch": 0.45591939546599497,
      "grad_norm": 0.12156279385089874,
      "learning_rate": 6.613918130405278e-05,
      "loss": 0.1682,
      "step": 2172
    },
    {
      "epoch": 0.4561293031066331,
      "grad_norm": 0.1219090148806572,
      "learning_rate": 6.61044973381364e-05,
      "loss": 0.1604,
      "step": 2173
    },
    {
      "epoch": 0.4563392107472712,
      "grad_norm": 0.10800813138484955,
      "learning_rate": 6.606980472373709e-05,
      "loss": 0.1334,
      "step": 2174
    },
    {
      "epoch": 0.45654911838790935,
      "grad_norm": 0.13099253177642822,
      "learning_rate": 6.603510347948556e-05,
      "loss": 0.1587,
      "step": 2175
    },
    {
      "epoch": 0.45675902602854745,
      "grad_norm": 0.1315564215183258,
      "learning_rate": 6.60003936240172e-05,
      "loss": 0.1729,
      "step": 2176
    },
    {
      "epoch": 0.45696893366918556,
      "grad_norm": 0.12284686416387558,
      "learning_rate": 6.596567517597196e-05,
      "loss": 0.1803,
      "step": 2177
    },
    {
      "epoch": 0.45717884130982367,
      "grad_norm": 0.1192544549703598,
      "learning_rate": 6.593094815399446e-05,
      "loss": 0.1685,
      "step": 2178
    },
    {
      "epoch": 0.4573887489504618,
      "grad_norm": 0.11488565057516098,
      "learning_rate": 6.58962125767339e-05,
      "loss": 0.1607,
      "step": 2179
    },
    {
      "epoch": 0.45759865659109994,
      "grad_norm": 0.12251152843236923,
      "learning_rate": 6.586146846284409e-05,
      "loss": 0.1671,
      "step": 2180
    },
    {
      "epoch": 0.45780856423173805,
      "grad_norm": 0.1218481957912445,
      "learning_rate": 6.582671583098337e-05,
      "loss": 0.1585,
      "step": 2181
    },
    {
      "epoch": 0.45801847187237615,
      "grad_norm": 0.11864989995956421,
      "learning_rate": 6.579195469981476e-05,
      "loss": 0.1597,
      "step": 2182
    },
    {
      "epoch": 0.45822837951301426,
      "grad_norm": 0.10890363156795502,
      "learning_rate": 6.575718508800574e-05,
      "loss": 0.1592,
      "step": 2183
    },
    {
      "epoch": 0.45843828715365237,
      "grad_norm": 0.11092708259820938,
      "learning_rate": 6.57224070142284e-05,
      "loss": 0.1653,
      "step": 2184
    },
    {
      "epoch": 0.45864819479429053,
      "grad_norm": 0.11390335112810135,
      "learning_rate": 6.568762049715933e-05,
      "loss": 0.1637,
      "step": 2185
    },
    {
      "epoch": 0.45885810243492864,
      "grad_norm": 0.12344920635223389,
      "learning_rate": 6.565282555547975e-05,
      "loss": 0.1569,
      "step": 2186
    },
    {
      "epoch": 0.45906801007556675,
      "grad_norm": 0.11731364578008652,
      "learning_rate": 6.561802220787526e-05,
      "loss": 0.167,
      "step": 2187
    },
    {
      "epoch": 0.45927791771620485,
      "grad_norm": 0.11007954180240631,
      "learning_rate": 6.558321047303612e-05,
      "loss": 0.1587,
      "step": 2188
    },
    {
      "epoch": 0.459487825356843,
      "grad_norm": 0.1331484168767929,
      "learning_rate": 6.554839036965696e-05,
      "loss": 0.139,
      "step": 2189
    },
    {
      "epoch": 0.4596977329974811,
      "grad_norm": 0.11050381511449814,
      "learning_rate": 6.551356191643705e-05,
      "loss": 0.1589,
      "step": 2190
    },
    {
      "epoch": 0.45990764063811923,
      "grad_norm": 0.12360599637031555,
      "learning_rate": 6.547872513208e-05,
      "loss": 0.1571,
      "step": 2191
    },
    {
      "epoch": 0.46011754827875734,
      "grad_norm": 0.11798024922609329,
      "learning_rate": 6.5443880035294e-05,
      "loss": 0.1574,
      "step": 2192
    },
    {
      "epoch": 0.46032745591939545,
      "grad_norm": 0.11354199051856995,
      "learning_rate": 6.540902664479164e-05,
      "loss": 0.1553,
      "step": 2193
    },
    {
      "epoch": 0.4605373635600336,
      "grad_norm": 0.131741002202034,
      "learning_rate": 6.537416497928998e-05,
      "loss": 0.1608,
      "step": 2194
    },
    {
      "epoch": 0.4607472712006717,
      "grad_norm": 0.122543103992939,
      "learning_rate": 6.533929505751055e-05,
      "loss": 0.1542,
      "step": 2195
    },
    {
      "epoch": 0.4609571788413098,
      "grad_norm": 0.11904080957174301,
      "learning_rate": 6.530441689817929e-05,
      "loss": 0.1704,
      "step": 2196
    },
    {
      "epoch": 0.46116708648194793,
      "grad_norm": 0.12378960847854614,
      "learning_rate": 6.526953052002657e-05,
      "loss": 0.1644,
      "step": 2197
    },
    {
      "epoch": 0.46137699412258604,
      "grad_norm": 0.11525364220142365,
      "learning_rate": 6.523463594178716e-05,
      "loss": 0.1706,
      "step": 2198
    },
    {
      "epoch": 0.4615869017632242,
      "grad_norm": 0.12415199726819992,
      "learning_rate": 6.519973318220025e-05,
      "loss": 0.1585,
      "step": 2199
    },
    {
      "epoch": 0.4617968094038623,
      "grad_norm": 0.12286436557769775,
      "learning_rate": 6.516482226000945e-05,
      "loss": 0.1526,
      "step": 2200
    },
    {
      "epoch": 0.4620067170445004,
      "grad_norm": 0.11900891363620758,
      "learning_rate": 6.512990319396266e-05,
      "loss": 0.1455,
      "step": 2201
    },
    {
      "epoch": 0.4622166246851385,
      "grad_norm": 0.12508201599121094,
      "learning_rate": 6.509497600281226e-05,
      "loss": 0.1577,
      "step": 2202
    },
    {
      "epoch": 0.4624265323257767,
      "grad_norm": 0.1479506492614746,
      "learning_rate": 6.506004070531494e-05,
      "loss": 0.1578,
      "step": 2203
    },
    {
      "epoch": 0.4626364399664148,
      "grad_norm": 0.1091904416680336,
      "learning_rate": 6.502509732023177e-05,
      "loss": 0.1482,
      "step": 2204
    },
    {
      "epoch": 0.4628463476070529,
      "grad_norm": 0.12727074325084686,
      "learning_rate": 6.499014586632812e-05,
      "loss": 0.1545,
      "step": 2205
    },
    {
      "epoch": 0.463056255247691,
      "grad_norm": 0.11460824310779572,
      "learning_rate": 6.495518636237373e-05,
      "loss": 0.1647,
      "step": 2206
    },
    {
      "epoch": 0.4632661628883291,
      "grad_norm": 0.12993770837783813,
      "learning_rate": 6.492021882714263e-05,
      "loss": 0.1628,
      "step": 2207
    },
    {
      "epoch": 0.4634760705289673,
      "grad_norm": 0.1012057512998581,
      "learning_rate": 6.488524327941323e-05,
      "loss": 0.1633,
      "step": 2208
    },
    {
      "epoch": 0.4636859781696054,
      "grad_norm": 0.09874848276376724,
      "learning_rate": 6.485025973796817e-05,
      "loss": 0.1434,
      "step": 2209
    },
    {
      "epoch": 0.4638958858102435,
      "grad_norm": 0.12227320671081543,
      "learning_rate": 6.48152682215944e-05,
      "loss": 0.1718,
      "step": 2210
    },
    {
      "epoch": 0.4641057934508816,
      "grad_norm": 0.11163394153118134,
      "learning_rate": 6.47802687490832e-05,
      "loss": 0.161,
      "step": 2211
    },
    {
      "epoch": 0.4643157010915197,
      "grad_norm": 0.12014634162187576,
      "learning_rate": 6.474526133923006e-05,
      "loss": 0.1624,
      "step": 2212
    },
    {
      "epoch": 0.46452560873215787,
      "grad_norm": 0.11568920314311981,
      "learning_rate": 6.471024601083475e-05,
      "loss": 0.1492,
      "step": 2213
    },
    {
      "epoch": 0.464735516372796,
      "grad_norm": 0.12237660586833954,
      "learning_rate": 6.467522278270133e-05,
      "loss": 0.1574,
      "step": 2214
    },
    {
      "epoch": 0.4649454240134341,
      "grad_norm": 0.0965287908911705,
      "learning_rate": 6.464019167363804e-05,
      "loss": 0.16,
      "step": 2215
    },
    {
      "epoch": 0.4651553316540722,
      "grad_norm": 0.12272460013628006,
      "learning_rate": 6.460515270245742e-05,
      "loss": 0.1649,
      "step": 2216
    },
    {
      "epoch": 0.46536523929471035,
      "grad_norm": 0.12838344275951385,
      "learning_rate": 6.457010588797619e-05,
      "loss": 0.1565,
      "step": 2217
    },
    {
      "epoch": 0.46557514693534846,
      "grad_norm": 0.13334384560585022,
      "learning_rate": 6.453505124901526e-05,
      "loss": 0.1387,
      "step": 2218
    },
    {
      "epoch": 0.46578505457598657,
      "grad_norm": 0.10292670875787735,
      "learning_rate": 6.44999888043998e-05,
      "loss": 0.1581,
      "step": 2219
    },
    {
      "epoch": 0.4659949622166247,
      "grad_norm": 0.1194579005241394,
      "learning_rate": 6.446491857295914e-05,
      "loss": 0.1673,
      "step": 2220
    },
    {
      "epoch": 0.4662048698572628,
      "grad_norm": 0.1133674830198288,
      "learning_rate": 6.442984057352679e-05,
      "loss": 0.1623,
      "step": 2221
    },
    {
      "epoch": 0.46641477749790095,
      "grad_norm": 0.10083403438329697,
      "learning_rate": 6.439475482494043e-05,
      "loss": 0.1565,
      "step": 2222
    },
    {
      "epoch": 0.46662468513853905,
      "grad_norm": 0.1386317014694214,
      "learning_rate": 6.435966134604195e-05,
      "loss": 0.18,
      "step": 2223
    },
    {
      "epoch": 0.46683459277917716,
      "grad_norm": 0.11142829060554504,
      "learning_rate": 6.432456015567728e-05,
      "loss": 0.167,
      "step": 2224
    },
    {
      "epoch": 0.46704450041981527,
      "grad_norm": 0.11028482764959335,
      "learning_rate": 6.428945127269663e-05,
      "loss": 0.1711,
      "step": 2225
    },
    {
      "epoch": 0.4672544080604534,
      "grad_norm": 0.11065543442964554,
      "learning_rate": 6.425433471595422e-05,
      "loss": 0.1526,
      "step": 2226
    },
    {
      "epoch": 0.46746431570109154,
      "grad_norm": 0.11224319040775299,
      "learning_rate": 6.421921050430849e-05,
      "loss": 0.1582,
      "step": 2227
    },
    {
      "epoch": 0.46767422334172964,
      "grad_norm": 0.12173019349575043,
      "learning_rate": 6.418407865662189e-05,
      "loss": 0.1714,
      "step": 2228
    },
    {
      "epoch": 0.46788413098236775,
      "grad_norm": 0.11253444850444794,
      "learning_rate": 6.414893919176106e-05,
      "loss": 0.1751,
      "step": 2229
    },
    {
      "epoch": 0.46809403862300586,
      "grad_norm": 0.1227567121386528,
      "learning_rate": 6.411379212859669e-05,
      "loss": 0.1558,
      "step": 2230
    },
    {
      "epoch": 0.468303946263644,
      "grad_norm": 0.12723511457443237,
      "learning_rate": 6.407863748600355e-05,
      "loss": 0.1574,
      "step": 2231
    },
    {
      "epoch": 0.46851385390428213,
      "grad_norm": 0.10584603250026703,
      "learning_rate": 6.40434752828605e-05,
      "loss": 0.156,
      "step": 2232
    },
    {
      "epoch": 0.46872376154492024,
      "grad_norm": 0.13455379009246826,
      "learning_rate": 6.400830553805043e-05,
      "loss": 0.157,
      "step": 2233
    },
    {
      "epoch": 0.46893366918555834,
      "grad_norm": 0.10209176689386368,
      "learning_rate": 6.397312827046029e-05,
      "loss": 0.161,
      "step": 2234
    },
    {
      "epoch": 0.46914357682619645,
      "grad_norm": 0.15381069481372833,
      "learning_rate": 6.393794349898111e-05,
      "loss": 0.1614,
      "step": 2235
    },
    {
      "epoch": 0.4693534844668346,
      "grad_norm": 0.12260998785495758,
      "learning_rate": 6.390275124250788e-05,
      "loss": 0.1451,
      "step": 2236
    },
    {
      "epoch": 0.4695633921074727,
      "grad_norm": 0.12964041531085968,
      "learning_rate": 6.386755151993967e-05,
      "loss": 0.1464,
      "step": 2237
    },
    {
      "epoch": 0.46977329974811083,
      "grad_norm": 0.14182746410369873,
      "learning_rate": 6.383234435017952e-05,
      "loss": 0.1666,
      "step": 2238
    },
    {
      "epoch": 0.46998320738874894,
      "grad_norm": 0.11773552000522614,
      "learning_rate": 6.37971297521345e-05,
      "loss": 0.1609,
      "step": 2239
    },
    {
      "epoch": 0.47019311502938704,
      "grad_norm": 0.12091661989688873,
      "learning_rate": 6.376190774471562e-05,
      "loss": 0.161,
      "step": 2240
    },
    {
      "epoch": 0.4704030226700252,
      "grad_norm": 0.12359412759542465,
      "learning_rate": 6.372667834683794e-05,
      "loss": 0.1407,
      "step": 2241
    },
    {
      "epoch": 0.4706129303106633,
      "grad_norm": 0.11386679857969284,
      "learning_rate": 6.369144157742045e-05,
      "loss": 0.1644,
      "step": 2242
    },
    {
      "epoch": 0.4708228379513014,
      "grad_norm": 0.12322553992271423,
      "learning_rate": 6.365619745538607e-05,
      "loss": 0.1661,
      "step": 2243
    },
    {
      "epoch": 0.47103274559193953,
      "grad_norm": 0.11949003487825394,
      "learning_rate": 6.362094599966171e-05,
      "loss": 0.1469,
      "step": 2244
    },
    {
      "epoch": 0.4712426532325777,
      "grad_norm": 0.11004175245761871,
      "learning_rate": 6.358568722917824e-05,
      "loss": 0.1452,
      "step": 2245
    },
    {
      "epoch": 0.4714525608732158,
      "grad_norm": 0.11459292471408844,
      "learning_rate": 6.355042116287038e-05,
      "loss": 0.166,
      "step": 2246
    },
    {
      "epoch": 0.4716624685138539,
      "grad_norm": 0.12747009098529816,
      "learning_rate": 6.351514781967682e-05,
      "loss": 0.1472,
      "step": 2247
    },
    {
      "epoch": 0.471872376154492,
      "grad_norm": 0.11906711012125015,
      "learning_rate": 6.347986721854018e-05,
      "loss": 0.1686,
      "step": 2248
    },
    {
      "epoch": 0.4720822837951301,
      "grad_norm": 0.12261734902858734,
      "learning_rate": 6.344457937840693e-05,
      "loss": 0.1578,
      "step": 2249
    },
    {
      "epoch": 0.4722921914357683,
      "grad_norm": 0.12668012082576752,
      "learning_rate": 6.340928431822744e-05,
      "loss": 0.1583,
      "step": 2250
    },
    {
      "epoch": 0.4725020990764064,
      "grad_norm": 0.14289303123950958,
      "learning_rate": 6.337398205695597e-05,
      "loss": 0.1392,
      "step": 2251
    },
    {
      "epoch": 0.4727120067170445,
      "grad_norm": 0.12961414456367493,
      "learning_rate": 6.333867261355064e-05,
      "loss": 0.1697,
      "step": 2252
    },
    {
      "epoch": 0.4729219143576826,
      "grad_norm": 0.11781413853168488,
      "learning_rate": 6.330335600697344e-05,
      "loss": 0.163,
      "step": 2253
    },
    {
      "epoch": 0.4731318219983207,
      "grad_norm": 0.13710007071495056,
      "learning_rate": 6.326803225619018e-05,
      "loss": 0.1705,
      "step": 2254
    },
    {
      "epoch": 0.4733417296389589,
      "grad_norm": 0.14093898236751556,
      "learning_rate": 6.323270138017052e-05,
      "loss": 0.1622,
      "step": 2255
    },
    {
      "epoch": 0.473551637279597,
      "grad_norm": 0.12070971727371216,
      "learning_rate": 6.319736339788795e-05,
      "loss": 0.1289,
      "step": 2256
    },
    {
      "epoch": 0.4737615449202351,
      "grad_norm": 0.12553520500659943,
      "learning_rate": 6.316201832831978e-05,
      "loss": 0.1514,
      "step": 2257
    },
    {
      "epoch": 0.4739714525608732,
      "grad_norm": 0.11646711826324463,
      "learning_rate": 6.312666619044712e-05,
      "loss": 0.1509,
      "step": 2258
    },
    {
      "epoch": 0.47418136020151136,
      "grad_norm": 0.12744106352329254,
      "learning_rate": 6.309130700325486e-05,
      "loss": 0.1599,
      "step": 2259
    },
    {
      "epoch": 0.47439126784214947,
      "grad_norm": 0.11816243827342987,
      "learning_rate": 6.305594078573172e-05,
      "loss": 0.1579,
      "step": 2260
    },
    {
      "epoch": 0.4746011754827876,
      "grad_norm": 0.1067761555314064,
      "learning_rate": 6.302056755687013e-05,
      "loss": 0.149,
      "step": 2261
    },
    {
      "epoch": 0.4748110831234257,
      "grad_norm": 0.1448090523481369,
      "learning_rate": 6.298518733566634e-05,
      "loss": 0.1471,
      "step": 2262
    },
    {
      "epoch": 0.4750209907640638,
      "grad_norm": 0.10772380977869034,
      "learning_rate": 6.294980014112035e-05,
      "loss": 0.1418,
      "step": 2263
    },
    {
      "epoch": 0.47523089840470195,
      "grad_norm": 0.11698709428310394,
      "learning_rate": 6.291440599223587e-05,
      "loss": 0.1535,
      "step": 2264
    },
    {
      "epoch": 0.47544080604534006,
      "grad_norm": 0.11132114380598068,
      "learning_rate": 6.287900490802038e-05,
      "loss": 0.1472,
      "step": 2265
    },
    {
      "epoch": 0.47565071368597817,
      "grad_norm": 0.10131499171257019,
      "learning_rate": 6.284359690748506e-05,
      "loss": 0.1649,
      "step": 2266
    },
    {
      "epoch": 0.4758606213266163,
      "grad_norm": 0.1225217878818512,
      "learning_rate": 6.280818200964484e-05,
      "loss": 0.1692,
      "step": 2267
    },
    {
      "epoch": 0.4760705289672544,
      "grad_norm": 0.11604775488376617,
      "learning_rate": 6.27727602335183e-05,
      "loss": 0.156,
      "step": 2268
    },
    {
      "epoch": 0.47628043660789254,
      "grad_norm": 0.09776950627565384,
      "learning_rate": 6.273733159812773e-05,
      "loss": 0.167,
      "step": 2269
    },
    {
      "epoch": 0.47649034424853065,
      "grad_norm": 0.127922922372818,
      "learning_rate": 6.27018961224992e-05,
      "loss": 0.1732,
      "step": 2270
    },
    {
      "epoch": 0.47670025188916876,
      "grad_norm": 0.1038019210100174,
      "learning_rate": 6.266645382566226e-05,
      "loss": 0.1538,
      "step": 2271
    },
    {
      "epoch": 0.47691015952980687,
      "grad_norm": 0.11374037712812424,
      "learning_rate": 6.26310047266503e-05,
      "loss": 0.1592,
      "step": 2272
    },
    {
      "epoch": 0.47712006717044503,
      "grad_norm": 0.1249479278922081,
      "learning_rate": 6.259554884450028e-05,
      "loss": 0.1638,
      "step": 2273
    },
    {
      "epoch": 0.47732997481108314,
      "grad_norm": 0.13736027479171753,
      "learning_rate": 6.25600861982528e-05,
      "loss": 0.181,
      "step": 2274
    },
    {
      "epoch": 0.47753988245172124,
      "grad_norm": 0.1172303855419159,
      "learning_rate": 6.252461680695214e-05,
      "loss": 0.1549,
      "step": 2275
    },
    {
      "epoch": 0.47774979009235935,
      "grad_norm": 0.1163245216012001,
      "learning_rate": 6.248914068964618e-05,
      "loss": 0.1821,
      "step": 2276
    },
    {
      "epoch": 0.47795969773299746,
      "grad_norm": 0.11509289592504501,
      "learning_rate": 6.245365786538636e-05,
      "loss": 0.1618,
      "step": 2277
    },
    {
      "epoch": 0.4781696053736356,
      "grad_norm": 0.11194191128015518,
      "learning_rate": 6.241816835322781e-05,
      "loss": 0.1384,
      "step": 2278
    },
    {
      "epoch": 0.4783795130142737,
      "grad_norm": 0.13771222531795502,
      "learning_rate": 6.238267217222919e-05,
      "loss": 0.1614,
      "step": 2279
    },
    {
      "epoch": 0.47858942065491183,
      "grad_norm": 0.1045057475566864,
      "learning_rate": 6.234716934145277e-05,
      "loss": 0.1522,
      "step": 2280
    },
    {
      "epoch": 0.47879932829554994,
      "grad_norm": 0.11645108461380005,
      "learning_rate": 6.231165987996436e-05,
      "loss": 0.1701,
      "step": 2281
    },
    {
      "epoch": 0.4790092359361881,
      "grad_norm": 0.12459120154380798,
      "learning_rate": 6.227614380683337e-05,
      "loss": 0.1764,
      "step": 2282
    },
    {
      "epoch": 0.4792191435768262,
      "grad_norm": 0.13852064311504364,
      "learning_rate": 6.224062114113275e-05,
      "loss": 0.1558,
      "step": 2283
    },
    {
      "epoch": 0.4794290512174643,
      "grad_norm": 0.10844726115465164,
      "learning_rate": 6.220509190193896e-05,
      "loss": 0.1665,
      "step": 2284
    },
    {
      "epoch": 0.4796389588581024,
      "grad_norm": 0.12302237004041672,
      "learning_rate": 6.216955610833203e-05,
      "loss": 0.1694,
      "step": 2285
    },
    {
      "epoch": 0.47984886649874053,
      "grad_norm": 0.11576231569051743,
      "learning_rate": 6.213401377939548e-05,
      "loss": 0.1683,
      "step": 2286
    },
    {
      "epoch": 0.4800587741393787,
      "grad_norm": 0.1322503685951233,
      "learning_rate": 6.209846493421637e-05,
      "loss": 0.167,
      "step": 2287
    },
    {
      "epoch": 0.4802686817800168,
      "grad_norm": 0.11987553536891937,
      "learning_rate": 6.206290959188523e-05,
      "loss": 0.1593,
      "step": 2288
    },
    {
      "epoch": 0.4804785894206549,
      "grad_norm": 0.11214146018028259,
      "learning_rate": 6.20273477714961e-05,
      "loss": 0.1829,
      "step": 2289
    },
    {
      "epoch": 0.480688497061293,
      "grad_norm": 0.11640086024999619,
      "learning_rate": 6.199177949214649e-05,
      "loss": 0.1542,
      "step": 2290
    },
    {
      "epoch": 0.4808984047019311,
      "grad_norm": 0.10756421089172363,
      "learning_rate": 6.195620477293734e-05,
      "loss": 0.1394,
      "step": 2291
    },
    {
      "epoch": 0.4811083123425693,
      "grad_norm": 0.11862246692180634,
      "learning_rate": 6.192062363297315e-05,
      "loss": 0.1585,
      "step": 2292
    },
    {
      "epoch": 0.4813182199832074,
      "grad_norm": 0.10187924653291702,
      "learning_rate": 6.188503609136175e-05,
      "loss": 0.1536,
      "step": 2293
    },
    {
      "epoch": 0.4815281276238455,
      "grad_norm": 0.1261720061302185,
      "learning_rate": 6.18494421672145e-05,
      "loss": 0.1619,
      "step": 2294
    },
    {
      "epoch": 0.4817380352644836,
      "grad_norm": 0.11360353976488113,
      "learning_rate": 6.181384187964613e-05,
      "loss": 0.1417,
      "step": 2295
    },
    {
      "epoch": 0.4819479429051218,
      "grad_norm": 0.1149376705288887,
      "learning_rate": 6.177823524777481e-05,
      "loss": 0.1569,
      "step": 2296
    },
    {
      "epoch": 0.4821578505457599,
      "grad_norm": 0.10513467341661453,
      "learning_rate": 6.174262229072212e-05,
      "loss": 0.138,
      "step": 2297
    },
    {
      "epoch": 0.482367758186398,
      "grad_norm": 0.10287132859230042,
      "learning_rate": 6.170700302761303e-05,
      "loss": 0.16,
      "step": 2298
    },
    {
      "epoch": 0.4825776658270361,
      "grad_norm": 0.12048831582069397,
      "learning_rate": 6.16713774775759e-05,
      "loss": 0.1764,
      "step": 2299
    },
    {
      "epoch": 0.4827875734676742,
      "grad_norm": 0.14703017473220825,
      "learning_rate": 6.163574565974246e-05,
      "loss": 0.1571,
      "step": 2300
    },
    {
      "epoch": 0.48299748110831237,
      "grad_norm": 0.10593768209218979,
      "learning_rate": 6.160010759324783e-05,
      "loss": 0.1544,
      "step": 2301
    },
    {
      "epoch": 0.4832073887489505,
      "grad_norm": 0.1123269572854042,
      "learning_rate": 6.156446329723042e-05,
      "loss": 0.1518,
      "step": 2302
    },
    {
      "epoch": 0.4834172963895886,
      "grad_norm": 0.12359534204006195,
      "learning_rate": 6.152881279083209e-05,
      "loss": 0.1573,
      "step": 2303
    },
    {
      "epoch": 0.4836272040302267,
      "grad_norm": 0.11115585267543793,
      "learning_rate": 6.149315609319793e-05,
      "loss": 0.1718,
      "step": 2304
    },
    {
      "epoch": 0.4838371116708648,
      "grad_norm": 0.12576039135456085,
      "learning_rate": 6.145749322347642e-05,
      "loss": 0.1581,
      "step": 2305
    },
    {
      "epoch": 0.48404701931150296,
      "grad_norm": 0.1161019429564476,
      "learning_rate": 6.142182420081931e-05,
      "loss": 0.1493,
      "step": 2306
    },
    {
      "epoch": 0.48425692695214106,
      "grad_norm": 0.10857407003641129,
      "learning_rate": 6.138614904438172e-05,
      "loss": 0.1642,
      "step": 2307
    },
    {
      "epoch": 0.48446683459277917,
      "grad_norm": 0.1394800990819931,
      "learning_rate": 6.1350467773322e-05,
      "loss": 0.1528,
      "step": 2308
    },
    {
      "epoch": 0.4846767422334173,
      "grad_norm": 0.11143230646848679,
      "learning_rate": 6.131478040680183e-05,
      "loss": 0.1509,
      "step": 2309
    },
    {
      "epoch": 0.48488664987405544,
      "grad_norm": 0.13411763310432434,
      "learning_rate": 6.12790869639861e-05,
      "loss": 0.1417,
      "step": 2310
    },
    {
      "epoch": 0.48509655751469355,
      "grad_norm": 0.1560184210538864,
      "learning_rate": 6.124338746404305e-05,
      "loss": 0.157,
      "step": 2311
    },
    {
      "epoch": 0.48530646515533166,
      "grad_norm": 0.13342639803886414,
      "learning_rate": 6.12076819261441e-05,
      "loss": 0.1601,
      "step": 2312
    },
    {
      "epoch": 0.48551637279596976,
      "grad_norm": 0.16688786447048187,
      "learning_rate": 6.117197036946394e-05,
      "loss": 0.168,
      "step": 2313
    },
    {
      "epoch": 0.48572628043660787,
      "grad_norm": 0.10276120901107788,
      "learning_rate": 6.11362528131805e-05,
      "loss": 0.1504,
      "step": 2314
    },
    {
      "epoch": 0.48593618807724603,
      "grad_norm": 0.1053510457277298,
      "learning_rate": 6.110052927647491e-05,
      "loss": 0.1517,
      "step": 2315
    },
    {
      "epoch": 0.48614609571788414,
      "grad_norm": 0.10654187202453613,
      "learning_rate": 6.106479977853154e-05,
      "loss": 0.1563,
      "step": 2316
    },
    {
      "epoch": 0.48635600335852225,
      "grad_norm": 0.11610613763332367,
      "learning_rate": 6.102906433853796e-05,
      "loss": 0.1549,
      "step": 2317
    },
    {
      "epoch": 0.48656591099916036,
      "grad_norm": 0.11274497210979462,
      "learning_rate": 6.099332297568488e-05,
      "loss": 0.1561,
      "step": 2318
    },
    {
      "epoch": 0.48677581863979846,
      "grad_norm": 0.12665775418281555,
      "learning_rate": 6.095757570916626e-05,
      "loss": 0.1618,
      "step": 2319
    },
    {
      "epoch": 0.4869857262804366,
      "grad_norm": 0.12388937175273895,
      "learning_rate": 6.092182255817918e-05,
      "loss": 0.1636,
      "step": 2320
    },
    {
      "epoch": 0.48719563392107473,
      "grad_norm": 0.09930404275655746,
      "learning_rate": 6.088606354192392e-05,
      "loss": 0.1542,
      "step": 2321
    },
    {
      "epoch": 0.48740554156171284,
      "grad_norm": 0.09602123498916626,
      "learning_rate": 6.0850298679603845e-05,
      "loss": 0.142,
      "step": 2322
    },
    {
      "epoch": 0.48761544920235095,
      "grad_norm": 0.1361406445503235,
      "learning_rate": 6.0814527990425555e-05,
      "loss": 0.1487,
      "step": 2323
    },
    {
      "epoch": 0.4878253568429891,
      "grad_norm": 0.11267097294330597,
      "learning_rate": 6.0778751493598686e-05,
      "loss": 0.1535,
      "step": 2324
    },
    {
      "epoch": 0.4880352644836272,
      "grad_norm": 0.11054635792970657,
      "learning_rate": 6.0742969208336074e-05,
      "loss": 0.1473,
      "step": 2325
    },
    {
      "epoch": 0.4882451721242653,
      "grad_norm": 0.12421461194753647,
      "learning_rate": 6.070718115385359e-05,
      "loss": 0.1659,
      "step": 2326
    },
    {
      "epoch": 0.48845507976490343,
      "grad_norm": 0.1250905990600586,
      "learning_rate": 6.0671387349370246e-05,
      "loss": 0.1605,
      "step": 2327
    },
    {
      "epoch": 0.48866498740554154,
      "grad_norm": 0.13226497173309326,
      "learning_rate": 6.0635587814108143e-05,
      "loss": 0.1671,
      "step": 2328
    },
    {
      "epoch": 0.4888748950461797,
      "grad_norm": 0.1366906613111496,
      "learning_rate": 6.059978256729244e-05,
      "loss": 0.1671,
      "step": 2329
    },
    {
      "epoch": 0.4890848026868178,
      "grad_norm": 0.11474860459566116,
      "learning_rate": 6.056397162815137e-05,
      "loss": 0.1408,
      "step": 2330
    },
    {
      "epoch": 0.4892947103274559,
      "grad_norm": 0.14364750683307648,
      "learning_rate": 6.052815501591622e-05,
      "loss": 0.1577,
      "step": 2331
    },
    {
      "epoch": 0.489504617968094,
      "grad_norm": 0.10247082263231277,
      "learning_rate": 6.0492332749821345e-05,
      "loss": 0.1471,
      "step": 2332
    },
    {
      "epoch": 0.48971452560873213,
      "grad_norm": 0.12132863700389862,
      "learning_rate": 6.045650484910412e-05,
      "loss": 0.1532,
      "step": 2333
    },
    {
      "epoch": 0.4899244332493703,
      "grad_norm": 0.11156227439641953,
      "learning_rate": 6.042067133300494e-05,
      "loss": 0.1455,
      "step": 2334
    },
    {
      "epoch": 0.4901343408900084,
      "grad_norm": 0.11364928632974625,
      "learning_rate": 6.038483222076722e-05,
      "loss": 0.166,
      "step": 2335
    },
    {
      "epoch": 0.4903442485306465,
      "grad_norm": 0.11692879348993301,
      "learning_rate": 6.034898753163738e-05,
      "loss": 0.1474,
      "step": 2336
    },
    {
      "epoch": 0.4905541561712846,
      "grad_norm": 0.13082164525985718,
      "learning_rate": 6.031313728486483e-05,
      "loss": 0.1598,
      "step": 2337
    },
    {
      "epoch": 0.4907640638119228,
      "grad_norm": 0.11700566112995148,
      "learning_rate": 6.0277281499701986e-05,
      "loss": 0.1662,
      "step": 2338
    },
    {
      "epoch": 0.4909739714525609,
      "grad_norm": 0.12339949607849121,
      "learning_rate": 6.0241420195404216e-05,
      "loss": 0.1474,
      "step": 2339
    },
    {
      "epoch": 0.491183879093199,
      "grad_norm": 0.12292884290218353,
      "learning_rate": 6.020555339122987e-05,
      "loss": 0.1556,
      "step": 2340
    },
    {
      "epoch": 0.4913937867338371,
      "grad_norm": 0.1513158679008484,
      "learning_rate": 6.0169681106440236e-05,
      "loss": 0.1432,
      "step": 2341
    },
    {
      "epoch": 0.4916036943744752,
      "grad_norm": 0.12808825075626373,
      "learning_rate": 6.0133803360299544e-05,
      "loss": 0.142,
      "step": 2342
    },
    {
      "epoch": 0.49181360201511337,
      "grad_norm": 0.10223434865474701,
      "learning_rate": 6.009792017207496e-05,
      "loss": 0.1619,
      "step": 2343
    },
    {
      "epoch": 0.4920235096557515,
      "grad_norm": 0.1480330377817154,
      "learning_rate": 6.006203156103659e-05,
      "loss": 0.1505,
      "step": 2344
    },
    {
      "epoch": 0.4922334172963896,
      "grad_norm": 0.11763409525156021,
      "learning_rate": 6.002613754645743e-05,
      "loss": 0.1559,
      "step": 2345
    },
    {
      "epoch": 0.4924433249370277,
      "grad_norm": 0.1263841837644577,
      "learning_rate": 5.99902381476134e-05,
      "loss": 0.147,
      "step": 2346
    },
    {
      "epoch": 0.4926532325776658,
      "grad_norm": 0.12160757929086685,
      "learning_rate": 5.9954333383783256e-05,
      "loss": 0.1588,
      "step": 2347
    },
    {
      "epoch": 0.49286314021830396,
      "grad_norm": 0.20746250450611115,
      "learning_rate": 5.991842327424871e-05,
      "loss": 0.1598,
      "step": 2348
    },
    {
      "epoch": 0.49307304785894207,
      "grad_norm": 0.12765191495418549,
      "learning_rate": 5.988250783829432e-05,
      "loss": 0.1618,
      "step": 2349
    },
    {
      "epoch": 0.4932829554995802,
      "grad_norm": 0.10818785429000854,
      "learning_rate": 5.9846587095207474e-05,
      "loss": 0.1521,
      "step": 2350
    },
    {
      "epoch": 0.4934928631402183,
      "grad_norm": 0.13110002875328064,
      "learning_rate": 5.981066106427843e-05,
      "loss": 0.1658,
      "step": 2351
    },
    {
      "epoch": 0.49370277078085645,
      "grad_norm": 0.11284149438142776,
      "learning_rate": 5.97747297648003e-05,
      "loss": 0.1578,
      "step": 2352
    },
    {
      "epoch": 0.49391267842149456,
      "grad_norm": 0.12948429584503174,
      "learning_rate": 5.973879321606901e-05,
      "loss": 0.1439,
      "step": 2353
    },
    {
      "epoch": 0.49412258606213266,
      "grad_norm": 0.1391395777463913,
      "learning_rate": 5.970285143738331e-05,
      "loss": 0.1697,
      "step": 2354
    },
    {
      "epoch": 0.49433249370277077,
      "grad_norm": 0.11676032096147537,
      "learning_rate": 5.966690444804474e-05,
      "loss": 0.1543,
      "step": 2355
    },
    {
      "epoch": 0.4945424013434089,
      "grad_norm": 0.1376785784959793,
      "learning_rate": 5.963095226735769e-05,
      "loss": 0.1547,
      "step": 2356
    },
    {
      "epoch": 0.49475230898404704,
      "grad_norm": 0.1433243304491043,
      "learning_rate": 5.9594994914629244e-05,
      "loss": 0.1516,
      "step": 2357
    },
    {
      "epoch": 0.49496221662468515,
      "grad_norm": 0.12082047760486603,
      "learning_rate": 5.955903240916938e-05,
      "loss": 0.1679,
      "step": 2358
    },
    {
      "epoch": 0.49517212426532325,
      "grad_norm": 0.0982949510216713,
      "learning_rate": 5.952306477029074e-05,
      "loss": 0.1559,
      "step": 2359
    },
    {
      "epoch": 0.49538203190596136,
      "grad_norm": 0.13121630251407623,
      "learning_rate": 5.9487092017308806e-05,
      "loss": 0.1696,
      "step": 2360
    },
    {
      "epoch": 0.49559193954659947,
      "grad_norm": 0.13477589190006256,
      "learning_rate": 5.9451114169541744e-05,
      "loss": 0.1565,
      "step": 2361
    },
    {
      "epoch": 0.49580184718723763,
      "grad_norm": 0.11858514696359634,
      "learning_rate": 5.94151312463105e-05,
      "loss": 0.1648,
      "step": 2362
    },
    {
      "epoch": 0.49601175482787574,
      "grad_norm": 0.0997430831193924,
      "learning_rate": 5.93791432669387e-05,
      "loss": 0.1547,
      "step": 2363
    },
    {
      "epoch": 0.49622166246851385,
      "grad_norm": 0.11332326382398605,
      "learning_rate": 5.934315025075272e-05,
      "loss": 0.1512,
      "step": 2364
    },
    {
      "epoch": 0.49643157010915195,
      "grad_norm": 0.1203446090221405,
      "learning_rate": 5.930715221708163e-05,
      "loss": 0.1677,
      "step": 2365
    },
    {
      "epoch": 0.4966414777497901,
      "grad_norm": 0.12282302975654602,
      "learning_rate": 5.927114918525722e-05,
      "loss": 0.1525,
      "step": 2366
    },
    {
      "epoch": 0.4968513853904282,
      "grad_norm": 0.10741288959980011,
      "learning_rate": 5.923514117461391e-05,
      "loss": 0.1345,
      "step": 2367
    },
    {
      "epoch": 0.49706129303106633,
      "grad_norm": 0.10739348083734512,
      "learning_rate": 5.919912820448884e-05,
      "loss": 0.1439,
      "step": 2368
    },
    {
      "epoch": 0.49727120067170444,
      "grad_norm": 0.11860400438308716,
      "learning_rate": 5.9163110294221804e-05,
      "loss": 0.1743,
      "step": 2369
    },
    {
      "epoch": 0.49748110831234255,
      "grad_norm": 0.1082897037267685,
      "learning_rate": 5.912708746315522e-05,
      "loss": 0.1542,
      "step": 2370
    },
    {
      "epoch": 0.4976910159529807,
      "grad_norm": 0.11466725915670395,
      "learning_rate": 5.909105973063419e-05,
      "loss": 0.16,
      "step": 2371
    },
    {
      "epoch": 0.4979009235936188,
      "grad_norm": 0.1288958340883255,
      "learning_rate": 5.905502711600641e-05,
      "loss": 0.1641,
      "step": 2372
    },
    {
      "epoch": 0.4981108312342569,
      "grad_norm": 0.1337483674287796,
      "learning_rate": 5.9018989638622234e-05,
      "loss": 0.1446,
      "step": 2373
    },
    {
      "epoch": 0.49832073887489503,
      "grad_norm": 0.12721973657608032,
      "learning_rate": 5.8982947317834615e-05,
      "loss": 0.1371,
      "step": 2374
    },
    {
      "epoch": 0.49853064651553314,
      "grad_norm": 0.13450993597507477,
      "learning_rate": 5.894690017299909e-05,
      "loss": 0.1605,
      "step": 2375
    },
    {
      "epoch": 0.4987405541561713,
      "grad_norm": 0.130893275141716,
      "learning_rate": 5.891084822347379e-05,
      "loss": 0.1582,
      "step": 2376
    },
    {
      "epoch": 0.4989504617968094,
      "grad_norm": 0.10963684320449829,
      "learning_rate": 5.887479148861948e-05,
      "loss": 0.17,
      "step": 2377
    },
    {
      "epoch": 0.4991603694374475,
      "grad_norm": 0.12141723185777664,
      "learning_rate": 5.883872998779939e-05,
      "loss": 0.1423,
      "step": 2378
    },
    {
      "epoch": 0.4993702770780856,
      "grad_norm": 0.12254346907138824,
      "learning_rate": 5.8802663740379414e-05,
      "loss": 0.1598,
      "step": 2379
    },
    {
      "epoch": 0.4995801847187238,
      "grad_norm": 0.1324474960565567,
      "learning_rate": 5.8766592765727934e-05,
      "loss": 0.1687,
      "step": 2380
    },
    {
      "epoch": 0.4997900923593619,
      "grad_norm": 0.13154703378677368,
      "learning_rate": 5.873051708321589e-05,
      "loss": 0.1434,
      "step": 2381
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.10656025260686874,
      "learning_rate": 5.869443671221673e-05,
      "loss": 0.1556,
      "step": 2382
    },
    {
      "epoch": 0.5002099076406381,
      "grad_norm": 0.11154552549123764,
      "learning_rate": 5.865835167210647e-05,
      "loss": 0.1674,
      "step": 2383
    },
    {
      "epoch": 0.5004198152812762,
      "grad_norm": 0.12540102005004883,
      "learning_rate": 5.862226198226357e-05,
      "loss": 0.1599,
      "step": 2384
    },
    {
      "epoch": 0.5006297229219143,
      "grad_norm": 0.10726683586835861,
      "learning_rate": 5.858616766206905e-05,
      "loss": 0.164,
      "step": 2385
    },
    {
      "epoch": 0.5008396305625524,
      "grad_norm": 0.10406230390071869,
      "learning_rate": 5.855006873090635e-05,
      "loss": 0.1492,
      "step": 2386
    },
    {
      "epoch": 0.5010495382031906,
      "grad_norm": 0.12061581760644913,
      "learning_rate": 5.851396520816146e-05,
      "loss": 0.166,
      "step": 2387
    },
    {
      "epoch": 0.5012594458438288,
      "grad_norm": 0.12048531323671341,
      "learning_rate": 5.8477857113222745e-05,
      "loss": 0.1409,
      "step": 2388
    },
    {
      "epoch": 0.5014693534844669,
      "grad_norm": 0.11254918575286865,
      "learning_rate": 5.8441744465481115e-05,
      "loss": 0.1502,
      "step": 2389
    },
    {
      "epoch": 0.501679261125105,
      "grad_norm": 0.1056586280465126,
      "learning_rate": 5.840562728432988e-05,
      "loss": 0.1677,
      "step": 2390
    },
    {
      "epoch": 0.5018891687657431,
      "grad_norm": 0.1021779254078865,
      "learning_rate": 5.836950558916481e-05,
      "loss": 0.1671,
      "step": 2391
    },
    {
      "epoch": 0.5020990764063812,
      "grad_norm": 0.1173490360379219,
      "learning_rate": 5.8333379399384026e-05,
      "loss": 0.1707,
      "step": 2392
    },
    {
      "epoch": 0.5023089840470193,
      "grad_norm": 0.11009548604488373,
      "learning_rate": 5.829724873438816e-05,
      "loss": 0.1395,
      "step": 2393
    },
    {
      "epoch": 0.5025188916876574,
      "grad_norm": 0.10494174063205719,
      "learning_rate": 5.82611136135802e-05,
      "loss": 0.1598,
      "step": 2394
    },
    {
      "epoch": 0.5027287993282955,
      "grad_norm": 0.1159316897392273,
      "learning_rate": 5.822497405636553e-05,
      "loss": 0.1812,
      "step": 2395
    },
    {
      "epoch": 0.5029387069689337,
      "grad_norm": 0.12583115696907043,
      "learning_rate": 5.81888300821519e-05,
      "loss": 0.148,
      "step": 2396
    },
    {
      "epoch": 0.5031486146095718,
      "grad_norm": 0.13148032128810883,
      "learning_rate": 5.815268171034947e-05,
      "loss": 0.1589,
      "step": 2397
    },
    {
      "epoch": 0.5033585222502099,
      "grad_norm": 0.10882681608200073,
      "learning_rate": 5.811652896037073e-05,
      "loss": 0.1567,
      "step": 2398
    },
    {
      "epoch": 0.503568429890848,
      "grad_norm": 0.12563377618789673,
      "learning_rate": 5.808037185163054e-05,
      "loss": 0.1635,
      "step": 2399
    },
    {
      "epoch": 0.5037783375314862,
      "grad_norm": 0.11974649131298065,
      "learning_rate": 5.8044210403546096e-05,
      "loss": 0.1517,
      "step": 2400
    },
    {
      "epoch": 0.5039882451721243,
      "grad_norm": 0.12729738652706146,
      "learning_rate": 5.8008044635536916e-05,
      "loss": 0.1369,
      "step": 2401
    },
    {
      "epoch": 0.5041981528127624,
      "grad_norm": 0.13505300879478455,
      "learning_rate": 5.797187456702483e-05,
      "loss": 0.1695,
      "step": 2402
    },
    {
      "epoch": 0.5044080604534005,
      "grad_norm": 0.13184520602226257,
      "learning_rate": 5.7935700217434017e-05,
      "loss": 0.15,
      "step": 2403
    },
    {
      "epoch": 0.5046179680940386,
      "grad_norm": 0.13112464547157288,
      "learning_rate": 5.789952160619092e-05,
      "loss": 0.147,
      "step": 2404
    },
    {
      "epoch": 0.5048278757346767,
      "grad_norm": 0.12043753266334534,
      "learning_rate": 5.7863338752724275e-05,
      "loss": 0.1529,
      "step": 2405
    },
    {
      "epoch": 0.5050377833753149,
      "grad_norm": 0.11811091750860214,
      "learning_rate": 5.782715167646508e-05,
      "loss": 0.1357,
      "step": 2406
    },
    {
      "epoch": 0.505247691015953,
      "grad_norm": 0.10397012531757355,
      "learning_rate": 5.779096039684667e-05,
      "loss": 0.1423,
      "step": 2407
    },
    {
      "epoch": 0.5054575986565911,
      "grad_norm": 0.1074468344449997,
      "learning_rate": 5.7754764933304554e-05,
      "loss": 0.1691,
      "step": 2408
    },
    {
      "epoch": 0.5056675062972292,
      "grad_norm": 0.11894194036722183,
      "learning_rate": 5.7718565305276526e-05,
      "loss": 0.159,
      "step": 2409
    },
    {
      "epoch": 0.5058774139378673,
      "grad_norm": 0.11252366006374359,
      "learning_rate": 5.7682361532202634e-05,
      "loss": 0.1545,
      "step": 2410
    },
    {
      "epoch": 0.5060873215785054,
      "grad_norm": 0.14150729775428772,
      "learning_rate": 5.76461536335251e-05,
      "loss": 0.1618,
      "step": 2411
    },
    {
      "epoch": 0.5062972292191436,
      "grad_norm": 0.16456341743469238,
      "learning_rate": 5.7609941628688424e-05,
      "loss": 0.1666,
      "step": 2412
    },
    {
      "epoch": 0.5065071368597817,
      "grad_norm": 0.10939100384712219,
      "learning_rate": 5.757372553713923e-05,
      "loss": 0.1688,
      "step": 2413
    },
    {
      "epoch": 0.5067170445004198,
      "grad_norm": 0.11484073102474213,
      "learning_rate": 5.753750537832645e-05,
      "loss": 0.1712,
      "step": 2414
    },
    {
      "epoch": 0.506926952141058,
      "grad_norm": 0.1294430047273636,
      "learning_rate": 5.7501281171701106e-05,
      "loss": 0.1646,
      "step": 2415
    },
    {
      "epoch": 0.5071368597816961,
      "grad_norm": 0.15343913435935974,
      "learning_rate": 5.7465052936716425e-05,
      "loss": 0.1496,
      "step": 2416
    },
    {
      "epoch": 0.5073467674223342,
      "grad_norm": 0.10801581293344498,
      "learning_rate": 5.7428820692827776e-05,
      "loss": 0.144,
      "step": 2417
    },
    {
      "epoch": 0.5075566750629723,
      "grad_norm": 0.12166569381952286,
      "learning_rate": 5.739258445949275e-05,
      "loss": 0.1625,
      "step": 2418
    },
    {
      "epoch": 0.5077665827036104,
      "grad_norm": 0.0971064567565918,
      "learning_rate": 5.735634425617098e-05,
      "loss": 0.1623,
      "step": 2419
    },
    {
      "epoch": 0.5079764903442485,
      "grad_norm": 0.13411295413970947,
      "learning_rate": 5.7320100102324325e-05,
      "loss": 0.1556,
      "step": 2420
    },
    {
      "epoch": 0.5081863979848866,
      "grad_norm": 0.12255419790744781,
      "learning_rate": 5.7283852017416696e-05,
      "loss": 0.1536,
      "step": 2421
    },
    {
      "epoch": 0.5083963056255247,
      "grad_norm": 0.10330315679311752,
      "learning_rate": 5.724760002091417e-05,
      "loss": 0.1439,
      "step": 2422
    },
    {
      "epoch": 0.5086062132661628,
      "grad_norm": 0.1231582835316658,
      "learning_rate": 5.721134413228486e-05,
      "loss": 0.1585,
      "step": 2423
    },
    {
      "epoch": 0.5088161209068011,
      "grad_norm": 0.11223944276571274,
      "learning_rate": 5.717508437099903e-05,
      "loss": 0.1564,
      "step": 2424
    },
    {
      "epoch": 0.5090260285474392,
      "grad_norm": 0.12430944293737411,
      "learning_rate": 5.713882075652901e-05,
      "loss": 0.1611,
      "step": 2425
    },
    {
      "epoch": 0.5092359361880773,
      "grad_norm": 0.11220264434814453,
      "learning_rate": 5.7102553308349174e-05,
      "loss": 0.1527,
      "step": 2426
    },
    {
      "epoch": 0.5094458438287154,
      "grad_norm": 0.1385231614112854,
      "learning_rate": 5.706628204593596e-05,
      "loss": 0.15,
      "step": 2427
    },
    {
      "epoch": 0.5096557514693535,
      "grad_norm": 0.12652051448822021,
      "learning_rate": 5.703000698876788e-05,
      "loss": 0.148,
      "step": 2428
    },
    {
      "epoch": 0.5098656591099916,
      "grad_norm": 0.11996873468160629,
      "learning_rate": 5.6993728156325466e-05,
      "loss": 0.1569,
      "step": 2429
    },
    {
      "epoch": 0.5100755667506297,
      "grad_norm": 0.1229998916387558,
      "learning_rate": 5.6957445568091284e-05,
      "loss": 0.152,
      "step": 2430
    },
    {
      "epoch": 0.5102854743912678,
      "grad_norm": 0.10919797420501709,
      "learning_rate": 5.692115924354989e-05,
      "loss": 0.1454,
      "step": 2431
    },
    {
      "epoch": 0.5104953820319059,
      "grad_norm": 0.13112346827983856,
      "learning_rate": 5.6884869202187906e-05,
      "loss": 0.1603,
      "step": 2432
    },
    {
      "epoch": 0.510705289672544,
      "grad_norm": 0.11570712178945541,
      "learning_rate": 5.684857546349388e-05,
      "loss": 0.1649,
      "step": 2433
    },
    {
      "epoch": 0.5109151973131822,
      "grad_norm": 0.12106207013130188,
      "learning_rate": 5.68122780469584e-05,
      "loss": 0.1573,
      "step": 2434
    },
    {
      "epoch": 0.5111251049538204,
      "grad_norm": 0.13014116883277893,
      "learning_rate": 5.6775976972073985e-05,
      "loss": 0.1487,
      "step": 2435
    },
    {
      "epoch": 0.5113350125944585,
      "grad_norm": 0.12581390142440796,
      "learning_rate": 5.673967225833519e-05,
      "loss": 0.1679,
      "step": 2436
    },
    {
      "epoch": 0.5115449202350966,
      "grad_norm": 0.15242688357830048,
      "learning_rate": 5.6703363925238404e-05,
      "loss": 0.1614,
      "step": 2437
    },
    {
      "epoch": 0.5117548278757347,
      "grad_norm": 0.11680883914232254,
      "learning_rate": 5.6667051992282094e-05,
      "loss": 0.16,
      "step": 2438
    },
    {
      "epoch": 0.5119647355163728,
      "grad_norm": 0.10956526547670364,
      "learning_rate": 5.663073647896656e-05,
      "loss": 0.1488,
      "step": 2439
    },
    {
      "epoch": 0.5121746431570109,
      "grad_norm": 0.1150960773229599,
      "learning_rate": 5.6594417404794086e-05,
      "loss": 0.1479,
      "step": 2440
    },
    {
      "epoch": 0.512384550797649,
      "grad_norm": 0.11409033834934235,
      "learning_rate": 5.655809478926881e-05,
      "loss": 0.162,
      "step": 2441
    },
    {
      "epoch": 0.5125944584382871,
      "grad_norm": 0.13178421556949615,
      "learning_rate": 5.6521768651896855e-05,
      "loss": 0.1529,
      "step": 2442
    },
    {
      "epoch": 0.5128043660789253,
      "grad_norm": 0.10844630002975464,
      "learning_rate": 5.648543901218615e-05,
      "loss": 0.1551,
      "step": 2443
    },
    {
      "epoch": 0.5130142737195634,
      "grad_norm": 0.10883074998855591,
      "learning_rate": 5.644910588964656e-05,
      "loss": 0.1396,
      "step": 2444
    },
    {
      "epoch": 0.5132241813602015,
      "grad_norm": 0.11679378896951675,
      "learning_rate": 5.641276930378978e-05,
      "loss": 0.1572,
      "step": 2445
    },
    {
      "epoch": 0.5134340890008396,
      "grad_norm": 0.11967083066701889,
      "learning_rate": 5.637642927412942e-05,
      "loss": 0.1731,
      "step": 2446
    },
    {
      "epoch": 0.5136439966414778,
      "grad_norm": 0.10097982734441757,
      "learning_rate": 5.634008582018086e-05,
      "loss": 0.1567,
      "step": 2447
    },
    {
      "epoch": 0.5138539042821159,
      "grad_norm": 0.11456773430109024,
      "learning_rate": 5.630373896146142e-05,
      "loss": 0.1783,
      "step": 2448
    },
    {
      "epoch": 0.514063811922754,
      "grad_norm": 0.12176181375980377,
      "learning_rate": 5.6267388717490144e-05,
      "loss": 0.1558,
      "step": 2449
    },
    {
      "epoch": 0.5142737195633921,
      "grad_norm": 0.11129914224147797,
      "learning_rate": 5.623103510778798e-05,
      "loss": 0.1558,
      "step": 2450
    },
    {
      "epoch": 0.5144836272040302,
      "grad_norm": 0.11947834491729736,
      "learning_rate": 5.619467815187762e-05,
      "loss": 0.1612,
      "step": 2451
    },
    {
      "epoch": 0.5146935348446684,
      "grad_norm": 0.11735592782497406,
      "learning_rate": 5.615831786928358e-05,
      "loss": 0.1593,
      "step": 2452
    },
    {
      "epoch": 0.5149034424853065,
      "grad_norm": 0.10720401257276535,
      "learning_rate": 5.6121954279532206e-05,
      "loss": 0.1529,
      "step": 2453
    },
    {
      "epoch": 0.5151133501259446,
      "grad_norm": 0.11391155421733856,
      "learning_rate": 5.608558740215152e-05,
      "loss": 0.1464,
      "step": 2454
    },
    {
      "epoch": 0.5153232577665827,
      "grad_norm": 0.12067031115293503,
      "learning_rate": 5.604921725667139e-05,
      "loss": 0.1541,
      "step": 2455
    },
    {
      "epoch": 0.5155331654072208,
      "grad_norm": 0.12708424031734467,
      "learning_rate": 5.601284386262342e-05,
      "loss": 0.1629,
      "step": 2456
    },
    {
      "epoch": 0.5157430730478589,
      "grad_norm": 0.11665681004524231,
      "learning_rate": 5.597646723954095e-05,
      "loss": 0.1685,
      "step": 2457
    },
    {
      "epoch": 0.515952980688497,
      "grad_norm": 0.11358939856290817,
      "learning_rate": 5.5940087406959054e-05,
      "loss": 0.147,
      "step": 2458
    },
    {
      "epoch": 0.5161628883291351,
      "grad_norm": 0.11850309371948242,
      "learning_rate": 5.5903704384414535e-05,
      "loss": 0.1663,
      "step": 2459
    },
    {
      "epoch": 0.5163727959697733,
      "grad_norm": 0.10734523087739944,
      "learning_rate": 5.586731819144591e-05,
      "loss": 0.1735,
      "step": 2460
    },
    {
      "epoch": 0.5165827036104114,
      "grad_norm": 0.11942029744386673,
      "learning_rate": 5.5830928847593376e-05,
      "loss": 0.1494,
      "step": 2461
    },
    {
      "epoch": 0.5167926112510496,
      "grad_norm": 0.0976463109254837,
      "learning_rate": 5.579453637239886e-05,
      "loss": 0.1525,
      "step": 2462
    },
    {
      "epoch": 0.5170025188916877,
      "grad_norm": 0.12405764311552048,
      "learning_rate": 5.5758140785405974e-05,
      "loss": 0.1342,
      "step": 2463
    },
    {
      "epoch": 0.5172124265323258,
      "grad_norm": 0.10457320511341095,
      "learning_rate": 5.5721742106159934e-05,
      "loss": 0.1381,
      "step": 2464
    },
    {
      "epoch": 0.5174223341729639,
      "grad_norm": 0.11121641099452972,
      "learning_rate": 5.568534035420771e-05,
      "loss": 0.1613,
      "step": 2465
    },
    {
      "epoch": 0.517632241813602,
      "grad_norm": 0.1370898187160492,
      "learning_rate": 5.564893554909781e-05,
      "loss": 0.16,
      "step": 2466
    },
    {
      "epoch": 0.5178421494542401,
      "grad_norm": 0.14508257806301117,
      "learning_rate": 5.561252771038052e-05,
      "loss": 0.159,
      "step": 2467
    },
    {
      "epoch": 0.5180520570948782,
      "grad_norm": 0.12698331475257874,
      "learning_rate": 5.557611685760763e-05,
      "loss": 0.155,
      "step": 2468
    },
    {
      "epoch": 0.5182619647355163,
      "grad_norm": 0.12996205687522888,
      "learning_rate": 5.5539703010332624e-05,
      "loss": 0.1562,
      "step": 2469
    },
    {
      "epoch": 0.5184718723761544,
      "grad_norm": 0.10742896795272827,
      "learning_rate": 5.550328618811055e-05,
      "loss": 0.1368,
      "step": 2470
    },
    {
      "epoch": 0.5186817800167927,
      "grad_norm": 0.11978666484355927,
      "learning_rate": 5.5466866410498095e-05,
      "loss": 0.1434,
      "step": 2471
    },
    {
      "epoch": 0.5188916876574308,
      "grad_norm": 0.1411794275045395,
      "learning_rate": 5.5430443697053494e-05,
      "loss": 0.1494,
      "step": 2472
    },
    {
      "epoch": 0.5191015952980689,
      "grad_norm": 0.12275288999080658,
      "learning_rate": 5.5394018067336586e-05,
      "loss": 0.145,
      "step": 2473
    },
    {
      "epoch": 0.519311502938707,
      "grad_norm": 0.1017669290304184,
      "learning_rate": 5.535758954090876e-05,
      "loss": 0.1433,
      "step": 2474
    },
    {
      "epoch": 0.5195214105793451,
      "grad_norm": 0.10941984504461288,
      "learning_rate": 5.5321158137332975e-05,
      "loss": 0.1615,
      "step": 2475
    },
    {
      "epoch": 0.5197313182199832,
      "grad_norm": 0.13621585071086884,
      "learning_rate": 5.5284723876173705e-05,
      "loss": 0.1633,
      "step": 2476
    },
    {
      "epoch": 0.5199412258606213,
      "grad_norm": 0.12817037105560303,
      "learning_rate": 5.524828677699703e-05,
      "loss": 0.1513,
      "step": 2477
    },
    {
      "epoch": 0.5201511335012594,
      "grad_norm": 0.125384122133255,
      "learning_rate": 5.521184685937044e-05,
      "loss": 0.1695,
      "step": 2478
    },
    {
      "epoch": 0.5203610411418975,
      "grad_norm": 0.12680883705615997,
      "learning_rate": 5.517540414286306e-05,
      "loss": 0.167,
      "step": 2479
    },
    {
      "epoch": 0.5205709487825357,
      "grad_norm": 0.11645618081092834,
      "learning_rate": 5.5138958647045426e-05,
      "loss": 0.1475,
      "step": 2480
    },
    {
      "epoch": 0.5207808564231738,
      "grad_norm": 0.11582057178020477,
      "learning_rate": 5.510251039148961e-05,
      "loss": 0.1616,
      "step": 2481
    },
    {
      "epoch": 0.520990764063812,
      "grad_norm": 0.11036519706249237,
      "learning_rate": 5.5066059395769164e-05,
      "loss": 0.1553,
      "step": 2482
    },
    {
      "epoch": 0.5212006717044501,
      "grad_norm": 0.16732360422611237,
      "learning_rate": 5.502960567945909e-05,
      "loss": 0.1551,
      "step": 2483
    },
    {
      "epoch": 0.5214105793450882,
      "grad_norm": 0.13218390941619873,
      "learning_rate": 5.49931492621359e-05,
      "loss": 0.1705,
      "step": 2484
    },
    {
      "epoch": 0.5216204869857263,
      "grad_norm": 0.13806398212909698,
      "learning_rate": 5.495669016337749e-05,
      "loss": 0.1754,
      "step": 2485
    },
    {
      "epoch": 0.5218303946263644,
      "grad_norm": 0.13826845586299896,
      "learning_rate": 5.492022840276324e-05,
      "loss": 0.1556,
      "step": 2486
    },
    {
      "epoch": 0.5220403022670025,
      "grad_norm": 0.14759372174739838,
      "learning_rate": 5.488376399987394e-05,
      "loss": 0.1565,
      "step": 2487
    },
    {
      "epoch": 0.5222502099076406,
      "grad_norm": 0.1314990073442459,
      "learning_rate": 5.484729697429183e-05,
      "loss": 0.1668,
      "step": 2488
    },
    {
      "epoch": 0.5224601175482787,
      "grad_norm": 0.1437479555606842,
      "learning_rate": 5.4810827345600516e-05,
      "loss": 0.162,
      "step": 2489
    },
    {
      "epoch": 0.5226700251889169,
      "grad_norm": 0.10261430591344833,
      "learning_rate": 5.4774355133385015e-05,
      "loss": 0.1673,
      "step": 2490
    },
    {
      "epoch": 0.522879932829555,
      "grad_norm": 0.10956492274999619,
      "learning_rate": 5.473788035723176e-05,
      "loss": 0.1699,
      "step": 2491
    },
    {
      "epoch": 0.5230898404701931,
      "grad_norm": 0.14587171375751495,
      "learning_rate": 5.470140303672853e-05,
      "loss": 0.1696,
      "step": 2492
    },
    {
      "epoch": 0.5232997481108312,
      "grad_norm": 0.10633595287799835,
      "learning_rate": 5.466492319146447e-05,
      "loss": 0.1651,
      "step": 2493
    },
    {
      "epoch": 0.5235096557514693,
      "grad_norm": 0.1117524653673172,
      "learning_rate": 5.4628440841030096e-05,
      "loss": 0.1566,
      "step": 2494
    },
    {
      "epoch": 0.5237195633921075,
      "grad_norm": 0.09613008797168732,
      "learning_rate": 5.4591956005017276e-05,
      "loss": 0.1673,
      "step": 2495
    },
    {
      "epoch": 0.5239294710327456,
      "grad_norm": 0.0960233211517334,
      "learning_rate": 5.455546870301919e-05,
      "loss": 0.1304,
      "step": 2496
    },
    {
      "epoch": 0.5241393786733837,
      "grad_norm": 0.1320691704750061,
      "learning_rate": 5.451897895463034e-05,
      "loss": 0.1479,
      "step": 2497
    },
    {
      "epoch": 0.5243492863140218,
      "grad_norm": 0.09648372232913971,
      "learning_rate": 5.448248677944658e-05,
      "loss": 0.1512,
      "step": 2498
    },
    {
      "epoch": 0.52455919395466,
      "grad_norm": 0.12556767463684082,
      "learning_rate": 5.444599219706502e-05,
      "loss": 0.1654,
      "step": 2499
    },
    {
      "epoch": 0.5247691015952981,
      "grad_norm": 0.11492988467216492,
      "learning_rate": 5.440949522708409e-05,
      "loss": 0.1388,
      "step": 2500
    },
    {
      "epoch": 0.5249790092359362,
      "grad_norm": 0.11009164899587631,
      "learning_rate": 5.4372995889103504e-05,
      "loss": 0.1525,
      "step": 2501
    },
    {
      "epoch": 0.5251889168765743,
      "grad_norm": 0.11773049831390381,
      "learning_rate": 5.433649420272425e-05,
      "loss": 0.1393,
      "step": 2502
    },
    {
      "epoch": 0.5253988245172124,
      "grad_norm": 0.14838874340057373,
      "learning_rate": 5.4299990187548535e-05,
      "loss": 0.1666,
      "step": 2503
    },
    {
      "epoch": 0.5256087321578505,
      "grad_norm": 0.12586016952991486,
      "learning_rate": 5.426348386317986e-05,
      "loss": 0.1461,
      "step": 2504
    },
    {
      "epoch": 0.5258186397984886,
      "grad_norm": 0.12975488603115082,
      "learning_rate": 5.422697524922299e-05,
      "loss": 0.1463,
      "step": 2505
    },
    {
      "epoch": 0.5260285474391267,
      "grad_norm": 0.11897221952676773,
      "learning_rate": 5.419046436528384e-05,
      "loss": 0.1606,
      "step": 2506
    },
    {
      "epoch": 0.5262384550797649,
      "grad_norm": 0.12332843244075775,
      "learning_rate": 5.4153951230969604e-05,
      "loss": 0.1395,
      "step": 2507
    },
    {
      "epoch": 0.5264483627204031,
      "grad_norm": 0.1560123860836029,
      "learning_rate": 5.411743586588868e-05,
      "loss": 0.1745,
      "step": 2508
    },
    {
      "epoch": 0.5266582703610412,
      "grad_norm": 0.1406356692314148,
      "learning_rate": 5.408091828965063e-05,
      "loss": 0.1603,
      "step": 2509
    },
    {
      "epoch": 0.5268681780016793,
      "grad_norm": 0.11687145382165909,
      "learning_rate": 5.404439852186624e-05,
      "loss": 0.1644,
      "step": 2510
    },
    {
      "epoch": 0.5270780856423174,
      "grad_norm": 0.09882459789514542,
      "learning_rate": 5.4007876582147465e-05,
      "loss": 0.1432,
      "step": 2511
    },
    {
      "epoch": 0.5272879932829555,
      "grad_norm": 0.14484454691410065,
      "learning_rate": 5.397135249010744e-05,
      "loss": 0.1574,
      "step": 2512
    },
    {
      "epoch": 0.5274979009235936,
      "grad_norm": 0.11317654699087143,
      "learning_rate": 5.393482626536038e-05,
      "loss": 0.1544,
      "step": 2513
    },
    {
      "epoch": 0.5277078085642317,
      "grad_norm": 0.12852750718593597,
      "learning_rate": 5.389829792752175e-05,
      "loss": 0.1545,
      "step": 2514
    },
    {
      "epoch": 0.5279177162048698,
      "grad_norm": 0.11933259665966034,
      "learning_rate": 5.3861767496208085e-05,
      "loss": 0.1527,
      "step": 2515
    },
    {
      "epoch": 0.5281276238455079,
      "grad_norm": 0.12715746462345123,
      "learning_rate": 5.382523499103705e-05,
      "loss": 0.1668,
      "step": 2516
    },
    {
      "epoch": 0.5283375314861462,
      "grad_norm": 0.12052647024393082,
      "learning_rate": 5.378870043162746e-05,
      "loss": 0.153,
      "step": 2517
    },
    {
      "epoch": 0.5285474391267843,
      "grad_norm": 0.17828109860420227,
      "learning_rate": 5.375216383759919e-05,
      "loss": 0.1625,
      "step": 2518
    },
    {
      "epoch": 0.5287573467674224,
      "grad_norm": 0.13423284888267517,
      "learning_rate": 5.371562522857323e-05,
      "loss": 0.1547,
      "step": 2519
    },
    {
      "epoch": 0.5289672544080605,
      "grad_norm": 0.13876746594905853,
      "learning_rate": 5.367908462417166e-05,
      "loss": 0.1533,
      "step": 2520
    },
    {
      "epoch": 0.5291771620486986,
      "grad_norm": 0.11023764312267303,
      "learning_rate": 5.3642542044017593e-05,
      "loss": 0.1598,
      "step": 2521
    },
    {
      "epoch": 0.5293870696893367,
      "grad_norm": 0.09991083294153214,
      "learning_rate": 5.360599750773526e-05,
      "loss": 0.1496,
      "step": 2522
    },
    {
      "epoch": 0.5295969773299748,
      "grad_norm": 0.11138133704662323,
      "learning_rate": 5.356945103494988e-05,
      "loss": 0.1579,
      "step": 2523
    },
    {
      "epoch": 0.5298068849706129,
      "grad_norm": 0.09804137051105499,
      "learning_rate": 5.353290264528776e-05,
      "loss": 0.1649,
      "step": 2524
    },
    {
      "epoch": 0.530016792611251,
      "grad_norm": 0.1129557341337204,
      "learning_rate": 5.3496352358376225e-05,
      "loss": 0.172,
      "step": 2525
    },
    {
      "epoch": 0.5302267002518891,
      "grad_norm": 0.15934525430202484,
      "learning_rate": 5.345980019384361e-05,
      "loss": 0.1357,
      "step": 2526
    },
    {
      "epoch": 0.5304366078925273,
      "grad_norm": 0.11331132799386978,
      "learning_rate": 5.3423246171319264e-05,
      "loss": 0.1515,
      "step": 2527
    },
    {
      "epoch": 0.5306465155331654,
      "grad_norm": 0.11778911203145981,
      "learning_rate": 5.338669031043351e-05,
      "loss": 0.1574,
      "step": 2528
    },
    {
      "epoch": 0.5308564231738035,
      "grad_norm": 0.11816948652267456,
      "learning_rate": 5.3350132630817715e-05,
      "loss": 0.1473,
      "step": 2529
    },
    {
      "epoch": 0.5310663308144417,
      "grad_norm": 0.10855350643396378,
      "learning_rate": 5.331357315210417e-05,
      "loss": 0.1506,
      "step": 2530
    },
    {
      "epoch": 0.5312762384550798,
      "grad_norm": 0.09619951248168945,
      "learning_rate": 5.327701189392616e-05,
      "loss": 0.1481,
      "step": 2531
    },
    {
      "epoch": 0.5314861460957179,
      "grad_norm": 0.26531854271888733,
      "learning_rate": 5.324044887591788e-05,
      "loss": 0.1677,
      "step": 2532
    },
    {
      "epoch": 0.531696053736356,
      "grad_norm": 0.10267012566328049,
      "learning_rate": 5.320388411771458e-05,
      "loss": 0.1474,
      "step": 2533
    },
    {
      "epoch": 0.5319059613769941,
      "grad_norm": 0.12348689883947372,
      "learning_rate": 5.3167317638952285e-05,
      "loss": 0.1638,
      "step": 2534
    },
    {
      "epoch": 0.5321158690176322,
      "grad_norm": 0.10946419090032578,
      "learning_rate": 5.313074945926808e-05,
      "loss": 0.163,
      "step": 2535
    },
    {
      "epoch": 0.5323257766582704,
      "grad_norm": 0.12162724882364273,
      "learning_rate": 5.3094179598299906e-05,
      "loss": 0.1661,
      "step": 2536
    },
    {
      "epoch": 0.5325356842989085,
      "grad_norm": 0.11680992692708969,
      "learning_rate": 5.30576080756866e-05,
      "loss": 0.1595,
      "step": 2537
    },
    {
      "epoch": 0.5327455919395466,
      "grad_norm": 0.1095302477478981,
      "learning_rate": 5.3021034911067926e-05,
      "loss": 0.155,
      "step": 2538
    },
    {
      "epoch": 0.5329554995801847,
      "grad_norm": 0.12963855266571045,
      "learning_rate": 5.298446012408449e-05,
      "loss": 0.1228,
      "step": 2539
    },
    {
      "epoch": 0.5331654072208228,
      "grad_norm": 0.10611241310834885,
      "learning_rate": 5.294788373437779e-05,
      "loss": 0.1619,
      "step": 2540
    },
    {
      "epoch": 0.533375314861461,
      "grad_norm": 0.12048359215259552,
      "learning_rate": 5.29113057615902e-05,
      "loss": 0.1526,
      "step": 2541
    },
    {
      "epoch": 0.533585222502099,
      "grad_norm": 0.1172405406832695,
      "learning_rate": 5.28747262253649e-05,
      "loss": 0.1708,
      "step": 2542
    },
    {
      "epoch": 0.5337951301427372,
      "grad_norm": 0.11585681885480881,
      "learning_rate": 5.2838145145345976e-05,
      "loss": 0.1613,
      "step": 2543
    },
    {
      "epoch": 0.5340050377833753,
      "grad_norm": 0.14023533463478088,
      "learning_rate": 5.280156254117825e-05,
      "loss": 0.1613,
      "step": 2544
    },
    {
      "epoch": 0.5342149454240135,
      "grad_norm": 0.13292895257472992,
      "learning_rate": 5.2764978432507464e-05,
      "loss": 0.1618,
      "step": 2545
    },
    {
      "epoch": 0.5344248530646516,
      "grad_norm": 0.13747252523899078,
      "learning_rate": 5.272839283898009e-05,
      "loss": 0.1577,
      "step": 2546
    },
    {
      "epoch": 0.5346347607052897,
      "grad_norm": 0.12026356905698776,
      "learning_rate": 5.2691805780243454e-05,
      "loss": 0.1552,
      "step": 2547
    },
    {
      "epoch": 0.5348446683459278,
      "grad_norm": 0.11890849471092224,
      "learning_rate": 5.26552172759456e-05,
      "loss": 0.1541,
      "step": 2548
    },
    {
      "epoch": 0.5350545759865659,
      "grad_norm": 0.1233121007680893,
      "learning_rate": 5.2618627345735436e-05,
      "loss": 0.159,
      "step": 2549
    },
    {
      "epoch": 0.535264483627204,
      "grad_norm": 0.11645440012216568,
      "learning_rate": 5.258203600926257e-05,
      "loss": 0.1693,
      "step": 2550
    },
    {
      "epoch": 0.5354743912678421,
      "grad_norm": 0.10614755749702454,
      "learning_rate": 5.25454432861774e-05,
      "loss": 0.1595,
      "step": 2551
    },
    {
      "epoch": 0.5356842989084802,
      "grad_norm": 0.10766322165727615,
      "learning_rate": 5.2508849196131016e-05,
      "loss": 0.1567,
      "step": 2552
    },
    {
      "epoch": 0.5358942065491183,
      "grad_norm": 0.13056086003780365,
      "learning_rate": 5.2472253758775327e-05,
      "loss": 0.1524,
      "step": 2553
    },
    {
      "epoch": 0.5361041141897565,
      "grad_norm": 0.10615919530391693,
      "learning_rate": 5.243565699376287e-05,
      "loss": 0.1624,
      "step": 2554
    },
    {
      "epoch": 0.5363140218303947,
      "grad_norm": 0.10946070402860641,
      "learning_rate": 5.239905892074699e-05,
      "loss": 0.1697,
      "step": 2555
    },
    {
      "epoch": 0.5365239294710328,
      "grad_norm": 0.13687758147716522,
      "learning_rate": 5.236245955938166e-05,
      "loss": 0.1644,
      "step": 2556
    },
    {
      "epoch": 0.5367338371116709,
      "grad_norm": 0.13054099678993225,
      "learning_rate": 5.23258589293216e-05,
      "loss": 0.1557,
      "step": 2557
    },
    {
      "epoch": 0.536943744752309,
      "grad_norm": 0.10445389896631241,
      "learning_rate": 5.228925705022215e-05,
      "loss": 0.1549,
      "step": 2558
    },
    {
      "epoch": 0.5371536523929471,
      "grad_norm": 0.14048103988170624,
      "learning_rate": 5.225265394173938e-05,
      "loss": 0.1556,
      "step": 2559
    },
    {
      "epoch": 0.5373635600335852,
      "grad_norm": 0.12132364511489868,
      "learning_rate": 5.221604962352999e-05,
      "loss": 0.1713,
      "step": 2560
    },
    {
      "epoch": 0.5375734676742233,
      "grad_norm": 0.1090470552444458,
      "learning_rate": 5.217944411525133e-05,
      "loss": 0.1695,
      "step": 2561
    },
    {
      "epoch": 0.5377833753148614,
      "grad_norm": 0.11678619682788849,
      "learning_rate": 5.21428374365614e-05,
      "loss": 0.1702,
      "step": 2562
    },
    {
      "epoch": 0.5379932829554995,
      "grad_norm": 0.13385826349258423,
      "learning_rate": 5.2106229607118805e-05,
      "loss": 0.1614,
      "step": 2563
    },
    {
      "epoch": 0.5382031905961377,
      "grad_norm": 0.12655134499073029,
      "learning_rate": 5.2069620646582805e-05,
      "loss": 0.1385,
      "step": 2564
    },
    {
      "epoch": 0.5384130982367759,
      "grad_norm": 0.13991332054138184,
      "learning_rate": 5.203301057461325e-05,
      "loss": 0.1655,
      "step": 2565
    },
    {
      "epoch": 0.538623005877414,
      "grad_norm": 0.12059177458286285,
      "learning_rate": 5.1996399410870576e-05,
      "loss": 0.1637,
      "step": 2566
    },
    {
      "epoch": 0.5388329135180521,
      "grad_norm": 0.1222875714302063,
      "learning_rate": 5.195978717501582e-05,
      "loss": 0.1477,
      "step": 2567
    },
    {
      "epoch": 0.5390428211586902,
      "grad_norm": 0.11073408275842667,
      "learning_rate": 5.192317388671058e-05,
      "loss": 0.1579,
      "step": 2568
    },
    {
      "epoch": 0.5392527287993283,
      "grad_norm": 0.1314646303653717,
      "learning_rate": 5.188655956561701e-05,
      "loss": 0.1479,
      "step": 2569
    },
    {
      "epoch": 0.5394626364399664,
      "grad_norm": 0.12713418900966644,
      "learning_rate": 5.184994423139788e-05,
      "loss": 0.1586,
      "step": 2570
    },
    {
      "epoch": 0.5396725440806045,
      "grad_norm": 0.1251654475927353,
      "learning_rate": 5.181332790371645e-05,
      "loss": 0.1563,
      "step": 2571
    },
    {
      "epoch": 0.5398824517212426,
      "grad_norm": 0.12144947052001953,
      "learning_rate": 5.17767106022365e-05,
      "loss": 0.1537,
      "step": 2572
    },
    {
      "epoch": 0.5400923593618808,
      "grad_norm": 0.11639848351478577,
      "learning_rate": 5.174009234662237e-05,
      "loss": 0.1411,
      "step": 2573
    },
    {
      "epoch": 0.5403022670025189,
      "grad_norm": 0.11065072566270828,
      "learning_rate": 5.170347315653893e-05,
      "loss": 0.1625,
      "step": 2574
    },
    {
      "epoch": 0.540512174643157,
      "grad_norm": 0.11936859041452408,
      "learning_rate": 5.1666853051651456e-05,
      "loss": 0.161,
      "step": 2575
    },
    {
      "epoch": 0.5407220822837951,
      "grad_norm": 0.11808102577924728,
      "learning_rate": 5.163023205162584e-05,
      "loss": 0.1552,
      "step": 2576
    },
    {
      "epoch": 0.5409319899244333,
      "grad_norm": 0.09937819093465805,
      "learning_rate": 5.159361017612836e-05,
      "loss": 0.1362,
      "step": 2577
    },
    {
      "epoch": 0.5411418975650714,
      "grad_norm": 0.13057714700698853,
      "learning_rate": 5.1556987444825836e-05,
      "loss": 0.1691,
      "step": 2578
    },
    {
      "epoch": 0.5413518052057095,
      "grad_norm": 0.11962155252695084,
      "learning_rate": 5.1520363877385466e-05,
      "loss": 0.1598,
      "step": 2579
    },
    {
      "epoch": 0.5415617128463476,
      "grad_norm": 0.10816418379545212,
      "learning_rate": 5.148373949347498e-05,
      "loss": 0.1354,
      "step": 2580
    },
    {
      "epoch": 0.5417716204869857,
      "grad_norm": 0.1253233402967453,
      "learning_rate": 5.144711431276249e-05,
      "loss": 0.1731,
      "step": 2581
    },
    {
      "epoch": 0.5419815281276238,
      "grad_norm": 0.11587586253881454,
      "learning_rate": 5.1410488354916555e-05,
      "loss": 0.1617,
      "step": 2582
    },
    {
      "epoch": 0.542191435768262,
      "grad_norm": 0.09815829247236252,
      "learning_rate": 5.1373861639606145e-05,
      "loss": 0.1668,
      "step": 2583
    },
    {
      "epoch": 0.5424013434089001,
      "grad_norm": 0.10858689993619919,
      "learning_rate": 5.133723418650067e-05,
      "loss": 0.175,
      "step": 2584
    },
    {
      "epoch": 0.5426112510495382,
      "grad_norm": 0.09952938556671143,
      "learning_rate": 5.130060601526988e-05,
      "loss": 0.1695,
      "step": 2585
    },
    {
      "epoch": 0.5428211586901763,
      "grad_norm": 0.11876317113637924,
      "learning_rate": 5.1263977145583964e-05,
      "loss": 0.1524,
      "step": 2586
    },
    {
      "epoch": 0.5430310663308144,
      "grad_norm": 0.13846537470817566,
      "learning_rate": 5.122734759711343e-05,
      "loss": 0.1654,
      "step": 2587
    },
    {
      "epoch": 0.5432409739714525,
      "grad_norm": 0.12535333633422852,
      "learning_rate": 5.119071738952922e-05,
      "loss": 0.1583,
      "step": 2588
    },
    {
      "epoch": 0.5434508816120907,
      "grad_norm": 0.11854858696460724,
      "learning_rate": 5.115408654250255e-05,
      "loss": 0.1542,
      "step": 2589
    },
    {
      "epoch": 0.5436607892527288,
      "grad_norm": 0.12108839303255081,
      "learning_rate": 5.111745507570507e-05,
      "loss": 0.1707,
      "step": 2590
    },
    {
      "epoch": 0.5438706968933669,
      "grad_norm": 0.11636438220739365,
      "learning_rate": 5.1080823008808665e-05,
      "loss": 0.1692,
      "step": 2591
    },
    {
      "epoch": 0.5440806045340051,
      "grad_norm": 0.12799912691116333,
      "learning_rate": 5.104419036148564e-05,
      "loss": 0.1615,
      "step": 2592
    },
    {
      "epoch": 0.5442905121746432,
      "grad_norm": 0.13550209999084473,
      "learning_rate": 5.1007557153408514e-05,
      "loss": 0.1735,
      "step": 2593
    },
    {
      "epoch": 0.5445004198152813,
      "grad_norm": 0.1329326629638672,
      "learning_rate": 5.09709234042502e-05,
      "loss": 0.1626,
      "step": 2594
    },
    {
      "epoch": 0.5447103274559194,
      "grad_norm": 0.11522291600704193,
      "learning_rate": 5.093428913368383e-05,
      "loss": 0.1679,
      "step": 2595
    },
    {
      "epoch": 0.5449202350965575,
      "grad_norm": 0.12600111961364746,
      "learning_rate": 5.089765436138286e-05,
      "loss": 0.1675,
      "step": 2596
    },
    {
      "epoch": 0.5451301427371956,
      "grad_norm": 0.09551798552274704,
      "learning_rate": 5.086101910702099e-05,
      "loss": 0.139,
      "step": 2597
    },
    {
      "epoch": 0.5453400503778337,
      "grad_norm": 0.1303727775812149,
      "learning_rate": 5.08243833902722e-05,
      "loss": 0.1529,
      "step": 2598
    },
    {
      "epoch": 0.5455499580184718,
      "grad_norm": 0.12384460121393204,
      "learning_rate": 5.078774723081069e-05,
      "loss": 0.1598,
      "step": 2599
    },
    {
      "epoch": 0.5457598656591099,
      "grad_norm": 0.14714710414409637,
      "learning_rate": 5.075111064831092e-05,
      "loss": 0.1645,
      "step": 2600
    },
    {
      "epoch": 0.5459697732997482,
      "grad_norm": 0.11978684365749359,
      "learning_rate": 5.071447366244758e-05,
      "loss": 0.1509,
      "step": 2601
    },
    {
      "epoch": 0.5461796809403863,
      "grad_norm": 0.1111048012971878,
      "learning_rate": 5.0677836292895567e-05,
      "loss": 0.1582,
      "step": 2602
    },
    {
      "epoch": 0.5463895885810244,
      "grad_norm": 0.13465799391269684,
      "learning_rate": 5.064119855932998e-05,
      "loss": 0.1609,
      "step": 2603
    },
    {
      "epoch": 0.5465994962216625,
      "grad_norm": 0.1144922599196434,
      "learning_rate": 5.060456048142611e-05,
      "loss": 0.1664,
      "step": 2604
    },
    {
      "epoch": 0.5468094038623006,
      "grad_norm": 0.11836301535367966,
      "learning_rate": 5.056792207885945e-05,
      "loss": 0.1602,
      "step": 2605
    },
    {
      "epoch": 0.5470193115029387,
      "grad_norm": 0.10931508988142014,
      "learning_rate": 5.053128337130568e-05,
      "loss": 0.1613,
      "step": 2606
    },
    {
      "epoch": 0.5472292191435768,
      "grad_norm": 0.1221369057893753,
      "learning_rate": 5.0494644378440595e-05,
      "loss": 0.1648,
      "step": 2607
    },
    {
      "epoch": 0.5474391267842149,
      "grad_norm": 0.10257924348115921,
      "learning_rate": 5.045800511994017e-05,
      "loss": 0.1468,
      "step": 2608
    },
    {
      "epoch": 0.547649034424853,
      "grad_norm": 0.1801408976316452,
      "learning_rate": 5.0421365615480556e-05,
      "loss": 0.1319,
      "step": 2609
    },
    {
      "epoch": 0.5478589420654912,
      "grad_norm": 0.11400793492794037,
      "learning_rate": 5.038472588473796e-05,
      "loss": 0.1465,
      "step": 2610
    },
    {
      "epoch": 0.5480688497061293,
      "grad_norm": 0.13004319369792938,
      "learning_rate": 5.034808594738878e-05,
      "loss": 0.1747,
      "step": 2611
    },
    {
      "epoch": 0.5482787573467675,
      "grad_norm": 0.13634802401065826,
      "learning_rate": 5.03114458231095e-05,
      "loss": 0.1789,
      "step": 2612
    },
    {
      "epoch": 0.5484886649874056,
      "grad_norm": 0.1203138455748558,
      "learning_rate": 5.02748055315767e-05,
      "loss": 0.1697,
      "step": 2613
    },
    {
      "epoch": 0.5486985726280437,
      "grad_norm": 0.09810598939657211,
      "learning_rate": 5.0238165092467037e-05,
      "loss": 0.1586,
      "step": 2614
    },
    {
      "epoch": 0.5489084802686818,
      "grad_norm": 0.1147264614701271,
      "learning_rate": 5.02015245254573e-05,
      "loss": 0.1738,
      "step": 2615
    },
    {
      "epoch": 0.5491183879093199,
      "grad_norm": 0.11263872683048248,
      "learning_rate": 5.0164883850224274e-05,
      "loss": 0.1583,
      "step": 2616
    },
    {
      "epoch": 0.549328295549958,
      "grad_norm": 0.12805059552192688,
      "learning_rate": 5.012824308644487e-05,
      "loss": 0.1538,
      "step": 2617
    },
    {
      "epoch": 0.5495382031905961,
      "grad_norm": 0.10417455434799194,
      "learning_rate": 5.009160225379598e-05,
      "loss": 0.1664,
      "step": 2618
    },
    {
      "epoch": 0.5497481108312342,
      "grad_norm": 0.12117290496826172,
      "learning_rate": 5.005496137195461e-05,
      "loss": 0.1587,
      "step": 2619
    },
    {
      "epoch": 0.5499580184718724,
      "grad_norm": 0.11718293279409409,
      "learning_rate": 5.00183204605977e-05,
      "loss": 0.1491,
      "step": 2620
    },
    {
      "epoch": 0.5501679261125105,
      "grad_norm": 0.12543454766273499,
      "learning_rate": 4.9981679539402316e-05,
      "loss": 0.1582,
      "step": 2621
    },
    {
      "epoch": 0.5503778337531486,
      "grad_norm": 0.09536810219287872,
      "learning_rate": 4.9945038628045414e-05,
      "loss": 0.1578,
      "step": 2622
    },
    {
      "epoch": 0.5505877413937867,
      "grad_norm": 0.11349605768918991,
      "learning_rate": 4.990839774620403e-05,
      "loss": 0.1595,
      "step": 2623
    },
    {
      "epoch": 0.5507976490344249,
      "grad_norm": 0.11620128154754639,
      "learning_rate": 4.9871756913555146e-05,
      "loss": 0.1535,
      "step": 2624
    },
    {
      "epoch": 0.551007556675063,
      "grad_norm": 0.11295798420906067,
      "learning_rate": 4.983511614977574e-05,
      "loss": 0.1662,
      "step": 2625
    },
    {
      "epoch": 0.5512174643157011,
      "grad_norm": 0.13581858575344086,
      "learning_rate": 4.979847547454271e-05,
      "loss": 0.1761,
      "step": 2626
    },
    {
      "epoch": 0.5514273719563392,
      "grad_norm": 0.12520180642604828,
      "learning_rate": 4.976183490753296e-05,
      "loss": 0.1394,
      "step": 2627
    },
    {
      "epoch": 0.5516372795969773,
      "grad_norm": 0.17680776119232178,
      "learning_rate": 4.972519446842333e-05,
      "loss": 0.1417,
      "step": 2628
    },
    {
      "epoch": 0.5518471872376155,
      "grad_norm": 0.11935435235500336,
      "learning_rate": 4.9688554176890523e-05,
      "loss": 0.1684,
      "step": 2629
    },
    {
      "epoch": 0.5520570948782536,
      "grad_norm": 0.12541987001895905,
      "learning_rate": 4.965191405261123e-05,
      "loss": 0.1568,
      "step": 2630
    },
    {
      "epoch": 0.5522670025188917,
      "grad_norm": 0.09537042677402496,
      "learning_rate": 4.9615274115262056e-05,
      "loss": 0.1452,
      "step": 2631
    },
    {
      "epoch": 0.5524769101595298,
      "grad_norm": 0.10122919082641602,
      "learning_rate": 4.9578634384519456e-05,
      "loss": 0.1543,
      "step": 2632
    },
    {
      "epoch": 0.5526868178001679,
      "grad_norm": 0.12309399247169495,
      "learning_rate": 4.954199488005983e-05,
      "loss": 0.16,
      "step": 2633
    },
    {
      "epoch": 0.552896725440806,
      "grad_norm": 0.12081638723611832,
      "learning_rate": 4.9505355621559416e-05,
      "loss": 0.1726,
      "step": 2634
    },
    {
      "epoch": 0.5531066330814441,
      "grad_norm": 0.12514087557792664,
      "learning_rate": 4.946871662869434e-05,
      "loss": 0.1374,
      "step": 2635
    },
    {
      "epoch": 0.5533165407220823,
      "grad_norm": 0.12766608595848083,
      "learning_rate": 4.943207792114055e-05,
      "loss": 0.1481,
      "step": 2636
    },
    {
      "epoch": 0.5535264483627204,
      "grad_norm": 0.1229751855134964,
      "learning_rate": 4.9395439518573904e-05,
      "loss": 0.1447,
      "step": 2637
    },
    {
      "epoch": 0.5537363560033586,
      "grad_norm": 0.12326975166797638,
      "learning_rate": 4.935880144067003e-05,
      "loss": 0.1579,
      "step": 2638
    },
    {
      "epoch": 0.5539462636439967,
      "grad_norm": 0.12600168585777283,
      "learning_rate": 4.932216370710444e-05,
      "loss": 0.1614,
      "step": 2639
    },
    {
      "epoch": 0.5541561712846348,
      "grad_norm": 0.10681076347827911,
      "learning_rate": 4.928552633755243e-05,
      "loss": 0.169,
      "step": 2640
    },
    {
      "epoch": 0.5543660789252729,
      "grad_norm": 0.12379185855388641,
      "learning_rate": 4.9248889351689074e-05,
      "loss": 0.1744,
      "step": 2641
    },
    {
      "epoch": 0.554575986565911,
      "grad_norm": 0.1300443708896637,
      "learning_rate": 4.9212252769189324e-05,
      "loss": 0.1698,
      "step": 2642
    },
    {
      "epoch": 0.5547858942065491,
      "grad_norm": 0.12122491747140884,
      "learning_rate": 4.917561660972781e-05,
      "loss": 0.1606,
      "step": 2643
    },
    {
      "epoch": 0.5549958018471872,
      "grad_norm": 0.10484204441308975,
      "learning_rate": 4.913898089297903e-05,
      "loss": 0.1355,
      "step": 2644
    },
    {
      "epoch": 0.5552057094878253,
      "grad_norm": 0.12357357144355774,
      "learning_rate": 4.9102345638617145e-05,
      "loss": 0.1752,
      "step": 2645
    },
    {
      "epoch": 0.5554156171284634,
      "grad_norm": 0.09464117139577866,
      "learning_rate": 4.9065710866316175e-05,
      "loss": 0.1591,
      "step": 2646
    },
    {
      "epoch": 0.5556255247691015,
      "grad_norm": 0.11045074462890625,
      "learning_rate": 4.9029076595749804e-05,
      "loss": 0.1484,
      "step": 2647
    },
    {
      "epoch": 0.5558354324097398,
      "grad_norm": 0.16476228833198547,
      "learning_rate": 4.899244284659148e-05,
      "loss": 0.147,
      "step": 2648
    },
    {
      "epoch": 0.5560453400503779,
      "grad_norm": 0.13166958093643188,
      "learning_rate": 4.895580963851438e-05,
      "loss": 0.1555,
      "step": 2649
    },
    {
      "epoch": 0.556255247691016,
      "grad_norm": 0.12306392192840576,
      "learning_rate": 4.8919176991191346e-05,
      "loss": 0.1481,
      "step": 2650
    },
    {
      "epoch": 0.5564651553316541,
      "grad_norm": 0.11133769154548645,
      "learning_rate": 4.888254492429494e-05,
      "loss": 0.138,
      "step": 2651
    },
    {
      "epoch": 0.5566750629722922,
      "grad_norm": 0.11845657974481583,
      "learning_rate": 4.884591345749746e-05,
      "loss": 0.15,
      "step": 2652
    },
    {
      "epoch": 0.5568849706129303,
      "grad_norm": 0.10165873169898987,
      "learning_rate": 4.880928261047079e-05,
      "loss": 0.1562,
      "step": 2653
    },
    {
      "epoch": 0.5570948782535684,
      "grad_norm": 0.11259020119905472,
      "learning_rate": 4.877265240288657e-05,
      "loss": 0.1548,
      "step": 2654
    },
    {
      "epoch": 0.5573047858942065,
      "grad_norm": 0.12641459703445435,
      "learning_rate": 4.873602285441604e-05,
      "loss": 0.1682,
      "step": 2655
    },
    {
      "epoch": 0.5575146935348446,
      "grad_norm": 0.11751089245080948,
      "learning_rate": 4.869939398473014e-05,
      "loss": 0.1488,
      "step": 2656
    },
    {
      "epoch": 0.5577246011754828,
      "grad_norm": 0.13346509635448456,
      "learning_rate": 4.866276581349934e-05,
      "loss": 0.1501,
      "step": 2657
    },
    {
      "epoch": 0.5579345088161209,
      "grad_norm": 0.10571077466011047,
      "learning_rate": 4.8626138360393867e-05,
      "loss": 0.1431,
      "step": 2658
    },
    {
      "epoch": 0.558144416456759,
      "grad_norm": 0.12459591776132584,
      "learning_rate": 4.858951164508346e-05,
      "loss": 0.1528,
      "step": 2659
    },
    {
      "epoch": 0.5583543240973972,
      "grad_norm": 0.15180979669094086,
      "learning_rate": 4.8552885687237525e-05,
      "loss": 0.1573,
      "step": 2660
    },
    {
      "epoch": 0.5585642317380353,
      "grad_norm": 0.1342502236366272,
      "learning_rate": 4.8516260506525024e-05,
      "loss": 0.1638,
      "step": 2661
    },
    {
      "epoch": 0.5587741393786734,
      "grad_norm": 0.11851493269205093,
      "learning_rate": 4.847963612261454e-05,
      "loss": 0.1759,
      "step": 2662
    },
    {
      "epoch": 0.5589840470193115,
      "grad_norm": 0.1613032966852188,
      "learning_rate": 4.844301255517419e-05,
      "loss": 0.1781,
      "step": 2663
    },
    {
      "epoch": 0.5591939546599496,
      "grad_norm": 0.11613575369119644,
      "learning_rate": 4.840638982387165e-05,
      "loss": 0.1406,
      "step": 2664
    },
    {
      "epoch": 0.5594038623005877,
      "grad_norm": 0.11966195702552795,
      "learning_rate": 4.8369767948374173e-05,
      "loss": 0.1617,
      "step": 2665
    },
    {
      "epoch": 0.5596137699412259,
      "grad_norm": 0.11400119960308075,
      "learning_rate": 4.833314694834855e-05,
      "loss": 0.1483,
      "step": 2666
    },
    {
      "epoch": 0.559823677581864,
      "grad_norm": 0.10645996034145355,
      "learning_rate": 4.829652684346109e-05,
      "loss": 0.1568,
      "step": 2667
    },
    {
      "epoch": 0.5600335852225021,
      "grad_norm": 0.1172577515244484,
      "learning_rate": 4.825990765337763e-05,
      "loss": 0.1661,
      "step": 2668
    },
    {
      "epoch": 0.5602434928631402,
      "grad_norm": 0.15190641582012177,
      "learning_rate": 4.8223289397763514e-05,
      "loss": 0.1571,
      "step": 2669
    },
    {
      "epoch": 0.5604534005037783,
      "grad_norm": 0.10944955050945282,
      "learning_rate": 4.818667209628358e-05,
      "loss": 0.1562,
      "step": 2670
    },
    {
      "epoch": 0.5606633081444165,
      "grad_norm": 0.11813446134328842,
      "learning_rate": 4.815005576860212e-05,
      "loss": 0.1637,
      "step": 2671
    },
    {
      "epoch": 0.5608732157850546,
      "grad_norm": 0.11116372793912888,
      "learning_rate": 4.8113440434383e-05,
      "loss": 0.1475,
      "step": 2672
    },
    {
      "epoch": 0.5610831234256927,
      "grad_norm": 0.14472845196723938,
      "learning_rate": 4.807682611328944e-05,
      "loss": 0.1577,
      "step": 2673
    },
    {
      "epoch": 0.5612930310663308,
      "grad_norm": 0.11451143026351929,
      "learning_rate": 4.80402128249842e-05,
      "loss": 0.1526,
      "step": 2674
    },
    {
      "epoch": 0.5615029387069689,
      "grad_norm": 0.09532547742128372,
      "learning_rate": 4.8003600589129435e-05,
      "loss": 0.1426,
      "step": 2675
    },
    {
      "epoch": 0.5617128463476071,
      "grad_norm": 0.11534955352544785,
      "learning_rate": 4.796698942538677e-05,
      "loss": 0.1606,
      "step": 2676
    },
    {
      "epoch": 0.5619227539882452,
      "grad_norm": 0.10720717161893845,
      "learning_rate": 4.7930379353417206e-05,
      "loss": 0.1667,
      "step": 2677
    },
    {
      "epoch": 0.5621326616288833,
      "grad_norm": 0.09807775169610977,
      "learning_rate": 4.789377039288121e-05,
      "loss": 0.1641,
      "step": 2678
    },
    {
      "epoch": 0.5623425692695214,
      "grad_norm": 0.12788937985897064,
      "learning_rate": 4.7857162563438614e-05,
      "loss": 0.1721,
      "step": 2679
    },
    {
      "epoch": 0.5625524769101595,
      "grad_norm": 0.11621350049972534,
      "learning_rate": 4.782055588474868e-05,
      "loss": 0.163,
      "step": 2680
    },
    {
      "epoch": 0.5627623845507976,
      "grad_norm": 0.10405416041612625,
      "learning_rate": 4.778395037647002e-05,
      "loss": 0.1434,
      "step": 2681
    },
    {
      "epoch": 0.5629722921914357,
      "grad_norm": 0.13004978001117706,
      "learning_rate": 4.7747346058260615e-05,
      "loss": 0.1514,
      "step": 2682
    },
    {
      "epoch": 0.5631821998320738,
      "grad_norm": 0.10238310694694519,
      "learning_rate": 4.771074294977786e-05,
      "loss": 0.1637,
      "step": 2683
    },
    {
      "epoch": 0.563392107472712,
      "grad_norm": 0.1132587268948555,
      "learning_rate": 4.767414107067841e-05,
      "loss": 0.1593,
      "step": 2684
    },
    {
      "epoch": 0.5636020151133502,
      "grad_norm": 0.12371500581502914,
      "learning_rate": 4.763754044061834e-05,
      "loss": 0.1538,
      "step": 2685
    },
    {
      "epoch": 0.5638119227539883,
      "grad_norm": 0.11944448202848434,
      "learning_rate": 4.7600941079253016e-05,
      "loss": 0.1496,
      "step": 2686
    },
    {
      "epoch": 0.5640218303946264,
      "grad_norm": 0.11663755774497986,
      "learning_rate": 4.7564343006237136e-05,
      "loss": 0.1606,
      "step": 2687
    },
    {
      "epoch": 0.5642317380352645,
      "grad_norm": 0.10922671854496002,
      "learning_rate": 4.7527746241224685e-05,
      "loss": 0.1746,
      "step": 2688
    },
    {
      "epoch": 0.5644416456759026,
      "grad_norm": 0.11024421453475952,
      "learning_rate": 4.749115080386899e-05,
      "loss": 0.1628,
      "step": 2689
    },
    {
      "epoch": 0.5646515533165407,
      "grad_norm": 0.12064459174871445,
      "learning_rate": 4.745455671382263e-05,
      "loss": 0.1449,
      "step": 2690
    },
    {
      "epoch": 0.5648614609571788,
      "grad_norm": 0.11907904595136642,
      "learning_rate": 4.7417963990737445e-05,
      "loss": 0.1435,
      "step": 2691
    },
    {
      "epoch": 0.5650713685978169,
      "grad_norm": 0.13502107560634613,
      "learning_rate": 4.738137265426457e-05,
      "loss": 0.149,
      "step": 2692
    },
    {
      "epoch": 0.565281276238455,
      "grad_norm": 0.13327385485172272,
      "learning_rate": 4.734478272405441e-05,
      "loss": 0.1581,
      "step": 2693
    },
    {
      "epoch": 0.5654911838790933,
      "grad_norm": 0.10342027992010117,
      "learning_rate": 4.730819421975656e-05,
      "loss": 0.1312,
      "step": 2694
    },
    {
      "epoch": 0.5657010915197314,
      "grad_norm": 0.1462603062391281,
      "learning_rate": 4.7271607161019915e-05,
      "loss": 0.1568,
      "step": 2695
    },
    {
      "epoch": 0.5659109991603695,
      "grad_norm": 0.10993596166372299,
      "learning_rate": 4.723502156749254e-05,
      "loss": 0.1525,
      "step": 2696
    },
    {
      "epoch": 0.5661209068010076,
      "grad_norm": 0.11225152015686035,
      "learning_rate": 4.719843745882177e-05,
      "loss": 0.1591,
      "step": 2697
    },
    {
      "epoch": 0.5663308144416457,
      "grad_norm": 0.1225719004869461,
      "learning_rate": 4.716185485465405e-05,
      "loss": 0.1305,
      "step": 2698
    },
    {
      "epoch": 0.5665407220822838,
      "grad_norm": 0.11367135494947433,
      "learning_rate": 4.712527377463511e-05,
      "loss": 0.1628,
      "step": 2699
    },
    {
      "epoch": 0.5667506297229219,
      "grad_norm": 0.13585880398750305,
      "learning_rate": 4.708869423840981e-05,
      "loss": 0.1495,
      "step": 2700
    },
    {
      "epoch": 0.56696053736356,
      "grad_norm": 0.14989891648292542,
      "learning_rate": 4.7052116265622214e-05,
      "loss": 0.1574,
      "step": 2701
    },
    {
      "epoch": 0.5671704450041981,
      "grad_norm": 0.1249023824930191,
      "learning_rate": 4.7015539875915515e-05,
      "loss": 0.1598,
      "step": 2702
    },
    {
      "epoch": 0.5673803526448362,
      "grad_norm": 0.10144059360027313,
      "learning_rate": 4.697896508893208e-05,
      "loss": 0.1474,
      "step": 2703
    },
    {
      "epoch": 0.5675902602854744,
      "grad_norm": 0.11371605843305588,
      "learning_rate": 4.694239192431341e-05,
      "loss": 0.153,
      "step": 2704
    },
    {
      "epoch": 0.5678001679261125,
      "grad_norm": 0.11912527680397034,
      "learning_rate": 4.690582040170012e-05,
      "loss": 0.1626,
      "step": 2705
    },
    {
      "epoch": 0.5680100755667506,
      "grad_norm": 0.11181552708148956,
      "learning_rate": 4.686925054073193e-05,
      "loss": 0.1519,
      "step": 2706
    },
    {
      "epoch": 0.5682199832073888,
      "grad_norm": 0.10603776574134827,
      "learning_rate": 4.683268236104773e-05,
      "loss": 0.1548,
      "step": 2707
    },
    {
      "epoch": 0.5684298908480269,
      "grad_norm": 0.12990587949752808,
      "learning_rate": 4.679611588228544e-05,
      "loss": 0.1661,
      "step": 2708
    },
    {
      "epoch": 0.568639798488665,
      "grad_norm": 0.10635796189308167,
      "learning_rate": 4.675955112408211e-05,
      "loss": 0.1363,
      "step": 2709
    },
    {
      "epoch": 0.5688497061293031,
      "grad_norm": 0.13790719211101532,
      "learning_rate": 4.6722988106073855e-05,
      "loss": 0.1739,
      "step": 2710
    },
    {
      "epoch": 0.5690596137699412,
      "grad_norm": 0.1612750142812729,
      "learning_rate": 4.668642684789585e-05,
      "loss": 0.1429,
      "step": 2711
    },
    {
      "epoch": 0.5692695214105793,
      "grad_norm": 0.09567128121852875,
      "learning_rate": 4.664986736918229e-05,
      "loss": 0.1346,
      "step": 2712
    },
    {
      "epoch": 0.5694794290512175,
      "grad_norm": 0.10926074534654617,
      "learning_rate": 4.6613309689566494e-05,
      "loss": 0.168,
      "step": 2713
    },
    {
      "epoch": 0.5696893366918556,
      "grad_norm": 0.12982285022735596,
      "learning_rate": 4.657675382868075e-05,
      "loss": 0.1559,
      "step": 2714
    },
    {
      "epoch": 0.5698992443324937,
      "grad_norm": 0.11957161128520966,
      "learning_rate": 4.65401998061564e-05,
      "loss": 0.1705,
      "step": 2715
    },
    {
      "epoch": 0.5701091519731318,
      "grad_norm": 0.1028459444642067,
      "learning_rate": 4.650364764162378e-05,
      "loss": 0.146,
      "step": 2716
    },
    {
      "epoch": 0.5703190596137699,
      "grad_norm": 0.1195150837302208,
      "learning_rate": 4.6467097354712244e-05,
      "loss": 0.1469,
      "step": 2717
    },
    {
      "epoch": 0.570528967254408,
      "grad_norm": 0.11445170640945435,
      "learning_rate": 4.6430548965050136e-05,
      "loss": 0.1748,
      "step": 2718
    },
    {
      "epoch": 0.5707388748950462,
      "grad_norm": 0.10902340710163116,
      "learning_rate": 4.6394002492264767e-05,
      "loss": 0.145,
      "step": 2719
    },
    {
      "epoch": 0.5709487825356843,
      "grad_norm": 0.09901425987482071,
      "learning_rate": 4.635745795598241e-05,
      "loss": 0.1384,
      "step": 2720
    },
    {
      "epoch": 0.5711586901763224,
      "grad_norm": 0.1242944598197937,
      "learning_rate": 4.6320915375828356e-05,
      "loss": 0.1573,
      "step": 2721
    },
    {
      "epoch": 0.5713685978169606,
      "grad_norm": 0.11540963500738144,
      "learning_rate": 4.6284374771426775e-05,
      "loss": 0.1307,
      "step": 2722
    },
    {
      "epoch": 0.5715785054575987,
      "grad_norm": 0.10758115351200104,
      "learning_rate": 4.624783616240081e-05,
      "loss": 0.165,
      "step": 2723
    },
    {
      "epoch": 0.5717884130982368,
      "grad_norm": 0.11912379413843155,
      "learning_rate": 4.6211299568372545e-05,
      "loss": 0.1502,
      "step": 2724
    },
    {
      "epoch": 0.5719983207388749,
      "grad_norm": 0.10736312717199326,
      "learning_rate": 4.617476500896296e-05,
      "loss": 0.1508,
      "step": 2725
    },
    {
      "epoch": 0.572208228379513,
      "grad_norm": 0.10583416372537613,
      "learning_rate": 4.613823250379194e-05,
      "loss": 0.1646,
      "step": 2726
    },
    {
      "epoch": 0.5724181360201511,
      "grad_norm": 0.10542289167642593,
      "learning_rate": 4.610170207247826e-05,
      "loss": 0.1598,
      "step": 2727
    },
    {
      "epoch": 0.5726280436607892,
      "grad_norm": 0.12792536616325378,
      "learning_rate": 4.606517373463963e-05,
      "loss": 0.1554,
      "step": 2728
    },
    {
      "epoch": 0.5728379513014273,
      "grad_norm": 0.10677836835384369,
      "learning_rate": 4.6028647509892574e-05,
      "loss": 0.1621,
      "step": 2729
    },
    {
      "epoch": 0.5730478589420654,
      "grad_norm": 0.10829446464776993,
      "learning_rate": 4.5992123417852526e-05,
      "loss": 0.1507,
      "step": 2730
    },
    {
      "epoch": 0.5732577665827037,
      "grad_norm": 0.1175784021615982,
      "learning_rate": 4.595560147813375e-05,
      "loss": 0.1704,
      "step": 2731
    },
    {
      "epoch": 0.5734676742233418,
      "grad_norm": 0.10802135616540909,
      "learning_rate": 4.5919081710349396e-05,
      "loss": 0.1707,
      "step": 2732
    },
    {
      "epoch": 0.5736775818639799,
      "grad_norm": 0.11632566154003143,
      "learning_rate": 4.588256413411134e-05,
      "loss": 0.1642,
      "step": 2733
    },
    {
      "epoch": 0.573887489504618,
      "grad_norm": 0.12177208811044693,
      "learning_rate": 4.5846048769030414e-05,
      "loss": 0.156,
      "step": 2734
    },
    {
      "epoch": 0.5740973971452561,
      "grad_norm": 0.11091190576553345,
      "learning_rate": 4.5809535634716175e-05,
      "loss": 0.1666,
      "step": 2735
    },
    {
      "epoch": 0.5743073047858942,
      "grad_norm": 0.11668343096971512,
      "learning_rate": 4.577302475077703e-05,
      "loss": 0.1381,
      "step": 2736
    },
    {
      "epoch": 0.5745172124265323,
      "grad_norm": 0.10720910876989365,
      "learning_rate": 4.573651613682013e-05,
      "loss": 0.1391,
      "step": 2737
    },
    {
      "epoch": 0.5747271200671704,
      "grad_norm": 0.12557905912399292,
      "learning_rate": 4.570000981245147e-05,
      "loss": 0.1548,
      "step": 2738
    },
    {
      "epoch": 0.5749370277078085,
      "grad_norm": 0.11367420852184296,
      "learning_rate": 4.566350579727577e-05,
      "loss": 0.1329,
      "step": 2739
    },
    {
      "epoch": 0.5751469353484466,
      "grad_norm": 0.11300312727689743,
      "learning_rate": 4.562700411089651e-05,
      "loss": 0.164,
      "step": 2740
    },
    {
      "epoch": 0.5753568429890848,
      "grad_norm": 0.1259835660457611,
      "learning_rate": 4.559050477291591e-05,
      "loss": 0.1534,
      "step": 2741
    },
    {
      "epoch": 0.575566750629723,
      "grad_norm": 0.10597500950098038,
      "learning_rate": 4.555400780293499e-05,
      "loss": 0.148,
      "step": 2742
    },
    {
      "epoch": 0.5757766582703611,
      "grad_norm": 0.11727175116539001,
      "learning_rate": 4.551751322055343e-05,
      "loss": 0.1628,
      "step": 2743
    },
    {
      "epoch": 0.5759865659109992,
      "grad_norm": 0.10633343458175659,
      "learning_rate": 4.5481021045369664e-05,
      "loss": 0.1497,
      "step": 2744
    },
    {
      "epoch": 0.5761964735516373,
      "grad_norm": 0.12353439629077911,
      "learning_rate": 4.5444531296980824e-05,
      "loss": 0.149,
      "step": 2745
    },
    {
      "epoch": 0.5764063811922754,
      "grad_norm": 0.1032649576663971,
      "learning_rate": 4.540804399498275e-05,
      "loss": 0.1385,
      "step": 2746
    },
    {
      "epoch": 0.5766162888329135,
      "grad_norm": 0.13375771045684814,
      "learning_rate": 4.537155915896991e-05,
      "loss": 0.1561,
      "step": 2747
    },
    {
      "epoch": 0.5768261964735516,
      "grad_norm": 0.11897927522659302,
      "learning_rate": 4.533507680853555e-05,
      "loss": 0.1527,
      "step": 2748
    },
    {
      "epoch": 0.5770361041141897,
      "grad_norm": 0.12122731655836105,
      "learning_rate": 4.5298596963271485e-05,
      "loss": 0.1456,
      "step": 2749
    },
    {
      "epoch": 0.5772460117548279,
      "grad_norm": 0.11372151225805283,
      "learning_rate": 4.5262119642768255e-05,
      "loss": 0.1489,
      "step": 2750
    },
    {
      "epoch": 0.577455919395466,
      "grad_norm": 0.10282732546329498,
      "learning_rate": 4.5225644866615e-05,
      "loss": 0.1572,
      "step": 2751
    },
    {
      "epoch": 0.5776658270361041,
      "grad_norm": 0.10694164037704468,
      "learning_rate": 4.518917265439949e-05,
      "loss": 0.1388,
      "step": 2752
    },
    {
      "epoch": 0.5778757346767422,
      "grad_norm": 0.11439422518014908,
      "learning_rate": 4.515270302570819e-05,
      "loss": 0.143,
      "step": 2753
    },
    {
      "epoch": 0.5780856423173804,
      "grad_norm": 0.11277041584253311,
      "learning_rate": 4.511623600012607e-05,
      "loss": 0.1586,
      "step": 2754
    },
    {
      "epoch": 0.5782955499580185,
      "grad_norm": 0.11806753277778625,
      "learning_rate": 4.507977159723677e-05,
      "loss": 0.1676,
      "step": 2755
    },
    {
      "epoch": 0.5785054575986566,
      "grad_norm": 0.1432192325592041,
      "learning_rate": 4.504330983662251e-05,
      "loss": 0.166,
      "step": 2756
    },
    {
      "epoch": 0.5787153652392947,
      "grad_norm": 0.11583569645881653,
      "learning_rate": 4.500685073786411e-05,
      "loss": 0.1629,
      "step": 2757
    },
    {
      "epoch": 0.5789252728799328,
      "grad_norm": 0.09797297418117523,
      "learning_rate": 4.49703943205409e-05,
      "loss": 0.1543,
      "step": 2758
    },
    {
      "epoch": 0.579135180520571,
      "grad_norm": 0.12834526598453522,
      "learning_rate": 4.493394060423084e-05,
      "loss": 0.1506,
      "step": 2759
    },
    {
      "epoch": 0.5793450881612091,
      "grad_norm": 0.1404927521944046,
      "learning_rate": 4.4897489608510406e-05,
      "loss": 0.1588,
      "step": 2760
    },
    {
      "epoch": 0.5795549958018472,
      "grad_norm": 0.11288987100124359,
      "learning_rate": 4.486104135295458e-05,
      "loss": 0.1622,
      "step": 2761
    },
    {
      "epoch": 0.5797649034424853,
      "grad_norm": 0.11807925254106522,
      "learning_rate": 4.482459585713695e-05,
      "loss": 0.1612,
      "step": 2762
    },
    {
      "epoch": 0.5799748110831234,
      "grad_norm": 0.142020121216774,
      "learning_rate": 4.4788153140629564e-05,
      "loss": 0.1351,
      "step": 2763
    },
    {
      "epoch": 0.5801847187237615,
      "grad_norm": 0.10823410749435425,
      "learning_rate": 4.4751713223002975e-05,
      "loss": 0.1501,
      "step": 2764
    },
    {
      "epoch": 0.5803946263643996,
      "grad_norm": 0.11584329605102539,
      "learning_rate": 4.4715276123826286e-05,
      "loss": 0.153,
      "step": 2765
    },
    {
      "epoch": 0.5806045340050378,
      "grad_norm": 0.11066281795501709,
      "learning_rate": 4.467884186266705e-05,
      "loss": 0.1457,
      "step": 2766
    },
    {
      "epoch": 0.5808144416456759,
      "grad_norm": 0.1182861477136612,
      "learning_rate": 4.464241045909125e-05,
      "loss": 0.1589,
      "step": 2767
    },
    {
      "epoch": 0.581024349286314,
      "grad_norm": 0.13916128873825073,
      "learning_rate": 4.460598193266343e-05,
      "loss": 0.1642,
      "step": 2768
    },
    {
      "epoch": 0.5812342569269522,
      "grad_norm": 0.12714353203773499,
      "learning_rate": 4.4569556302946525e-05,
      "loss": 0.1486,
      "step": 2769
    },
    {
      "epoch": 0.5814441645675903,
      "grad_norm": 0.10147153586149216,
      "learning_rate": 4.4533133589501916e-05,
      "loss": 0.1613,
      "step": 2770
    },
    {
      "epoch": 0.5816540722082284,
      "grad_norm": 0.1153976321220398,
      "learning_rate": 4.4496713811889455e-05,
      "loss": 0.1864,
      "step": 2771
    },
    {
      "epoch": 0.5818639798488665,
      "grad_norm": 0.0999862402677536,
      "learning_rate": 4.446029698966738e-05,
      "loss": 0.1459,
      "step": 2772
    },
    {
      "epoch": 0.5820738874895046,
      "grad_norm": 0.11894406378269196,
      "learning_rate": 4.442388314239239e-05,
      "loss": 0.151,
      "step": 2773
    },
    {
      "epoch": 0.5822837951301427,
      "grad_norm": 0.13212604820728302,
      "learning_rate": 4.4387472289619496e-05,
      "loss": 0.1583,
      "step": 2774
    },
    {
      "epoch": 0.5824937027707808,
      "grad_norm": 0.11380186676979065,
      "learning_rate": 4.43510644509022e-05,
      "loss": 0.1678,
      "step": 2775
    },
    {
      "epoch": 0.5827036104114189,
      "grad_norm": 0.11516433954238892,
      "learning_rate": 4.431465964579231e-05,
      "loss": 0.1676,
      "step": 2776
    },
    {
      "epoch": 0.582913518052057,
      "grad_norm": 0.14080291986465454,
      "learning_rate": 4.427825789384007e-05,
      "loss": 0.1407,
      "step": 2777
    },
    {
      "epoch": 0.5831234256926953,
      "grad_norm": 0.13508716225624084,
      "learning_rate": 4.424185921459403e-05,
      "loss": 0.1608,
      "step": 2778
    },
    {
      "epoch": 0.5833333333333334,
      "grad_norm": 0.11875400692224503,
      "learning_rate": 4.4205463627601135e-05,
      "loss": 0.1561,
      "step": 2779
    },
    {
      "epoch": 0.5835432409739715,
      "grad_norm": 0.12027221918106079,
      "learning_rate": 4.4169071152406636e-05,
      "loss": 0.1537,
      "step": 2780
    },
    {
      "epoch": 0.5837531486146096,
      "grad_norm": 0.09931481629610062,
      "learning_rate": 4.4132681808554124e-05,
      "loss": 0.1623,
      "step": 2781
    },
    {
      "epoch": 0.5839630562552477,
      "grad_norm": 0.12637794017791748,
      "learning_rate": 4.4096295615585483e-05,
      "loss": 0.1591,
      "step": 2782
    },
    {
      "epoch": 0.5841729638958858,
      "grad_norm": 0.11060679703950882,
      "learning_rate": 4.4059912593040964e-05,
      "loss": 0.1452,
      "step": 2783
    },
    {
      "epoch": 0.5843828715365239,
      "grad_norm": 0.12010272592306137,
      "learning_rate": 4.4023532760459055e-05,
      "loss": 0.179,
      "step": 2784
    },
    {
      "epoch": 0.584592779177162,
      "grad_norm": 0.10742159187793732,
      "learning_rate": 4.3987156137376585e-05,
      "loss": 0.1607,
      "step": 2785
    },
    {
      "epoch": 0.5848026868178001,
      "grad_norm": 0.11892808228731155,
      "learning_rate": 4.395078274332862e-05,
      "loss": 0.1546,
      "step": 2786
    },
    {
      "epoch": 0.5850125944584383,
      "grad_norm": 0.12276134639978409,
      "learning_rate": 4.3914412597848506e-05,
      "loss": 0.159,
      "step": 2787
    },
    {
      "epoch": 0.5852225020990764,
      "grad_norm": 0.10355465859174728,
      "learning_rate": 4.387804572046781e-05,
      "loss": 0.1518,
      "step": 2788
    },
    {
      "epoch": 0.5854324097397146,
      "grad_norm": 0.1348765641450882,
      "learning_rate": 4.384168213071642e-05,
      "loss": 0.1628,
      "step": 2789
    },
    {
      "epoch": 0.5856423173803527,
      "grad_norm": 0.10627730190753937,
      "learning_rate": 4.380532184812239e-05,
      "loss": 0.1587,
      "step": 2790
    },
    {
      "epoch": 0.5858522250209908,
      "grad_norm": 0.13369868695735931,
      "learning_rate": 4.376896489221203e-05,
      "loss": 0.155,
      "step": 2791
    },
    {
      "epoch": 0.5860621326616289,
      "grad_norm": 0.10591692477464676,
      "learning_rate": 4.373261128250987e-05,
      "loss": 0.1605,
      "step": 2792
    },
    {
      "epoch": 0.586272040302267,
      "grad_norm": 0.09910458326339722,
      "learning_rate": 4.369626103853859e-05,
      "loss": 0.1597,
      "step": 2793
    },
    {
      "epoch": 0.5864819479429051,
      "grad_norm": 0.1167030781507492,
      "learning_rate": 4.365991417981915e-05,
      "loss": 0.1627,
      "step": 2794
    },
    {
      "epoch": 0.5866918555835432,
      "grad_norm": 0.10456610471010208,
      "learning_rate": 4.362357072587061e-05,
      "loss": 0.1488,
      "step": 2795
    },
    {
      "epoch": 0.5869017632241813,
      "grad_norm": 0.11257223784923553,
      "learning_rate": 4.3587230696210224e-05,
      "loss": 0.1537,
      "step": 2796
    },
    {
      "epoch": 0.5871116708648195,
      "grad_norm": 0.1202881932258606,
      "learning_rate": 4.355089411035346e-05,
      "loss": 0.1685,
      "step": 2797
    },
    {
      "epoch": 0.5873215785054576,
      "grad_norm": 0.1485619693994522,
      "learning_rate": 4.3514560987813864e-05,
      "loss": 0.16,
      "step": 2798
    },
    {
      "epoch": 0.5875314861460957,
      "grad_norm": 0.12521955370903015,
      "learning_rate": 4.347823134810315e-05,
      "loss": 0.1631,
      "step": 2799
    },
    {
      "epoch": 0.5877413937867338,
      "grad_norm": 0.09790458530187607,
      "learning_rate": 4.344190521073119e-05,
      "loss": 0.1359,
      "step": 2800
    },
    {
      "epoch": 0.587951301427372,
      "grad_norm": 0.11199049651622772,
      "learning_rate": 4.340558259520594e-05,
      "loss": 0.1458,
      "step": 2801
    },
    {
      "epoch": 0.5881612090680101,
      "grad_norm": 0.1020960882306099,
      "learning_rate": 4.336926352103345e-05,
      "loss": 0.1559,
      "step": 2802
    },
    {
      "epoch": 0.5883711167086482,
      "grad_norm": 0.12979748845100403,
      "learning_rate": 4.3332948007717925e-05,
      "loss": 0.1654,
      "step": 2803
    },
    {
      "epoch": 0.5885810243492863,
      "grad_norm": 0.1190989539027214,
      "learning_rate": 4.32966360747616e-05,
      "loss": 0.1682,
      "step": 2804
    },
    {
      "epoch": 0.5887909319899244,
      "grad_norm": 0.10461704432964325,
      "learning_rate": 4.3260327741664824e-05,
      "loss": 0.1759,
      "step": 2805
    },
    {
      "epoch": 0.5890008396305626,
      "grad_norm": 0.11184418946504593,
      "learning_rate": 4.3224023027926007e-05,
      "loss": 0.1606,
      "step": 2806
    },
    {
      "epoch": 0.5892107472712007,
      "grad_norm": 0.10270063579082489,
      "learning_rate": 4.31877219530416e-05,
      "loss": 0.1488,
      "step": 2807
    },
    {
      "epoch": 0.5894206549118388,
      "grad_norm": 0.12192819267511368,
      "learning_rate": 4.315142453650613e-05,
      "loss": 0.1723,
      "step": 2808
    },
    {
      "epoch": 0.5896305625524769,
      "grad_norm": 0.11914799362421036,
      "learning_rate": 4.311513079781211e-05,
      "loss": 0.1651,
      "step": 2809
    },
    {
      "epoch": 0.589840470193115,
      "grad_norm": 0.11079401522874832,
      "learning_rate": 4.307884075645012e-05,
      "loss": 0.1475,
      "step": 2810
    },
    {
      "epoch": 0.5900503778337531,
      "grad_norm": 0.13855847716331482,
      "learning_rate": 4.3042554431908735e-05,
      "loss": 0.1488,
      "step": 2811
    },
    {
      "epoch": 0.5902602854743912,
      "grad_norm": 0.12921065092086792,
      "learning_rate": 4.3006271843674546e-05,
      "loss": 0.1688,
      "step": 2812
    },
    {
      "epoch": 0.5904701931150294,
      "grad_norm": 0.15366600453853607,
      "learning_rate": 4.296999301123212e-05,
      "loss": 0.1582,
      "step": 2813
    },
    {
      "epoch": 0.5906801007556675,
      "grad_norm": 0.12695199251174927,
      "learning_rate": 4.293371795406404e-05,
      "loss": 0.1645,
      "step": 2814
    },
    {
      "epoch": 0.5908900083963057,
      "grad_norm": 0.13746286928653717,
      "learning_rate": 4.289744669165085e-05,
      "loss": 0.1688,
      "step": 2815
    },
    {
      "epoch": 0.5910999160369438,
      "grad_norm": 0.09618178009986877,
      "learning_rate": 4.286117924347101e-05,
      "loss": 0.1746,
      "step": 2816
    },
    {
      "epoch": 0.5913098236775819,
      "grad_norm": 0.12498153746128082,
      "learning_rate": 4.282491562900097e-05,
      "loss": 0.1399,
      "step": 2817
    },
    {
      "epoch": 0.59151973131822,
      "grad_norm": 0.11736780405044556,
      "learning_rate": 4.278865586771515e-05,
      "loss": 0.145,
      "step": 2818
    },
    {
      "epoch": 0.5917296389588581,
      "grad_norm": 0.12359510362148285,
      "learning_rate": 4.2752399979085836e-05,
      "loss": 0.1406,
      "step": 2819
    },
    {
      "epoch": 0.5919395465994962,
      "grad_norm": 0.13295608758926392,
      "learning_rate": 4.27161479825833e-05,
      "loss": 0.1425,
      "step": 2820
    },
    {
      "epoch": 0.5921494542401343,
      "grad_norm": 0.12067288160324097,
      "learning_rate": 4.267989989767568e-05,
      "loss": 0.1526,
      "step": 2821
    },
    {
      "epoch": 0.5923593618807724,
      "grad_norm": 0.12987972795963287,
      "learning_rate": 4.264365574382903e-05,
      "loss": 0.1488,
      "step": 2822
    },
    {
      "epoch": 0.5925692695214105,
      "grad_norm": 0.13814137876033783,
      "learning_rate": 4.2607415540507264e-05,
      "loss": 0.1575,
      "step": 2823
    },
    {
      "epoch": 0.5927791771620488,
      "grad_norm": 0.10198640078306198,
      "learning_rate": 4.2571179307172235e-05,
      "loss": 0.165,
      "step": 2824
    },
    {
      "epoch": 0.5929890848026869,
      "grad_norm": 0.1260635405778885,
      "learning_rate": 4.2534947063283594e-05,
      "loss": 0.1578,
      "step": 2825
    },
    {
      "epoch": 0.593198992443325,
      "grad_norm": 0.11537054181098938,
      "learning_rate": 4.2498718828298906e-05,
      "loss": 0.1623,
      "step": 2826
    },
    {
      "epoch": 0.5934089000839631,
      "grad_norm": 0.11723814159631729,
      "learning_rate": 4.246249462167356e-05,
      "loss": 0.1391,
      "step": 2827
    },
    {
      "epoch": 0.5936188077246012,
      "grad_norm": 0.11966104060411453,
      "learning_rate": 4.242627446286076e-05,
      "loss": 0.1635,
      "step": 2828
    },
    {
      "epoch": 0.5938287153652393,
      "grad_norm": 0.11630399525165558,
      "learning_rate": 4.23900583713116e-05,
      "loss": 0.1578,
      "step": 2829
    },
    {
      "epoch": 0.5940386230058774,
      "grad_norm": 0.14845402538776398,
      "learning_rate": 4.235384636647492e-05,
      "loss": 0.1327,
      "step": 2830
    },
    {
      "epoch": 0.5942485306465155,
      "grad_norm": 0.1327534019947052,
      "learning_rate": 4.2317638467797384e-05,
      "loss": 0.1416,
      "step": 2831
    },
    {
      "epoch": 0.5944584382871536,
      "grad_norm": 0.11541690677404404,
      "learning_rate": 4.228143469472348e-05,
      "loss": 0.1712,
      "step": 2832
    },
    {
      "epoch": 0.5946683459277917,
      "grad_norm": 0.11513253301382065,
      "learning_rate": 4.224523506669545e-05,
      "loss": 0.1337,
      "step": 2833
    },
    {
      "epoch": 0.5948782535684299,
      "grad_norm": 0.127321258187294,
      "learning_rate": 4.220903960315333e-05,
      "loss": 0.1663,
      "step": 2834
    },
    {
      "epoch": 0.595088161209068,
      "grad_norm": 0.12505748867988586,
      "learning_rate": 4.2172848323534915e-05,
      "loss": 0.1545,
      "step": 2835
    },
    {
      "epoch": 0.5952980688497062,
      "grad_norm": 0.10772762447595596,
      "learning_rate": 4.213666124727575e-05,
      "loss": 0.1606,
      "step": 2836
    },
    {
      "epoch": 0.5955079764903443,
      "grad_norm": 0.12374143302440643,
      "learning_rate": 4.210047839380909e-05,
      "loss": 0.1502,
      "step": 2837
    },
    {
      "epoch": 0.5957178841309824,
      "grad_norm": 0.11770914494991302,
      "learning_rate": 4.206429978256599e-05,
      "loss": 0.1639,
      "step": 2838
    },
    {
      "epoch": 0.5959277917716205,
      "grad_norm": 0.12015975266695023,
      "learning_rate": 4.2028125432975175e-05,
      "loss": 0.1571,
      "step": 2839
    },
    {
      "epoch": 0.5961376994122586,
      "grad_norm": 0.10522496700286865,
      "learning_rate": 4.1991955364463096e-05,
      "loss": 0.1538,
      "step": 2840
    },
    {
      "epoch": 0.5963476070528967,
      "grad_norm": 0.1105801910161972,
      "learning_rate": 4.195578959645391e-05,
      "loss": 0.1461,
      "step": 2841
    },
    {
      "epoch": 0.5965575146935348,
      "grad_norm": 0.11777419596910477,
      "learning_rate": 4.191962814836945e-05,
      "loss": 0.1573,
      "step": 2842
    },
    {
      "epoch": 0.596767422334173,
      "grad_norm": 0.12069468200206757,
      "learning_rate": 4.188347103962928e-05,
      "loss": 0.1567,
      "step": 2843
    },
    {
      "epoch": 0.5969773299748111,
      "grad_norm": 0.11869499832391739,
      "learning_rate": 4.184731828965054e-05,
      "loss": 0.1645,
      "step": 2844
    },
    {
      "epoch": 0.5971872376154492,
      "grad_norm": 0.1150764673948288,
      "learning_rate": 4.181116991784811e-05,
      "loss": 0.1523,
      "step": 2845
    },
    {
      "epoch": 0.5973971452560873,
      "grad_norm": 0.1119702160358429,
      "learning_rate": 4.177502594363448e-05,
      "loss": 0.144,
      "step": 2846
    },
    {
      "epoch": 0.5976070528967254,
      "grad_norm": 0.17013654112815857,
      "learning_rate": 4.173888638641981e-05,
      "loss": 0.1726,
      "step": 2847
    },
    {
      "epoch": 0.5978169605373636,
      "grad_norm": 0.10498031228780746,
      "learning_rate": 4.1702751265611835e-05,
      "loss": 0.1413,
      "step": 2848
    },
    {
      "epoch": 0.5980268681780017,
      "grad_norm": 0.11212137341499329,
      "learning_rate": 4.166662060061597e-05,
      "loss": 0.137,
      "step": 2849
    },
    {
      "epoch": 0.5982367758186398,
      "grad_norm": 0.12234228104352951,
      "learning_rate": 4.163049441083522e-05,
      "loss": 0.1719,
      "step": 2850
    },
    {
      "epoch": 0.5984466834592779,
      "grad_norm": 0.10817383229732513,
      "learning_rate": 4.1594372715670124e-05,
      "loss": 0.1584,
      "step": 2851
    },
    {
      "epoch": 0.5986565910999161,
      "grad_norm": 0.13407209515571594,
      "learning_rate": 4.155825553451889e-05,
      "loss": 0.1636,
      "step": 2852
    },
    {
      "epoch": 0.5988664987405542,
      "grad_norm": 0.15183603763580322,
      "learning_rate": 4.152214288677726e-05,
      "loss": 0.1723,
      "step": 2853
    },
    {
      "epoch": 0.5990764063811923,
      "grad_norm": 0.11314138770103455,
      "learning_rate": 4.1486034791838554e-05,
      "loss": 0.1527,
      "step": 2854
    },
    {
      "epoch": 0.5992863140218304,
      "grad_norm": 0.11711379885673523,
      "learning_rate": 4.144993126909365e-05,
      "loss": 0.1605,
      "step": 2855
    },
    {
      "epoch": 0.5994962216624685,
      "grad_norm": 0.13769389688968658,
      "learning_rate": 4.141383233793097e-05,
      "loss": 0.1506,
      "step": 2856
    },
    {
      "epoch": 0.5997061293031066,
      "grad_norm": 0.16857290267944336,
      "learning_rate": 4.137773801773644e-05,
      "loss": 0.1744,
      "step": 2857
    },
    {
      "epoch": 0.5999160369437447,
      "grad_norm": 0.1259475201368332,
      "learning_rate": 4.1341648327893544e-05,
      "loss": 0.1668,
      "step": 2858
    },
    {
      "epoch": 0.6001259445843828,
      "grad_norm": 0.10995698720216751,
      "learning_rate": 4.1305563287783284e-05,
      "loss": 0.146,
      "step": 2859
    },
    {
      "epoch": 0.600335852225021,
      "grad_norm": 0.10539478063583374,
      "learning_rate": 4.1269482916784124e-05,
      "loss": 0.146,
      "step": 2860
    },
    {
      "epoch": 0.6005457598656591,
      "grad_norm": 0.11225578933954239,
      "learning_rate": 4.1233407234272084e-05,
      "loss": 0.1416,
      "step": 2861
    },
    {
      "epoch": 0.6007556675062973,
      "grad_norm": 0.11747094988822937,
      "learning_rate": 4.11973362596206e-05,
      "loss": 0.1529,
      "step": 2862
    },
    {
      "epoch": 0.6009655751469354,
      "grad_norm": 0.13106967508792877,
      "learning_rate": 4.116127001220063e-05,
      "loss": 0.1593,
      "step": 2863
    },
    {
      "epoch": 0.6011754827875735,
      "grad_norm": 0.1032341942191124,
      "learning_rate": 4.112520851138054e-05,
      "loss": 0.1539,
      "step": 2864
    },
    {
      "epoch": 0.6013853904282116,
      "grad_norm": 0.11255158483982086,
      "learning_rate": 4.1089151776526215e-05,
      "loss": 0.1546,
      "step": 2865
    },
    {
      "epoch": 0.6015952980688497,
      "grad_norm": 0.11748593300580978,
      "learning_rate": 4.1053099827000915e-05,
      "loss": 0.1536,
      "step": 2866
    },
    {
      "epoch": 0.6018052057094878,
      "grad_norm": 0.13722585141658783,
      "learning_rate": 4.10170526821654e-05,
      "loss": 0.1613,
      "step": 2867
    },
    {
      "epoch": 0.6020151133501259,
      "grad_norm": 0.12346901744604111,
      "learning_rate": 4.098101036137778e-05,
      "loss": 0.1535,
      "step": 2868
    },
    {
      "epoch": 0.602225020990764,
      "grad_norm": 0.11565665900707245,
      "learning_rate": 4.0944972883993594e-05,
      "loss": 0.1492,
      "step": 2869
    },
    {
      "epoch": 0.6024349286314021,
      "grad_norm": 0.12450307607650757,
      "learning_rate": 4.0908940269365835e-05,
      "loss": 0.1507,
      "step": 2870
    },
    {
      "epoch": 0.6026448362720404,
      "grad_norm": 0.10263592004776001,
      "learning_rate": 4.0872912536844805e-05,
      "loss": 0.1552,
      "step": 2871
    },
    {
      "epoch": 0.6028547439126785,
      "grad_norm": 0.10213368386030197,
      "learning_rate": 4.0836889705778215e-05,
      "loss": 0.1548,
      "step": 2872
    },
    {
      "epoch": 0.6030646515533166,
      "grad_norm": 0.12833476066589355,
      "learning_rate": 4.0800871795511164e-05,
      "loss": 0.1657,
      "step": 2873
    },
    {
      "epoch": 0.6032745591939547,
      "grad_norm": 0.13922633230686188,
      "learning_rate": 4.07648588253861e-05,
      "loss": 0.14,
      "step": 2874
    },
    {
      "epoch": 0.6034844668345928,
      "grad_norm": 0.12405204772949219,
      "learning_rate": 4.072885081474278e-05,
      "loss": 0.1756,
      "step": 2875
    },
    {
      "epoch": 0.6036943744752309,
      "grad_norm": 0.12056063115596771,
      "learning_rate": 4.069284778291837e-05,
      "loss": 0.1625,
      "step": 2876
    },
    {
      "epoch": 0.603904282115869,
      "grad_norm": 0.1193803995847702,
      "learning_rate": 4.06568497492473e-05,
      "loss": 0.162,
      "step": 2877
    },
    {
      "epoch": 0.6041141897565071,
      "grad_norm": 0.1092044785618782,
      "learning_rate": 4.062085673306132e-05,
      "loss": 0.1538,
      "step": 2878
    },
    {
      "epoch": 0.6043240973971452,
      "grad_norm": 0.1371615082025528,
      "learning_rate": 4.058486875368952e-05,
      "loss": 0.1605,
      "step": 2879
    },
    {
      "epoch": 0.6045340050377834,
      "grad_norm": 0.11028730124235153,
      "learning_rate": 4.054888583045827e-05,
      "loss": 0.1572,
      "step": 2880
    },
    {
      "epoch": 0.6047439126784215,
      "grad_norm": 0.13068102300167084,
      "learning_rate": 4.05129079826912e-05,
      "loss": 0.1578,
      "step": 2881
    },
    {
      "epoch": 0.6049538203190596,
      "grad_norm": 0.10060226172208786,
      "learning_rate": 4.0476935229709265e-05,
      "loss": 0.1732,
      "step": 2882
    },
    {
      "epoch": 0.6051637279596978,
      "grad_norm": 0.10314224660396576,
      "learning_rate": 4.0440967590830625e-05,
      "loss": 0.1646,
      "step": 2883
    },
    {
      "epoch": 0.6053736356003359,
      "grad_norm": 0.13616135716438293,
      "learning_rate": 4.040500508537077e-05,
      "loss": 0.1778,
      "step": 2884
    },
    {
      "epoch": 0.605583543240974,
      "grad_norm": 0.21042796969413757,
      "learning_rate": 4.036904773264234e-05,
      "loss": 0.1501,
      "step": 2885
    },
    {
      "epoch": 0.6057934508816121,
      "grad_norm": 0.115667924284935,
      "learning_rate": 4.033309555195527e-05,
      "loss": 0.1707,
      "step": 2886
    },
    {
      "epoch": 0.6060033585222502,
      "grad_norm": 0.17195206880569458,
      "learning_rate": 4.02971485626167e-05,
      "loss": 0.1539,
      "step": 2887
    },
    {
      "epoch": 0.6062132661628883,
      "grad_norm": 0.10804500430822372,
      "learning_rate": 4.0261206783931e-05,
      "loss": 0.1494,
      "step": 2888
    },
    {
      "epoch": 0.6064231738035264,
      "grad_norm": 0.1175246611237526,
      "learning_rate": 4.02252702351997e-05,
      "loss": 0.1579,
      "step": 2889
    },
    {
      "epoch": 0.6066330814441646,
      "grad_norm": 0.11100801080465317,
      "learning_rate": 4.018933893572157e-05,
      "loss": 0.1525,
      "step": 2890
    },
    {
      "epoch": 0.6068429890848027,
      "grad_norm": 0.1287776231765747,
      "learning_rate": 4.015341290479255e-05,
      "loss": 0.1573,
      "step": 2891
    },
    {
      "epoch": 0.6070528967254408,
      "grad_norm": 0.0996914654970169,
      "learning_rate": 4.01174921617057e-05,
      "loss": 0.1667,
      "step": 2892
    },
    {
      "epoch": 0.6072628043660789,
      "grad_norm": 0.11938051134347916,
      "learning_rate": 4.0081576725751294e-05,
      "loss": 0.1585,
      "step": 2893
    },
    {
      "epoch": 0.607472712006717,
      "grad_norm": 0.13560616970062256,
      "learning_rate": 4.004566661621676e-05,
      "loss": 0.1533,
      "step": 2894
    },
    {
      "epoch": 0.6076826196473551,
      "grad_norm": 0.15640297532081604,
      "learning_rate": 4.000976185238662e-05,
      "loss": 0.1637,
      "step": 2895
    },
    {
      "epoch": 0.6078925272879933,
      "grad_norm": 0.12788182497024536,
      "learning_rate": 3.9973862453542575e-05,
      "loss": 0.1464,
      "step": 2896
    },
    {
      "epoch": 0.6081024349286314,
      "grad_norm": 0.1404847800731659,
      "learning_rate": 3.9937968438963416e-05,
      "loss": 0.1727,
      "step": 2897
    },
    {
      "epoch": 0.6083123425692695,
      "grad_norm": 0.1205708235502243,
      "learning_rate": 3.990207982792505e-05,
      "loss": 0.1554,
      "step": 2898
    },
    {
      "epoch": 0.6085222502099077,
      "grad_norm": 0.13461466133594513,
      "learning_rate": 3.986619663970047e-05,
      "loss": 0.1713,
      "step": 2899
    },
    {
      "epoch": 0.6087321578505458,
      "grad_norm": 0.12866820394992828,
      "learning_rate": 3.983031889355978e-05,
      "loss": 0.1588,
      "step": 2900
    },
    {
      "epoch": 0.6089420654911839,
      "grad_norm": 0.11548592895269394,
      "learning_rate": 3.9794446608770134e-05,
      "loss": 0.1507,
      "step": 2901
    },
    {
      "epoch": 0.609151973131822,
      "grad_norm": 0.12713722884655,
      "learning_rate": 3.975857980459579e-05,
      "loss": 0.1499,
      "step": 2902
    },
    {
      "epoch": 0.6093618807724601,
      "grad_norm": 0.13451896607875824,
      "learning_rate": 3.9722718500298026e-05,
      "loss": 0.1675,
      "step": 2903
    },
    {
      "epoch": 0.6095717884130982,
      "grad_norm": 0.12704823911190033,
      "learning_rate": 3.9686862715135176e-05,
      "loss": 0.1434,
      "step": 2904
    },
    {
      "epoch": 0.6097816960537363,
      "grad_norm": 0.11345957219600677,
      "learning_rate": 3.965101246836265e-05,
      "loss": 0.1603,
      "step": 2905
    },
    {
      "epoch": 0.6099916036943744,
      "grad_norm": 0.10510886460542679,
      "learning_rate": 3.96151677792328e-05,
      "loss": 0.1541,
      "step": 2906
    },
    {
      "epoch": 0.6102015113350125,
      "grad_norm": 0.1235707551240921,
      "learning_rate": 3.957932866699508e-05,
      "loss": 0.1409,
      "step": 2907
    },
    {
      "epoch": 0.6104114189756508,
      "grad_norm": 0.11852505803108215,
      "learning_rate": 3.954349515089589e-05,
      "loss": 0.167,
      "step": 2908
    },
    {
      "epoch": 0.6106213266162889,
      "grad_norm": 0.10151386260986328,
      "learning_rate": 3.950766725017866e-05,
      "loss": 0.1484,
      "step": 2909
    },
    {
      "epoch": 0.610831234256927,
      "grad_norm": 0.1326039880514145,
      "learning_rate": 3.947184498408378e-05,
      "loss": 0.1552,
      "step": 2910
    },
    {
      "epoch": 0.6110411418975651,
      "grad_norm": 0.14236897230148315,
      "learning_rate": 3.9436028371848646e-05,
      "loss": 0.1583,
      "step": 2911
    },
    {
      "epoch": 0.6112510495382032,
      "grad_norm": 0.1296806037425995,
      "learning_rate": 3.940021743270759e-05,
      "loss": 0.1542,
      "step": 2912
    },
    {
      "epoch": 0.6114609571788413,
      "grad_norm": 0.10085690766572952,
      "learning_rate": 3.936441218589187e-05,
      "loss": 0.1518,
      "step": 2913
    },
    {
      "epoch": 0.6116708648194794,
      "grad_norm": 0.1088392585515976,
      "learning_rate": 3.9328612650629765e-05,
      "loss": 0.1605,
      "step": 2914
    },
    {
      "epoch": 0.6118807724601175,
      "grad_norm": 0.09953225404024124,
      "learning_rate": 3.929281884614643e-05,
      "loss": 0.1409,
      "step": 2915
    },
    {
      "epoch": 0.6120906801007556,
      "grad_norm": 0.10369337350130081,
      "learning_rate": 3.925703079166394e-05,
      "loss": 0.1579,
      "step": 2916
    },
    {
      "epoch": 0.6123005877413937,
      "grad_norm": 0.10897854715585709,
      "learning_rate": 3.922124850640131e-05,
      "loss": 0.1493,
      "step": 2917
    },
    {
      "epoch": 0.612510495382032,
      "grad_norm": 0.09800262004137039,
      "learning_rate": 3.918547200957445e-05,
      "loss": 0.1606,
      "step": 2918
    },
    {
      "epoch": 0.6127204030226701,
      "grad_norm": 0.09800999611616135,
      "learning_rate": 3.914970132039616e-05,
      "loss": 0.1465,
      "step": 2919
    },
    {
      "epoch": 0.6129303106633082,
      "grad_norm": 0.11520504951477051,
      "learning_rate": 3.9113936458076106e-05,
      "loss": 0.1544,
      "step": 2920
    },
    {
      "epoch": 0.6131402183039463,
      "grad_norm": 0.12162473052740097,
      "learning_rate": 3.9078177441820834e-05,
      "loss": 0.1511,
      "step": 2921
    },
    {
      "epoch": 0.6133501259445844,
      "grad_norm": 0.14628733694553375,
      "learning_rate": 3.9042424290833746e-05,
      "loss": 0.1564,
      "step": 2922
    },
    {
      "epoch": 0.6135600335852225,
      "grad_norm": 0.11655198037624359,
      "learning_rate": 3.900667702431512e-05,
      "loss": 0.1665,
      "step": 2923
    },
    {
      "epoch": 0.6137699412258606,
      "grad_norm": 0.11280401051044464,
      "learning_rate": 3.897093566146204e-05,
      "loss": 0.1672,
      "step": 2924
    },
    {
      "epoch": 0.6139798488664987,
      "grad_norm": 0.13265039026737213,
      "learning_rate": 3.8935200221468446e-05,
      "loss": 0.1703,
      "step": 2925
    },
    {
      "epoch": 0.6141897565071368,
      "grad_norm": 0.10153202712535858,
      "learning_rate": 3.88994707235251e-05,
      "loss": 0.1498,
      "step": 2926
    },
    {
      "epoch": 0.614399664147775,
      "grad_norm": 0.1125868558883667,
      "learning_rate": 3.886374718681952e-05,
      "loss": 0.1737,
      "step": 2927
    },
    {
      "epoch": 0.6146095717884131,
      "grad_norm": 0.1055431067943573,
      "learning_rate": 3.882802963053607e-05,
      "loss": 0.1531,
      "step": 2928
    },
    {
      "epoch": 0.6148194794290512,
      "grad_norm": 0.12137924879789352,
      "learning_rate": 3.8792318073855913e-05,
      "loss": 0.1498,
      "step": 2929
    },
    {
      "epoch": 0.6150293870696893,
      "grad_norm": 0.11451124399900436,
      "learning_rate": 3.8756612535956954e-05,
      "loss": 0.1613,
      "step": 2930
    },
    {
      "epoch": 0.6152392947103275,
      "grad_norm": 0.10448075830936432,
      "learning_rate": 3.87209130360139e-05,
      "loss": 0.1565,
      "step": 2931
    },
    {
      "epoch": 0.6154492023509656,
      "grad_norm": 0.11644694209098816,
      "learning_rate": 3.868521959319817e-05,
      "loss": 0.1691,
      "step": 2932
    },
    {
      "epoch": 0.6156591099916037,
      "grad_norm": 0.11724112182855606,
      "learning_rate": 3.8649532226678015e-05,
      "loss": 0.1637,
      "step": 2933
    },
    {
      "epoch": 0.6158690176322418,
      "grad_norm": 0.1371581256389618,
      "learning_rate": 3.861385095561829e-05,
      "loss": 0.143,
      "step": 2934
    },
    {
      "epoch": 0.6160789252728799,
      "grad_norm": 0.12549670040607452,
      "learning_rate": 3.85781757991807e-05,
      "loss": 0.1523,
      "step": 2935
    },
    {
      "epoch": 0.6162888329135181,
      "grad_norm": 0.10449302196502686,
      "learning_rate": 3.85425067765236e-05,
      "loss": 0.1508,
      "step": 2936
    },
    {
      "epoch": 0.6164987405541562,
      "grad_norm": 0.12108724564313889,
      "learning_rate": 3.8506843906802085e-05,
      "loss": 0.1502,
      "step": 2937
    },
    {
      "epoch": 0.6167086481947943,
      "grad_norm": 0.10048652440309525,
      "learning_rate": 3.8471187209167925e-05,
      "loss": 0.1491,
      "step": 2938
    },
    {
      "epoch": 0.6169185558354324,
      "grad_norm": 0.13422608375549316,
      "learning_rate": 3.843553670276957e-05,
      "loss": 0.1629,
      "step": 2939
    },
    {
      "epoch": 0.6171284634760705,
      "grad_norm": 0.14751432836055756,
      "learning_rate": 3.8399892406752183e-05,
      "loss": 0.1563,
      "step": 2940
    },
    {
      "epoch": 0.6173383711167086,
      "grad_norm": 0.11447716504335403,
      "learning_rate": 3.836425434025754e-05,
      "loss": 0.1604,
      "step": 2941
    },
    {
      "epoch": 0.6175482787573467,
      "grad_norm": 0.13063320517539978,
      "learning_rate": 3.83286225224241e-05,
      "loss": 0.1692,
      "step": 2942
    },
    {
      "epoch": 0.6177581863979849,
      "grad_norm": 0.13792036473751068,
      "learning_rate": 3.8292996972386976e-05,
      "loss": 0.1506,
      "step": 2943
    },
    {
      "epoch": 0.617968094038623,
      "grad_norm": 0.12366560101509094,
      "learning_rate": 3.825737770927789e-05,
      "loss": 0.1476,
      "step": 2944
    },
    {
      "epoch": 0.6181780016792612,
      "grad_norm": 0.1087699830532074,
      "learning_rate": 3.8221764752225195e-05,
      "loss": 0.1565,
      "step": 2945
    },
    {
      "epoch": 0.6183879093198993,
      "grad_norm": 0.1199265792965889,
      "learning_rate": 3.8186158120353884e-05,
      "loss": 0.174,
      "step": 2946
    },
    {
      "epoch": 0.6185978169605374,
      "grad_norm": 0.10412439703941345,
      "learning_rate": 3.815055783278553e-05,
      "loss": 0.1285,
      "step": 2947
    },
    {
      "epoch": 0.6188077246011755,
      "grad_norm": 0.12668044865131378,
      "learning_rate": 3.8114963908638266e-05,
      "loss": 0.1644,
      "step": 2948
    },
    {
      "epoch": 0.6190176322418136,
      "grad_norm": 0.13107618689537048,
      "learning_rate": 3.807937636702688e-05,
      "loss": 0.1654,
      "step": 2949
    },
    {
      "epoch": 0.6192275398824517,
      "grad_norm": 0.11218323558568954,
      "learning_rate": 3.8043795227062675e-05,
      "loss": 0.1364,
      "step": 2950
    },
    {
      "epoch": 0.6194374475230898,
      "grad_norm": 0.10924772173166275,
      "learning_rate": 3.800822050785353e-05,
      "loss": 0.1565,
      "step": 2951
    },
    {
      "epoch": 0.6196473551637279,
      "grad_norm": 0.12488337606191635,
      "learning_rate": 3.797265222850391e-05,
      "loss": 0.1632,
      "step": 2952
    },
    {
      "epoch": 0.619857262804366,
      "grad_norm": 0.12105172872543335,
      "learning_rate": 3.7937090408114784e-05,
      "loss": 0.1601,
      "step": 2953
    },
    {
      "epoch": 0.6200671704450041,
      "grad_norm": 0.13021235167980194,
      "learning_rate": 3.7901535065783633e-05,
      "loss": 0.1697,
      "step": 2954
    },
    {
      "epoch": 0.6202770780856424,
      "grad_norm": 0.12432187795639038,
      "learning_rate": 3.786598622060453e-05,
      "loss": 0.1648,
      "step": 2955
    },
    {
      "epoch": 0.6204869857262805,
      "grad_norm": 0.10438645631074905,
      "learning_rate": 3.7830443891667986e-05,
      "loss": 0.1515,
      "step": 2956
    },
    {
      "epoch": 0.6206968933669186,
      "grad_norm": 0.10381833463907242,
      "learning_rate": 3.7794908098061046e-05,
      "loss": 0.1398,
      "step": 2957
    },
    {
      "epoch": 0.6209068010075567,
      "grad_norm": 0.1131964772939682,
      "learning_rate": 3.775937885886726e-05,
      "loss": 0.1674,
      "step": 2958
    },
    {
      "epoch": 0.6211167086481948,
      "grad_norm": 0.09784296900033951,
      "learning_rate": 3.772385619316663e-05,
      "loss": 0.1519,
      "step": 2959
    },
    {
      "epoch": 0.6213266162888329,
      "grad_norm": 0.10954361408948898,
      "learning_rate": 3.7688340120035654e-05,
      "loss": 0.1376,
      "step": 2960
    },
    {
      "epoch": 0.621536523929471,
      "grad_norm": 0.10720100998878479,
      "learning_rate": 3.7652830658547254e-05,
      "loss": 0.1494,
      "step": 2961
    },
    {
      "epoch": 0.6217464315701091,
      "grad_norm": 0.11651554703712463,
      "learning_rate": 3.761732782777083e-05,
      "loss": 0.1522,
      "step": 2962
    },
    {
      "epoch": 0.6219563392107472,
      "grad_norm": 0.12341861426830292,
      "learning_rate": 3.7581831646772205e-05,
      "loss": 0.1486,
      "step": 2963
    },
    {
      "epoch": 0.6221662468513854,
      "grad_norm": 0.12296581268310547,
      "learning_rate": 3.754634213461365e-05,
      "loss": 0.1553,
      "step": 2964
    },
    {
      "epoch": 0.6223761544920235,
      "grad_norm": 0.11492616683244705,
      "learning_rate": 3.7510859310353836e-05,
      "loss": 0.1526,
      "step": 2965
    },
    {
      "epoch": 0.6225860621326617,
      "grad_norm": 0.13258479535579681,
      "learning_rate": 3.7475383193047866e-05,
      "loss": 0.1521,
      "step": 2966
    },
    {
      "epoch": 0.6227959697732998,
      "grad_norm": 0.11529724299907684,
      "learning_rate": 3.7439913801747214e-05,
      "loss": 0.1423,
      "step": 2967
    },
    {
      "epoch": 0.6230058774139379,
      "grad_norm": 0.12614604830741882,
      "learning_rate": 3.7404451155499755e-05,
      "loss": 0.1416,
      "step": 2968
    },
    {
      "epoch": 0.623215785054576,
      "grad_norm": 0.13627269864082336,
      "learning_rate": 3.7368995273349716e-05,
      "loss": 0.1701,
      "step": 2969
    },
    {
      "epoch": 0.6234256926952141,
      "grad_norm": 0.09332132339477539,
      "learning_rate": 3.733354617433776e-05,
      "loss": 0.1567,
      "step": 2970
    },
    {
      "epoch": 0.6236356003358522,
      "grad_norm": 0.1580103486776352,
      "learning_rate": 3.729810387750082e-05,
      "loss": 0.1779,
      "step": 2971
    },
    {
      "epoch": 0.6238455079764903,
      "grad_norm": 0.10894285887479782,
      "learning_rate": 3.726266840187226e-05,
      "loss": 0.1634,
      "step": 2972
    },
    {
      "epoch": 0.6240554156171285,
      "grad_norm": 0.14514589309692383,
      "learning_rate": 3.72272397664817e-05,
      "loss": 0.1551,
      "step": 2973
    },
    {
      "epoch": 0.6242653232577666,
      "grad_norm": 0.13061858713626862,
      "learning_rate": 3.719181799035519e-05,
      "loss": 0.1612,
      "step": 2974
    },
    {
      "epoch": 0.6244752308984047,
      "grad_norm": 0.1479208767414093,
      "learning_rate": 3.715640309251495e-05,
      "loss": 0.159,
      "step": 2975
    },
    {
      "epoch": 0.6246851385390428,
      "grad_norm": 0.11024679988622665,
      "learning_rate": 3.712099509197964e-05,
      "loss": 0.1415,
      "step": 2976
    },
    {
      "epoch": 0.624895046179681,
      "grad_norm": 0.12445060163736343,
      "learning_rate": 3.7085594007764134e-05,
      "loss": 0.1511,
      "step": 2977
    },
    {
      "epoch": 0.625104953820319,
      "grad_norm": 0.10473346710205078,
      "learning_rate": 3.705019985887966e-05,
      "loss": 0.1575,
      "step": 2978
    },
    {
      "epoch": 0.6253148614609572,
      "grad_norm": 0.11598935723304749,
      "learning_rate": 3.701481266433367e-05,
      "loss": 0.1517,
      "step": 2979
    },
    {
      "epoch": 0.6255247691015953,
      "grad_norm": 0.1171158105134964,
      "learning_rate": 3.697943244312988e-05,
      "loss": 0.1755,
      "step": 2980
    },
    {
      "epoch": 0.6257346767422334,
      "grad_norm": 0.12433173507452011,
      "learning_rate": 3.694405921426831e-05,
      "loss": 0.1409,
      "step": 2981
    },
    {
      "epoch": 0.6259445843828715,
      "grad_norm": 0.13838061690330505,
      "learning_rate": 3.690869299674515e-05,
      "loss": 0.1632,
      "step": 2982
    },
    {
      "epoch": 0.6261544920235097,
      "grad_norm": 0.1209079921245575,
      "learning_rate": 3.687333380955289e-05,
      "loss": 0.15,
      "step": 2983
    },
    {
      "epoch": 0.6263643996641478,
      "grad_norm": 0.11533726751804352,
      "learning_rate": 3.683798167168023e-05,
      "loss": 0.1593,
      "step": 2984
    },
    {
      "epoch": 0.6265743073047859,
      "grad_norm": 0.11141890287399292,
      "learning_rate": 3.680263660211206e-05,
      "loss": 0.1448,
      "step": 2985
    },
    {
      "epoch": 0.626784214945424,
      "grad_norm": 0.10416781902313232,
      "learning_rate": 3.676729861982948e-05,
      "loss": 0.1492,
      "step": 2986
    },
    {
      "epoch": 0.6269941225860621,
      "grad_norm": 0.13024356961250305,
      "learning_rate": 3.673196774380983e-05,
      "loss": 0.15,
      "step": 2987
    },
    {
      "epoch": 0.6272040302267002,
      "grad_norm": 0.1203446090221405,
      "learning_rate": 3.669664399302658e-05,
      "loss": 0.1673,
      "step": 2988
    },
    {
      "epoch": 0.6274139378673383,
      "grad_norm": 0.11524245887994766,
      "learning_rate": 3.6661327386449365e-05,
      "loss": 0.1347,
      "step": 2989
    },
    {
      "epoch": 0.6276238455079765,
      "grad_norm": 0.1355539858341217,
      "learning_rate": 3.662601794304404e-05,
      "loss": 0.1643,
      "step": 2990
    },
    {
      "epoch": 0.6278337531486146,
      "grad_norm": 0.11982898414134979,
      "learning_rate": 3.659071568177258e-05,
      "loss": 0.1396,
      "step": 2991
    },
    {
      "epoch": 0.6280436607892528,
      "grad_norm": 0.11782204359769821,
      "learning_rate": 3.655542062159308e-05,
      "loss": 0.1611,
      "step": 2992
    },
    {
      "epoch": 0.6282535684298909,
      "grad_norm": 0.13236179947853088,
      "learning_rate": 3.652013278145983e-05,
      "loss": 0.1534,
      "step": 2993
    },
    {
      "epoch": 0.628463476070529,
      "grad_norm": 0.11964340507984161,
      "learning_rate": 3.648485218032317e-05,
      "loss": 0.1508,
      "step": 2994
    },
    {
      "epoch": 0.6286733837111671,
      "grad_norm": 0.11381798982620239,
      "learning_rate": 3.644957883712964e-05,
      "loss": 0.1849,
      "step": 2995
    },
    {
      "epoch": 0.6288832913518052,
      "grad_norm": 0.13884767889976501,
      "learning_rate": 3.641431277082178e-05,
      "loss": 0.1621,
      "step": 2996
    },
    {
      "epoch": 0.6290931989924433,
      "grad_norm": 0.13538014888763428,
      "learning_rate": 3.63790540003383e-05,
      "loss": 0.1598,
      "step": 2997
    },
    {
      "epoch": 0.6293031066330814,
      "grad_norm": 0.11793238669633865,
      "learning_rate": 3.6343802544613936e-05,
      "loss": 0.1484,
      "step": 2998
    },
    {
      "epoch": 0.6295130142737195,
      "grad_norm": 0.15728533267974854,
      "learning_rate": 3.6308558422579565e-05,
      "loss": 0.1581,
      "step": 2999
    },
    {
      "epoch": 0.6297229219143576,
      "grad_norm": 0.10782770812511444,
      "learning_rate": 3.6273321653162055e-05,
      "loss": 0.1559,
      "step": 3000
    },
    {
      "epoch": 0.6299328295549959,
      "grad_norm": 0.11745624989271164,
      "learning_rate": 3.623809225528438e-05,
      "loss": 0.1651,
      "step": 3001
    },
    {
      "epoch": 0.630142737195634,
      "grad_norm": 0.1334666609764099,
      "learning_rate": 3.620287024786553e-05,
      "loss": 0.1458,
      "step": 3002
    },
    {
      "epoch": 0.6303526448362721,
      "grad_norm": 0.1259807050228119,
      "learning_rate": 3.6167655649820495e-05,
      "loss": 0.1555,
      "step": 3003
    },
    {
      "epoch": 0.6305625524769102,
      "grad_norm": 0.13792045414447784,
      "learning_rate": 3.6132448480060346e-05,
      "loss": 0.1523,
      "step": 3004
    },
    {
      "epoch": 0.6307724601175483,
      "grad_norm": 0.10497776418924332,
      "learning_rate": 3.609724875749213e-05,
      "loss": 0.1446,
      "step": 3005
    },
    {
      "epoch": 0.6309823677581864,
      "grad_norm": 0.11485005170106888,
      "learning_rate": 3.606205650101889e-05,
      "loss": 0.15,
      "step": 3006
    },
    {
      "epoch": 0.6311922753988245,
      "grad_norm": 0.12223370373249054,
      "learning_rate": 3.602687172953971e-05,
      "loss": 0.1607,
      "step": 3007
    },
    {
      "epoch": 0.6314021830394626,
      "grad_norm": 0.12655122578144073,
      "learning_rate": 3.599169446194958e-05,
      "loss": 0.1636,
      "step": 3008
    },
    {
      "epoch": 0.6316120906801007,
      "grad_norm": 0.11615359783172607,
      "learning_rate": 3.595652471713953e-05,
      "loss": 0.1651,
      "step": 3009
    },
    {
      "epoch": 0.6318219983207388,
      "grad_norm": 0.11752969771623611,
      "learning_rate": 3.592136251399646e-05,
      "loss": 0.1485,
      "step": 3010
    },
    {
      "epoch": 0.632031905961377,
      "grad_norm": 0.11900593340396881,
      "learning_rate": 3.5886207871403326e-05,
      "loss": 0.159,
      "step": 3011
    },
    {
      "epoch": 0.6322418136020151,
      "grad_norm": 0.10577517747879028,
      "learning_rate": 3.5851060808238944e-05,
      "loss": 0.1333,
      "step": 3012
    },
    {
      "epoch": 0.6324517212426533,
      "grad_norm": 0.104424849152565,
      "learning_rate": 3.5815921343378125e-05,
      "loss": 0.1541,
      "step": 3013
    },
    {
      "epoch": 0.6326616288832914,
      "grad_norm": 0.1216193214058876,
      "learning_rate": 3.5780789495691526e-05,
      "loss": 0.1536,
      "step": 3014
    },
    {
      "epoch": 0.6328715365239295,
      "grad_norm": 0.11798959970474243,
      "learning_rate": 3.574566528404577e-05,
      "loss": 0.1673,
      "step": 3015
    },
    {
      "epoch": 0.6330814441645676,
      "grad_norm": 0.15715515613555908,
      "learning_rate": 3.5710548727303384e-05,
      "loss": 0.1482,
      "step": 3016
    },
    {
      "epoch": 0.6332913518052057,
      "grad_norm": 0.10306575894355774,
      "learning_rate": 3.567543984432272e-05,
      "loss": 0.1473,
      "step": 3017
    },
    {
      "epoch": 0.6335012594458438,
      "grad_norm": 0.1256084144115448,
      "learning_rate": 3.5640338653958064e-05,
      "loss": 0.1462,
      "step": 3018
    },
    {
      "epoch": 0.6337111670864819,
      "grad_norm": 0.11586036533117294,
      "learning_rate": 3.5605245175059564e-05,
      "loss": 0.1608,
      "step": 3019
    },
    {
      "epoch": 0.6339210747271201,
      "grad_norm": 0.13672485947608948,
      "learning_rate": 3.5570159426473217e-05,
      "loss": 0.1662,
      "step": 3020
    },
    {
      "epoch": 0.6341309823677582,
      "grad_norm": 0.1345447152853012,
      "learning_rate": 3.553508142704086e-05,
      "loss": 0.1672,
      "step": 3021
    },
    {
      "epoch": 0.6343408900083963,
      "grad_norm": 0.11291158199310303,
      "learning_rate": 3.55000111956002e-05,
      "loss": 0.1299,
      "step": 3022
    },
    {
      "epoch": 0.6345507976490344,
      "grad_norm": 0.1844807267189026,
      "learning_rate": 3.546494875098476e-05,
      "loss": 0.153,
      "step": 3023
    },
    {
      "epoch": 0.6347607052896725,
      "grad_norm": 0.11099909991025925,
      "learning_rate": 3.542989411202383e-05,
      "loss": 0.1448,
      "step": 3024
    },
    {
      "epoch": 0.6349706129303107,
      "grad_norm": 0.1145697608590126,
      "learning_rate": 3.539484729754259e-05,
      "loss": 0.1591,
      "step": 3025
    },
    {
      "epoch": 0.6351805205709488,
      "grad_norm": 0.11213824152946472,
      "learning_rate": 3.5359808326361966e-05,
      "loss": 0.1506,
      "step": 3026
    },
    {
      "epoch": 0.6353904282115869,
      "grad_norm": 0.11795682460069656,
      "learning_rate": 3.5324777217298674e-05,
      "loss": 0.167,
      "step": 3027
    },
    {
      "epoch": 0.635600335852225,
      "grad_norm": 0.10587096959352493,
      "learning_rate": 3.528975398916525e-05,
      "loss": 0.152,
      "step": 3028
    },
    {
      "epoch": 0.6358102434928632,
      "grad_norm": 0.1358242630958557,
      "learning_rate": 3.5254738660769945e-05,
      "loss": 0.1631,
      "step": 3029
    },
    {
      "epoch": 0.6360201511335013,
      "grad_norm": 0.1284114420413971,
      "learning_rate": 3.521973125091681e-05,
      "loss": 0.1721,
      "step": 3030
    },
    {
      "epoch": 0.6362300587741394,
      "grad_norm": 0.11894943565130234,
      "learning_rate": 3.5184731778405606e-05,
      "loss": 0.1525,
      "step": 3031
    },
    {
      "epoch": 0.6364399664147775,
      "grad_norm": 0.13244931399822235,
      "learning_rate": 3.514974026203185e-05,
      "loss": 0.1662,
      "step": 3032
    },
    {
      "epoch": 0.6366498740554156,
      "grad_norm": 0.1697516143321991,
      "learning_rate": 3.5114756720586775e-05,
      "loss": 0.1602,
      "step": 3033
    },
    {
      "epoch": 0.6368597816960537,
      "grad_norm": 0.13253948092460632,
      "learning_rate": 3.507978117285737e-05,
      "loss": 0.1611,
      "step": 3034
    },
    {
      "epoch": 0.6370696893366918,
      "grad_norm": 0.09245211631059647,
      "learning_rate": 3.5044813637626287e-05,
      "loss": 0.143,
      "step": 3035
    },
    {
      "epoch": 0.6372795969773299,
      "grad_norm": 0.10185031592845917,
      "learning_rate": 3.5009854133671895e-05,
      "loss": 0.1428,
      "step": 3036
    },
    {
      "epoch": 0.637489504617968,
      "grad_norm": 0.11988876760005951,
      "learning_rate": 3.497490267976825e-05,
      "loss": 0.1596,
      "step": 3037
    },
    {
      "epoch": 0.6376994122586063,
      "grad_norm": 0.10822691023349762,
      "learning_rate": 3.4939959294685075e-05,
      "loss": 0.1648,
      "step": 3038
    },
    {
      "epoch": 0.6379093198992444,
      "grad_norm": 0.12224753201007843,
      "learning_rate": 3.490502399718775e-05,
      "loss": 0.1777,
      "step": 3039
    },
    {
      "epoch": 0.6381192275398825,
      "grad_norm": 0.11879666894674301,
      "learning_rate": 3.487009680603736e-05,
      "loss": 0.1783,
      "step": 3040
    },
    {
      "epoch": 0.6383291351805206,
      "grad_norm": 0.1191924661397934,
      "learning_rate": 3.483517773999057e-05,
      "loss": 0.1533,
      "step": 3041
    },
    {
      "epoch": 0.6385390428211587,
      "grad_norm": 0.10266808420419693,
      "learning_rate": 3.480026681779975e-05,
      "loss": 0.1683,
      "step": 3042
    },
    {
      "epoch": 0.6387489504617968,
      "grad_norm": 0.11044494807720184,
      "learning_rate": 3.476536405821286e-05,
      "loss": 0.1344,
      "step": 3043
    },
    {
      "epoch": 0.6389588581024349,
      "grad_norm": 0.12145400047302246,
      "learning_rate": 3.473046947997345e-05,
      "loss": 0.1396,
      "step": 3044
    },
    {
      "epoch": 0.639168765743073,
      "grad_norm": 0.13350744545459747,
      "learning_rate": 3.469558310182072e-05,
      "loss": 0.1552,
      "step": 3045
    },
    {
      "epoch": 0.6393786733837111,
      "grad_norm": 0.12254119664430618,
      "learning_rate": 3.4660704942489463e-05,
      "loss": 0.1474,
      "step": 3046
    },
    {
      "epoch": 0.6395885810243492,
      "grad_norm": 0.11156780272722244,
      "learning_rate": 3.462583502071002e-05,
      "loss": 0.153,
      "step": 3047
    },
    {
      "epoch": 0.6397984886649875,
      "grad_norm": 0.10157320648431778,
      "learning_rate": 3.459097335520837e-05,
      "loss": 0.15,
      "step": 3048
    },
    {
      "epoch": 0.6400083963056256,
      "grad_norm": 0.11385181546211243,
      "learning_rate": 3.4556119964706e-05,
      "loss": 0.1669,
      "step": 3049
    },
    {
      "epoch": 0.6402183039462637,
      "grad_norm": 0.12311366945505142,
      "learning_rate": 3.4521274867920014e-05,
      "loss": 0.1607,
      "step": 3050
    },
    {
      "epoch": 0.6404282115869018,
      "grad_norm": 0.11515744030475616,
      "learning_rate": 3.4486438083562964e-05,
      "loss": 0.1471,
      "step": 3051
    },
    {
      "epoch": 0.6406381192275399,
      "grad_norm": 0.1235782578587532,
      "learning_rate": 3.445160963034304e-05,
      "loss": 0.1597,
      "step": 3052
    },
    {
      "epoch": 0.640848026868178,
      "grad_norm": 0.1385941505432129,
      "learning_rate": 3.44167895269639e-05,
      "loss": 0.1529,
      "step": 3053
    },
    {
      "epoch": 0.6410579345088161,
      "grad_norm": 0.24027882516384125,
      "learning_rate": 3.438197779212475e-05,
      "loss": 0.1648,
      "step": 3054
    },
    {
      "epoch": 0.6412678421494542,
      "grad_norm": 0.10963878035545349,
      "learning_rate": 3.4347174444520266e-05,
      "loss": 0.1673,
      "step": 3055
    },
    {
      "epoch": 0.6414777497900923,
      "grad_norm": 0.12933169305324554,
      "learning_rate": 3.4312379502840665e-05,
      "loss": 0.1478,
      "step": 3056
    },
    {
      "epoch": 0.6416876574307305,
      "grad_norm": 0.11227238178253174,
      "learning_rate": 3.427759298577163e-05,
      "loss": 0.1534,
      "step": 3057
    },
    {
      "epoch": 0.6418975650713686,
      "grad_norm": 0.12470339983701706,
      "learning_rate": 3.4242814911994284e-05,
      "loss": 0.1521,
      "step": 3058
    },
    {
      "epoch": 0.6421074727120067,
      "grad_norm": 0.12207306921482086,
      "learning_rate": 3.4208045300185256e-05,
      "loss": 0.1701,
      "step": 3059
    },
    {
      "epoch": 0.6423173803526449,
      "grad_norm": 0.1703091263771057,
      "learning_rate": 3.417328416901663e-05,
      "loss": 0.1528,
      "step": 3060
    },
    {
      "epoch": 0.642527287993283,
      "grad_norm": 0.12895244359970093,
      "learning_rate": 3.413853153715593e-05,
      "loss": 0.1616,
      "step": 3061
    },
    {
      "epoch": 0.6427371956339211,
      "grad_norm": 0.10771409422159195,
      "learning_rate": 3.4103787423266106e-05,
      "loss": 0.1451,
      "step": 3062
    },
    {
      "epoch": 0.6429471032745592,
      "grad_norm": 0.1198326051235199,
      "learning_rate": 3.406905184600555e-05,
      "loss": 0.1571,
      "step": 3063
    },
    {
      "epoch": 0.6431570109151973,
      "grad_norm": 0.11176072061061859,
      "learning_rate": 3.403432482402806e-05,
      "loss": 0.1602,
      "step": 3064
    },
    {
      "epoch": 0.6433669185558354,
      "grad_norm": 0.11805977672338486,
      "learning_rate": 3.3999606375982816e-05,
      "loss": 0.1562,
      "step": 3065
    },
    {
      "epoch": 0.6435768261964736,
      "grad_norm": 0.12517598271369934,
      "learning_rate": 3.3964896520514445e-05,
      "loss": 0.1566,
      "step": 3066
    },
    {
      "epoch": 0.6437867338371117,
      "grad_norm": 0.09705498069524765,
      "learning_rate": 3.393019527626292e-05,
      "loss": 0.154,
      "step": 3067
    },
    {
      "epoch": 0.6439966414777498,
      "grad_norm": 0.13115854561328888,
      "learning_rate": 3.38955026618636e-05,
      "loss": 0.156,
      "step": 3068
    },
    {
      "epoch": 0.6442065491183879,
      "grad_norm": 0.10653633624315262,
      "learning_rate": 3.386081869594723e-05,
      "loss": 0.146,
      "step": 3069
    },
    {
      "epoch": 0.644416456759026,
      "grad_norm": 0.12678882479667664,
      "learning_rate": 3.382614339713986e-05,
      "loss": 0.1642,
      "step": 3070
    },
    {
      "epoch": 0.6446263643996641,
      "grad_norm": 0.10640235245227814,
      "learning_rate": 3.379147678406296e-05,
      "loss": 0.1381,
      "step": 3071
    },
    {
      "epoch": 0.6448362720403022,
      "grad_norm": 0.13540606200695038,
      "learning_rate": 3.3756818875333264e-05,
      "loss": 0.1519,
      "step": 3072
    },
    {
      "epoch": 0.6450461796809404,
      "grad_norm": 0.14184518158435822,
      "learning_rate": 3.372216968956286e-05,
      "loss": 0.1577,
      "step": 3073
    },
    {
      "epoch": 0.6452560873215785,
      "grad_norm": 0.1309628039598465,
      "learning_rate": 3.368752924535914e-05,
      "loss": 0.1513,
      "step": 3074
    },
    {
      "epoch": 0.6454659949622166,
      "grad_norm": 0.11215526610612869,
      "learning_rate": 3.3652897561324854e-05,
      "loss": 0.1553,
      "step": 3075
    },
    {
      "epoch": 0.6456759026028548,
      "grad_norm": 0.13833361864089966,
      "learning_rate": 3.361827465605797e-05,
      "loss": 0.1704,
      "step": 3076
    },
    {
      "epoch": 0.6458858102434929,
      "grad_norm": 0.10265146195888519,
      "learning_rate": 3.35836605481518e-05,
      "loss": 0.1483,
      "step": 3077
    },
    {
      "epoch": 0.646095717884131,
      "grad_norm": 0.10432995110750198,
      "learning_rate": 3.354905525619492e-05,
      "loss": 0.1549,
      "step": 3078
    },
    {
      "epoch": 0.6463056255247691,
      "grad_norm": 0.15146160125732422,
      "learning_rate": 3.3514458798771135e-05,
      "loss": 0.1563,
      "step": 3079
    },
    {
      "epoch": 0.6465155331654072,
      "grad_norm": 0.13524983823299408,
      "learning_rate": 3.347987119445953e-05,
      "loss": 0.1638,
      "step": 3080
    },
    {
      "epoch": 0.6467254408060453,
      "grad_norm": 0.11744452267885208,
      "learning_rate": 3.344529246183448e-05,
      "loss": 0.1476,
      "step": 3081
    },
    {
      "epoch": 0.6469353484466834,
      "grad_norm": 0.11828684061765671,
      "learning_rate": 3.341072261946549e-05,
      "loss": 0.1505,
      "step": 3082
    },
    {
      "epoch": 0.6471452560873215,
      "grad_norm": 0.1145751029253006,
      "learning_rate": 3.337616168591743e-05,
      "loss": 0.1522,
      "step": 3083
    },
    {
      "epoch": 0.6473551637279596,
      "grad_norm": 0.10468050092458725,
      "learning_rate": 3.334160967975025e-05,
      "loss": 0.1472,
      "step": 3084
    },
    {
      "epoch": 0.6475650713685979,
      "grad_norm": 0.10044942796230316,
      "learning_rate": 3.3307066619519225e-05,
      "loss": 0.1608,
      "step": 3085
    },
    {
      "epoch": 0.647774979009236,
      "grad_norm": 0.11021267622709274,
      "learning_rate": 3.327253252377469e-05,
      "loss": 0.1476,
      "step": 3086
    },
    {
      "epoch": 0.6479848866498741,
      "grad_norm": 0.11376648396253586,
      "learning_rate": 3.323800741106231e-05,
      "loss": 0.1584,
      "step": 3087
    },
    {
      "epoch": 0.6481947942905122,
      "grad_norm": 0.12721943855285645,
      "learning_rate": 3.320349129992282e-05,
      "loss": 0.1635,
      "step": 3088
    },
    {
      "epoch": 0.6484047019311503,
      "grad_norm": 0.11218689382076263,
      "learning_rate": 3.3168984208892186e-05,
      "loss": 0.1543,
      "step": 3089
    },
    {
      "epoch": 0.6486146095717884,
      "grad_norm": 0.10654529184103012,
      "learning_rate": 3.3134486156501474e-05,
      "loss": 0.1484,
      "step": 3090
    },
    {
      "epoch": 0.6488245172124265,
      "grad_norm": 0.12575599551200867,
      "learning_rate": 3.309999716127696e-05,
      "loss": 0.1611,
      "step": 3091
    },
    {
      "epoch": 0.6490344248530646,
      "grad_norm": 0.1281367838382721,
      "learning_rate": 3.306551724174002e-05,
      "loss": 0.1555,
      "step": 3092
    },
    {
      "epoch": 0.6492443324937027,
      "grad_norm": 0.12584231793880463,
      "learning_rate": 3.303104641640714e-05,
      "loss": 0.143,
      "step": 3093
    },
    {
      "epoch": 0.6494542401343409,
      "grad_norm": 0.11681025475263596,
      "learning_rate": 3.299658470378993e-05,
      "loss": 0.1556,
      "step": 3094
    },
    {
      "epoch": 0.649664147774979,
      "grad_norm": 0.12307055294513702,
      "learning_rate": 3.2962132122395144e-05,
      "loss": 0.1556,
      "step": 3095
    },
    {
      "epoch": 0.6498740554156172,
      "grad_norm": 0.123899906873703,
      "learning_rate": 3.2927688690724584e-05,
      "loss": 0.1494,
      "step": 3096
    },
    {
      "epoch": 0.6500839630562553,
      "grad_norm": 0.13147427141666412,
      "learning_rate": 3.2893254427275186e-05,
      "loss": 0.1506,
      "step": 3097
    },
    {
      "epoch": 0.6502938706968934,
      "grad_norm": 0.11978549510240555,
      "learning_rate": 3.285882935053891e-05,
      "loss": 0.1497,
      "step": 3098
    },
    {
      "epoch": 0.6505037783375315,
      "grad_norm": 0.12688401341438293,
      "learning_rate": 3.2824413479002837e-05,
      "loss": 0.162,
      "step": 3099
    },
    {
      "epoch": 0.6507136859781696,
      "grad_norm": 0.13381405174732208,
      "learning_rate": 3.279000683114903e-05,
      "loss": 0.1558,
      "step": 3100
    },
    {
      "epoch": 0.6509235936188077,
      "grad_norm": 0.11726606637239456,
      "learning_rate": 3.2755609425454695e-05,
      "loss": 0.1566,
      "step": 3101
    },
    {
      "epoch": 0.6511335012594458,
      "grad_norm": 0.14988724887371063,
      "learning_rate": 3.272122128039198e-05,
      "loss": 0.1487,
      "step": 3102
    },
    {
      "epoch": 0.6513434089000839,
      "grad_norm": 0.15168358385562897,
      "learning_rate": 3.268684241442814e-05,
      "loss": 0.1637,
      "step": 3103
    },
    {
      "epoch": 0.6515533165407221,
      "grad_norm": 0.10664239525794983,
      "learning_rate": 3.2652472846025394e-05,
      "loss": 0.1514,
      "step": 3104
    },
    {
      "epoch": 0.6517632241813602,
      "grad_norm": 0.12004858255386353,
      "learning_rate": 3.261811259364098e-05,
      "loss": 0.177,
      "step": 3105
    },
    {
      "epoch": 0.6519731318219983,
      "grad_norm": 0.11165560036897659,
      "learning_rate": 3.258376167572718e-05,
      "loss": 0.1578,
      "step": 3106
    },
    {
      "epoch": 0.6521830394626364,
      "grad_norm": 0.1488139033317566,
      "learning_rate": 3.2549420110731166e-05,
      "loss": 0.1524,
      "step": 3107
    },
    {
      "epoch": 0.6523929471032746,
      "grad_norm": 0.12690328061580658,
      "learning_rate": 3.251508791709517e-05,
      "loss": 0.171,
      "step": 3108
    },
    {
      "epoch": 0.6526028547439127,
      "grad_norm": 0.12381751090288162,
      "learning_rate": 3.248076511325635e-05,
      "loss": 0.1619,
      "step": 3109
    },
    {
      "epoch": 0.6528127623845508,
      "grad_norm": 0.10168858617544174,
      "learning_rate": 3.244645171764685e-05,
      "loss": 0.1493,
      "step": 3110
    },
    {
      "epoch": 0.6530226700251889,
      "grad_norm": 0.1144690215587616,
      "learning_rate": 3.2412147748693746e-05,
      "loss": 0.1627,
      "step": 3111
    },
    {
      "epoch": 0.653232577665827,
      "grad_norm": 0.12090957164764404,
      "learning_rate": 3.2377853224819056e-05,
      "loss": 0.1716,
      "step": 3112
    },
    {
      "epoch": 0.6534424853064652,
      "grad_norm": 0.11143061518669128,
      "learning_rate": 3.2343568164439716e-05,
      "loss": 0.1552,
      "step": 3113
    },
    {
      "epoch": 0.6536523929471033,
      "grad_norm": 0.12665286660194397,
      "learning_rate": 3.2309292585967585e-05,
      "loss": 0.1642,
      "step": 3114
    },
    {
      "epoch": 0.6538623005877414,
      "grad_norm": 0.12189696729183197,
      "learning_rate": 3.227502650780943e-05,
      "loss": 0.1598,
      "step": 3115
    },
    {
      "epoch": 0.6540722082283795,
      "grad_norm": 0.12147743999958038,
      "learning_rate": 3.224076994836692e-05,
      "loss": 0.1518,
      "step": 3116
    },
    {
      "epoch": 0.6542821158690176,
      "grad_norm": 0.10486441850662231,
      "learning_rate": 3.2206522926036616e-05,
      "loss": 0.1327,
      "step": 3117
    },
    {
      "epoch": 0.6544920235096557,
      "grad_norm": 0.13366806507110596,
      "learning_rate": 3.217228545920994e-05,
      "loss": 0.169,
      "step": 3118
    },
    {
      "epoch": 0.6547019311502938,
      "grad_norm": 0.12645749747753143,
      "learning_rate": 3.21380575662732e-05,
      "loss": 0.1449,
      "step": 3119
    },
    {
      "epoch": 0.654911838790932,
      "grad_norm": 0.11253198981285095,
      "learning_rate": 3.210383926560759e-05,
      "loss": 0.1596,
      "step": 3120
    },
    {
      "epoch": 0.6551217464315701,
      "grad_norm": 0.1114022433757782,
      "learning_rate": 3.2069630575589046e-05,
      "loss": 0.1436,
      "step": 3121
    },
    {
      "epoch": 0.6553316540722083,
      "grad_norm": 0.11162886768579483,
      "learning_rate": 3.203543151458847e-05,
      "loss": 0.1565,
      "step": 3122
    },
    {
      "epoch": 0.6555415617128464,
      "grad_norm": 0.09162106364965439,
      "learning_rate": 3.200124210097152e-05,
      "loss": 0.1511,
      "step": 3123
    },
    {
      "epoch": 0.6557514693534845,
      "grad_norm": 0.10009972006082535,
      "learning_rate": 3.196706235309871e-05,
      "loss": 0.1409,
      "step": 3124
    },
    {
      "epoch": 0.6559613769941226,
      "grad_norm": 0.11142247170209885,
      "learning_rate": 3.193289228932531e-05,
      "loss": 0.1459,
      "step": 3125
    },
    {
      "epoch": 0.6561712846347607,
      "grad_norm": 0.11290930956602097,
      "learning_rate": 3.189873192800147e-05,
      "loss": 0.171,
      "step": 3126
    },
    {
      "epoch": 0.6563811922753988,
      "grad_norm": 0.11958450824022293,
      "learning_rate": 3.186458128747206e-05,
      "loss": 0.1613,
      "step": 3127
    },
    {
      "epoch": 0.6565910999160369,
      "grad_norm": 0.1151280552148819,
      "learning_rate": 3.1830440386076745e-05,
      "loss": 0.1648,
      "step": 3128
    },
    {
      "epoch": 0.656801007556675,
      "grad_norm": 0.11343618482351303,
      "learning_rate": 3.179630924214998e-05,
      "loss": 0.1578,
      "step": 3129
    },
    {
      "epoch": 0.6570109151973131,
      "grad_norm": 0.1188909262418747,
      "learning_rate": 3.176218787402097e-05,
      "loss": 0.1682,
      "step": 3130
    },
    {
      "epoch": 0.6572208228379512,
      "grad_norm": 0.13589932024478912,
      "learning_rate": 3.172807630001366e-05,
      "loss": 0.1731,
      "step": 3131
    },
    {
      "epoch": 0.6574307304785895,
      "grad_norm": 0.12900803983211517,
      "learning_rate": 3.1693974538446756e-05,
      "loss": 0.1457,
      "step": 3132
    },
    {
      "epoch": 0.6576406381192276,
      "grad_norm": 0.12908852100372314,
      "learning_rate": 3.165988260763369e-05,
      "loss": 0.159,
      "step": 3133
    },
    {
      "epoch": 0.6578505457598657,
      "grad_norm": 0.14849267899990082,
      "learning_rate": 3.16258005258826e-05,
      "loss": 0.1688,
      "step": 3134
    },
    {
      "epoch": 0.6580604534005038,
      "grad_norm": 0.1258804351091385,
      "learning_rate": 3.159172831149631e-05,
      "loss": 0.1725,
      "step": 3135
    },
    {
      "epoch": 0.6582703610411419,
      "grad_norm": 0.12530116736888885,
      "learning_rate": 3.155766598277243e-05,
      "loss": 0.1408,
      "step": 3136
    },
    {
      "epoch": 0.65848026868178,
      "grad_norm": 0.14081692695617676,
      "learning_rate": 3.1523613558003174e-05,
      "loss": 0.1622,
      "step": 3137
    },
    {
      "epoch": 0.6586901763224181,
      "grad_norm": 0.1500757932662964,
      "learning_rate": 3.1489571055475486e-05,
      "loss": 0.1671,
      "step": 3138
    },
    {
      "epoch": 0.6589000839630562,
      "grad_norm": 0.11542218178510666,
      "learning_rate": 3.145553849347097e-05,
      "loss": 0.1495,
      "step": 3139
    },
    {
      "epoch": 0.6591099916036943,
      "grad_norm": 0.12322347611188889,
      "learning_rate": 3.1421515890265893e-05,
      "loss": 0.139,
      "step": 3140
    },
    {
      "epoch": 0.6593198992443325,
      "grad_norm": 0.1026599332690239,
      "learning_rate": 3.1387503264131137e-05,
      "loss": 0.1599,
      "step": 3141
    },
    {
      "epoch": 0.6595298068849706,
      "grad_norm": 0.13048797845840454,
      "learning_rate": 3.135350063333229e-05,
      "loss": 0.177,
      "step": 3142
    },
    {
      "epoch": 0.6597397145256088,
      "grad_norm": 0.11457457393407822,
      "learning_rate": 3.1319508016129526e-05,
      "loss": 0.176,
      "step": 3143
    },
    {
      "epoch": 0.6599496221662469,
      "grad_norm": 0.11056394875049591,
      "learning_rate": 3.1285525430777686e-05,
      "loss": 0.1572,
      "step": 3144
    },
    {
      "epoch": 0.660159529806885,
      "grad_norm": 0.12389667332172394,
      "learning_rate": 3.125155289552616e-05,
      "loss": 0.1479,
      "step": 3145
    },
    {
      "epoch": 0.6603694374475231,
      "grad_norm": 0.1127491444349289,
      "learning_rate": 3.121759042861898e-05,
      "loss": 0.1665,
      "step": 3146
    },
    {
      "epoch": 0.6605793450881612,
      "grad_norm": 0.11531811952590942,
      "learning_rate": 3.11836380482948e-05,
      "loss": 0.1625,
      "step": 3147
    },
    {
      "epoch": 0.6607892527287993,
      "grad_norm": 0.1351030021905899,
      "learning_rate": 3.114969577278679e-05,
      "loss": 0.1437,
      "step": 3148
    },
    {
      "epoch": 0.6609991603694374,
      "grad_norm": 0.13135118782520294,
      "learning_rate": 3.111576362032273e-05,
      "loss": 0.1598,
      "step": 3149
    },
    {
      "epoch": 0.6612090680100756,
      "grad_norm": 0.112535759806633,
      "learning_rate": 3.1081841609124964e-05,
      "loss": 0.1584,
      "step": 3150
    },
    {
      "epoch": 0.6614189756507137,
      "grad_norm": 0.1222812607884407,
      "learning_rate": 3.10479297574104e-05,
      "loss": 0.1773,
      "step": 3151
    },
    {
      "epoch": 0.6616288832913518,
      "grad_norm": 0.11697451025247574,
      "learning_rate": 3.1014028083390456e-05,
      "loss": 0.1705,
      "step": 3152
    },
    {
      "epoch": 0.6618387909319899,
      "grad_norm": 0.12584251165390015,
      "learning_rate": 3.0980136605271127e-05,
      "loss": 0.1519,
      "step": 3153
    },
    {
      "epoch": 0.662048698572628,
      "grad_norm": 0.12311432510614395,
      "learning_rate": 3.0946255341252914e-05,
      "loss": 0.1631,
      "step": 3154
    },
    {
      "epoch": 0.6622586062132662,
      "grad_norm": 0.1087951511144638,
      "learning_rate": 3.09123843095308e-05,
      "loss": 0.1443,
      "step": 3155
    },
    {
      "epoch": 0.6624685138539043,
      "grad_norm": 0.11787470430135727,
      "learning_rate": 3.087852352829431e-05,
      "loss": 0.1655,
      "step": 3156
    },
    {
      "epoch": 0.6626784214945424,
      "grad_norm": 0.13332132995128632,
      "learning_rate": 3.0844673015727476e-05,
      "loss": 0.1495,
      "step": 3157
    },
    {
      "epoch": 0.6628883291351805,
      "grad_norm": 0.09370341897010803,
      "learning_rate": 3.081083279000878e-05,
      "loss": 0.1664,
      "step": 3158
    },
    {
      "epoch": 0.6630982367758187,
      "grad_norm": 0.1299756020307541,
      "learning_rate": 3.0777002869311214e-05,
      "loss": 0.1561,
      "step": 3159
    },
    {
      "epoch": 0.6633081444164568,
      "grad_norm": 0.1356882005929947,
      "learning_rate": 3.074318327180219e-05,
      "loss": 0.1468,
      "step": 3160
    },
    {
      "epoch": 0.6635180520570949,
      "grad_norm": 0.12208224087953568,
      "learning_rate": 3.070937401564365e-05,
      "loss": 0.1571,
      "step": 3161
    },
    {
      "epoch": 0.663727959697733,
      "grad_norm": 0.1734291911125183,
      "learning_rate": 3.0675575118991886e-05,
      "loss": 0.1561,
      "step": 3162
    },
    {
      "epoch": 0.6639378673383711,
      "grad_norm": 0.11419927328824997,
      "learning_rate": 3.064178659999771e-05,
      "loss": 0.1735,
      "step": 3163
    },
    {
      "epoch": 0.6641477749790092,
      "grad_norm": 0.10793764144182205,
      "learning_rate": 3.06080084768063e-05,
      "loss": 0.1581,
      "step": 3164
    },
    {
      "epoch": 0.6643576826196473,
      "grad_norm": 0.12163906544446945,
      "learning_rate": 3.0574240767557315e-05,
      "loss": 0.1681,
      "step": 3165
    },
    {
      "epoch": 0.6645675902602854,
      "grad_norm": 0.12725725769996643,
      "learning_rate": 3.054048349038476e-05,
      "loss": 0.1486,
      "step": 3166
    },
    {
      "epoch": 0.6647774979009236,
      "grad_norm": 0.1115790456533432,
      "learning_rate": 3.050673666341708e-05,
      "loss": 0.1458,
      "step": 3167
    },
    {
      "epoch": 0.6649874055415617,
      "grad_norm": 0.12533406913280487,
      "learning_rate": 3.04730003047771e-05,
      "loss": 0.1533,
      "step": 3168
    },
    {
      "epoch": 0.6651973131821999,
      "grad_norm": 0.11061311513185501,
      "learning_rate": 3.0439274432581987e-05,
      "loss": 0.157,
      "step": 3169
    },
    {
      "epoch": 0.665407220822838,
      "grad_norm": 0.11690240353345871,
      "learning_rate": 3.04055590649433e-05,
      "loss": 0.1543,
      "step": 3170
    },
    {
      "epoch": 0.6656171284634761,
      "grad_norm": 0.11588296294212341,
      "learning_rate": 3.0371854219967e-05,
      "loss": 0.15,
      "step": 3171
    },
    {
      "epoch": 0.6658270361041142,
      "grad_norm": 0.13207639753818512,
      "learning_rate": 3.0338159915753328e-05,
      "loss": 0.1495,
      "step": 3172
    },
    {
      "epoch": 0.6660369437447523,
      "grad_norm": 0.14322279393672943,
      "learning_rate": 3.0304476170396912e-05,
      "loss": 0.1324,
      "step": 3173
    },
    {
      "epoch": 0.6662468513853904,
      "grad_norm": 0.1106521412730217,
      "learning_rate": 3.0270803001986682e-05,
      "loss": 0.134,
      "step": 3174
    },
    {
      "epoch": 0.6664567590260285,
      "grad_norm": 0.1531291902065277,
      "learning_rate": 3.0237140428605913e-05,
      "loss": 0.1617,
      "step": 3175
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.1315673142671585,
      "learning_rate": 3.0203488468332154e-05,
      "loss": 0.1656,
      "step": 3176
    },
    {
      "epoch": 0.6668765743073047,
      "grad_norm": 0.1372612863779068,
      "learning_rate": 3.0169847139237294e-05,
      "loss": 0.1671,
      "step": 3177
    },
    {
      "epoch": 0.667086481947943,
      "grad_norm": 0.10959561169147491,
      "learning_rate": 3.0136216459387477e-05,
      "loss": 0.1617,
      "step": 3178
    },
    {
      "epoch": 0.6672963895885811,
      "grad_norm": 0.11745753139257431,
      "learning_rate": 3.010259644684317e-05,
      "loss": 0.1597,
      "step": 3179
    },
    {
      "epoch": 0.6675062972292192,
      "grad_norm": 0.12485745549201965,
      "learning_rate": 3.006898711965907e-05,
      "loss": 0.14,
      "step": 3180
    },
    {
      "epoch": 0.6677162048698573,
      "grad_norm": 0.12715327739715576,
      "learning_rate": 3.0035388495884154e-05,
      "loss": 0.1428,
      "step": 3181
    },
    {
      "epoch": 0.6679261125104954,
      "grad_norm": 0.12533673644065857,
      "learning_rate": 3.000180059356168e-05,
      "loss": 0.1351,
      "step": 3182
    },
    {
      "epoch": 0.6681360201511335,
      "grad_norm": 0.11587972939014435,
      "learning_rate": 2.9968223430729076e-05,
      "loss": 0.168,
      "step": 3183
    },
    {
      "epoch": 0.6683459277917716,
      "grad_norm": 0.12353124469518661,
      "learning_rate": 2.9934657025418055e-05,
      "loss": 0.1468,
      "step": 3184
    },
    {
      "epoch": 0.6685558354324097,
      "grad_norm": 0.13154150545597076,
      "learning_rate": 2.990110139565457e-05,
      "loss": 0.1688,
      "step": 3185
    },
    {
      "epoch": 0.6687657430730478,
      "grad_norm": 0.13129670917987823,
      "learning_rate": 2.9867556559458738e-05,
      "loss": 0.1636,
      "step": 3186
    },
    {
      "epoch": 0.668975650713686,
      "grad_norm": 0.1262674629688263,
      "learning_rate": 2.9834022534844897e-05,
      "loss": 0.178,
      "step": 3187
    },
    {
      "epoch": 0.6691855583543241,
      "grad_norm": 0.13294339179992676,
      "learning_rate": 2.9800499339821603e-05,
      "loss": 0.1688,
      "step": 3188
    },
    {
      "epoch": 0.6693954659949622,
      "grad_norm": 0.11901882290840149,
      "learning_rate": 2.9766986992391583e-05,
      "loss": 0.1786,
      "step": 3189
    },
    {
      "epoch": 0.6696053736356004,
      "grad_norm": 0.13393592834472656,
      "learning_rate": 2.9733485510551706e-05,
      "loss": 0.1481,
      "step": 3190
    },
    {
      "epoch": 0.6698152812762385,
      "grad_norm": 0.1557946652173996,
      "learning_rate": 2.9699994912293038e-05,
      "loss": 0.1651,
      "step": 3191
    },
    {
      "epoch": 0.6700251889168766,
      "grad_norm": 0.13933785259723663,
      "learning_rate": 2.966651521560081e-05,
      "loss": 0.1512,
      "step": 3192
    },
    {
      "epoch": 0.6702350965575147,
      "grad_norm": 0.12215977162122726,
      "learning_rate": 2.963304643845437e-05,
      "loss": 0.1748,
      "step": 3193
    },
    {
      "epoch": 0.6704450041981528,
      "grad_norm": 0.13210149109363556,
      "learning_rate": 2.9599588598827233e-05,
      "loss": 0.176,
      "step": 3194
    },
    {
      "epoch": 0.6706549118387909,
      "grad_norm": 0.14156967401504517,
      "learning_rate": 2.9566141714687e-05,
      "loss": 0.1388,
      "step": 3195
    },
    {
      "epoch": 0.670864819479429,
      "grad_norm": 0.12155076861381531,
      "learning_rate": 2.9532705803995464e-05,
      "loss": 0.1399,
      "step": 3196
    },
    {
      "epoch": 0.6710747271200672,
      "grad_norm": 0.12318848073482513,
      "learning_rate": 2.9499280884708393e-05,
      "loss": 0.1541,
      "step": 3197
    },
    {
      "epoch": 0.6712846347607053,
      "grad_norm": 0.1175316795706749,
      "learning_rate": 2.946586697477579e-05,
      "loss": 0.1486,
      "step": 3198
    },
    {
      "epoch": 0.6714945424013434,
      "grad_norm": 0.12203275412321091,
      "learning_rate": 2.9432464092141655e-05,
      "loss": 0.1647,
      "step": 3199
    },
    {
      "epoch": 0.6717044500419815,
      "grad_norm": 0.15709438920021057,
      "learning_rate": 2.939907225474413e-05,
      "loss": 0.1512,
      "step": 3200
    },
    {
      "epoch": 0.6719143576826196,
      "grad_norm": 0.11067797243595123,
      "learning_rate": 2.9365691480515368e-05,
      "loss": 0.1693,
      "step": 3201
    },
    {
      "epoch": 0.6721242653232578,
      "grad_norm": 0.0939958468079567,
      "learning_rate": 2.9332321787381622e-05,
      "loss": 0.1538,
      "step": 3202
    },
    {
      "epoch": 0.6723341729638959,
      "grad_norm": 0.12024837732315063,
      "learning_rate": 2.9298963193263195e-05,
      "loss": 0.1575,
      "step": 3203
    },
    {
      "epoch": 0.672544080604534,
      "grad_norm": 0.10305708646774292,
      "learning_rate": 2.926561571607438e-05,
      "loss": 0.1482,
      "step": 3204
    },
    {
      "epoch": 0.6727539882451721,
      "grad_norm": 0.12283535301685333,
      "learning_rate": 2.923227937372354e-05,
      "loss": 0.1531,
      "step": 3205
    },
    {
      "epoch": 0.6729638958858103,
      "grad_norm": 0.12914074957370758,
      "learning_rate": 2.9198954184113074e-05,
      "loss": 0.1724,
      "step": 3206
    },
    {
      "epoch": 0.6731738035264484,
      "grad_norm": 0.12399044632911682,
      "learning_rate": 2.9165640165139346e-05,
      "loss": 0.1426,
      "step": 3207
    },
    {
      "epoch": 0.6733837111670865,
      "grad_norm": 0.12031276524066925,
      "learning_rate": 2.9132337334692767e-05,
      "loss": 0.1614,
      "step": 3208
    },
    {
      "epoch": 0.6735936188077246,
      "grad_norm": 0.10775694251060486,
      "learning_rate": 2.9099045710657702e-05,
      "loss": 0.1346,
      "step": 3209
    },
    {
      "epoch": 0.6738035264483627,
      "grad_norm": 0.11117519438266754,
      "learning_rate": 2.9065765310912523e-05,
      "loss": 0.1516,
      "step": 3210
    },
    {
      "epoch": 0.6740134340890008,
      "grad_norm": 0.12548020482063293,
      "learning_rate": 2.9032496153329558e-05,
      "loss": 0.1512,
      "step": 3211
    },
    {
      "epoch": 0.6742233417296389,
      "grad_norm": 0.11416387557983398,
      "learning_rate": 2.8999238255775113e-05,
      "loss": 0.1363,
      "step": 3212
    },
    {
      "epoch": 0.674433249370277,
      "grad_norm": 0.12334737926721573,
      "learning_rate": 2.896599163610943e-05,
      "loss": 0.157,
      "step": 3213
    },
    {
      "epoch": 0.6746431570109152,
      "grad_norm": 0.13314662873744965,
      "learning_rate": 2.8932756312186703e-05,
      "loss": 0.1614,
      "step": 3214
    },
    {
      "epoch": 0.6748530646515534,
      "grad_norm": 0.1167728453874588,
      "learning_rate": 2.889953230185505e-05,
      "loss": 0.1456,
      "step": 3215
    },
    {
      "epoch": 0.6750629722921915,
      "grad_norm": 0.10373178124427795,
      "learning_rate": 2.886631962295655e-05,
      "loss": 0.1516,
      "step": 3216
    },
    {
      "epoch": 0.6752728799328296,
      "grad_norm": 0.12552447617053986,
      "learning_rate": 2.8833118293327177e-05,
      "loss": 0.1517,
      "step": 3217
    },
    {
      "epoch": 0.6754827875734677,
      "grad_norm": 0.12713772058486938,
      "learning_rate": 2.8799928330796742e-05,
      "loss": 0.1542,
      "step": 3218
    },
    {
      "epoch": 0.6756926952141058,
      "grad_norm": 0.13052749633789062,
      "learning_rate": 2.876674975318908e-05,
      "loss": 0.1585,
      "step": 3219
    },
    {
      "epoch": 0.6759026028547439,
      "grad_norm": 0.1303415447473526,
      "learning_rate": 2.8733582578321817e-05,
      "loss": 0.164,
      "step": 3220
    },
    {
      "epoch": 0.676112510495382,
      "grad_norm": 0.11603204160928726,
      "learning_rate": 2.8700426824006487e-05,
      "loss": 0.1487,
      "step": 3221
    },
    {
      "epoch": 0.6763224181360201,
      "grad_norm": 0.11492080241441727,
      "learning_rate": 2.8667282508048488e-05,
      "loss": 0.1541,
      "step": 3222
    },
    {
      "epoch": 0.6765323257766582,
      "grad_norm": 0.1157570406794548,
      "learning_rate": 2.8634149648247045e-05,
      "loss": 0.1717,
      "step": 3223
    },
    {
      "epoch": 0.6767422334172963,
      "grad_norm": 0.11512907594442368,
      "learning_rate": 2.8601028262395336e-05,
      "loss": 0.1701,
      "step": 3224
    },
    {
      "epoch": 0.6769521410579346,
      "grad_norm": 0.10905471444129944,
      "learning_rate": 2.8567918368280233e-05,
      "loss": 0.1583,
      "step": 3225
    },
    {
      "epoch": 0.6771620486985727,
      "grad_norm": 0.12866859138011932,
      "learning_rate": 2.8534819983682525e-05,
      "loss": 0.1519,
      "step": 3226
    },
    {
      "epoch": 0.6773719563392108,
      "grad_norm": 0.12510065734386444,
      "learning_rate": 2.8501733126376783e-05,
      "loss": 0.1347,
      "step": 3227
    },
    {
      "epoch": 0.6775818639798489,
      "grad_norm": 0.10416650027036667,
      "learning_rate": 2.8468657814131443e-05,
      "loss": 0.149,
      "step": 3228
    },
    {
      "epoch": 0.677791771620487,
      "grad_norm": 0.12856334447860718,
      "learning_rate": 2.843559406470869e-05,
      "loss": 0.1694,
      "step": 3229
    },
    {
      "epoch": 0.6780016792611251,
      "grad_norm": 0.13767604529857635,
      "learning_rate": 2.84025418958645e-05,
      "loss": 0.1659,
      "step": 3230
    },
    {
      "epoch": 0.6782115869017632,
      "grad_norm": 0.12578238546848297,
      "learning_rate": 2.8369501325348658e-05,
      "loss": 0.1686,
      "step": 3231
    },
    {
      "epoch": 0.6784214945424013,
      "grad_norm": 0.11950796842575073,
      "learning_rate": 2.8336472370904688e-05,
      "loss": 0.1435,
      "step": 3232
    },
    {
      "epoch": 0.6786314021830394,
      "grad_norm": 0.1251773238182068,
      "learning_rate": 2.8303455050269912e-05,
      "loss": 0.154,
      "step": 3233
    },
    {
      "epoch": 0.6788413098236776,
      "grad_norm": 0.13571852445602417,
      "learning_rate": 2.8270449381175367e-05,
      "loss": 0.1417,
      "step": 3234
    },
    {
      "epoch": 0.6790512174643157,
      "grad_norm": 0.12125960737466812,
      "learning_rate": 2.8237455381345846e-05,
      "loss": 0.1635,
      "step": 3235
    },
    {
      "epoch": 0.6792611251049538,
      "grad_norm": 0.1407996267080307,
      "learning_rate": 2.8204473068499914e-05,
      "loss": 0.1663,
      "step": 3236
    },
    {
      "epoch": 0.679471032745592,
      "grad_norm": 0.09970130771398544,
      "learning_rate": 2.8171502460349825e-05,
      "loss": 0.1593,
      "step": 3237
    },
    {
      "epoch": 0.6796809403862301,
      "grad_norm": 0.10902740061283112,
      "learning_rate": 2.8138543574601506e-05,
      "loss": 0.1619,
      "step": 3238
    },
    {
      "epoch": 0.6798908480268682,
      "grad_norm": 0.10807805508375168,
      "learning_rate": 2.8105596428954645e-05,
      "loss": 0.1743,
      "step": 3239
    },
    {
      "epoch": 0.6801007556675063,
      "grad_norm": 0.11692219227552414,
      "learning_rate": 2.807266104110264e-05,
      "loss": 0.1482,
      "step": 3240
    },
    {
      "epoch": 0.6803106633081444,
      "grad_norm": 0.13072606921195984,
      "learning_rate": 2.8039737428732526e-05,
      "loss": 0.1614,
      "step": 3241
    },
    {
      "epoch": 0.6805205709487825,
      "grad_norm": 0.12482010573148727,
      "learning_rate": 2.800682560952504e-05,
      "loss": 0.1431,
      "step": 3242
    },
    {
      "epoch": 0.6807304785894207,
      "grad_norm": 0.11568538844585419,
      "learning_rate": 2.797392560115456e-05,
      "loss": 0.1525,
      "step": 3243
    },
    {
      "epoch": 0.6809403862300588,
      "grad_norm": 0.12819020450115204,
      "learning_rate": 2.7941037421289197e-05,
      "loss": 0.134,
      "step": 3244
    },
    {
      "epoch": 0.6811502938706969,
      "grad_norm": 0.13712668418884277,
      "learning_rate": 2.79081610875906e-05,
      "loss": 0.1565,
      "step": 3245
    },
    {
      "epoch": 0.681360201511335,
      "grad_norm": 0.1392073929309845,
      "learning_rate": 2.787529661771413e-05,
      "loss": 0.1685,
      "step": 3246
    },
    {
      "epoch": 0.6815701091519731,
      "grad_norm": 0.12498795986175537,
      "learning_rate": 2.7842444029308746e-05,
      "loss": 0.1477,
      "step": 3247
    },
    {
      "epoch": 0.6817800167926112,
      "grad_norm": 0.13243268430233002,
      "learning_rate": 2.7809603340017064e-05,
      "loss": 0.1549,
      "step": 3248
    },
    {
      "epoch": 0.6819899244332494,
      "grad_norm": 0.1170286163687706,
      "learning_rate": 2.7776774567475273e-05,
      "loss": 0.143,
      "step": 3249
    },
    {
      "epoch": 0.6821998320738875,
      "grad_norm": 0.10857724398374557,
      "learning_rate": 2.7743957729313165e-05,
      "loss": 0.1534,
      "step": 3250
    },
    {
      "epoch": 0.6824097397145256,
      "grad_norm": 0.13726240396499634,
      "learning_rate": 2.771115284315414e-05,
      "loss": 0.1608,
      "step": 3251
    },
    {
      "epoch": 0.6826196473551638,
      "grad_norm": 0.1267462968826294,
      "learning_rate": 2.7678359926615173e-05,
      "loss": 0.1562,
      "step": 3252
    },
    {
      "epoch": 0.6828295549958019,
      "grad_norm": 0.12552182376384735,
      "learning_rate": 2.764557899730681e-05,
      "loss": 0.1796,
      "step": 3253
    },
    {
      "epoch": 0.68303946263644,
      "grad_norm": 0.15919016301631927,
      "learning_rate": 2.7612810072833146e-05,
      "loss": 0.1457,
      "step": 3254
    },
    {
      "epoch": 0.6832493702770781,
      "grad_norm": 0.1230967566370964,
      "learning_rate": 2.7580053170791853e-05,
      "loss": 0.1374,
      "step": 3255
    },
    {
      "epoch": 0.6834592779177162,
      "grad_norm": 0.11543905735015869,
      "learning_rate": 2.7547308308774112e-05,
      "loss": 0.1452,
      "step": 3256
    },
    {
      "epoch": 0.6836691855583543,
      "grad_norm": 0.11295941472053528,
      "learning_rate": 2.7514575504364702e-05,
      "loss": 0.164,
      "step": 3257
    },
    {
      "epoch": 0.6838790931989924,
      "grad_norm": 0.11988019198179245,
      "learning_rate": 2.7481854775141896e-05,
      "loss": 0.1611,
      "step": 3258
    },
    {
      "epoch": 0.6840890008396305,
      "grad_norm": 0.1507832258939743,
      "learning_rate": 2.7449146138677397e-05,
      "loss": 0.1701,
      "step": 3259
    },
    {
      "epoch": 0.6842989084802686,
      "grad_norm": 0.12370573729276657,
      "learning_rate": 2.7416449612536566e-05,
      "loss": 0.1511,
      "step": 3260
    },
    {
      "epoch": 0.6845088161209067,
      "grad_norm": 0.11700747907161713,
      "learning_rate": 2.7383765214278144e-05,
      "loss": 0.158,
      "step": 3261
    },
    {
      "epoch": 0.684718723761545,
      "grad_norm": 0.10759387165307999,
      "learning_rate": 2.7351092961454418e-05,
      "loss": 0.152,
      "step": 3262
    },
    {
      "epoch": 0.6849286314021831,
      "grad_norm": 0.10964877158403397,
      "learning_rate": 2.7318432871611126e-05,
      "loss": 0.1713,
      "step": 3263
    },
    {
      "epoch": 0.6851385390428212,
      "grad_norm": 0.12055815756320953,
      "learning_rate": 2.7285784962287465e-05,
      "loss": 0.1623,
      "step": 3264
    },
    {
      "epoch": 0.6853484466834593,
      "grad_norm": 0.10874463617801666,
      "learning_rate": 2.7253149251016163e-05,
      "loss": 0.1453,
      "step": 3265
    },
    {
      "epoch": 0.6855583543240974,
      "grad_norm": 0.1362292468547821,
      "learning_rate": 2.722052575532329e-05,
      "loss": 0.1497,
      "step": 3266
    },
    {
      "epoch": 0.6857682619647355,
      "grad_norm": 0.11515046656131744,
      "learning_rate": 2.718791449272843e-05,
      "loss": 0.1507,
      "step": 3267
    },
    {
      "epoch": 0.6859781696053736,
      "grad_norm": 0.10474392026662827,
      "learning_rate": 2.7155315480744542e-05,
      "loss": 0.1558,
      "step": 3268
    },
    {
      "epoch": 0.6861880772460117,
      "grad_norm": 0.10753299295902252,
      "learning_rate": 2.7122728736878088e-05,
      "loss": 0.164,
      "step": 3269
    },
    {
      "epoch": 0.6863979848866498,
      "grad_norm": 0.11624909192323685,
      "learning_rate": 2.7090154278628884e-05,
      "loss": 0.1512,
      "step": 3270
    },
    {
      "epoch": 0.686607892527288,
      "grad_norm": 0.1127047911286354,
      "learning_rate": 2.7057592123490138e-05,
      "loss": 0.1813,
      "step": 3271
    },
    {
      "epoch": 0.6868178001679262,
      "grad_norm": 0.18497790396213531,
      "learning_rate": 2.702504228894849e-05,
      "loss": 0.1857,
      "step": 3272
    },
    {
      "epoch": 0.6870277078085643,
      "grad_norm": 0.12313918769359589,
      "learning_rate": 2.699250479248393e-05,
      "loss": 0.1577,
      "step": 3273
    },
    {
      "epoch": 0.6872376154492024,
      "grad_norm": 0.10727359354496002,
      "learning_rate": 2.695997965156985e-05,
      "loss": 0.1543,
      "step": 3274
    },
    {
      "epoch": 0.6874475230898405,
      "grad_norm": 0.13982883095741272,
      "learning_rate": 2.692746688367298e-05,
      "loss": 0.1453,
      "step": 3275
    },
    {
      "epoch": 0.6876574307304786,
      "grad_norm": 0.12415811419487,
      "learning_rate": 2.689496650625341e-05,
      "loss": 0.1614,
      "step": 3276
    },
    {
      "epoch": 0.6878673383711167,
      "grad_norm": 0.13363641500473022,
      "learning_rate": 2.6862478536764625e-05,
      "loss": 0.1731,
      "step": 3277
    },
    {
      "epoch": 0.6880772460117548,
      "grad_norm": 0.12584188580513,
      "learning_rate": 2.6830002992653375e-05,
      "loss": 0.1698,
      "step": 3278
    },
    {
      "epoch": 0.6882871536523929,
      "grad_norm": 0.11803454160690308,
      "learning_rate": 2.6797539891359803e-05,
      "loss": 0.1713,
      "step": 3279
    },
    {
      "epoch": 0.6884970612930311,
      "grad_norm": 0.11334547400474548,
      "learning_rate": 2.6765089250317266e-05,
      "loss": 0.1586,
      "step": 3280
    },
    {
      "epoch": 0.6887069689336692,
      "grad_norm": 0.12662799656391144,
      "learning_rate": 2.673265108695256e-05,
      "loss": 0.165,
      "step": 3281
    },
    {
      "epoch": 0.6889168765743073,
      "grad_norm": 0.11635055392980576,
      "learning_rate": 2.670022541868571e-05,
      "loss": 0.1566,
      "step": 3282
    },
    {
      "epoch": 0.6891267842149454,
      "grad_norm": 0.12273656576871872,
      "learning_rate": 2.6667812262930027e-05,
      "loss": 0.1591,
      "step": 3283
    },
    {
      "epoch": 0.6893366918555835,
      "grad_norm": 0.10126462578773499,
      "learning_rate": 2.66354116370921e-05,
      "loss": 0.1567,
      "step": 3284
    },
    {
      "epoch": 0.6895465994962217,
      "grad_norm": 0.10264413803815842,
      "learning_rate": 2.660302355857186e-05,
      "loss": 0.1605,
      "step": 3285
    },
    {
      "epoch": 0.6897565071368598,
      "grad_norm": 0.11424882709980011,
      "learning_rate": 2.6570648044762426e-05,
      "loss": 0.1309,
      "step": 3286
    },
    {
      "epoch": 0.6899664147774979,
      "grad_norm": 0.135597363114357,
      "learning_rate": 2.653828511305015e-05,
      "loss": 0.1575,
      "step": 3287
    },
    {
      "epoch": 0.690176322418136,
      "grad_norm": 0.1263936311006546,
      "learning_rate": 2.650593478081468e-05,
      "loss": 0.1459,
      "step": 3288
    },
    {
      "epoch": 0.6903862300587741,
      "grad_norm": 0.10977193713188171,
      "learning_rate": 2.6473597065428903e-05,
      "loss": 0.163,
      "step": 3289
    },
    {
      "epoch": 0.6905961376994123,
      "grad_norm": 0.13382400572299957,
      "learning_rate": 2.644127198425891e-05,
      "loss": 0.1532,
      "step": 3290
    },
    {
      "epoch": 0.6908060453400504,
      "grad_norm": 0.12279796600341797,
      "learning_rate": 2.6408959554663993e-05,
      "loss": 0.1564,
      "step": 3291
    },
    {
      "epoch": 0.6910159529806885,
      "grad_norm": 0.09982291609048843,
      "learning_rate": 2.6376659793996676e-05,
      "loss": 0.1521,
      "step": 3292
    },
    {
      "epoch": 0.6912258606213266,
      "grad_norm": 0.11847012490034103,
      "learning_rate": 2.6344372719602666e-05,
      "loss": 0.1573,
      "step": 3293
    },
    {
      "epoch": 0.6914357682619647,
      "grad_norm": 0.10027361661195755,
      "learning_rate": 2.6312098348820856e-05,
      "loss": 0.1491,
      "step": 3294
    },
    {
      "epoch": 0.6916456759026028,
      "grad_norm": 0.10242442041635513,
      "learning_rate": 2.6279836698983318e-05,
      "loss": 0.164,
      "step": 3295
    },
    {
      "epoch": 0.691855583543241,
      "grad_norm": 0.11035119742155075,
      "learning_rate": 2.6247587787415306e-05,
      "loss": 0.1445,
      "step": 3296
    },
    {
      "epoch": 0.6920654911838791,
      "grad_norm": 0.12222526222467422,
      "learning_rate": 2.621535163143519e-05,
      "loss": 0.1509,
      "step": 3297
    },
    {
      "epoch": 0.6922753988245172,
      "grad_norm": 0.11721225827932358,
      "learning_rate": 2.6183128248354566e-05,
      "loss": 0.1492,
      "step": 3298
    },
    {
      "epoch": 0.6924853064651554,
      "grad_norm": 0.09510528296232224,
      "learning_rate": 2.6150917655478107e-05,
      "loss": 0.1509,
      "step": 3299
    },
    {
      "epoch": 0.6926952141057935,
      "grad_norm": 0.13749557733535767,
      "learning_rate": 2.6118719870103638e-05,
      "loss": 0.1633,
      "step": 3300
    },
    {
      "epoch": 0.6929051217464316,
      "grad_norm": 0.1405005007982254,
      "learning_rate": 2.608653490952211e-05,
      "loss": 0.1508,
      "step": 3301
    },
    {
      "epoch": 0.6931150293870697,
      "grad_norm": 0.13257022202014923,
      "learning_rate": 2.6054362791017573e-05,
      "loss": 0.1527,
      "step": 3302
    },
    {
      "epoch": 0.6933249370277078,
      "grad_norm": 0.12121597677469254,
      "learning_rate": 2.6022203531867195e-05,
      "loss": 0.1613,
      "step": 3303
    },
    {
      "epoch": 0.6935348446683459,
      "grad_norm": 0.11893527954816818,
      "learning_rate": 2.599005714934123e-05,
      "loss": 0.1416,
      "step": 3304
    },
    {
      "epoch": 0.693744752308984,
      "grad_norm": 0.12664799392223358,
      "learning_rate": 2.5957923660703e-05,
      "loss": 0.1539,
      "step": 3305
    },
    {
      "epoch": 0.6939546599496221,
      "grad_norm": 0.14884023368358612,
      "learning_rate": 2.5925803083208966e-05,
      "loss": 0.1489,
      "step": 3306
    },
    {
      "epoch": 0.6941645675902602,
      "grad_norm": 0.13241392374038696,
      "learning_rate": 2.58936954341086e-05,
      "loss": 0.1585,
      "step": 3307
    },
    {
      "epoch": 0.6943744752308985,
      "grad_norm": 0.12439943104982376,
      "learning_rate": 2.586160073064442e-05,
      "loss": 0.1614,
      "step": 3308
    },
    {
      "epoch": 0.6945843828715366,
      "grad_norm": 0.10723230987787247,
      "learning_rate": 2.582951899005201e-05,
      "loss": 0.1402,
      "step": 3309
    },
    {
      "epoch": 0.6947942905121747,
      "grad_norm": 0.12096869945526123,
      "learning_rate": 2.579745022956003e-05,
      "loss": 0.1592,
      "step": 3310
    },
    {
      "epoch": 0.6950041981528128,
      "grad_norm": 0.11293628066778183,
      "learning_rate": 2.5765394466390126e-05,
      "loss": 0.1835,
      "step": 3311
    },
    {
      "epoch": 0.6952141057934509,
      "grad_norm": 0.11762647330760956,
      "learning_rate": 2.573335171775697e-05,
      "loss": 0.1518,
      "step": 3312
    },
    {
      "epoch": 0.695424013434089,
      "grad_norm": 0.13241271674633026,
      "learning_rate": 2.5701322000868223e-05,
      "loss": 0.161,
      "step": 3313
    },
    {
      "epoch": 0.6956339210747271,
      "grad_norm": 0.13515324890613556,
      "learning_rate": 2.566930533292465e-05,
      "loss": 0.1468,
      "step": 3314
    },
    {
      "epoch": 0.6958438287153652,
      "grad_norm": 0.11079118400812149,
      "learning_rate": 2.5637301731119862e-05,
      "loss": 0.145,
      "step": 3315
    },
    {
      "epoch": 0.6960537363560033,
      "grad_norm": 0.16726547479629517,
      "learning_rate": 2.5605311212640547e-05,
      "loss": 0.1358,
      "step": 3316
    },
    {
      "epoch": 0.6962636439966414,
      "grad_norm": 0.1353970617055893,
      "learning_rate": 2.5573333794666326e-05,
      "loss": 0.1537,
      "step": 3317
    },
    {
      "epoch": 0.6964735516372796,
      "grad_norm": 0.10820521414279938,
      "learning_rate": 2.554136949436984e-05,
      "loss": 0.1382,
      "step": 3318
    },
    {
      "epoch": 0.6966834592779177,
      "grad_norm": 0.20719845592975616,
      "learning_rate": 2.5509418328916635e-05,
      "loss": 0.1446,
      "step": 3319
    },
    {
      "epoch": 0.6968933669185559,
      "grad_norm": 0.12657080590724945,
      "learning_rate": 2.5477480315465217e-05,
      "loss": 0.1575,
      "step": 3320
    },
    {
      "epoch": 0.697103274559194,
      "grad_norm": 0.14084367454051971,
      "learning_rate": 2.5445555471167e-05,
      "loss": 0.136,
      "step": 3321
    },
    {
      "epoch": 0.6973131821998321,
      "grad_norm": 0.12968450784683228,
      "learning_rate": 2.541364381316639e-05,
      "loss": 0.1644,
      "step": 3322
    },
    {
      "epoch": 0.6975230898404702,
      "grad_norm": 0.1417001336812973,
      "learning_rate": 2.5381745358600662e-05,
      "loss": 0.155,
      "step": 3323
    },
    {
      "epoch": 0.6977329974811083,
      "grad_norm": 0.13210801780223846,
      "learning_rate": 2.5349860124600024e-05,
      "loss": 0.1715,
      "step": 3324
    },
    {
      "epoch": 0.6979429051217464,
      "grad_norm": 0.11423230171203613,
      "learning_rate": 2.531798812828754e-05,
      "loss": 0.1522,
      "step": 3325
    },
    {
      "epoch": 0.6981528127623845,
      "grad_norm": 0.1215028241276741,
      "learning_rate": 2.5286129386779256e-05,
      "loss": 0.1583,
      "step": 3326
    },
    {
      "epoch": 0.6983627204030227,
      "grad_norm": 0.1341346949338913,
      "learning_rate": 2.525428391718403e-05,
      "loss": 0.1684,
      "step": 3327
    },
    {
      "epoch": 0.6985726280436608,
      "grad_norm": 0.1249391958117485,
      "learning_rate": 2.5222451736603575e-05,
      "loss": 0.1583,
      "step": 3328
    },
    {
      "epoch": 0.6987825356842989,
      "grad_norm": 0.12421222776174545,
      "learning_rate": 2.5190632862132495e-05,
      "loss": 0.155,
      "step": 3329
    },
    {
      "epoch": 0.698992443324937,
      "grad_norm": 0.10891323536634445,
      "learning_rate": 2.5158827310858297e-05,
      "loss": 0.1539,
      "step": 3330
    },
    {
      "epoch": 0.6992023509655751,
      "grad_norm": 0.11101838946342468,
      "learning_rate": 2.5127035099861262e-05,
      "loss": 0.1493,
      "step": 3331
    },
    {
      "epoch": 0.6994122586062133,
      "grad_norm": 0.12890756130218506,
      "learning_rate": 2.5095256246214537e-05,
      "loss": 0.159,
      "step": 3332
    },
    {
      "epoch": 0.6996221662468514,
      "grad_norm": 0.10890180617570877,
      "learning_rate": 2.50634907669841e-05,
      "loss": 0.1376,
      "step": 3333
    },
    {
      "epoch": 0.6998320738874895,
      "grad_norm": 0.11944495141506195,
      "learning_rate": 2.503173867922872e-05,
      "loss": 0.1747,
      "step": 3334
    },
    {
      "epoch": 0.7000419815281276,
      "grad_norm": 0.10909661650657654,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.1483,
      "step": 3335
    },
    {
      "epoch": 0.7002518891687658,
      "grad_norm": 0.12965939939022064,
      "learning_rate": 2.496827474634236e-05,
      "loss": 0.1651,
      "step": 3336
    },
    {
      "epoch": 0.7004617968094039,
      "grad_norm": 0.13666345179080963,
      "learning_rate": 2.4936562935292958e-05,
      "loss": 0.1501,
      "step": 3337
    },
    {
      "epoch": 0.700671704450042,
      "grad_norm": 0.112842857837677,
      "learning_rate": 2.490486458388175e-05,
      "loss": 0.1633,
      "step": 3338
    },
    {
      "epoch": 0.7008816120906801,
      "grad_norm": 0.11788749694824219,
      "learning_rate": 2.487317970913151e-05,
      "loss": 0.1611,
      "step": 3339
    },
    {
      "epoch": 0.7010915197313182,
      "grad_norm": 0.11510729044675827,
      "learning_rate": 2.484150832805773e-05,
      "loss": 0.155,
      "step": 3340
    },
    {
      "epoch": 0.7013014273719563,
      "grad_norm": 0.1199810653924942,
      "learning_rate": 2.4809850457668653e-05,
      "loss": 0.1567,
      "step": 3341
    },
    {
      "epoch": 0.7015113350125944,
      "grad_norm": 0.13416971266269684,
      "learning_rate": 2.4778206114965284e-05,
      "loss": 0.146,
      "step": 3342
    },
    {
      "epoch": 0.7017212426532325,
      "grad_norm": 0.09663829207420349,
      "learning_rate": 2.474657531694136e-05,
      "loss": 0.1597,
      "step": 3343
    },
    {
      "epoch": 0.7019311502938707,
      "grad_norm": 0.1087060421705246,
      "learning_rate": 2.4714958080583338e-05,
      "loss": 0.1514,
      "step": 3344
    },
    {
      "epoch": 0.7021410579345088,
      "grad_norm": 0.12510518729686737,
      "learning_rate": 2.4683354422870403e-05,
      "loss": 0.1674,
      "step": 3345
    },
    {
      "epoch": 0.702350965575147,
      "grad_norm": 0.11586936563253403,
      "learning_rate": 2.465176436077441e-05,
      "loss": 0.1534,
      "step": 3346
    },
    {
      "epoch": 0.7025608732157851,
      "grad_norm": 0.15616261959075928,
      "learning_rate": 2.4620187911259996e-05,
      "loss": 0.1715,
      "step": 3347
    },
    {
      "epoch": 0.7027707808564232,
      "grad_norm": 0.11882652342319489,
      "learning_rate": 2.4588625091284428e-05,
      "loss": 0.1618,
      "step": 3348
    },
    {
      "epoch": 0.7029806884970613,
      "grad_norm": 0.11008409410715103,
      "learning_rate": 2.4557075917797624e-05,
      "loss": 0.1489,
      "step": 3349
    },
    {
      "epoch": 0.7031905961376994,
      "grad_norm": 0.11380871385335922,
      "learning_rate": 2.452554040774221e-05,
      "loss": 0.1601,
      "step": 3350
    },
    {
      "epoch": 0.7034005037783375,
      "grad_norm": 0.12558501958847046,
      "learning_rate": 2.4494018578053514e-05,
      "loss": 0.1624,
      "step": 3351
    },
    {
      "epoch": 0.7036104114189756,
      "grad_norm": 0.13026200234889984,
      "learning_rate": 2.446251044565946e-05,
      "loss": 0.1596,
      "step": 3352
    },
    {
      "epoch": 0.7038203190596137,
      "grad_norm": 0.10429004579782486,
      "learning_rate": 2.443101602748063e-05,
      "loss": 0.1502,
      "step": 3353
    },
    {
      "epoch": 0.7040302267002518,
      "grad_norm": 0.13567504286766052,
      "learning_rate": 2.439953534043024e-05,
      "loss": 0.151,
      "step": 3354
    },
    {
      "epoch": 0.7042401343408901,
      "grad_norm": 0.12509050965309143,
      "learning_rate": 2.436806840141418e-05,
      "loss": 0.155,
      "step": 3355
    },
    {
      "epoch": 0.7044500419815282,
      "grad_norm": 0.13032452762126923,
      "learning_rate": 2.4336615227330868e-05,
      "loss": 0.1724,
      "step": 3356
    },
    {
      "epoch": 0.7046599496221663,
      "grad_norm": 0.11473406106233597,
      "learning_rate": 2.4305175835071393e-05,
      "loss": 0.1438,
      "step": 3357
    },
    {
      "epoch": 0.7048698572628044,
      "grad_norm": 0.12625393271446228,
      "learning_rate": 2.427375024151941e-05,
      "loss": 0.1428,
      "step": 3358
    },
    {
      "epoch": 0.7050797649034425,
      "grad_norm": 0.12183254957199097,
      "learning_rate": 2.4242338463551212e-05,
      "loss": 0.1523,
      "step": 3359
    },
    {
      "epoch": 0.7052896725440806,
      "grad_norm": 0.11879599094390869,
      "learning_rate": 2.4210940518035628e-05,
      "loss": 0.1682,
      "step": 3360
    },
    {
      "epoch": 0.7054995801847187,
      "grad_norm": 0.1198955848813057,
      "learning_rate": 2.417955642183407e-05,
      "loss": 0.1378,
      "step": 3361
    },
    {
      "epoch": 0.7057094878253568,
      "grad_norm": 0.11626074463129044,
      "learning_rate": 2.4148186191800516e-05,
      "loss": 0.155,
      "step": 3362
    },
    {
      "epoch": 0.7059193954659949,
      "grad_norm": 0.12635177373886108,
      "learning_rate": 2.411682984478149e-05,
      "loss": 0.1619,
      "step": 3363
    },
    {
      "epoch": 0.7061293031066331,
      "grad_norm": 0.11214794963598251,
      "learning_rate": 2.4085487397616076e-05,
      "loss": 0.1556,
      "step": 3364
    },
    {
      "epoch": 0.7063392107472712,
      "grad_norm": 0.11064722388982773,
      "learning_rate": 2.4054158867135878e-05,
      "loss": 0.1426,
      "step": 3365
    },
    {
      "epoch": 0.7065491183879093,
      "grad_norm": 0.12300856411457062,
      "learning_rate": 2.4022844270165018e-05,
      "loss": 0.1535,
      "step": 3366
    },
    {
      "epoch": 0.7067590260285475,
      "grad_norm": 0.10280261933803558,
      "learning_rate": 2.3991543623520187e-05,
      "loss": 0.1531,
      "step": 3367
    },
    {
      "epoch": 0.7069689336691856,
      "grad_norm": 0.12246986478567123,
      "learning_rate": 2.3960256944010516e-05,
      "loss": 0.1625,
      "step": 3368
    },
    {
      "epoch": 0.7071788413098237,
      "grad_norm": 0.11043044179677963,
      "learning_rate": 2.39289842484377e-05,
      "loss": 0.1463,
      "step": 3369
    },
    {
      "epoch": 0.7073887489504618,
      "grad_norm": 0.11431458592414856,
      "learning_rate": 2.3897725553595833e-05,
      "loss": 0.163,
      "step": 3370
    },
    {
      "epoch": 0.7075986565910999,
      "grad_norm": 0.1473897248506546,
      "learning_rate": 2.38664808762716e-05,
      "loss": 0.1758,
      "step": 3371
    },
    {
      "epoch": 0.707808564231738,
      "grad_norm": 0.12143264710903168,
      "learning_rate": 2.3835250233244093e-05,
      "loss": 0.1776,
      "step": 3372
    },
    {
      "epoch": 0.7080184718723762,
      "grad_norm": 0.11493881046772003,
      "learning_rate": 2.380403364128487e-05,
      "loss": 0.1555,
      "step": 3373
    },
    {
      "epoch": 0.7082283795130143,
      "grad_norm": 0.12290722876787186,
      "learning_rate": 2.3772831117157967e-05,
      "loss": 0.1693,
      "step": 3374
    },
    {
      "epoch": 0.7084382871536524,
      "grad_norm": 0.09523230046033859,
      "learning_rate": 2.3741642677619824e-05,
      "loss": 0.1408,
      "step": 3375
    },
    {
      "epoch": 0.7086481947942905,
      "grad_norm": 0.1283012479543686,
      "learning_rate": 2.3710468339419406e-05,
      "loss": 0.1473,
      "step": 3376
    },
    {
      "epoch": 0.7088581024349286,
      "grad_norm": 0.10908805578947067,
      "learning_rate": 2.3679308119297988e-05,
      "loss": 0.1586,
      "step": 3377
    },
    {
      "epoch": 0.7090680100755667,
      "grad_norm": 0.12288475781679153,
      "learning_rate": 2.364816203398934e-05,
      "loss": 0.1382,
      "step": 3378
    },
    {
      "epoch": 0.7092779177162049,
      "grad_norm": 0.11164416372776031,
      "learning_rate": 2.36170301002196e-05,
      "loss": 0.1518,
      "step": 3379
    },
    {
      "epoch": 0.709487825356843,
      "grad_norm": 0.1325451135635376,
      "learning_rate": 2.358591233470736e-05,
      "loss": 0.152,
      "step": 3380
    },
    {
      "epoch": 0.7096977329974811,
      "grad_norm": 0.11789526045322418,
      "learning_rate": 2.3554808754163566e-05,
      "loss": 0.1634,
      "step": 3381
    },
    {
      "epoch": 0.7099076406381192,
      "grad_norm": 0.11117194592952728,
      "learning_rate": 2.3523719375291535e-05,
      "loss": 0.1599,
      "step": 3382
    },
    {
      "epoch": 0.7101175482787574,
      "grad_norm": 0.12263230234384537,
      "learning_rate": 2.349264421478699e-05,
      "loss": 0.146,
      "step": 3383
    },
    {
      "epoch": 0.7103274559193955,
      "grad_norm": 0.1571977734565735,
      "learning_rate": 2.346158328933799e-05,
      "loss": 0.1534,
      "step": 3384
    },
    {
      "epoch": 0.7105373635600336,
      "grad_norm": 0.1306907832622528,
      "learning_rate": 2.3430536615624966e-05,
      "loss": 0.1579,
      "step": 3385
    },
    {
      "epoch": 0.7107472712006717,
      "grad_norm": 0.12317048013210297,
      "learning_rate": 2.339950421032069e-05,
      "loss": 0.138,
      "step": 3386
    },
    {
      "epoch": 0.7109571788413098,
      "grad_norm": 0.1279805451631546,
      "learning_rate": 2.3368486090090248e-05,
      "loss": 0.158,
      "step": 3387
    },
    {
      "epoch": 0.7111670864819479,
      "grad_norm": 0.11604040861129761,
      "learning_rate": 2.3337482271591125e-05,
      "loss": 0.1573,
      "step": 3388
    },
    {
      "epoch": 0.711376994122586,
      "grad_norm": 0.10710416734218597,
      "learning_rate": 2.3306492771473064e-05,
      "loss": 0.1292,
      "step": 3389
    },
    {
      "epoch": 0.7115869017632241,
      "grad_norm": 0.11916560679674149,
      "learning_rate": 2.3275517606378145e-05,
      "loss": 0.1635,
      "step": 3390
    },
    {
      "epoch": 0.7117968094038623,
      "grad_norm": 0.1225900799036026,
      "learning_rate": 2.3244556792940685e-05,
      "loss": 0.1712,
      "step": 3391
    },
    {
      "epoch": 0.7120067170445005,
      "grad_norm": 0.14129529893398285,
      "learning_rate": 2.3213610347787406e-05,
      "loss": 0.1793,
      "step": 3392
    },
    {
      "epoch": 0.7122166246851386,
      "grad_norm": 0.11461693048477173,
      "learning_rate": 2.318267828753724e-05,
      "loss": 0.154,
      "step": 3393
    },
    {
      "epoch": 0.7124265323257767,
      "grad_norm": 0.10360202938318253,
      "learning_rate": 2.31517606288014e-05,
      "loss": 0.1342,
      "step": 3394
    },
    {
      "epoch": 0.7126364399664148,
      "grad_norm": 0.1024770587682724,
      "learning_rate": 2.3120857388183365e-05,
      "loss": 0.1675,
      "step": 3395
    },
    {
      "epoch": 0.7128463476070529,
      "grad_norm": 0.1129373162984848,
      "learning_rate": 2.3089968582278915e-05,
      "loss": 0.1643,
      "step": 3396
    },
    {
      "epoch": 0.713056255247691,
      "grad_norm": 0.13601639866828918,
      "learning_rate": 2.3059094227676047e-05,
      "loss": 0.1457,
      "step": 3397
    },
    {
      "epoch": 0.7132661628883291,
      "grad_norm": 0.109441839158535,
      "learning_rate": 2.3028234340954953e-05,
      "loss": 0.1654,
      "step": 3398
    },
    {
      "epoch": 0.7134760705289672,
      "grad_norm": 0.12384672462940216,
      "learning_rate": 2.2997388938688104e-05,
      "loss": 0.1711,
      "step": 3399
    },
    {
      "epoch": 0.7136859781696053,
      "grad_norm": 0.11933963000774384,
      "learning_rate": 2.296655803744021e-05,
      "loss": 0.1605,
      "step": 3400
    },
    {
      "epoch": 0.7138958858102435,
      "grad_norm": 0.11621814221143723,
      "learning_rate": 2.2935741653768167e-05,
      "loss": 0.1433,
      "step": 3401
    },
    {
      "epoch": 0.7141057934508817,
      "grad_norm": 0.14022648334503174,
      "learning_rate": 2.2904939804221064e-05,
      "loss": 0.1663,
      "step": 3402
    },
    {
      "epoch": 0.7143157010915198,
      "grad_norm": 0.11107058823108673,
      "learning_rate": 2.2874152505340206e-05,
      "loss": 0.1586,
      "step": 3403
    },
    {
      "epoch": 0.7145256087321579,
      "grad_norm": 0.10179081559181213,
      "learning_rate": 2.2843379773659075e-05,
      "loss": 0.162,
      "step": 3404
    },
    {
      "epoch": 0.714735516372796,
      "grad_norm": 0.12742018699645996,
      "learning_rate": 2.2812621625703323e-05,
      "loss": 0.1654,
      "step": 3405
    },
    {
      "epoch": 0.7149454240134341,
      "grad_norm": 0.12928232550621033,
      "learning_rate": 2.278187807799079e-05,
      "loss": 0.1617,
      "step": 3406
    },
    {
      "epoch": 0.7151553316540722,
      "grad_norm": 0.12862341105937958,
      "learning_rate": 2.2751149147031432e-05,
      "loss": 0.1622,
      "step": 3407
    },
    {
      "epoch": 0.7153652392947103,
      "grad_norm": 0.11547853797674179,
      "learning_rate": 2.2720434849327427e-05,
      "loss": 0.1541,
      "step": 3408
    },
    {
      "epoch": 0.7155751469353484,
      "grad_norm": 0.14156000316143036,
      "learning_rate": 2.2689735201373047e-05,
      "loss": 0.1556,
      "step": 3409
    },
    {
      "epoch": 0.7157850545759865,
      "grad_norm": 0.10587289184331894,
      "learning_rate": 2.2659050219654692e-05,
      "loss": 0.156,
      "step": 3410
    },
    {
      "epoch": 0.7159949622166247,
      "grad_norm": 0.11178750544786453,
      "learning_rate": 2.2628379920650906e-05,
      "loss": 0.1572,
      "step": 3411
    },
    {
      "epoch": 0.7162048698572628,
      "grad_norm": 0.16399018466472626,
      "learning_rate": 2.2597724320832338e-05,
      "loss": 0.1598,
      "step": 3412
    },
    {
      "epoch": 0.716414777497901,
      "grad_norm": 0.14096342027187347,
      "learning_rate": 2.256708343666174e-05,
      "loss": 0.1566,
      "step": 3413
    },
    {
      "epoch": 0.716624685138539,
      "grad_norm": 0.12449552118778229,
      "learning_rate": 2.253645728459397e-05,
      "loss": 0.1673,
      "step": 3414
    },
    {
      "epoch": 0.7168345927791772,
      "grad_norm": 0.12170739471912384,
      "learning_rate": 2.250584588107597e-05,
      "loss": 0.1691,
      "step": 3415
    },
    {
      "epoch": 0.7170445004198153,
      "grad_norm": 0.12910914421081543,
      "learning_rate": 2.2475249242546755e-05,
      "loss": 0.1628,
      "step": 3416
    },
    {
      "epoch": 0.7172544080604534,
      "grad_norm": 0.10747946798801422,
      "learning_rate": 2.2444667385437455e-05,
      "loss": 0.1692,
      "step": 3417
    },
    {
      "epoch": 0.7174643157010915,
      "grad_norm": 0.1221894845366478,
      "learning_rate": 2.2414100326171188e-05,
      "loss": 0.1652,
      "step": 3418
    },
    {
      "epoch": 0.7176742233417296,
      "grad_norm": 0.11467088758945465,
      "learning_rate": 2.238354808116316e-05,
      "loss": 0.1522,
      "step": 3419
    },
    {
      "epoch": 0.7178841309823678,
      "grad_norm": 0.12537196278572083,
      "learning_rate": 2.2353010666820645e-05,
      "loss": 0.1529,
      "step": 3420
    },
    {
      "epoch": 0.7180940386230059,
      "grad_norm": 0.11724396049976349,
      "learning_rate": 2.2322488099542938e-05,
      "loss": 0.1618,
      "step": 3421
    },
    {
      "epoch": 0.718303946263644,
      "grad_norm": 0.11824364215135574,
      "learning_rate": 2.2291980395721336e-05,
      "loss": 0.1535,
      "step": 3422
    },
    {
      "epoch": 0.7185138539042821,
      "grad_norm": 0.13034173846244812,
      "learning_rate": 2.226148757173918e-05,
      "loss": 0.1525,
      "step": 3423
    },
    {
      "epoch": 0.7187237615449202,
      "grad_norm": 0.1236094981431961,
      "learning_rate": 2.2231009643971817e-05,
      "loss": 0.1649,
      "step": 3424
    },
    {
      "epoch": 0.7189336691855583,
      "grad_norm": 0.1695818454027176,
      "learning_rate": 2.2200546628786585e-05,
      "loss": 0.1362,
      "step": 3425
    },
    {
      "epoch": 0.7191435768261965,
      "grad_norm": 0.12300975620746613,
      "learning_rate": 2.2170098542542815e-05,
      "loss": 0.1496,
      "step": 3426
    },
    {
      "epoch": 0.7193534844668346,
      "grad_norm": 0.11368977278470993,
      "learning_rate": 2.2139665401591837e-05,
      "loss": 0.1646,
      "step": 3427
    },
    {
      "epoch": 0.7195633921074727,
      "grad_norm": 0.10919873416423798,
      "learning_rate": 2.2109247222276912e-05,
      "loss": 0.1437,
      "step": 3428
    },
    {
      "epoch": 0.7197732997481109,
      "grad_norm": 0.11324533820152283,
      "learning_rate": 2.207884402093334e-05,
      "loss": 0.1471,
      "step": 3429
    },
    {
      "epoch": 0.719983207388749,
      "grad_norm": 0.11208245158195496,
      "learning_rate": 2.204845581388832e-05,
      "loss": 0.1497,
      "step": 3430
    },
    {
      "epoch": 0.7201931150293871,
      "grad_norm": 0.1129378080368042,
      "learning_rate": 2.201808261746102e-05,
      "loss": 0.1445,
      "step": 3431
    },
    {
      "epoch": 0.7204030226700252,
      "grad_norm": 0.12027370184659958,
      "learning_rate": 2.19877244479625e-05,
      "loss": 0.1725,
      "step": 3432
    },
    {
      "epoch": 0.7206129303106633,
      "grad_norm": 0.11807487159967422,
      "learning_rate": 2.195738132169583e-05,
      "loss": 0.1366,
      "step": 3433
    },
    {
      "epoch": 0.7208228379513014,
      "grad_norm": 0.11906223744153976,
      "learning_rate": 2.1927053254955955e-05,
      "loss": 0.1593,
      "step": 3434
    },
    {
      "epoch": 0.7210327455919395,
      "grad_norm": 0.1283073127269745,
      "learning_rate": 2.1896740264029737e-05,
      "loss": 0.1578,
      "step": 3435
    },
    {
      "epoch": 0.7212426532325776,
      "grad_norm": 0.1256849616765976,
      "learning_rate": 2.1866442365195927e-05,
      "loss": 0.1289,
      "step": 3436
    },
    {
      "epoch": 0.7214525608732157,
      "grad_norm": 0.11481695622205734,
      "learning_rate": 2.1836159574725218e-05,
      "loss": 0.1416,
      "step": 3437
    },
    {
      "epoch": 0.7216624685138538,
      "grad_norm": 0.10437635332345963,
      "learning_rate": 2.1805891908880172e-05,
      "loss": 0.1439,
      "step": 3438
    },
    {
      "epoch": 0.7218723761544921,
      "grad_norm": 0.13188210129737854,
      "learning_rate": 2.1775639383915168e-05,
      "loss": 0.1446,
      "step": 3439
    },
    {
      "epoch": 0.7220822837951302,
      "grad_norm": 0.10859008878469467,
      "learning_rate": 2.1745402016076515e-05,
      "loss": 0.1658,
      "step": 3440
    },
    {
      "epoch": 0.7222921914357683,
      "grad_norm": 0.12801185250282288,
      "learning_rate": 2.17151798216024e-05,
      "loss": 0.1517,
      "step": 3441
    },
    {
      "epoch": 0.7225020990764064,
      "grad_norm": 0.12271418422460556,
      "learning_rate": 2.1684972816722815e-05,
      "loss": 0.1544,
      "step": 3442
    },
    {
      "epoch": 0.7227120067170445,
      "grad_norm": 0.11636804044246674,
      "learning_rate": 2.165478101765961e-05,
      "loss": 0.1609,
      "step": 3443
    },
    {
      "epoch": 0.7229219143576826,
      "grad_norm": 0.11120007187128067,
      "learning_rate": 2.1624604440626472e-05,
      "loss": 0.1422,
      "step": 3444
    },
    {
      "epoch": 0.7231318219983207,
      "grad_norm": 0.11704123020172119,
      "learning_rate": 2.159444310182891e-05,
      "loss": 0.1685,
      "step": 3445
    },
    {
      "epoch": 0.7233417296389588,
      "grad_norm": 0.12351725250482559,
      "learning_rate": 2.156429701746426e-05,
      "loss": 0.1706,
      "step": 3446
    },
    {
      "epoch": 0.7235516372795969,
      "grad_norm": 0.13515010476112366,
      "learning_rate": 2.1534166203721644e-05,
      "loss": 0.1618,
      "step": 3447
    },
    {
      "epoch": 0.7237615449202351,
      "grad_norm": 0.13099972903728485,
      "learning_rate": 2.150405067678199e-05,
      "loss": 0.1554,
      "step": 3448
    },
    {
      "epoch": 0.7239714525608733,
      "grad_norm": 0.11865659058094025,
      "learning_rate": 2.147395045281806e-05,
      "loss": 0.1353,
      "step": 3449
    },
    {
      "epoch": 0.7241813602015114,
      "grad_norm": 0.10976529866456985,
      "learning_rate": 2.1443865547994336e-05,
      "loss": 0.1384,
      "step": 3450
    },
    {
      "epoch": 0.7243912678421495,
      "grad_norm": 0.1451970487833023,
      "learning_rate": 2.1413795978467105e-05,
      "loss": 0.1531,
      "step": 3451
    },
    {
      "epoch": 0.7246011754827876,
      "grad_norm": 0.1169748529791832,
      "learning_rate": 2.138374176038442e-05,
      "loss": 0.1566,
      "step": 3452
    },
    {
      "epoch": 0.7248110831234257,
      "grad_norm": 0.12926514446735382,
      "learning_rate": 2.1353702909886066e-05,
      "loss": 0.1495,
      "step": 3453
    },
    {
      "epoch": 0.7250209907640638,
      "grad_norm": 0.11117440462112427,
      "learning_rate": 2.13236794431036e-05,
      "loss": 0.1586,
      "step": 3454
    },
    {
      "epoch": 0.7252308984047019,
      "grad_norm": 0.12877735495567322,
      "learning_rate": 2.1293671376160308e-05,
      "loss": 0.1579,
      "step": 3455
    },
    {
      "epoch": 0.72544080604534,
      "grad_norm": 0.10876293480396271,
      "learning_rate": 2.1263678725171216e-05,
      "loss": 0.1548,
      "step": 3456
    },
    {
      "epoch": 0.7256507136859782,
      "grad_norm": 0.1247049942612648,
      "learning_rate": 2.123370150624303e-05,
      "loss": 0.1566,
      "step": 3457
    },
    {
      "epoch": 0.7258606213266163,
      "grad_norm": 0.12170092016458511,
      "learning_rate": 2.1203739735474248e-05,
      "loss": 0.1521,
      "step": 3458
    },
    {
      "epoch": 0.7260705289672544,
      "grad_norm": 0.1315077245235443,
      "learning_rate": 2.1173793428955023e-05,
      "loss": 0.1673,
      "step": 3459
    },
    {
      "epoch": 0.7262804366078925,
      "grad_norm": 0.12094591557979584,
      "learning_rate": 2.114386260276715e-05,
      "loss": 0.1611,
      "step": 3460
    },
    {
      "epoch": 0.7264903442485307,
      "grad_norm": 0.1278723180294037,
      "learning_rate": 2.1113947272984224e-05,
      "loss": 0.1564,
      "step": 3461
    },
    {
      "epoch": 0.7267002518891688,
      "grad_norm": 0.1344061940908432,
      "learning_rate": 2.1084047455671435e-05,
      "loss": 0.1566,
      "step": 3462
    },
    {
      "epoch": 0.7269101595298069,
      "grad_norm": 0.11192744970321655,
      "learning_rate": 2.1054163166885675e-05,
      "loss": 0.17,
      "step": 3463
    },
    {
      "epoch": 0.727120067170445,
      "grad_norm": 0.12571950256824493,
      "learning_rate": 2.102429442267549e-05,
      "loss": 0.1429,
      "step": 3464
    },
    {
      "epoch": 0.7273299748110831,
      "grad_norm": 0.13342975080013275,
      "learning_rate": 2.099444123908106e-05,
      "loss": 0.1457,
      "step": 3465
    },
    {
      "epoch": 0.7275398824517213,
      "grad_norm": 0.1195952519774437,
      "learning_rate": 2.0964603632134265e-05,
      "loss": 0.1709,
      "step": 3466
    },
    {
      "epoch": 0.7277497900923594,
      "grad_norm": 0.1186537891626358,
      "learning_rate": 2.093478161785855e-05,
      "loss": 0.1511,
      "step": 3467
    },
    {
      "epoch": 0.7279596977329975,
      "grad_norm": 0.09371523559093475,
      "learning_rate": 2.0904975212269024e-05,
      "loss": 0.146,
      "step": 3468
    },
    {
      "epoch": 0.7281696053736356,
      "grad_norm": 0.10283226519823074,
      "learning_rate": 2.087518443137239e-05,
      "loss": 0.1652,
      "step": 3469
    },
    {
      "epoch": 0.7283795130142737,
      "grad_norm": 0.11636362224817276,
      "learning_rate": 2.0845409291167023e-05,
      "loss": 0.1381,
      "step": 3470
    },
    {
      "epoch": 0.7285894206549118,
      "grad_norm": 0.12790684401988983,
      "learning_rate": 2.0815649807642824e-05,
      "loss": 0.1555,
      "step": 3471
    },
    {
      "epoch": 0.7287993282955499,
      "grad_norm": 0.1255161166191101,
      "learning_rate": 2.0785905996781312e-05,
      "loss": 0.1752,
      "step": 3472
    },
    {
      "epoch": 0.729009235936188,
      "grad_norm": 0.1269216686487198,
      "learning_rate": 2.075617787455561e-05,
      "loss": 0.1698,
      "step": 3473
    },
    {
      "epoch": 0.7292191435768262,
      "grad_norm": 0.11415970325469971,
      "learning_rate": 2.072646545693038e-05,
      "loss": 0.157,
      "step": 3474
    },
    {
      "epoch": 0.7294290512174643,
      "grad_norm": 0.12151431292295456,
      "learning_rate": 2.0696768759861884e-05,
      "loss": 0.1682,
      "step": 3475
    },
    {
      "epoch": 0.7296389588581025,
      "grad_norm": 0.13548152148723602,
      "learning_rate": 2.0667087799297923e-05,
      "loss": 0.1282,
      "step": 3476
    },
    {
      "epoch": 0.7298488664987406,
      "grad_norm": 0.1168590858578682,
      "learning_rate": 2.063742259117783e-05,
      "loss": 0.149,
      "step": 3477
    },
    {
      "epoch": 0.7300587741393787,
      "grad_norm": 0.1430933028459549,
      "learning_rate": 2.060777315143253e-05,
      "loss": 0.1624,
      "step": 3478
    },
    {
      "epoch": 0.7302686817800168,
      "grad_norm": 0.1568112075328827,
      "learning_rate": 2.0578139495984437e-05,
      "loss": 0.1584,
      "step": 3479
    },
    {
      "epoch": 0.7304785894206549,
      "grad_norm": 0.13265225291252136,
      "learning_rate": 2.0548521640747525e-05,
      "loss": 0.1516,
      "step": 3480
    },
    {
      "epoch": 0.730688497061293,
      "grad_norm": 0.13450318574905396,
      "learning_rate": 2.0518919601627197e-05,
      "loss": 0.1408,
      "step": 3481
    },
    {
      "epoch": 0.7308984047019311,
      "grad_norm": 0.11844109743833542,
      "learning_rate": 2.0489333394520483e-05,
      "loss": 0.1582,
      "step": 3482
    },
    {
      "epoch": 0.7311083123425692,
      "grad_norm": 0.13041557371616364,
      "learning_rate": 2.0459763035315828e-05,
      "loss": 0.1563,
      "step": 3483
    },
    {
      "epoch": 0.7313182199832073,
      "grad_norm": 0.1173725575208664,
      "learning_rate": 2.043020853989319e-05,
      "loss": 0.1603,
      "step": 3484
    },
    {
      "epoch": 0.7315281276238456,
      "grad_norm": 0.12801702320575714,
      "learning_rate": 2.0400669924124016e-05,
      "loss": 0.1765,
      "step": 3485
    },
    {
      "epoch": 0.7317380352644837,
      "grad_norm": 0.11436662822961807,
      "learning_rate": 2.0371147203871195e-05,
      "loss": 0.1773,
      "step": 3486
    },
    {
      "epoch": 0.7319479429051218,
      "grad_norm": 0.10988820344209671,
      "learning_rate": 2.0341640394989154e-05,
      "loss": 0.1451,
      "step": 3487
    },
    {
      "epoch": 0.7321578505457599,
      "grad_norm": 0.13320259749889374,
      "learning_rate": 2.031214951332367e-05,
      "loss": 0.1696,
      "step": 3488
    },
    {
      "epoch": 0.732367758186398,
      "grad_norm": 0.11017803847789764,
      "learning_rate": 2.0282674574712024e-05,
      "loss": 0.1373,
      "step": 3489
    },
    {
      "epoch": 0.7325776658270361,
      "grad_norm": 0.1179022565484047,
      "learning_rate": 2.0253215594982965e-05,
      "loss": 0.1649,
      "step": 3490
    },
    {
      "epoch": 0.7327875734676742,
      "grad_norm": 0.11736958473920822,
      "learning_rate": 2.0223772589956623e-05,
      "loss": 0.1576,
      "step": 3491
    },
    {
      "epoch": 0.7329974811083123,
      "grad_norm": 0.1266537308692932,
      "learning_rate": 2.0194345575444556e-05,
      "loss": 0.1714,
      "step": 3492
    },
    {
      "epoch": 0.7332073887489504,
      "grad_norm": 0.12094676494598389,
      "learning_rate": 2.0164934567249756e-05,
      "loss": 0.1656,
      "step": 3493
    },
    {
      "epoch": 0.7334172963895886,
      "grad_norm": 0.10691839456558228,
      "learning_rate": 2.0135539581166592e-05,
      "loss": 0.1653,
      "step": 3494
    },
    {
      "epoch": 0.7336272040302267,
      "grad_norm": 0.1287030428647995,
      "learning_rate": 2.0106160632980848e-05,
      "loss": 0.1445,
      "step": 3495
    },
    {
      "epoch": 0.7338371116708649,
      "grad_norm": 0.14120900630950928,
      "learning_rate": 2.0076797738469692e-05,
      "loss": 0.1625,
      "step": 3496
    },
    {
      "epoch": 0.734047019311503,
      "grad_norm": 0.1141500473022461,
      "learning_rate": 2.0047450913401672e-05,
      "loss": 0.1485,
      "step": 3497
    },
    {
      "epoch": 0.7342569269521411,
      "grad_norm": 0.11120381951332092,
      "learning_rate": 2.001812017353668e-05,
      "loss": 0.1557,
      "step": 3498
    },
    {
      "epoch": 0.7344668345927792,
      "grad_norm": 0.14555326104164124,
      "learning_rate": 1.998880553462604e-05,
      "loss": 0.1643,
      "step": 3499
    },
    {
      "epoch": 0.7346767422334173,
      "grad_norm": 0.14281868934631348,
      "learning_rate": 1.9959507012412343e-05,
      "loss": 0.1435,
      "step": 3500
    },
    {
      "epoch": 0.7348866498740554,
      "grad_norm": 0.11729402840137482,
      "learning_rate": 1.9930224622629583e-05,
      "loss": 0.145,
      "step": 3501
    },
    {
      "epoch": 0.7350965575146935,
      "grad_norm": 0.1215299442410469,
      "learning_rate": 1.990095838100307e-05,
      "loss": 0.1502,
      "step": 3502
    },
    {
      "epoch": 0.7353064651553316,
      "grad_norm": 0.13514041900634766,
      "learning_rate": 1.9871708303249437e-05,
      "loss": 0.153,
      "step": 3503
    },
    {
      "epoch": 0.7355163727959698,
      "grad_norm": 0.12567518651485443,
      "learning_rate": 1.9842474405076655e-05,
      "loss": 0.1486,
      "step": 3504
    },
    {
      "epoch": 0.7357262804366079,
      "grad_norm": 0.11353006213903427,
      "learning_rate": 1.981325670218398e-05,
      "loss": 0.1442,
      "step": 3505
    },
    {
      "epoch": 0.735936188077246,
      "grad_norm": 0.09958070516586304,
      "learning_rate": 1.9784055210261983e-05,
      "loss": 0.1425,
      "step": 3506
    },
    {
      "epoch": 0.7361460957178841,
      "grad_norm": 0.13688147068023682,
      "learning_rate": 1.9754869944992577e-05,
      "loss": 0.1615,
      "step": 3507
    },
    {
      "epoch": 0.7363560033585222,
      "grad_norm": 0.11260102689266205,
      "learning_rate": 1.972570092204886e-05,
      "loss": 0.1495,
      "step": 3508
    },
    {
      "epoch": 0.7365659109991604,
      "grad_norm": 0.12669405341148376,
      "learning_rate": 1.969654815709529e-05,
      "loss": 0.1451,
      "step": 3509
    },
    {
      "epoch": 0.7367758186397985,
      "grad_norm": 0.1202203780412674,
      "learning_rate": 1.9667411665787545e-05,
      "loss": 0.1581,
      "step": 3510
    },
    {
      "epoch": 0.7369857262804366,
      "grad_norm": 0.1210731789469719,
      "learning_rate": 1.9638291463772623e-05,
      "loss": 0.1559,
      "step": 3511
    },
    {
      "epoch": 0.7371956339210747,
      "grad_norm": 0.11393287032842636,
      "learning_rate": 1.9609187566688724e-05,
      "loss": 0.1584,
      "step": 3512
    },
    {
      "epoch": 0.7374055415617129,
      "grad_norm": 0.11847749352455139,
      "learning_rate": 1.9580099990165296e-05,
      "loss": 0.1697,
      "step": 3513
    },
    {
      "epoch": 0.737615449202351,
      "grad_norm": 0.11857393383979797,
      "learning_rate": 1.955102874982304e-05,
      "loss": 0.1537,
      "step": 3514
    },
    {
      "epoch": 0.7378253568429891,
      "grad_norm": 0.12403122335672379,
      "learning_rate": 1.9521973861273875e-05,
      "loss": 0.1598,
      "step": 3515
    },
    {
      "epoch": 0.7380352644836272,
      "grad_norm": 0.12605972588062286,
      "learning_rate": 1.9492935340120933e-05,
      "loss": 0.1617,
      "step": 3516
    },
    {
      "epoch": 0.7382451721242653,
      "grad_norm": 0.1054360643029213,
      "learning_rate": 1.946391320195857e-05,
      "loss": 0.156,
      "step": 3517
    },
    {
      "epoch": 0.7384550797649034,
      "grad_norm": 0.13782882690429688,
      "learning_rate": 1.943490746237232e-05,
      "loss": 0.1562,
      "step": 3518
    },
    {
      "epoch": 0.7386649874055415,
      "grad_norm": 0.12123388051986694,
      "learning_rate": 1.9405918136938962e-05,
      "loss": 0.1664,
      "step": 3519
    },
    {
      "epoch": 0.7388748950461796,
      "grad_norm": 0.11024918407201767,
      "learning_rate": 1.937694524122641e-05,
      "loss": 0.1486,
      "step": 3520
    },
    {
      "epoch": 0.7390848026868178,
      "grad_norm": 0.11437639594078064,
      "learning_rate": 1.934798879079378e-05,
      "loss": 0.1682,
      "step": 3521
    },
    {
      "epoch": 0.739294710327456,
      "grad_norm": 0.10866285115480423,
      "learning_rate": 1.9319048801191304e-05,
      "loss": 0.1601,
      "step": 3522
    },
    {
      "epoch": 0.7395046179680941,
      "grad_norm": 0.1192096620798111,
      "learning_rate": 1.929012528796046e-05,
      "loss": 0.1647,
      "step": 3523
    },
    {
      "epoch": 0.7397145256087322,
      "grad_norm": 0.1007518619298935,
      "learning_rate": 1.926121826663383e-05,
      "loss": 0.1511,
      "step": 3524
    },
    {
      "epoch": 0.7399244332493703,
      "grad_norm": 0.13826365768909454,
      "learning_rate": 1.9232327752735137e-05,
      "loss": 0.1582,
      "step": 3525
    },
    {
      "epoch": 0.7401343408900084,
      "grad_norm": 0.11610054224729538,
      "learning_rate": 1.920345376177925e-05,
      "loss": 0.1566,
      "step": 3526
    },
    {
      "epoch": 0.7403442485306465,
      "grad_norm": 0.10525956749916077,
      "learning_rate": 1.9174596309272143e-05,
      "loss": 0.161,
      "step": 3527
    },
    {
      "epoch": 0.7405541561712846,
      "grad_norm": 0.1056981310248375,
      "learning_rate": 1.914575541071098e-05,
      "loss": 0.1545,
      "step": 3528
    },
    {
      "epoch": 0.7407640638119227,
      "grad_norm": 0.12345774471759796,
      "learning_rate": 1.9116931081583928e-05,
      "loss": 0.1598,
      "step": 3529
    },
    {
      "epoch": 0.7409739714525608,
      "grad_norm": 0.10237154364585876,
      "learning_rate": 1.9088123337370313e-05,
      "loss": 0.1456,
      "step": 3530
    },
    {
      "epoch": 0.7411838790931989,
      "grad_norm": 0.1328197717666626,
      "learning_rate": 1.905933219354058e-05,
      "loss": 0.1298,
      "step": 3531
    },
    {
      "epoch": 0.7413937867338372,
      "grad_norm": 0.10656782239675522,
      "learning_rate": 1.9030557665556226e-05,
      "loss": 0.1612,
      "step": 3532
    },
    {
      "epoch": 0.7416036943744753,
      "grad_norm": 0.1428254395723343,
      "learning_rate": 1.900179976886982e-05,
      "loss": 0.1552,
      "step": 3533
    },
    {
      "epoch": 0.7418136020151134,
      "grad_norm": 0.12702079117298126,
      "learning_rate": 1.8973058518925013e-05,
      "loss": 0.1551,
      "step": 3534
    },
    {
      "epoch": 0.7420235096557515,
      "grad_norm": 0.1408890336751938,
      "learning_rate": 1.894433393115651e-05,
      "loss": 0.143,
      "step": 3535
    },
    {
      "epoch": 0.7422334172963896,
      "grad_norm": 0.11404120922088623,
      "learning_rate": 1.8915626020990073e-05,
      "loss": 0.1605,
      "step": 3536
    },
    {
      "epoch": 0.7424433249370277,
      "grad_norm": 0.13507436215877533,
      "learning_rate": 1.8886934803842504e-05,
      "loss": 0.1709,
      "step": 3537
    },
    {
      "epoch": 0.7426532325776658,
      "grad_norm": 0.11132592707872391,
      "learning_rate": 1.8858260295121643e-05,
      "loss": 0.1439,
      "step": 3538
    },
    {
      "epoch": 0.7428631402183039,
      "grad_norm": 0.10579460859298706,
      "learning_rate": 1.882960251022634e-05,
      "loss": 0.1681,
      "step": 3539
    },
    {
      "epoch": 0.743073047858942,
      "grad_norm": 0.12509001791477203,
      "learning_rate": 1.8800961464546518e-05,
      "loss": 0.1598,
      "step": 3540
    },
    {
      "epoch": 0.7432829554995802,
      "grad_norm": 0.1258598417043686,
      "learning_rate": 1.8772337173463045e-05,
      "loss": 0.1458,
      "step": 3541
    },
    {
      "epoch": 0.7434928631402183,
      "grad_norm": 0.11713558435440063,
      "learning_rate": 1.874372965234783e-05,
      "loss": 0.1467,
      "step": 3542
    },
    {
      "epoch": 0.7437027707808564,
      "grad_norm": 0.15114378929138184,
      "learning_rate": 1.8715138916563758e-05,
      "loss": 0.1623,
      "step": 3543
    },
    {
      "epoch": 0.7439126784214946,
      "grad_norm": 0.12825213372707367,
      "learning_rate": 1.8686564981464706e-05,
      "loss": 0.1508,
      "step": 3544
    },
    {
      "epoch": 0.7441225860621327,
      "grad_norm": 0.11679855734109879,
      "learning_rate": 1.865800786239554e-05,
      "loss": 0.1595,
      "step": 3545
    },
    {
      "epoch": 0.7443324937027708,
      "grad_norm": 0.12408013641834259,
      "learning_rate": 1.8629467574692073e-05,
      "loss": 0.1753,
      "step": 3546
    },
    {
      "epoch": 0.7445424013434089,
      "grad_norm": 0.10703358054161072,
      "learning_rate": 1.860094413368108e-05,
      "loss": 0.153,
      "step": 3547
    },
    {
      "epoch": 0.744752308984047,
      "grad_norm": 0.10585933923721313,
      "learning_rate": 1.857243755468034e-05,
      "loss": 0.1504,
      "step": 3548
    },
    {
      "epoch": 0.7449622166246851,
      "grad_norm": 0.12033756822347641,
      "learning_rate": 1.8543947852998528e-05,
      "loss": 0.1558,
      "step": 3549
    },
    {
      "epoch": 0.7451721242653233,
      "grad_norm": 0.1374211460351944,
      "learning_rate": 1.8515475043935237e-05,
      "loss": 0.163,
      "step": 3550
    },
    {
      "epoch": 0.7453820319059614,
      "grad_norm": 0.11834236234426498,
      "learning_rate": 1.848701914278101e-05,
      "loss": 0.1593,
      "step": 3551
    },
    {
      "epoch": 0.7455919395465995,
      "grad_norm": 0.11231713742017746,
      "learning_rate": 1.8458580164817357e-05,
      "loss": 0.1468,
      "step": 3552
    },
    {
      "epoch": 0.7458018471872376,
      "grad_norm": 0.12200900167226791,
      "learning_rate": 1.843015812531663e-05,
      "loss": 0.1597,
      "step": 3553
    },
    {
      "epoch": 0.7460117548278757,
      "grad_norm": 0.10779359191656113,
      "learning_rate": 1.840175303954213e-05,
      "loss": 0.1655,
      "step": 3554
    },
    {
      "epoch": 0.7462216624685138,
      "grad_norm": 0.12323888391256332,
      "learning_rate": 1.8373364922748027e-05,
      "loss": 0.1652,
      "step": 3555
    },
    {
      "epoch": 0.746431570109152,
      "grad_norm": 0.12674176692962646,
      "learning_rate": 1.8344993790179387e-05,
      "loss": 0.1576,
      "step": 3556
    },
    {
      "epoch": 0.7466414777497901,
      "grad_norm": 0.12774962186813354,
      "learning_rate": 1.8316639657072155e-05,
      "loss": 0.1443,
      "step": 3557
    },
    {
      "epoch": 0.7468513853904282,
      "grad_norm": 0.1553429812192917,
      "learning_rate": 1.828830253865316e-05,
      "loss": 0.1389,
      "step": 3558
    },
    {
      "epoch": 0.7470612930310663,
      "grad_norm": 0.15105120837688446,
      "learning_rate": 1.825998245014005e-05,
      "loss": 0.1585,
      "step": 3559
    },
    {
      "epoch": 0.7472712006717045,
      "grad_norm": 0.11354130506515503,
      "learning_rate": 1.8231679406741402e-05,
      "loss": 0.1407,
      "step": 3560
    },
    {
      "epoch": 0.7474811083123426,
      "grad_norm": 0.12431042641401291,
      "learning_rate": 1.820339342365658e-05,
      "loss": 0.1613,
      "step": 3561
    },
    {
      "epoch": 0.7476910159529807,
      "grad_norm": 0.11633086949586868,
      "learning_rate": 1.8175124516075796e-05,
      "loss": 0.1369,
      "step": 3562
    },
    {
      "epoch": 0.7479009235936188,
      "grad_norm": 0.11281375586986542,
      "learning_rate": 1.8146872699180108e-05,
      "loss": 0.1599,
      "step": 3563
    },
    {
      "epoch": 0.7481108312342569,
      "grad_norm": 0.14105822145938873,
      "learning_rate": 1.811863798814138e-05,
      "loss": 0.1585,
      "step": 3564
    },
    {
      "epoch": 0.748320738874895,
      "grad_norm": 0.1503167301416397,
      "learning_rate": 1.8090420398122304e-05,
      "loss": 0.1506,
      "step": 3565
    },
    {
      "epoch": 0.7485306465155331,
      "grad_norm": 0.12038581073284149,
      "learning_rate": 1.8062219944276365e-05,
      "loss": 0.1624,
      "step": 3566
    },
    {
      "epoch": 0.7487405541561712,
      "grad_norm": 0.12901602685451508,
      "learning_rate": 1.8034036641747847e-05,
      "loss": 0.1505,
      "step": 3567
    },
    {
      "epoch": 0.7489504617968094,
      "grad_norm": 0.12743479013442993,
      "learning_rate": 1.8005870505671823e-05,
      "loss": 0.1685,
      "step": 3568
    },
    {
      "epoch": 0.7491603694374476,
      "grad_norm": 0.13382984697818756,
      "learning_rate": 1.797772155117417e-05,
      "loss": 0.1702,
      "step": 3569
    },
    {
      "epoch": 0.7493702770780857,
      "grad_norm": 0.10113348811864853,
      "learning_rate": 1.7949589793371535e-05,
      "loss": 0.1412,
      "step": 3570
    },
    {
      "epoch": 0.7495801847187238,
      "grad_norm": 0.13199184834957123,
      "learning_rate": 1.792147524737125e-05,
      "loss": 0.1587,
      "step": 3571
    },
    {
      "epoch": 0.7497900923593619,
      "grad_norm": 0.11780549585819244,
      "learning_rate": 1.7893377928271526e-05,
      "loss": 0.1481,
      "step": 3572
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.12867914140224457,
      "learning_rate": 1.786529785116125e-05,
      "loss": 0.1594,
      "step": 3573
    },
    {
      "epoch": 0.7502099076406381,
      "grad_norm": 0.11833992600440979,
      "learning_rate": 1.7837235031120063e-05,
      "loss": 0.148,
      "step": 3574
    },
    {
      "epoch": 0.7504198152812762,
      "grad_norm": 0.12939533591270447,
      "learning_rate": 1.7809189483218336e-05,
      "loss": 0.1623,
      "step": 3575
    },
    {
      "epoch": 0.7506297229219143,
      "grad_norm": 0.11650772392749786,
      "learning_rate": 1.7781161222517163e-05,
      "loss": 0.1444,
      "step": 3576
    },
    {
      "epoch": 0.7508396305625524,
      "grad_norm": 0.13511453568935394,
      "learning_rate": 1.7753150264068398e-05,
      "loss": 0.1568,
      "step": 3577
    },
    {
      "epoch": 0.7510495382031906,
      "grad_norm": 0.13567636907100677,
      "learning_rate": 1.7725156622914518e-05,
      "loss": 0.157,
      "step": 3578
    },
    {
      "epoch": 0.7512594458438288,
      "grad_norm": 0.1252739429473877,
      "learning_rate": 1.7697180314088767e-05,
      "loss": 0.1641,
      "step": 3579
    },
    {
      "epoch": 0.7514693534844669,
      "grad_norm": 0.11738979071378708,
      "learning_rate": 1.766922135261505e-05,
      "loss": 0.1527,
      "step": 3580
    },
    {
      "epoch": 0.751679261125105,
      "grad_norm": 0.13429807126522064,
      "learning_rate": 1.7641279753507993e-05,
      "loss": 0.1539,
      "step": 3581
    },
    {
      "epoch": 0.7518891687657431,
      "grad_norm": 0.11173707246780396,
      "learning_rate": 1.761335553177286e-05,
      "loss": 0.1585,
      "step": 3582
    },
    {
      "epoch": 0.7520990764063812,
      "grad_norm": 0.12978778779506683,
      "learning_rate": 1.75854487024056e-05,
      "loss": 0.1543,
      "step": 3583
    },
    {
      "epoch": 0.7523089840470193,
      "grad_norm": 0.11753234267234802,
      "learning_rate": 1.7557559280392806e-05,
      "loss": 0.1657,
      "step": 3584
    },
    {
      "epoch": 0.7525188916876574,
      "grad_norm": 0.11019283533096313,
      "learning_rate": 1.7529687280711747e-05,
      "loss": 0.16,
      "step": 3585
    },
    {
      "epoch": 0.7527287993282955,
      "grad_norm": 0.120496965944767,
      "learning_rate": 1.750183271833032e-05,
      "loss": 0.1477,
      "step": 3586
    },
    {
      "epoch": 0.7529387069689337,
      "grad_norm": 0.1287451982498169,
      "learning_rate": 1.7473995608207056e-05,
      "loss": 0.1575,
      "step": 3587
    },
    {
      "epoch": 0.7531486146095718,
      "grad_norm": 0.11368725448846817,
      "learning_rate": 1.744617596529111e-05,
      "loss": 0.1489,
      "step": 3588
    },
    {
      "epoch": 0.7533585222502099,
      "grad_norm": 0.12659108638763428,
      "learning_rate": 1.7418373804522293e-05,
      "loss": 0.1504,
      "step": 3589
    },
    {
      "epoch": 0.753568429890848,
      "grad_norm": 0.1425846368074417,
      "learning_rate": 1.7390589140830987e-05,
      "loss": 0.1535,
      "step": 3590
    },
    {
      "epoch": 0.7537783375314862,
      "grad_norm": 0.11686164885759354,
      "learning_rate": 1.7362821989138204e-05,
      "loss": 0.1522,
      "step": 3591
    },
    {
      "epoch": 0.7539882451721243,
      "grad_norm": 0.12282045185565948,
      "learning_rate": 1.7335072364355498e-05,
      "loss": 0.1387,
      "step": 3592
    },
    {
      "epoch": 0.7541981528127624,
      "grad_norm": 0.11782212555408478,
      "learning_rate": 1.7307340281385093e-05,
      "loss": 0.1715,
      "step": 3593
    },
    {
      "epoch": 0.7544080604534005,
      "grad_norm": 0.10659587383270264,
      "learning_rate": 1.7279625755119733e-05,
      "loss": 0.1621,
      "step": 3594
    },
    {
      "epoch": 0.7546179680940386,
      "grad_norm": 0.13287796080112457,
      "learning_rate": 1.7251928800442757e-05,
      "loss": 0.1571,
      "step": 3595
    },
    {
      "epoch": 0.7548278757346767,
      "grad_norm": 0.10035673528909683,
      "learning_rate": 1.7224249432228063e-05,
      "loss": 0.146,
      "step": 3596
    },
    {
      "epoch": 0.7550377833753149,
      "grad_norm": 0.12282916158437729,
      "learning_rate": 1.719658766534008e-05,
      "loss": 0.1582,
      "step": 3597
    },
    {
      "epoch": 0.755247691015953,
      "grad_norm": 0.10245541483163834,
      "learning_rate": 1.7168943514633863e-05,
      "loss": 0.1493,
      "step": 3598
    },
    {
      "epoch": 0.7554575986565911,
      "grad_norm": 0.14495441317558289,
      "learning_rate": 1.7141316994954904e-05,
      "loss": 0.158,
      "step": 3599
    },
    {
      "epoch": 0.7556675062972292,
      "grad_norm": 0.11242242157459259,
      "learning_rate": 1.7113708121139278e-05,
      "loss": 0.1492,
      "step": 3600
    },
    {
      "epoch": 0.7558774139378673,
      "grad_norm": 0.12160933762788773,
      "learning_rate": 1.7086116908013606e-05,
      "loss": 0.1626,
      "step": 3601
    },
    {
      "epoch": 0.7560873215785054,
      "grad_norm": 0.12219791114330292,
      "learning_rate": 1.7058543370394986e-05,
      "loss": 0.1785,
      "step": 3602
    },
    {
      "epoch": 0.7562972292191436,
      "grad_norm": 0.13026191294193268,
      "learning_rate": 1.7030987523091042e-05,
      "loss": 0.1561,
      "step": 3603
    },
    {
      "epoch": 0.7565071368597817,
      "grad_norm": 0.10388700664043427,
      "learning_rate": 1.700344938089989e-05,
      "loss": 0.158,
      "step": 3604
    },
    {
      "epoch": 0.7567170445004198,
      "grad_norm": 0.10586515814065933,
      "learning_rate": 1.697592895861014e-05,
      "loss": 0.1507,
      "step": 3605
    },
    {
      "epoch": 0.756926952141058,
      "grad_norm": 0.11730484664440155,
      "learning_rate": 1.694842627100089e-05,
      "loss": 0.1702,
      "step": 3606
    },
    {
      "epoch": 0.7571368597816961,
      "grad_norm": 0.1099337711930275,
      "learning_rate": 1.6920941332841706e-05,
      "loss": 0.1574,
      "step": 3607
    },
    {
      "epoch": 0.7573467674223342,
      "grad_norm": 0.11451583355665207,
      "learning_rate": 1.6893474158892636e-05,
      "loss": 0.1607,
      "step": 3608
    },
    {
      "epoch": 0.7575566750629723,
      "grad_norm": 0.10995732247829437,
      "learning_rate": 1.6866024763904158e-05,
      "loss": 0.1393,
      "step": 3609
    },
    {
      "epoch": 0.7577665827036104,
      "grad_norm": 0.1097119078040123,
      "learning_rate": 1.683859316261726e-05,
      "loss": 0.1537,
      "step": 3610
    },
    {
      "epoch": 0.7579764903442485,
      "grad_norm": 0.09406376630067825,
      "learning_rate": 1.6811179369763334e-05,
      "loss": 0.1495,
      "step": 3611
    },
    {
      "epoch": 0.7581863979848866,
      "grad_norm": 0.12109964340925217,
      "learning_rate": 1.6783783400064168e-05,
      "loss": 0.1702,
      "step": 3612
    },
    {
      "epoch": 0.7583963056255247,
      "grad_norm": 0.1099427118897438,
      "learning_rate": 1.6756405268232073e-05,
      "loss": 0.1639,
      "step": 3613
    },
    {
      "epoch": 0.7586062132661628,
      "grad_norm": 0.11676996946334839,
      "learning_rate": 1.672904498896971e-05,
      "loss": 0.1662,
      "step": 3614
    },
    {
      "epoch": 0.7588161209068011,
      "grad_norm": 0.1413038820028305,
      "learning_rate": 1.670170257697018e-05,
      "loss": 0.1614,
      "step": 3615
    },
    {
      "epoch": 0.7590260285474392,
      "grad_norm": 0.11984378844499588,
      "learning_rate": 1.6674378046916983e-05,
      "loss": 0.1526,
      "step": 3616
    },
    {
      "epoch": 0.7592359361880773,
      "grad_norm": 0.13852879405021667,
      "learning_rate": 1.6647071413483988e-05,
      "loss": 0.1705,
      "step": 3617
    },
    {
      "epoch": 0.7594458438287154,
      "grad_norm": 0.12046951800584793,
      "learning_rate": 1.6619782691335546e-05,
      "loss": 0.1518,
      "step": 3618
    },
    {
      "epoch": 0.7596557514693535,
      "grad_norm": 0.1124371588230133,
      "learning_rate": 1.6592511895126263e-05,
      "loss": 0.1517,
      "step": 3619
    },
    {
      "epoch": 0.7598656591099916,
      "grad_norm": 0.12519319355487823,
      "learning_rate": 1.6565259039501195e-05,
      "loss": 0.1564,
      "step": 3620
    },
    {
      "epoch": 0.7600755667506297,
      "grad_norm": 0.11457381397485733,
      "learning_rate": 1.6538024139095748e-05,
      "loss": 0.1466,
      "step": 3621
    },
    {
      "epoch": 0.7602854743912678,
      "grad_norm": 0.12174105644226074,
      "learning_rate": 1.65108072085357e-05,
      "loss": 0.1555,
      "step": 3622
    },
    {
      "epoch": 0.7604953820319059,
      "grad_norm": 0.11587201803922653,
      "learning_rate": 1.6483608262437154e-05,
      "loss": 0.1525,
      "step": 3623
    },
    {
      "epoch": 0.760705289672544,
      "grad_norm": 0.10113554447889328,
      "learning_rate": 1.645642731540657e-05,
      "loss": 0.1589,
      "step": 3624
    },
    {
      "epoch": 0.7609151973131822,
      "grad_norm": 0.12906788289546967,
      "learning_rate": 1.6429264382040737e-05,
      "loss": 0.1556,
      "step": 3625
    },
    {
      "epoch": 0.7611251049538204,
      "grad_norm": 0.1424580067396164,
      "learning_rate": 1.6402119476926765e-05,
      "loss": 0.1488,
      "step": 3626
    },
    {
      "epoch": 0.7613350125944585,
      "grad_norm": 0.12937603890895844,
      "learning_rate": 1.637499261464209e-05,
      "loss": 0.15,
      "step": 3627
    },
    {
      "epoch": 0.7615449202350966,
      "grad_norm": 0.1439439356327057,
      "learning_rate": 1.6347883809754455e-05,
      "loss": 0.1582,
      "step": 3628
    },
    {
      "epoch": 0.7617548278757347,
      "grad_norm": 0.15245719254016876,
      "learning_rate": 1.6320793076821893e-05,
      "loss": 0.1646,
      "step": 3629
    },
    {
      "epoch": 0.7619647355163728,
      "grad_norm": 0.11606140434741974,
      "learning_rate": 1.6293720430392783e-05,
      "loss": 0.1685,
      "step": 3630
    },
    {
      "epoch": 0.7621746431570109,
      "grad_norm": 0.1192961186170578,
      "learning_rate": 1.6266665885005738e-05,
      "loss": 0.1885,
      "step": 3631
    },
    {
      "epoch": 0.762384550797649,
      "grad_norm": 0.15543906390666962,
      "learning_rate": 1.6239629455189677e-05,
      "loss": 0.1428,
      "step": 3632
    },
    {
      "epoch": 0.7625944584382871,
      "grad_norm": 0.1320749819278717,
      "learning_rate": 1.621261115546373e-05,
      "loss": 0.1582,
      "step": 3633
    },
    {
      "epoch": 0.7628043660789253,
      "grad_norm": 0.1201024129986763,
      "learning_rate": 1.6185611000337397e-05,
      "loss": 0.1593,
      "step": 3634
    },
    {
      "epoch": 0.7630142737195634,
      "grad_norm": 0.13097235560417175,
      "learning_rate": 1.6158629004310362e-05,
      "loss": 0.1602,
      "step": 3635
    },
    {
      "epoch": 0.7632241813602015,
      "grad_norm": 0.14049144089221954,
      "learning_rate": 1.6131665181872567e-05,
      "loss": 0.1642,
      "step": 3636
    },
    {
      "epoch": 0.7634340890008396,
      "grad_norm": 0.11801209300756454,
      "learning_rate": 1.610471954750421e-05,
      "loss": 0.1708,
      "step": 3637
    },
    {
      "epoch": 0.7636439966414778,
      "grad_norm": 0.11156057566404343,
      "learning_rate": 1.6077792115675683e-05,
      "loss": 0.1614,
      "step": 3638
    },
    {
      "epoch": 0.7638539042821159,
      "grad_norm": 0.12610188126564026,
      "learning_rate": 1.605088290084769e-05,
      "loss": 0.1494,
      "step": 3639
    },
    {
      "epoch": 0.764063811922754,
      "grad_norm": 0.11332854628562927,
      "learning_rate": 1.602399191747104e-05,
      "loss": 0.1535,
      "step": 3640
    },
    {
      "epoch": 0.7642737195633921,
      "grad_norm": 0.12335386872291565,
      "learning_rate": 1.599711917998682e-05,
      "loss": 0.1569,
      "step": 3641
    },
    {
      "epoch": 0.7644836272040302,
      "grad_norm": 0.13249602913856506,
      "learning_rate": 1.5970264702826317e-05,
      "loss": 0.164,
      "step": 3642
    },
    {
      "epoch": 0.7646935348446684,
      "grad_norm": 0.10993882268667221,
      "learning_rate": 1.5943428500410986e-05,
      "loss": 0.1546,
      "step": 3643
    },
    {
      "epoch": 0.7649034424853065,
      "grad_norm": 0.11593450605869293,
      "learning_rate": 1.5916610587152485e-05,
      "loss": 0.1827,
      "step": 3644
    },
    {
      "epoch": 0.7651133501259446,
      "grad_norm": 0.1301153302192688,
      "learning_rate": 1.5889810977452652e-05,
      "loss": 0.1627,
      "step": 3645
    },
    {
      "epoch": 0.7653232577665827,
      "grad_norm": 0.13022619485855103,
      "learning_rate": 1.5863029685703477e-05,
      "loss": 0.158,
      "step": 3646
    },
    {
      "epoch": 0.7655331654072208,
      "grad_norm": 0.13365799188613892,
      "learning_rate": 1.583626672628714e-05,
      "loss": 0.1686,
      "step": 3647
    },
    {
      "epoch": 0.7657430730478589,
      "grad_norm": 0.12435203790664673,
      "learning_rate": 1.5809522113575946e-05,
      "loss": 0.1338,
      "step": 3648
    },
    {
      "epoch": 0.765952980688497,
      "grad_norm": 0.12281453609466553,
      "learning_rate": 1.5782795861932376e-05,
      "loss": 0.1623,
      "step": 3649
    },
    {
      "epoch": 0.7661628883291351,
      "grad_norm": 0.14482922852039337,
      "learning_rate": 1.5756087985709013e-05,
      "loss": 0.1542,
      "step": 3650
    },
    {
      "epoch": 0.7663727959697733,
      "grad_norm": 0.1194080114364624,
      "learning_rate": 1.5729398499248626e-05,
      "loss": 0.1378,
      "step": 3651
    },
    {
      "epoch": 0.7665827036104114,
      "grad_norm": 0.12180230766534805,
      "learning_rate": 1.570272741688408e-05,
      "loss": 0.1674,
      "step": 3652
    },
    {
      "epoch": 0.7667926112510496,
      "grad_norm": 0.1125471219420433,
      "learning_rate": 1.5676074752938337e-05,
      "loss": 0.1681,
      "step": 3653
    },
    {
      "epoch": 0.7670025188916877,
      "grad_norm": 0.13706545531749725,
      "learning_rate": 1.564944052172449e-05,
      "loss": 0.1549,
      "step": 3654
    },
    {
      "epoch": 0.7672124265323258,
      "grad_norm": 0.10688710957765579,
      "learning_rate": 1.5622824737545734e-05,
      "loss": 0.1571,
      "step": 3655
    },
    {
      "epoch": 0.7674223341729639,
      "grad_norm": 0.12797637283802032,
      "learning_rate": 1.5596227414695353e-05,
      "loss": 0.1723,
      "step": 3656
    },
    {
      "epoch": 0.767632241813602,
      "grad_norm": 0.12048980593681335,
      "learning_rate": 1.5569648567456718e-05,
      "loss": 0.1547,
      "step": 3657
    },
    {
      "epoch": 0.7678421494542401,
      "grad_norm": 0.1186976432800293,
      "learning_rate": 1.5543088210103257e-05,
      "loss": 0.1509,
      "step": 3658
    },
    {
      "epoch": 0.7680520570948782,
      "grad_norm": 0.1264502853155136,
      "learning_rate": 1.5516546356898527e-05,
      "loss": 0.1488,
      "step": 3659
    },
    {
      "epoch": 0.7682619647355163,
      "grad_norm": 0.11536482721567154,
      "learning_rate": 1.5490023022096097e-05,
      "loss": 0.1657,
      "step": 3660
    },
    {
      "epoch": 0.7684718723761544,
      "grad_norm": 0.12747111916542053,
      "learning_rate": 1.5463518219939584e-05,
      "loss": 0.1364,
      "step": 3661
    },
    {
      "epoch": 0.7686817800167927,
      "grad_norm": 0.12515799701213837,
      "learning_rate": 1.5437031964662662e-05,
      "loss": 0.1437,
      "step": 3662
    },
    {
      "epoch": 0.7688916876574308,
      "grad_norm": 0.10816127061843872,
      "learning_rate": 1.5410564270489102e-05,
      "loss": 0.1616,
      "step": 3663
    },
    {
      "epoch": 0.7691015952980689,
      "grad_norm": 0.12977659702301025,
      "learning_rate": 1.5384115151632628e-05,
      "loss": 0.1623,
      "step": 3664
    },
    {
      "epoch": 0.769311502938707,
      "grad_norm": 0.1383851319551468,
      "learning_rate": 1.535768462229703e-05,
      "loss": 0.1327,
      "step": 3665
    },
    {
      "epoch": 0.7695214105793451,
      "grad_norm": 0.11733153462409973,
      "learning_rate": 1.533127269667609e-05,
      "loss": 0.156,
      "step": 3666
    },
    {
      "epoch": 0.7697313182199832,
      "grad_norm": 0.12266400456428528,
      "learning_rate": 1.5304879388953663e-05,
      "loss": 0.1505,
      "step": 3667
    },
    {
      "epoch": 0.7699412258606213,
      "grad_norm": 0.13448049128055573,
      "learning_rate": 1.5278504713303505e-05,
      "loss": 0.1525,
      "step": 3668
    },
    {
      "epoch": 0.7701511335012594,
      "grad_norm": 0.12451707571744919,
      "learning_rate": 1.5252148683889439e-05,
      "loss": 0.1507,
      "step": 3669
    },
    {
      "epoch": 0.7703610411418975,
      "grad_norm": 0.12748034298419952,
      "learning_rate": 1.5225811314865241e-05,
      "loss": 0.1613,
      "step": 3670
    },
    {
      "epoch": 0.7705709487825357,
      "grad_norm": 0.11085929721593857,
      "learning_rate": 1.5199492620374705e-05,
      "loss": 0.1532,
      "step": 3671
    },
    {
      "epoch": 0.7707808564231738,
      "grad_norm": 0.11221485584974289,
      "learning_rate": 1.5173192614551562e-05,
      "loss": 0.1467,
      "step": 3672
    },
    {
      "epoch": 0.770990764063812,
      "grad_norm": 0.12698894739151,
      "learning_rate": 1.5146911311519518e-05,
      "loss": 0.1642,
      "step": 3673
    },
    {
      "epoch": 0.7712006717044501,
      "grad_norm": 0.12469375878572464,
      "learning_rate": 1.5120648725392223e-05,
      "loss": 0.1677,
      "step": 3674
    },
    {
      "epoch": 0.7714105793450882,
      "grad_norm": 0.13764609396457672,
      "learning_rate": 1.5094404870273293e-05,
      "loss": 0.166,
      "step": 3675
    },
    {
      "epoch": 0.7716204869857263,
      "grad_norm": 0.16062691807746887,
      "learning_rate": 1.5068179760256273e-05,
      "loss": 0.168,
      "step": 3676
    },
    {
      "epoch": 0.7718303946263644,
      "grad_norm": 0.11727982014417648,
      "learning_rate": 1.5041973409424653e-05,
      "loss": 0.1555,
      "step": 3677
    },
    {
      "epoch": 0.7720403022670025,
      "grad_norm": 0.11587007343769073,
      "learning_rate": 1.5015785831851837e-05,
      "loss": 0.1606,
      "step": 3678
    },
    {
      "epoch": 0.7722502099076406,
      "grad_norm": 0.10293179005384445,
      "learning_rate": 1.498961704160114e-05,
      "loss": 0.1486,
      "step": 3679
    },
    {
      "epoch": 0.7724601175482787,
      "grad_norm": 0.11385567486286163,
      "learning_rate": 1.4963467052725827e-05,
      "loss": 0.1658,
      "step": 3680
    },
    {
      "epoch": 0.7726700251889169,
      "grad_norm": 0.11889761686325073,
      "learning_rate": 1.4937335879269038e-05,
      "loss": 0.1651,
      "step": 3681
    },
    {
      "epoch": 0.772879932829555,
      "grad_norm": 0.10438660532236099,
      "learning_rate": 1.4911223535263774e-05,
      "loss": 0.1619,
      "step": 3682
    },
    {
      "epoch": 0.7730898404701931,
      "grad_norm": 0.11689276248216629,
      "learning_rate": 1.4885130034732992e-05,
      "loss": 0.1635,
      "step": 3683
    },
    {
      "epoch": 0.7732997481108312,
      "grad_norm": 0.10763343423604965,
      "learning_rate": 1.4859055391689497e-05,
      "loss": 0.1542,
      "step": 3684
    },
    {
      "epoch": 0.7735096557514693,
      "grad_norm": 0.11056458950042725,
      "learning_rate": 1.4832999620135957e-05,
      "loss": 0.1441,
      "step": 3685
    },
    {
      "epoch": 0.7737195633921075,
      "grad_norm": 0.13629314303398132,
      "learning_rate": 1.4806962734064917e-05,
      "loss": 0.1298,
      "step": 3686
    },
    {
      "epoch": 0.7739294710327456,
      "grad_norm": 0.12725608050823212,
      "learning_rate": 1.4780944747458774e-05,
      "loss": 0.1537,
      "step": 3687
    },
    {
      "epoch": 0.7741393786733837,
      "grad_norm": 0.1429300606250763,
      "learning_rate": 1.4754945674289817e-05,
      "loss": 0.1577,
      "step": 3688
    },
    {
      "epoch": 0.7743492863140218,
      "grad_norm": 0.12305251508951187,
      "learning_rate": 1.47289655285201e-05,
      "loss": 0.1615,
      "step": 3689
    },
    {
      "epoch": 0.77455919395466,
      "grad_norm": 0.13106046617031097,
      "learning_rate": 1.470300432410157e-05,
      "loss": 0.1672,
      "step": 3690
    },
    {
      "epoch": 0.7747691015952981,
      "grad_norm": 0.14498285949230194,
      "learning_rate": 1.4677062074975972e-05,
      "loss": 0.1691,
      "step": 3691
    },
    {
      "epoch": 0.7749790092359362,
      "grad_norm": 0.11447465419769287,
      "learning_rate": 1.465113879507492e-05,
      "loss": 0.1575,
      "step": 3692
    },
    {
      "epoch": 0.7751889168765743,
      "grad_norm": 0.14588482677936554,
      "learning_rate": 1.4625234498319784e-05,
      "loss": 0.1709,
      "step": 3693
    },
    {
      "epoch": 0.7753988245172124,
      "grad_norm": 0.12316252291202545,
      "learning_rate": 1.459934919862177e-05,
      "loss": 0.1603,
      "step": 3694
    },
    {
      "epoch": 0.7756087321578505,
      "grad_norm": 0.12461654841899872,
      "learning_rate": 1.457348290988187e-05,
      "loss": 0.1408,
      "step": 3695
    },
    {
      "epoch": 0.7758186397984886,
      "grad_norm": 0.10895299166440964,
      "learning_rate": 1.4547635645990875e-05,
      "loss": 0.1746,
      "step": 3696
    },
    {
      "epoch": 0.7760285474391267,
      "grad_norm": 0.11470239609479904,
      "learning_rate": 1.452180742082936e-05,
      "loss": 0.1567,
      "step": 3697
    },
    {
      "epoch": 0.7762384550797649,
      "grad_norm": 0.1124269962310791,
      "learning_rate": 1.4495998248267662e-05,
      "loss": 0.1804,
      "step": 3698
    },
    {
      "epoch": 0.7764483627204031,
      "grad_norm": 0.11764495819807053,
      "learning_rate": 1.4470208142165892e-05,
      "loss": 0.1562,
      "step": 3699
    },
    {
      "epoch": 0.7766582703610412,
      "grad_norm": 0.1293368935585022,
      "learning_rate": 1.444443711637395e-05,
      "loss": 0.1509,
      "step": 3700
    },
    {
      "epoch": 0.7768681780016793,
      "grad_norm": 0.11314057558774948,
      "learning_rate": 1.4418685184731462e-05,
      "loss": 0.1511,
      "step": 3701
    },
    {
      "epoch": 0.7770780856423174,
      "grad_norm": 0.12856042385101318,
      "learning_rate": 1.4392952361067785e-05,
      "loss": 0.1564,
      "step": 3702
    },
    {
      "epoch": 0.7772879932829555,
      "grad_norm": 0.1180645078420639,
      "learning_rate": 1.436723865920202e-05,
      "loss": 0.1642,
      "step": 3703
    },
    {
      "epoch": 0.7774979009235936,
      "grad_norm": 0.11847218871116638,
      "learning_rate": 1.4341544092943055e-05,
      "loss": 0.149,
      "step": 3704
    },
    {
      "epoch": 0.7777078085642317,
      "grad_norm": 0.12634143233299255,
      "learning_rate": 1.4315868676089427e-05,
      "loss": 0.166,
      "step": 3705
    },
    {
      "epoch": 0.7779177162048698,
      "grad_norm": 0.10073088109493256,
      "learning_rate": 1.4290212422429444e-05,
      "loss": 0.1506,
      "step": 3706
    },
    {
      "epoch": 0.7781276238455079,
      "grad_norm": 0.12493392080068588,
      "learning_rate": 1.4264575345741065e-05,
      "loss": 0.1443,
      "step": 3707
    },
    {
      "epoch": 0.7783375314861462,
      "grad_norm": 0.15417474508285522,
      "learning_rate": 1.4238957459792041e-05,
      "loss": 0.1462,
      "step": 3708
    },
    {
      "epoch": 0.7785474391267843,
      "grad_norm": 0.10922432690858841,
      "learning_rate": 1.4213358778339719e-05,
      "loss": 0.154,
      "step": 3709
    },
    {
      "epoch": 0.7787573467674224,
      "grad_norm": 0.11463958770036697,
      "learning_rate": 1.4187779315131184e-05,
      "loss": 0.1524,
      "step": 3710
    },
    {
      "epoch": 0.7789672544080605,
      "grad_norm": 0.11828161031007767,
      "learning_rate": 1.4162219083903172e-05,
      "loss": 0.1409,
      "step": 3711
    },
    {
      "epoch": 0.7791771620486986,
      "grad_norm": 0.09642691910266876,
      "learning_rate": 1.4136678098382156e-05,
      "loss": 0.1422,
      "step": 3712
    },
    {
      "epoch": 0.7793870696893367,
      "grad_norm": 0.12066272646188736,
      "learning_rate": 1.4111156372284207e-05,
      "loss": 0.1521,
      "step": 3713
    },
    {
      "epoch": 0.7795969773299748,
      "grad_norm": 0.12618005275726318,
      "learning_rate": 1.4085653919315072e-05,
      "loss": 0.1534,
      "step": 3714
    },
    {
      "epoch": 0.7798068849706129,
      "grad_norm": 0.14962835609912872,
      "learning_rate": 1.4060170753170154e-05,
      "loss": 0.1598,
      "step": 3715
    },
    {
      "epoch": 0.780016792611251,
      "grad_norm": 0.12119120359420776,
      "learning_rate": 1.4034706887534493e-05,
      "loss": 0.1585,
      "step": 3716
    },
    {
      "epoch": 0.7802267002518891,
      "grad_norm": 0.1072816252708435,
      "learning_rate": 1.4009262336082767e-05,
      "loss": 0.153,
      "step": 3717
    },
    {
      "epoch": 0.7804366078925273,
      "grad_norm": 0.1178685873746872,
      "learning_rate": 1.3983837112479282e-05,
      "loss": 0.1649,
      "step": 3718
    },
    {
      "epoch": 0.7806465155331654,
      "grad_norm": 0.11183042824268341,
      "learning_rate": 1.3958431230377966e-05,
      "loss": 0.1508,
      "step": 3719
    },
    {
      "epoch": 0.7808564231738035,
      "grad_norm": 0.12451529502868652,
      "learning_rate": 1.3933044703422337e-05,
      "loss": 0.1617,
      "step": 3720
    },
    {
      "epoch": 0.7810663308144417,
      "grad_norm": 0.132112517952919,
      "learning_rate": 1.3907677545245579e-05,
      "loss": 0.1497,
      "step": 3721
    },
    {
      "epoch": 0.7812762384550798,
      "grad_norm": 0.12987057864665985,
      "learning_rate": 1.3882329769470425e-05,
      "loss": 0.1535,
      "step": 3722
    },
    {
      "epoch": 0.7814861460957179,
      "grad_norm": 0.11654311418533325,
      "learning_rate": 1.3857001389709174e-05,
      "loss": 0.1488,
      "step": 3723
    },
    {
      "epoch": 0.781696053736356,
      "grad_norm": 0.12595154345035553,
      "learning_rate": 1.3831692419563786e-05,
      "loss": 0.1533,
      "step": 3724
    },
    {
      "epoch": 0.7819059613769941,
      "grad_norm": 0.14091216027736664,
      "learning_rate": 1.380640287262574e-05,
      "loss": 0.1734,
      "step": 3725
    },
    {
      "epoch": 0.7821158690176322,
      "grad_norm": 0.14180079102516174,
      "learning_rate": 1.3781132762476096e-05,
      "loss": 0.1464,
      "step": 3726
    },
    {
      "epoch": 0.7823257766582704,
      "grad_norm": 0.12052421271800995,
      "learning_rate": 1.3755882102685486e-05,
      "loss": 0.1564,
      "step": 3727
    },
    {
      "epoch": 0.7825356842989085,
      "grad_norm": 0.1349646896123886,
      "learning_rate": 1.373065090681408e-05,
      "loss": 0.1486,
      "step": 3728
    },
    {
      "epoch": 0.7827455919395466,
      "grad_norm": 0.14333775639533997,
      "learning_rate": 1.3705439188411645e-05,
      "loss": 0.1527,
      "step": 3729
    },
    {
      "epoch": 0.7829554995801847,
      "grad_norm": 0.13959290087223053,
      "learning_rate": 1.3680246961017401e-05,
      "loss": 0.1659,
      "step": 3730
    },
    {
      "epoch": 0.7831654072208228,
      "grad_norm": 0.11381187289953232,
      "learning_rate": 1.3655074238160175e-05,
      "loss": 0.1587,
      "step": 3731
    },
    {
      "epoch": 0.783375314861461,
      "grad_norm": 0.14192262291908264,
      "learning_rate": 1.3629921033358272e-05,
      "loss": 0.1367,
      "step": 3732
    },
    {
      "epoch": 0.783585222502099,
      "grad_norm": 0.10846276581287384,
      "learning_rate": 1.3604787360119574e-05,
      "loss": 0.1476,
      "step": 3733
    },
    {
      "epoch": 0.7837951301427372,
      "grad_norm": 0.10947103053331375,
      "learning_rate": 1.3579673231941415e-05,
      "loss": 0.1479,
      "step": 3734
    },
    {
      "epoch": 0.7840050377833753,
      "grad_norm": 0.11812829971313477,
      "learning_rate": 1.3554578662310657e-05,
      "loss": 0.1748,
      "step": 3735
    },
    {
      "epoch": 0.7842149454240135,
      "grad_norm": 0.13791900873184204,
      "learning_rate": 1.3529503664703664e-05,
      "loss": 0.1586,
      "step": 3736
    },
    {
      "epoch": 0.7844248530646516,
      "grad_norm": 0.12275446206331253,
      "learning_rate": 1.3504448252586272e-05,
      "loss": 0.1587,
      "step": 3737
    },
    {
      "epoch": 0.7846347607052897,
      "grad_norm": 0.1119173988699913,
      "learning_rate": 1.347941243941382e-05,
      "loss": 0.1353,
      "step": 3738
    },
    {
      "epoch": 0.7848446683459278,
      "grad_norm": 0.1312474012374878,
      "learning_rate": 1.3454396238631101e-05,
      "loss": 0.1613,
      "step": 3739
    },
    {
      "epoch": 0.7850545759865659,
      "grad_norm": 0.1218884214758873,
      "learning_rate": 1.3429399663672382e-05,
      "loss": 0.1622,
      "step": 3740
    },
    {
      "epoch": 0.785264483627204,
      "grad_norm": 0.12461695820093155,
      "learning_rate": 1.3404422727961413e-05,
      "loss": 0.1413,
      "step": 3741
    },
    {
      "epoch": 0.7854743912678421,
      "grad_norm": 0.12566153705120087,
      "learning_rate": 1.3379465444911376e-05,
      "loss": 0.1357,
      "step": 3742
    },
    {
      "epoch": 0.7856842989084802,
      "grad_norm": 0.10288352519273758,
      "learning_rate": 1.3354527827924912e-05,
      "loss": 0.1728,
      "step": 3743
    },
    {
      "epoch": 0.7858942065491183,
      "grad_norm": 0.11198544502258301,
      "learning_rate": 1.3329609890394046e-05,
      "loss": 0.1342,
      "step": 3744
    },
    {
      "epoch": 0.7861041141897565,
      "grad_norm": 0.10878617316484451,
      "learning_rate": 1.3304711645700324e-05,
      "loss": 0.1562,
      "step": 3745
    },
    {
      "epoch": 0.7863140218303947,
      "grad_norm": 0.13761012256145477,
      "learning_rate": 1.327983310721465e-05,
      "loss": 0.1676,
      "step": 3746
    },
    {
      "epoch": 0.7865239294710328,
      "grad_norm": 0.10911747813224792,
      "learning_rate": 1.3254974288297384e-05,
      "loss": 0.1426,
      "step": 3747
    },
    {
      "epoch": 0.7867338371116709,
      "grad_norm": 0.10680964589118958,
      "learning_rate": 1.3230135202298256e-05,
      "loss": 0.1717,
      "step": 3748
    },
    {
      "epoch": 0.786943744752309,
      "grad_norm": 0.12579379975795746,
      "learning_rate": 1.3205315862556444e-05,
      "loss": 0.1505,
      "step": 3749
    },
    {
      "epoch": 0.7871536523929471,
      "grad_norm": 0.11058539152145386,
      "learning_rate": 1.3180516282400513e-05,
      "loss": 0.1435,
      "step": 3750
    },
    {
      "epoch": 0.7873635600335852,
      "grad_norm": 0.14114023745059967,
      "learning_rate": 1.3155736475148372e-05,
      "loss": 0.1517,
      "step": 3751
    },
    {
      "epoch": 0.7875734676742233,
      "grad_norm": 0.11331699043512344,
      "learning_rate": 1.3130976454107336e-05,
      "loss": 0.1407,
      "step": 3752
    },
    {
      "epoch": 0.7877833753148614,
      "grad_norm": 0.1149875819683075,
      "learning_rate": 1.3106236232574138e-05,
      "loss": 0.1596,
      "step": 3753
    },
    {
      "epoch": 0.7879932829554995,
      "grad_norm": 0.13585644960403442,
      "learning_rate": 1.3081515823834828e-05,
      "loss": 0.1616,
      "step": 3754
    },
    {
      "epoch": 0.7882031905961377,
      "grad_norm": 0.12654295563697815,
      "learning_rate": 1.3056815241164826e-05,
      "loss": 0.166,
      "step": 3755
    },
    {
      "epoch": 0.7884130982367759,
      "grad_norm": 0.1386784017086029,
      "learning_rate": 1.3032134497828913e-05,
      "loss": 0.1668,
      "step": 3756
    },
    {
      "epoch": 0.788623005877414,
      "grad_norm": 0.14663977921009064,
      "learning_rate": 1.300747360708121e-05,
      "loss": 0.1635,
      "step": 3757
    },
    {
      "epoch": 0.7888329135180521,
      "grad_norm": 0.14930498600006104,
      "learning_rate": 1.2982832582165177e-05,
      "loss": 0.1521,
      "step": 3758
    },
    {
      "epoch": 0.7890428211586902,
      "grad_norm": 0.10799439996480942,
      "learning_rate": 1.295821143631361e-05,
      "loss": 0.1494,
      "step": 3759
    },
    {
      "epoch": 0.7892527287993283,
      "grad_norm": 0.1401253193616867,
      "learning_rate": 1.2933610182748629e-05,
      "loss": 0.1447,
      "step": 3760
    },
    {
      "epoch": 0.7894626364399664,
      "grad_norm": 0.12463638931512833,
      "learning_rate": 1.2909028834681646e-05,
      "loss": 0.1575,
      "step": 3761
    },
    {
      "epoch": 0.7896725440806045,
      "grad_norm": 0.10640936344861984,
      "learning_rate": 1.2884467405313444e-05,
      "loss": 0.1617,
      "step": 3762
    },
    {
      "epoch": 0.7898824517212426,
      "grad_norm": 0.11918281018733978,
      "learning_rate": 1.2859925907834053e-05,
      "loss": 0.1667,
      "step": 3763
    },
    {
      "epoch": 0.7900923593618808,
      "grad_norm": 0.12711642682552338,
      "learning_rate": 1.2835404355422814e-05,
      "loss": 0.1687,
      "step": 3764
    },
    {
      "epoch": 0.7903022670025189,
      "grad_norm": 0.11989998072385788,
      "learning_rate": 1.2810902761248373e-05,
      "loss": 0.1669,
      "step": 3765
    },
    {
      "epoch": 0.790512174643157,
      "grad_norm": 0.15876778960227966,
      "learning_rate": 1.2786421138468636e-05,
      "loss": 0.1621,
      "step": 3766
    },
    {
      "epoch": 0.7907220822837951,
      "grad_norm": 0.11356843262910843,
      "learning_rate": 1.2761959500230791e-05,
      "loss": 0.1512,
      "step": 3767
    },
    {
      "epoch": 0.7909319899244333,
      "grad_norm": 0.1321927160024643,
      "learning_rate": 1.2737517859671305e-05,
      "loss": 0.1434,
      "step": 3768
    },
    {
      "epoch": 0.7911418975650714,
      "grad_norm": 0.1526881903409958,
      "learning_rate": 1.2713096229915883e-05,
      "loss": 0.1544,
      "step": 3769
    },
    {
      "epoch": 0.7913518052057095,
      "grad_norm": 0.12897861003875732,
      "learning_rate": 1.2688694624079516e-05,
      "loss": 0.1725,
      "step": 3770
    },
    {
      "epoch": 0.7915617128463476,
      "grad_norm": 0.1343359798192978,
      "learning_rate": 1.2664313055266436e-05,
      "loss": 0.1426,
      "step": 3771
    },
    {
      "epoch": 0.7917716204869857,
      "grad_norm": 0.1096104085445404,
      "learning_rate": 1.2639951536570066e-05,
      "loss": 0.1809,
      "step": 3772
    },
    {
      "epoch": 0.7919815281276238,
      "grad_norm": 0.1288156807422638,
      "learning_rate": 1.2615610081073104e-05,
      "loss": 0.1642,
      "step": 3773
    },
    {
      "epoch": 0.792191435768262,
      "grad_norm": 0.1227651983499527,
      "learning_rate": 1.2591288701847487e-05,
      "loss": 0.1484,
      "step": 3774
    },
    {
      "epoch": 0.7924013434089001,
      "grad_norm": 0.1349715292453766,
      "learning_rate": 1.2566987411954345e-05,
      "loss": 0.1542,
      "step": 3775
    },
    {
      "epoch": 0.7926112510495382,
      "grad_norm": 0.1394587904214859,
      "learning_rate": 1.2542706224444023e-05,
      "loss": 0.1617,
      "step": 3776
    },
    {
      "epoch": 0.7928211586901763,
      "grad_norm": 0.11542592942714691,
      "learning_rate": 1.2518445152356062e-05,
      "loss": 0.1578,
      "step": 3777
    },
    {
      "epoch": 0.7930310663308144,
      "grad_norm": 0.11972792446613312,
      "learning_rate": 1.2494204208719251e-05,
      "loss": 0.1503,
      "step": 3778
    },
    {
      "epoch": 0.7932409739714525,
      "grad_norm": 0.11570564657449722,
      "learning_rate": 1.2469983406551484e-05,
      "loss": 0.1738,
      "step": 3779
    },
    {
      "epoch": 0.7934508816120907,
      "grad_norm": 0.12285133451223373,
      "learning_rate": 1.2445782758859909e-05,
      "loss": 0.1399,
      "step": 3780
    },
    {
      "epoch": 0.7936607892527288,
      "grad_norm": 0.12845146656036377,
      "learning_rate": 1.2421602278640804e-05,
      "loss": 0.1371,
      "step": 3781
    },
    {
      "epoch": 0.7938706968933669,
      "grad_norm": 0.12982353568077087,
      "learning_rate": 1.2397441978879676e-05,
      "loss": 0.1551,
      "step": 3782
    },
    {
      "epoch": 0.7940806045340051,
      "grad_norm": 0.12580209970474243,
      "learning_rate": 1.2373301872551134e-05,
      "loss": 0.1566,
      "step": 3783
    },
    {
      "epoch": 0.7942905121746432,
      "grad_norm": 0.18834897875785828,
      "learning_rate": 1.2349181972618973e-05,
      "loss": 0.1583,
      "step": 3784
    },
    {
      "epoch": 0.7945004198152813,
      "grad_norm": 0.11271581053733826,
      "learning_rate": 1.2325082292036127e-05,
      "loss": 0.146,
      "step": 3785
    },
    {
      "epoch": 0.7947103274559194,
      "grad_norm": 0.12085813283920288,
      "learning_rate": 1.230100284374468e-05,
      "loss": 0.1465,
      "step": 3786
    },
    {
      "epoch": 0.7949202350965575,
      "grad_norm": 0.11245040595531464,
      "learning_rate": 1.2276943640675842e-05,
      "loss": 0.1355,
      "step": 3787
    },
    {
      "epoch": 0.7951301427371956,
      "grad_norm": 0.14471422135829926,
      "learning_rate": 1.2252904695749951e-05,
      "loss": 0.1607,
      "step": 3788
    },
    {
      "epoch": 0.7953400503778337,
      "grad_norm": 0.11694999784231186,
      "learning_rate": 1.2228886021876463e-05,
      "loss": 0.1618,
      "step": 3789
    },
    {
      "epoch": 0.7955499580184718,
      "grad_norm": 0.10486170649528503,
      "learning_rate": 1.2204887631953975e-05,
      "loss": 0.1332,
      "step": 3790
    },
    {
      "epoch": 0.7957598656591099,
      "grad_norm": 0.10166893154382706,
      "learning_rate": 1.2180909538870177e-05,
      "loss": 0.1565,
      "step": 3791
    },
    {
      "epoch": 0.7959697732997482,
      "grad_norm": 0.13252127170562744,
      "learning_rate": 1.2156951755501822e-05,
      "loss": 0.1525,
      "step": 3792
    },
    {
      "epoch": 0.7961796809403863,
      "grad_norm": 0.13535131514072418,
      "learning_rate": 1.213301429471479e-05,
      "loss": 0.1439,
      "step": 3793
    },
    {
      "epoch": 0.7963895885810244,
      "grad_norm": 0.1082453727722168,
      "learning_rate": 1.2109097169364064e-05,
      "loss": 0.1517,
      "step": 3794
    },
    {
      "epoch": 0.7965994962216625,
      "grad_norm": 0.11083705723285675,
      "learning_rate": 1.2085200392293683e-05,
      "loss": 0.1339,
      "step": 3795
    },
    {
      "epoch": 0.7968094038623006,
      "grad_norm": 0.12013557553291321,
      "learning_rate": 1.2061323976336752e-05,
      "loss": 0.1669,
      "step": 3796
    },
    {
      "epoch": 0.7970193115029387,
      "grad_norm": 0.12751804292201996,
      "learning_rate": 1.203746793431546e-05,
      "loss": 0.1488,
      "step": 3797
    },
    {
      "epoch": 0.7972292191435768,
      "grad_norm": 0.1310308873653412,
      "learning_rate": 1.2013632279041042e-05,
      "loss": 0.1618,
      "step": 3798
    },
    {
      "epoch": 0.7974391267842149,
      "grad_norm": 0.12405902147293091,
      "learning_rate": 1.1989817023313792e-05,
      "loss": 0.1654,
      "step": 3799
    },
    {
      "epoch": 0.797649034424853,
      "grad_norm": 0.11220216006040573,
      "learning_rate": 1.1966022179923043e-05,
      "loss": 0.1552,
      "step": 3800
    },
    {
      "epoch": 0.7978589420654912,
      "grad_norm": 0.11725794523954391,
      "learning_rate": 1.1942247761647174e-05,
      "loss": 0.1732,
      "step": 3801
    },
    {
      "epoch": 0.7980688497061293,
      "grad_norm": 0.11368782073259354,
      "learning_rate": 1.191849378125357e-05,
      "loss": 0.168,
      "step": 3802
    },
    {
      "epoch": 0.7982787573467675,
      "grad_norm": 0.13916854560375214,
      "learning_rate": 1.1894760251498699e-05,
      "loss": 0.1479,
      "step": 3803
    },
    {
      "epoch": 0.7984886649874056,
      "grad_norm": 0.10956691950559616,
      "learning_rate": 1.1871047185127987e-05,
      "loss": 0.146,
      "step": 3804
    },
    {
      "epoch": 0.7986985726280437,
      "grad_norm": 0.1180625706911087,
      "learning_rate": 1.1847354594875893e-05,
      "loss": 0.156,
      "step": 3805
    },
    {
      "epoch": 0.7989084802686818,
      "grad_norm": 0.11851587146520615,
      "learning_rate": 1.1823682493465876e-05,
      "loss": 0.1718,
      "step": 3806
    },
    {
      "epoch": 0.7991183879093199,
      "grad_norm": 0.10954640805721283,
      "learning_rate": 1.1800030893610404e-05,
      "loss": 0.1582,
      "step": 3807
    },
    {
      "epoch": 0.799328295549958,
      "grad_norm": 0.16035374999046326,
      "learning_rate": 1.1776399808010924e-05,
      "loss": 0.1674,
      "step": 3808
    },
    {
      "epoch": 0.7995382031905961,
      "grad_norm": 0.12425439804792404,
      "learning_rate": 1.1752789249357865e-05,
      "loss": 0.1442,
      "step": 3809
    },
    {
      "epoch": 0.7997481108312342,
      "grad_norm": 0.12865933775901794,
      "learning_rate": 1.1729199230330617e-05,
      "loss": 0.1663,
      "step": 3810
    },
    {
      "epoch": 0.7999580184718724,
      "grad_norm": 0.11566450446844101,
      "learning_rate": 1.1705629763597603e-05,
      "loss": 0.1532,
      "step": 3811
    },
    {
      "epoch": 0.8001679261125105,
      "grad_norm": 0.1077750027179718,
      "learning_rate": 1.1682080861816152e-05,
      "loss": 0.1584,
      "step": 3812
    },
    {
      "epoch": 0.8003778337531486,
      "grad_norm": 0.11821168661117554,
      "learning_rate": 1.165855253763254e-05,
      "loss": 0.1609,
      "step": 3813
    },
    {
      "epoch": 0.8005877413937867,
      "grad_norm": 0.11459922045469284,
      "learning_rate": 1.1635044803682011e-05,
      "loss": 0.1458,
      "step": 3814
    },
    {
      "epoch": 0.8007976490344249,
      "grad_norm": 0.11377918720245361,
      "learning_rate": 1.1611557672588785e-05,
      "loss": 0.1485,
      "step": 3815
    },
    {
      "epoch": 0.801007556675063,
      "grad_norm": 0.1579093039035797,
      "learning_rate": 1.1588091156965975e-05,
      "loss": 0.1708,
      "step": 3816
    },
    {
      "epoch": 0.8012174643157011,
      "grad_norm": 0.1275099813938141,
      "learning_rate": 1.1564645269415637e-05,
      "loss": 0.1549,
      "step": 3817
    },
    {
      "epoch": 0.8014273719563392,
      "grad_norm": 0.1207624152302742,
      "learning_rate": 1.1541220022528725e-05,
      "loss": 0.1688,
      "step": 3818
    },
    {
      "epoch": 0.8016372795969773,
      "grad_norm": 0.13440079987049103,
      "learning_rate": 1.1517815428885186e-05,
      "loss": 0.1575,
      "step": 3819
    },
    {
      "epoch": 0.8018471872376155,
      "grad_norm": 0.1259511113166809,
      "learning_rate": 1.149443150105377e-05,
      "loss": 0.1534,
      "step": 3820
    },
    {
      "epoch": 0.8020570948782536,
      "grad_norm": 0.10427981615066528,
      "learning_rate": 1.1471068251592204e-05,
      "loss": 0.148,
      "step": 3821
    },
    {
      "epoch": 0.8022670025188917,
      "grad_norm": 0.10866251587867737,
      "learning_rate": 1.1447725693047062e-05,
      "loss": 0.1467,
      "step": 3822
    },
    {
      "epoch": 0.8024769101595298,
      "grad_norm": 0.11648157238960266,
      "learning_rate": 1.1424403837953862e-05,
      "loss": 0.1399,
      "step": 3823
    },
    {
      "epoch": 0.8026868178001679,
      "grad_norm": 0.12202499806880951,
      "learning_rate": 1.1401102698836968e-05,
      "loss": 0.1513,
      "step": 3824
    },
    {
      "epoch": 0.802896725440806,
      "grad_norm": 0.1321491301059723,
      "learning_rate": 1.1377822288209611e-05,
      "loss": 0.1674,
      "step": 3825
    },
    {
      "epoch": 0.8031066330814441,
      "grad_norm": 0.11231104284524918,
      "learning_rate": 1.135456261857391e-05,
      "loss": 0.1501,
      "step": 3826
    },
    {
      "epoch": 0.8033165407220823,
      "grad_norm": 0.14203539490699768,
      "learning_rate": 1.1331323702420842e-05,
      "loss": 0.1549,
      "step": 3827
    },
    {
      "epoch": 0.8035264483627204,
      "grad_norm": 0.11880168318748474,
      "learning_rate": 1.1308105552230231e-05,
      "loss": 0.1549,
      "step": 3828
    },
    {
      "epoch": 0.8037363560033586,
      "grad_norm": 0.11236574500799179,
      "learning_rate": 1.1284908180470748e-05,
      "loss": 0.1486,
      "step": 3829
    },
    {
      "epoch": 0.8039462636439967,
      "grad_norm": 0.12606748938560486,
      "learning_rate": 1.1261731599599911e-05,
      "loss": 0.1531,
      "step": 3830
    },
    {
      "epoch": 0.8041561712846348,
      "grad_norm": 0.13274842500686646,
      "learning_rate": 1.1238575822064096e-05,
      "loss": 0.1598,
      "step": 3831
    },
    {
      "epoch": 0.8043660789252729,
      "grad_norm": 0.1181090697646141,
      "learning_rate": 1.1215440860298465e-05,
      "loss": 0.1613,
      "step": 3832
    },
    {
      "epoch": 0.804575986565911,
      "grad_norm": 0.12809668481349945,
      "learning_rate": 1.1192326726727042e-05,
      "loss": 0.1592,
      "step": 3833
    },
    {
      "epoch": 0.8047858942065491,
      "grad_norm": 0.11602126806974411,
      "learning_rate": 1.11692334337626e-05,
      "loss": 0.1514,
      "step": 3834
    },
    {
      "epoch": 0.8049958018471872,
      "grad_norm": 0.12484683841466904,
      "learning_rate": 1.1146160993806804e-05,
      "loss": 0.1589,
      "step": 3835
    },
    {
      "epoch": 0.8052057094878253,
      "grad_norm": 0.1220654621720314,
      "learning_rate": 1.1123109419250072e-05,
      "loss": 0.1499,
      "step": 3836
    },
    {
      "epoch": 0.8054156171284634,
      "grad_norm": 0.12533478438854218,
      "learning_rate": 1.1100078722471618e-05,
      "loss": 0.157,
      "step": 3837
    },
    {
      "epoch": 0.8056255247691015,
      "grad_norm": 0.15392319858074188,
      "learning_rate": 1.1077068915839467e-05,
      "loss": 0.1609,
      "step": 3838
    },
    {
      "epoch": 0.8058354324097398,
      "grad_norm": 0.16898441314697266,
      "learning_rate": 1.1054080011710383e-05,
      "loss": 0.1617,
      "step": 3839
    },
    {
      "epoch": 0.8060453400503779,
      "grad_norm": 0.1096329316496849,
      "learning_rate": 1.1031112022429984e-05,
      "loss": 0.1393,
      "step": 3840
    },
    {
      "epoch": 0.806255247691016,
      "grad_norm": 0.1226249560713768,
      "learning_rate": 1.1008164960332556e-05,
      "loss": 0.1633,
      "step": 3841
    },
    {
      "epoch": 0.8064651553316541,
      "grad_norm": 0.10499352961778641,
      "learning_rate": 1.0985238837741213e-05,
      "loss": 0.1456,
      "step": 3842
    },
    {
      "epoch": 0.8066750629722922,
      "grad_norm": 0.14240071177482605,
      "learning_rate": 1.0962333666967789e-05,
      "loss": 0.1511,
      "step": 3843
    },
    {
      "epoch": 0.8068849706129303,
      "grad_norm": 0.11764491349458694,
      "learning_rate": 1.0939449460312918e-05,
      "loss": 0.1681,
      "step": 3844
    },
    {
      "epoch": 0.8070948782535684,
      "grad_norm": 0.1263282299041748,
      "learning_rate": 1.0916586230065922e-05,
      "loss": 0.135,
      "step": 3845
    },
    {
      "epoch": 0.8073047858942065,
      "grad_norm": 0.14005941152572632,
      "learning_rate": 1.0893743988504884e-05,
      "loss": 0.1527,
      "step": 3846
    },
    {
      "epoch": 0.8075146935348446,
      "grad_norm": 0.1181284487247467,
      "learning_rate": 1.08709227478966e-05,
      "loss": 0.1388,
      "step": 3847
    },
    {
      "epoch": 0.8077246011754828,
      "grad_norm": 0.10536040365695953,
      "learning_rate": 1.0848122520496607e-05,
      "loss": 0.1736,
      "step": 3848
    },
    {
      "epoch": 0.8079345088161209,
      "grad_norm": 0.10224374383687973,
      "learning_rate": 1.0825343318549141e-05,
      "loss": 0.1439,
      "step": 3849
    },
    {
      "epoch": 0.808144416456759,
      "grad_norm": 0.10331021994352341,
      "learning_rate": 1.080258515428716e-05,
      "loss": 0.1465,
      "step": 3850
    },
    {
      "epoch": 0.8083543240973972,
      "grad_norm": 0.1074858009815216,
      "learning_rate": 1.07798480399323e-05,
      "loss": 0.1531,
      "step": 3851
    },
    {
      "epoch": 0.8085642317380353,
      "grad_norm": 0.1292872577905655,
      "learning_rate": 1.075713198769494e-05,
      "loss": 0.1535,
      "step": 3852
    },
    {
      "epoch": 0.8087741393786734,
      "grad_norm": 0.1513303965330124,
      "learning_rate": 1.0734437009774106e-05,
      "loss": 0.1667,
      "step": 3853
    },
    {
      "epoch": 0.8089840470193115,
      "grad_norm": 0.12170423567295074,
      "learning_rate": 1.0711763118357527e-05,
      "loss": 0.16,
      "step": 3854
    },
    {
      "epoch": 0.8091939546599496,
      "grad_norm": 0.12069430202245712,
      "learning_rate": 1.068911032562157e-05,
      "loss": 0.1503,
      "step": 3855
    },
    {
      "epoch": 0.8094038623005877,
      "grad_norm": 0.11853886395692825,
      "learning_rate": 1.0666478643731332e-05,
      "loss": 0.1626,
      "step": 3856
    },
    {
      "epoch": 0.8096137699412259,
      "grad_norm": 0.1364051252603531,
      "learning_rate": 1.0643868084840542e-05,
      "loss": 0.1507,
      "step": 3857
    },
    {
      "epoch": 0.809823677581864,
      "grad_norm": 0.11953744292259216,
      "learning_rate": 1.0621278661091571e-05,
      "loss": 0.1444,
      "step": 3858
    },
    {
      "epoch": 0.8100335852225021,
      "grad_norm": 0.14346711337566376,
      "learning_rate": 1.0598710384615457e-05,
      "loss": 0.1534,
      "step": 3859
    },
    {
      "epoch": 0.8102434928631402,
      "grad_norm": 0.14622189104557037,
      "learning_rate": 1.0576163267531901e-05,
      "loss": 0.1648,
      "step": 3860
    },
    {
      "epoch": 0.8104534005037783,
      "grad_norm": 0.1627495288848877,
      "learning_rate": 1.0553637321949217e-05,
      "loss": 0.1608,
      "step": 3861
    },
    {
      "epoch": 0.8106633081444165,
      "grad_norm": 0.10121377557516098,
      "learning_rate": 1.0531132559964329e-05,
      "loss": 0.1524,
      "step": 3862
    },
    {
      "epoch": 0.8108732157850546,
      "grad_norm": 0.13579495251178741,
      "learning_rate": 1.0508648993662806e-05,
      "loss": 0.1506,
      "step": 3863
    },
    {
      "epoch": 0.8110831234256927,
      "grad_norm": 0.14970248937606812,
      "learning_rate": 1.0486186635118866e-05,
      "loss": 0.16,
      "step": 3864
    },
    {
      "epoch": 0.8112930310663308,
      "grad_norm": 0.14687329530715942,
      "learning_rate": 1.0463745496395294e-05,
      "loss": 0.1777,
      "step": 3865
    },
    {
      "epoch": 0.8115029387069689,
      "grad_norm": 0.1286310851573944,
      "learning_rate": 1.0441325589543494e-05,
      "loss": 0.1405,
      "step": 3866
    },
    {
      "epoch": 0.8117128463476071,
      "grad_norm": 0.12971775233745575,
      "learning_rate": 1.0418926926603467e-05,
      "loss": 0.1536,
      "step": 3867
    },
    {
      "epoch": 0.8119227539882452,
      "grad_norm": 0.11588292568922043,
      "learning_rate": 1.039654951960381e-05,
      "loss": 0.1585,
      "step": 3868
    },
    {
      "epoch": 0.8121326616288833,
      "grad_norm": 0.11025537550449371,
      "learning_rate": 1.0374193380561704e-05,
      "loss": 0.158,
      "step": 3869
    },
    {
      "epoch": 0.8123425692695214,
      "grad_norm": 0.11403115838766098,
      "learning_rate": 1.0351858521482893e-05,
      "loss": 0.1537,
      "step": 3870
    },
    {
      "epoch": 0.8125524769101595,
      "grad_norm": 0.11629325896501541,
      "learning_rate": 1.0329544954361709e-05,
      "loss": 0.1495,
      "step": 3871
    },
    {
      "epoch": 0.8127623845507976,
      "grad_norm": 0.12189958244562149,
      "learning_rate": 1.0307252691181063e-05,
      "loss": 0.166,
      "step": 3872
    },
    {
      "epoch": 0.8129722921914357,
      "grad_norm": 0.11428286880254745,
      "learning_rate": 1.0284981743912397e-05,
      "loss": 0.1617,
      "step": 3873
    },
    {
      "epoch": 0.8131821998320738,
      "grad_norm": 0.13367648422718048,
      "learning_rate": 1.0262732124515729e-05,
      "loss": 0.1518,
      "step": 3874
    },
    {
      "epoch": 0.813392107472712,
      "grad_norm": 0.09968928247690201,
      "learning_rate": 1.0240503844939603e-05,
      "loss": 0.1431,
      "step": 3875
    },
    {
      "epoch": 0.8136020151133502,
      "grad_norm": 0.12892545759677887,
      "learning_rate": 1.0218296917121111e-05,
      "loss": 0.1501,
      "step": 3876
    },
    {
      "epoch": 0.8138119227539883,
      "grad_norm": 0.14232294261455536,
      "learning_rate": 1.0196111352985887e-05,
      "loss": 0.1384,
      "step": 3877
    },
    {
      "epoch": 0.8140218303946264,
      "grad_norm": 0.12474681437015533,
      "learning_rate": 1.0173947164448089e-05,
      "loss": 0.1441,
      "step": 3878
    },
    {
      "epoch": 0.8142317380352645,
      "grad_norm": 0.11398620903491974,
      "learning_rate": 1.0151804363410383e-05,
      "loss": 0.1494,
      "step": 3879
    },
    {
      "epoch": 0.8144416456759026,
      "grad_norm": 0.1219918429851532,
      "learning_rate": 1.0129682961763948e-05,
      "loss": 0.1552,
      "step": 3880
    },
    {
      "epoch": 0.8146515533165407,
      "grad_norm": 0.12299709022045135,
      "learning_rate": 1.0107582971388524e-05,
      "loss": 0.1359,
      "step": 3881
    },
    {
      "epoch": 0.8148614609571788,
      "grad_norm": 0.12771455943584442,
      "learning_rate": 1.0085504404152273e-05,
      "loss": 0.1404,
      "step": 3882
    },
    {
      "epoch": 0.8150713685978169,
      "grad_norm": 0.12867510318756104,
      "learning_rate": 1.0063447271911897e-05,
      "loss": 0.1677,
      "step": 3883
    },
    {
      "epoch": 0.815281276238455,
      "grad_norm": 0.12489780783653259,
      "learning_rate": 1.0041411586512572e-05,
      "loss": 0.1458,
      "step": 3884
    },
    {
      "epoch": 0.8154911838790933,
      "grad_norm": 0.11988552659749985,
      "learning_rate": 1.0019397359787991e-05,
      "loss": 0.1367,
      "step": 3885
    },
    {
      "epoch": 0.8157010915197314,
      "grad_norm": 0.11586282402276993,
      "learning_rate": 9.997404603560278e-06,
      "loss": 0.1635,
      "step": 3886
    },
    {
      "epoch": 0.8159109991603695,
      "grad_norm": 0.11998885124921799,
      "learning_rate": 9.975433329640054e-06,
      "loss": 0.1705,
      "step": 3887
    },
    {
      "epoch": 0.8161209068010076,
      "grad_norm": 0.12188997864723206,
      "learning_rate": 9.953483549826387e-06,
      "loss": 0.1435,
      "step": 3888
    },
    {
      "epoch": 0.8163308144416457,
      "grad_norm": 0.12801359593868256,
      "learning_rate": 9.931555275906812e-06,
      "loss": 0.155,
      "step": 3889
    },
    {
      "epoch": 0.8165407220822838,
      "grad_norm": 0.12145894020795822,
      "learning_rate": 9.909648519657316e-06,
      "loss": 0.1516,
      "step": 3890
    },
    {
      "epoch": 0.8167506297229219,
      "grad_norm": 0.13459019362926483,
      "learning_rate": 9.887763292842323e-06,
      "loss": 0.1392,
      "step": 3891
    },
    {
      "epoch": 0.81696053736356,
      "grad_norm": 0.10860946029424667,
      "learning_rate": 9.86589960721469e-06,
      "loss": 0.1655,
      "step": 3892
    },
    {
      "epoch": 0.8171704450041981,
      "grad_norm": 0.15670441091060638,
      "learning_rate": 9.844057474515745e-06,
      "loss": 0.1706,
      "step": 3893
    },
    {
      "epoch": 0.8173803526448362,
      "grad_norm": 0.10645842552185059,
      "learning_rate": 9.822236906475191e-06,
      "loss": 0.154,
      "step": 3894
    },
    {
      "epoch": 0.8175902602854744,
      "grad_norm": 0.11920606344938278,
      "learning_rate": 9.800437914811189e-06,
      "loss": 0.1782,
      "step": 3895
    },
    {
      "epoch": 0.8178001679261125,
      "grad_norm": 0.11042717099189758,
      "learning_rate": 9.778660511230253e-06,
      "loss": 0.1434,
      "step": 3896
    },
    {
      "epoch": 0.8180100755667506,
      "grad_norm": 0.13635507225990295,
      "learning_rate": 9.756904707427395e-06,
      "loss": 0.1564,
      "step": 3897
    },
    {
      "epoch": 0.8182199832073888,
      "grad_norm": 0.11731656640768051,
      "learning_rate": 9.73517051508595e-06,
      "loss": 0.1452,
      "step": 3898
    },
    {
      "epoch": 0.8184298908480269,
      "grad_norm": 0.1272083967924118,
      "learning_rate": 9.713457945877691e-06,
      "loss": 0.1511,
      "step": 3899
    },
    {
      "epoch": 0.818639798488665,
      "grad_norm": 0.11312077194452286,
      "learning_rate": 9.691767011462743e-06,
      "loss": 0.1552,
      "step": 3900
    },
    {
      "epoch": 0.8188497061293031,
      "grad_norm": 0.12853741645812988,
      "learning_rate": 9.670097723489662e-06,
      "loss": 0.1598,
      "step": 3901
    },
    {
      "epoch": 0.8190596137699412,
      "grad_norm": 0.10277300328016281,
      "learning_rate": 9.648450093595346e-06,
      "loss": 0.1612,
      "step": 3902
    },
    {
      "epoch": 0.8192695214105793,
      "grad_norm": 0.1202654018998146,
      "learning_rate": 9.626824133405043e-06,
      "loss": 0.1625,
      "step": 3903
    },
    {
      "epoch": 0.8194794290512175,
      "grad_norm": 0.1286875158548355,
      "learning_rate": 9.605219854532393e-06,
      "loss": 0.1555,
      "step": 3904
    },
    {
      "epoch": 0.8196893366918556,
      "grad_norm": 0.1292945295572281,
      "learning_rate": 9.583637268579405e-06,
      "loss": 0.1494,
      "step": 3905
    },
    {
      "epoch": 0.8198992443324937,
      "grad_norm": 0.12690283358097076,
      "learning_rate": 9.562076387136414e-06,
      "loss": 0.1494,
      "step": 3906
    },
    {
      "epoch": 0.8201091519731318,
      "grad_norm": 0.0949103832244873,
      "learning_rate": 9.540537221782098e-06,
      "loss": 0.1495,
      "step": 3907
    },
    {
      "epoch": 0.8203190596137699,
      "grad_norm": 0.11503642797470093,
      "learning_rate": 9.519019784083488e-06,
      "loss": 0.1485,
      "step": 3908
    },
    {
      "epoch": 0.820528967254408,
      "grad_norm": 0.12526684999465942,
      "learning_rate": 9.497524085595943e-06,
      "loss": 0.1641,
      "step": 3909
    },
    {
      "epoch": 0.8207388748950462,
      "grad_norm": 0.1197986826300621,
      "learning_rate": 9.476050137863136e-06,
      "loss": 0.1546,
      "step": 3910
    },
    {
      "epoch": 0.8209487825356843,
      "grad_norm": 0.13118648529052734,
      "learning_rate": 9.454597952417066e-06,
      "loss": 0.153,
      "step": 3911
    },
    {
      "epoch": 0.8211586901763224,
      "grad_norm": 0.11788372695446014,
      "learning_rate": 9.433167540778037e-06,
      "loss": 0.151,
      "step": 3912
    },
    {
      "epoch": 0.8213685978169606,
      "grad_norm": 0.12314417958259583,
      "learning_rate": 9.411758914454698e-06,
      "loss": 0.1612,
      "step": 3913
    },
    {
      "epoch": 0.8215785054575987,
      "grad_norm": 0.12039327621459961,
      "learning_rate": 9.39037208494396e-06,
      "loss": 0.1559,
      "step": 3914
    },
    {
      "epoch": 0.8217884130982368,
      "grad_norm": 0.11974295228719711,
      "learning_rate": 9.369007063731029e-06,
      "loss": 0.1501,
      "step": 3915
    },
    {
      "epoch": 0.8219983207388749,
      "grad_norm": 0.11983851343393326,
      "learning_rate": 9.34766386228942e-06,
      "loss": 0.1537,
      "step": 3916
    },
    {
      "epoch": 0.822208228379513,
      "grad_norm": 0.1342703104019165,
      "learning_rate": 9.326342492080908e-06,
      "loss": 0.1545,
      "step": 3917
    },
    {
      "epoch": 0.8224181360201511,
      "grad_norm": 0.13480761647224426,
      "learning_rate": 9.305042964555567e-06,
      "loss": 0.1538,
      "step": 3918
    },
    {
      "epoch": 0.8226280436607892,
      "grad_norm": 0.1398279070854187,
      "learning_rate": 9.283765291151724e-06,
      "loss": 0.1599,
      "step": 3919
    },
    {
      "epoch": 0.8228379513014273,
      "grad_norm": 0.12116552144289017,
      "learning_rate": 9.262509483295977e-06,
      "loss": 0.1463,
      "step": 3920
    },
    {
      "epoch": 0.8230478589420654,
      "grad_norm": 0.11443023383617401,
      "learning_rate": 9.241275552403168e-06,
      "loss": 0.1425,
      "step": 3921
    },
    {
      "epoch": 0.8232577665827037,
      "grad_norm": 0.13521820306777954,
      "learning_rate": 9.220063509876431e-06,
      "loss": 0.1679,
      "step": 3922
    },
    {
      "epoch": 0.8234676742233418,
      "grad_norm": 0.13272807002067566,
      "learning_rate": 9.198873367107109e-06,
      "loss": 0.1609,
      "step": 3923
    },
    {
      "epoch": 0.8236775818639799,
      "grad_norm": 0.1287853866815567,
      "learning_rate": 9.177705135474774e-06,
      "loss": 0.1554,
      "step": 3924
    },
    {
      "epoch": 0.823887489504618,
      "grad_norm": 0.11695458739995956,
      "learning_rate": 9.15655882634725e-06,
      "loss": 0.1451,
      "step": 3925
    },
    {
      "epoch": 0.8240973971452561,
      "grad_norm": 0.11539758741855621,
      "learning_rate": 9.13543445108061e-06,
      "loss": 0.1518,
      "step": 3926
    },
    {
      "epoch": 0.8243073047858942,
      "grad_norm": 0.1678249090909958,
      "learning_rate": 9.114332021019117e-06,
      "loss": 0.1645,
      "step": 3927
    },
    {
      "epoch": 0.8245172124265323,
      "grad_norm": 0.13638179004192352,
      "learning_rate": 9.093251547495257e-06,
      "loss": 0.1731,
      "step": 3928
    },
    {
      "epoch": 0.8247271200671704,
      "grad_norm": 0.12491890788078308,
      "learning_rate": 9.072193041829708e-06,
      "loss": 0.1337,
      "step": 3929
    },
    {
      "epoch": 0.8249370277078085,
      "grad_norm": 0.1244591474533081,
      "learning_rate": 9.051156515331417e-06,
      "loss": 0.1621,
      "step": 3930
    },
    {
      "epoch": 0.8251469353484466,
      "grad_norm": 0.11335837841033936,
      "learning_rate": 9.030141979297429e-06,
      "loss": 0.1443,
      "step": 3931
    },
    {
      "epoch": 0.8253568429890848,
      "grad_norm": 0.12867414951324463,
      "learning_rate": 9.009149445013054e-06,
      "loss": 0.1667,
      "step": 3932
    },
    {
      "epoch": 0.825566750629723,
      "grad_norm": 0.09924521297216415,
      "learning_rate": 8.988178923751745e-06,
      "loss": 0.1433,
      "step": 3933
    },
    {
      "epoch": 0.8257766582703611,
      "grad_norm": 0.11446885764598846,
      "learning_rate": 8.96723042677517e-06,
      "loss": 0.1481,
      "step": 3934
    },
    {
      "epoch": 0.8259865659109992,
      "grad_norm": 0.15089714527130127,
      "learning_rate": 8.946303965333147e-06,
      "loss": 0.1504,
      "step": 3935
    },
    {
      "epoch": 0.8261964735516373,
      "grad_norm": 0.12573710083961487,
      "learning_rate": 8.925399550663661e-06,
      "loss": 0.1535,
      "step": 3936
    },
    {
      "epoch": 0.8264063811922754,
      "grad_norm": 0.10890112817287445,
      "learning_rate": 8.904517193992862e-06,
      "loss": 0.1465,
      "step": 3937
    },
    {
      "epoch": 0.8266162888329135,
      "grad_norm": 0.12931612133979797,
      "learning_rate": 8.883656906535042e-06,
      "loss": 0.1582,
      "step": 3938
    },
    {
      "epoch": 0.8268261964735516,
      "grad_norm": 0.12019342184066772,
      "learning_rate": 8.86281869949267e-06,
      "loss": 0.1655,
      "step": 3939
    },
    {
      "epoch": 0.8270361041141897,
      "grad_norm": 0.12204300612211227,
      "learning_rate": 8.842002584056325e-06,
      "loss": 0.1582,
      "step": 3940
    },
    {
      "epoch": 0.8272460117548279,
      "grad_norm": 0.11083068698644638,
      "learning_rate": 8.821208571404726e-06,
      "loss": 0.139,
      "step": 3941
    },
    {
      "epoch": 0.827455919395466,
      "grad_norm": 0.10661565512418747,
      "learning_rate": 8.800436672704759e-06,
      "loss": 0.1469,
      "step": 3942
    },
    {
      "epoch": 0.8276658270361041,
      "grad_norm": 0.13081884384155273,
      "learning_rate": 8.7796868991114e-06,
      "loss": 0.1551,
      "step": 3943
    },
    {
      "epoch": 0.8278757346767422,
      "grad_norm": 0.09872268885374069,
      "learning_rate": 8.758959261767758e-06,
      "loss": 0.1479,
      "step": 3944
    },
    {
      "epoch": 0.8280856423173804,
      "grad_norm": 0.11020808666944504,
      "learning_rate": 8.73825377180501e-06,
      "loss": 0.1536,
      "step": 3945
    },
    {
      "epoch": 0.8282955499580185,
      "grad_norm": 0.11102128773927689,
      "learning_rate": 8.717570440342521e-06,
      "loss": 0.1503,
      "step": 3946
    },
    {
      "epoch": 0.8285054575986566,
      "grad_norm": 0.11287751793861389,
      "learning_rate": 8.696909278487697e-06,
      "loss": 0.144,
      "step": 3947
    },
    {
      "epoch": 0.8287153652392947,
      "grad_norm": 0.13401024043560028,
      "learning_rate": 8.676270297336054e-06,
      "loss": 0.1656,
      "step": 3948
    },
    {
      "epoch": 0.8289252728799328,
      "grad_norm": 0.13312119245529175,
      "learning_rate": 8.65565350797119e-06,
      "loss": 0.1551,
      "step": 3949
    },
    {
      "epoch": 0.829135180520571,
      "grad_norm": 0.1267576813697815,
      "learning_rate": 8.63505892146479e-06,
      "loss": 0.1588,
      "step": 3950
    },
    {
      "epoch": 0.8293450881612091,
      "grad_norm": 0.11618844419717789,
      "learning_rate": 8.614486548876644e-06,
      "loss": 0.1496,
      "step": 3951
    },
    {
      "epoch": 0.8295549958018472,
      "grad_norm": 0.12536628544330597,
      "learning_rate": 8.59393640125456e-06,
      "loss": 0.1727,
      "step": 3952
    },
    {
      "epoch": 0.8297649034424853,
      "grad_norm": 0.13767768442630768,
      "learning_rate": 8.573408489634427e-06,
      "loss": 0.1791,
      "step": 3953
    },
    {
      "epoch": 0.8299748110831234,
      "grad_norm": 0.12117113918066025,
      "learning_rate": 8.552902825040233e-06,
      "loss": 0.1372,
      "step": 3954
    },
    {
      "epoch": 0.8301847187237615,
      "grad_norm": 0.1230069175362587,
      "learning_rate": 8.532419418483972e-06,
      "loss": 0.1698,
      "step": 3955
    },
    {
      "epoch": 0.8303946263643996,
      "grad_norm": 0.11761808395385742,
      "learning_rate": 8.511958280965703e-06,
      "loss": 0.1598,
      "step": 3956
    },
    {
      "epoch": 0.8306045340050378,
      "grad_norm": 0.11687527596950531,
      "learning_rate": 8.491519423473526e-06,
      "loss": 0.1521,
      "step": 3957
    },
    {
      "epoch": 0.8308144416456759,
      "grad_norm": 0.12229825556278229,
      "learning_rate": 8.47110285698357e-06,
      "loss": 0.1605,
      "step": 3958
    },
    {
      "epoch": 0.831024349286314,
      "grad_norm": 0.11104927211999893,
      "learning_rate": 8.450708592459999e-06,
      "loss": 0.1315,
      "step": 3959
    },
    {
      "epoch": 0.8312342569269522,
      "grad_norm": 0.11745012551546097,
      "learning_rate": 8.430336640855007e-06,
      "loss": 0.1548,
      "step": 3960
    },
    {
      "epoch": 0.8314441645675903,
      "grad_norm": 0.1281951665878296,
      "learning_rate": 8.409987013108784e-06,
      "loss": 0.1455,
      "step": 3961
    },
    {
      "epoch": 0.8316540722082284,
      "grad_norm": 0.11386459320783615,
      "learning_rate": 8.389659720149546e-06,
      "loss": 0.159,
      "step": 3962
    },
    {
      "epoch": 0.8318639798488665,
      "grad_norm": 0.13125190138816833,
      "learning_rate": 8.369354772893523e-06,
      "loss": 0.1468,
      "step": 3963
    },
    {
      "epoch": 0.8320738874895046,
      "grad_norm": 0.1297435313463211,
      "learning_rate": 8.349072182244939e-06,
      "loss": 0.167,
      "step": 3964
    },
    {
      "epoch": 0.8322837951301427,
      "grad_norm": 0.12241479754447937,
      "learning_rate": 8.328811959095995e-06,
      "loss": 0.1691,
      "step": 3965
    },
    {
      "epoch": 0.8324937027707808,
      "grad_norm": 0.16956380009651184,
      "learning_rate": 8.308574114326906e-06,
      "loss": 0.1632,
      "step": 3966
    },
    {
      "epoch": 0.8327036104114189,
      "grad_norm": 0.1364806890487671,
      "learning_rate": 8.288358658805845e-06,
      "loss": 0.1449,
      "step": 3967
    },
    {
      "epoch": 0.832913518052057,
      "grad_norm": 0.12498215585947037,
      "learning_rate": 8.268165603388983e-06,
      "loss": 0.1588,
      "step": 3968
    },
    {
      "epoch": 0.8331234256926953,
      "grad_norm": 0.12091932445764542,
      "learning_rate": 8.247994958920446e-06,
      "loss": 0.1536,
      "step": 3969
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.11790355294942856,
      "learning_rate": 8.22784673623232e-06,
      "loss": 0.1526,
      "step": 3970
    },
    {
      "epoch": 0.8335432409739715,
      "grad_norm": 0.12362008541822433,
      "learning_rate": 8.207720946144687e-06,
      "loss": 0.1575,
      "step": 3971
    },
    {
      "epoch": 0.8337531486146096,
      "grad_norm": 0.11366129666566849,
      "learning_rate": 8.187617599465558e-06,
      "loss": 0.15,
      "step": 3972
    },
    {
      "epoch": 0.8339630562552477,
      "grad_norm": 0.12639477849006653,
      "learning_rate": 8.167536706990858e-06,
      "loss": 0.1588,
      "step": 3973
    },
    {
      "epoch": 0.8341729638958858,
      "grad_norm": 0.10803087800741196,
      "learning_rate": 8.147478279504489e-06,
      "loss": 0.1534,
      "step": 3974
    },
    {
      "epoch": 0.8343828715365239,
      "grad_norm": 0.11064185947179794,
      "learning_rate": 8.127442327778306e-06,
      "loss": 0.1582,
      "step": 3975
    },
    {
      "epoch": 0.834592779177162,
      "grad_norm": 0.12956248223781586,
      "learning_rate": 8.10742886257207e-06,
      "loss": 0.1537,
      "step": 3976
    },
    {
      "epoch": 0.8348026868178001,
      "grad_norm": 0.1277211457490921,
      "learning_rate": 8.087437894633459e-06,
      "loss": 0.1398,
      "step": 3977
    },
    {
      "epoch": 0.8350125944584383,
      "grad_norm": 0.1317525953054428,
      "learning_rate": 8.067469434698083e-06,
      "loss": 0.1636,
      "step": 3978
    },
    {
      "epoch": 0.8352225020990764,
      "grad_norm": 0.10670098662376404,
      "learning_rate": 8.047523493489462e-06,
      "loss": 0.1457,
      "step": 3979
    },
    {
      "epoch": 0.8354324097397146,
      "grad_norm": 0.11436796188354492,
      "learning_rate": 8.027600081719017e-06,
      "loss": 0.1444,
      "step": 3980
    },
    {
      "epoch": 0.8356423173803527,
      "grad_norm": 0.10503967851400375,
      "learning_rate": 8.007699210086084e-06,
      "loss": 0.1571,
      "step": 3981
    },
    {
      "epoch": 0.8358522250209908,
      "grad_norm": 0.1279638260602951,
      "learning_rate": 7.98782088927787e-06,
      "loss": 0.1493,
      "step": 3982
    },
    {
      "epoch": 0.8360621326616289,
      "grad_norm": 0.1345963031053543,
      "learning_rate": 7.967965129969506e-06,
      "loss": 0.1509,
      "step": 3983
    },
    {
      "epoch": 0.836272040302267,
      "grad_norm": 0.12100664526224136,
      "learning_rate": 7.948131942823984e-06,
      "loss": 0.1646,
      "step": 3984
    },
    {
      "epoch": 0.8364819479429051,
      "grad_norm": 0.11500248312950134,
      "learning_rate": 7.928321338492185e-06,
      "loss": 0.1506,
      "step": 3985
    },
    {
      "epoch": 0.8366918555835432,
      "grad_norm": 0.10120087116956711,
      "learning_rate": 7.90853332761282e-06,
      "loss": 0.1467,
      "step": 3986
    },
    {
      "epoch": 0.8369017632241813,
      "grad_norm": 0.13616779446601868,
      "learning_rate": 7.888767920812545e-06,
      "loss": 0.1419,
      "step": 3987
    },
    {
      "epoch": 0.8371116708648195,
      "grad_norm": 0.20254676043987274,
      "learning_rate": 7.869025128705815e-06,
      "loss": 0.1465,
      "step": 3988
    },
    {
      "epoch": 0.8373215785054576,
      "grad_norm": 0.13580451905727386,
      "learning_rate": 7.849304961894966e-06,
      "loss": 0.1713,
      "step": 3989
    },
    {
      "epoch": 0.8375314861460957,
      "grad_norm": 0.13808149099349976,
      "learning_rate": 7.829607430970177e-06,
      "loss": 0.1477,
      "step": 3990
    },
    {
      "epoch": 0.8377413937867338,
      "grad_norm": 0.12043146789073944,
      "learning_rate": 7.809932546509463e-06,
      "loss": 0.1573,
      "step": 3991
    },
    {
      "epoch": 0.837951301427372,
      "grad_norm": 0.12428410351276398,
      "learning_rate": 7.790280319078713e-06,
      "loss": 0.1787,
      "step": 3992
    },
    {
      "epoch": 0.8381612090680101,
      "grad_norm": 0.16309167444705963,
      "learning_rate": 7.77065075923159e-06,
      "loss": 0.1482,
      "step": 3993
    },
    {
      "epoch": 0.8383711167086482,
      "grad_norm": 0.10672210156917572,
      "learning_rate": 7.751043877509623e-06,
      "loss": 0.1473,
      "step": 3994
    },
    {
      "epoch": 0.8385810243492863,
      "grad_norm": 0.11658763885498047,
      "learning_rate": 7.731459684442171e-06,
      "loss": 0.1476,
      "step": 3995
    },
    {
      "epoch": 0.8387909319899244,
      "grad_norm": 0.13712359964847565,
      "learning_rate": 7.71189819054639e-06,
      "loss": 0.1629,
      "step": 3996
    },
    {
      "epoch": 0.8390008396305626,
      "grad_norm": 0.11185979843139648,
      "learning_rate": 7.692359406327238e-06,
      "loss": 0.1694,
      "step": 3997
    },
    {
      "epoch": 0.8392107472712007,
      "grad_norm": 0.12185225635766983,
      "learning_rate": 7.672843342277498e-06,
      "loss": 0.1613,
      "step": 3998
    },
    {
      "epoch": 0.8394206549118388,
      "grad_norm": 0.1134178638458252,
      "learning_rate": 7.65335000887773e-06,
      "loss": 0.1537,
      "step": 3999
    },
    {
      "epoch": 0.8396305625524769,
      "grad_norm": 0.11603078246116638,
      "learning_rate": 7.633879416596312e-06,
      "loss": 0.1592,
      "step": 4000
    },
    {
      "epoch": 0.839840470193115,
      "grad_norm": 0.11461564153432846,
      "learning_rate": 7.6144315758893916e-06,
      "loss": 0.1722,
      "step": 4001
    },
    {
      "epoch": 0.8400503778337531,
      "grad_norm": 0.12421731650829315,
      "learning_rate": 7.595006497200907e-06,
      "loss": 0.1532,
      "step": 4002
    },
    {
      "epoch": 0.8402602854743912,
      "grad_norm": 0.12341675162315369,
      "learning_rate": 7.575604190962549e-06,
      "loss": 0.1564,
      "step": 4003
    },
    {
      "epoch": 0.8404701931150294,
      "grad_norm": 0.12224044650793076,
      "learning_rate": 7.556224667593831e-06,
      "loss": 0.1335,
      "step": 4004
    },
    {
      "epoch": 0.8406801007556675,
      "grad_norm": 0.12325417250394821,
      "learning_rate": 7.536867937501985e-06,
      "loss": 0.1567,
      "step": 4005
    },
    {
      "epoch": 0.8408900083963057,
      "grad_norm": 0.15603864192962646,
      "learning_rate": 7.517534011082017e-06,
      "loss": 0.1484,
      "step": 4006
    },
    {
      "epoch": 0.8410999160369438,
      "grad_norm": 0.1234329491853714,
      "learning_rate": 7.498222898716683e-06,
      "loss": 0.1608,
      "step": 4007
    },
    {
      "epoch": 0.8413098236775819,
      "grad_norm": 0.16258041560649872,
      "learning_rate": 7.4789346107765e-06,
      "loss": 0.149,
      "step": 4008
    },
    {
      "epoch": 0.84151973131822,
      "grad_norm": 0.10971637070178986,
      "learning_rate": 7.4596691576197145e-06,
      "loss": 0.1506,
      "step": 4009
    },
    {
      "epoch": 0.8417296389588581,
      "grad_norm": 0.1123458594083786,
      "learning_rate": 7.440426549592316e-06,
      "loss": 0.1536,
      "step": 4010
    },
    {
      "epoch": 0.8419395465994962,
      "grad_norm": 0.11952552199363708,
      "learning_rate": 7.421206797028013e-06,
      "loss": 0.1577,
      "step": 4011
    },
    {
      "epoch": 0.8421494542401343,
      "grad_norm": 0.12449429929256439,
      "learning_rate": 7.402009910248276e-06,
      "loss": 0.1518,
      "step": 4012
    },
    {
      "epoch": 0.8423593618807724,
      "grad_norm": 0.11453278362751007,
      "learning_rate": 7.382835899562263e-06,
      "loss": 0.1605,
      "step": 4013
    },
    {
      "epoch": 0.8425692695214105,
      "grad_norm": 0.10088256746530533,
      "learning_rate": 7.363684775266844e-06,
      "loss": 0.1573,
      "step": 4014
    },
    {
      "epoch": 0.8427791771620488,
      "grad_norm": 0.11597353219985962,
      "learning_rate": 7.344556547646608e-06,
      "loss": 0.1635,
      "step": 4015
    },
    {
      "epoch": 0.8429890848026869,
      "grad_norm": 0.11462254077196121,
      "learning_rate": 7.325451226973867e-06,
      "loss": 0.1471,
      "step": 4016
    },
    {
      "epoch": 0.843198992443325,
      "grad_norm": 0.10850072652101517,
      "learning_rate": 7.306368823508608e-06,
      "loss": 0.1614,
      "step": 4017
    },
    {
      "epoch": 0.8434089000839631,
      "grad_norm": 0.1240755096077919,
      "learning_rate": 7.287309347498517e-06,
      "loss": 0.1563,
      "step": 4018
    },
    {
      "epoch": 0.8436188077246012,
      "grad_norm": 0.1336059272289276,
      "learning_rate": 7.2682728091789665e-06,
      "loss": 0.1644,
      "step": 4019
    },
    {
      "epoch": 0.8438287153652393,
      "grad_norm": 0.11125699430704117,
      "learning_rate": 7.2492592187730136e-06,
      "loss": 0.1544,
      "step": 4020
    },
    {
      "epoch": 0.8440386230058774,
      "grad_norm": 0.12155328691005707,
      "learning_rate": 7.23026858649139e-06,
      "loss": 0.1489,
      "step": 4021
    },
    {
      "epoch": 0.8442485306465155,
      "grad_norm": 0.14981696009635925,
      "learning_rate": 7.211300922532505e-06,
      "loss": 0.1455,
      "step": 4022
    },
    {
      "epoch": 0.8444584382871536,
      "grad_norm": 0.11799201369285583,
      "learning_rate": 7.192356237082404e-06,
      "loss": 0.141,
      "step": 4023
    },
    {
      "epoch": 0.8446683459277917,
      "grad_norm": 0.12192690372467041,
      "learning_rate": 7.17343454031485e-06,
      "loss": 0.1511,
      "step": 4024
    },
    {
      "epoch": 0.8448782535684299,
      "grad_norm": 0.12700971961021423,
      "learning_rate": 7.154535842391208e-06,
      "loss": 0.1443,
      "step": 4025
    },
    {
      "epoch": 0.845088161209068,
      "grad_norm": 0.12561286985874176,
      "learning_rate": 7.1356601534605105e-06,
      "loss": 0.1444,
      "step": 4026
    },
    {
      "epoch": 0.8452980688497062,
      "grad_norm": 0.10910797864198685,
      "learning_rate": 7.1168074836594375e-06,
      "loss": 0.1626,
      "step": 4027
    },
    {
      "epoch": 0.8455079764903443,
      "grad_norm": 0.13664831221103668,
      "learning_rate": 7.097977843112302e-06,
      "loss": 0.164,
      "step": 4028
    },
    {
      "epoch": 0.8457178841309824,
      "grad_norm": 0.131288543343544,
      "learning_rate": 7.0791712419310545e-06,
      "loss": 0.1524,
      "step": 4029
    },
    {
      "epoch": 0.8459277917716205,
      "grad_norm": 0.14658468961715698,
      "learning_rate": 7.060387690215258e-06,
      "loss": 0.1503,
      "step": 4030
    },
    {
      "epoch": 0.8461376994122586,
      "grad_norm": 0.1412951648235321,
      "learning_rate": 7.04162719805212e-06,
      "loss": 0.1398,
      "step": 4031
    },
    {
      "epoch": 0.8463476070528967,
      "grad_norm": 0.1180790513753891,
      "learning_rate": 7.022889775516439e-06,
      "loss": 0.146,
      "step": 4032
    },
    {
      "epoch": 0.8465575146935348,
      "grad_norm": 0.1320628523826599,
      "learning_rate": 7.0041754326706625e-06,
      "loss": 0.1591,
      "step": 4033
    },
    {
      "epoch": 0.846767422334173,
      "grad_norm": 0.1340452879667282,
      "learning_rate": 6.9854841795648166e-06,
      "loss": 0.138,
      "step": 4034
    },
    {
      "epoch": 0.8469773299748111,
      "grad_norm": 0.10865829885005951,
      "learning_rate": 6.9668160262365e-06,
      "loss": 0.1525,
      "step": 4035
    },
    {
      "epoch": 0.8471872376154492,
      "grad_norm": 0.1448470950126648,
      "learning_rate": 6.948170982710972e-06,
      "loss": 0.1516,
      "step": 4036
    },
    {
      "epoch": 0.8473971452560873,
      "grad_norm": 0.13538743555545807,
      "learning_rate": 6.929549059001028e-06,
      "loss": 0.1629,
      "step": 4037
    },
    {
      "epoch": 0.8476070528967254,
      "grad_norm": 0.10407985746860504,
      "learning_rate": 6.91095026510708e-06,
      "loss": 0.1506,
      "step": 4038
    },
    {
      "epoch": 0.8478169605373636,
      "grad_norm": 0.11444611102342606,
      "learning_rate": 6.892374611017094e-06,
      "loss": 0.1356,
      "step": 4039
    },
    {
      "epoch": 0.8480268681780017,
      "grad_norm": 0.11654292047023773,
      "learning_rate": 6.873822106706612e-06,
      "loss": 0.1708,
      "step": 4040
    },
    {
      "epoch": 0.8482367758186398,
      "grad_norm": 0.11343565583229065,
      "learning_rate": 6.855292762138793e-06,
      "loss": 0.1588,
      "step": 4041
    },
    {
      "epoch": 0.8484466834592779,
      "grad_norm": 0.11720237135887146,
      "learning_rate": 6.836786587264271e-06,
      "loss": 0.1587,
      "step": 4042
    },
    {
      "epoch": 0.8486565910999161,
      "grad_norm": 0.11687421053647995,
      "learning_rate": 6.818303592021308e-06,
      "loss": 0.1495,
      "step": 4043
    },
    {
      "epoch": 0.8488664987405542,
      "grad_norm": 0.1309545636177063,
      "learning_rate": 6.79984378633568e-06,
      "loss": 0.1577,
      "step": 4044
    },
    {
      "epoch": 0.8490764063811923,
      "grad_norm": 0.13915085792541504,
      "learning_rate": 6.7814071801207464e-06,
      "loss": 0.1716,
      "step": 4045
    },
    {
      "epoch": 0.8492863140218304,
      "grad_norm": 0.1628192812204361,
      "learning_rate": 6.762993783277371e-06,
      "loss": 0.1372,
      "step": 4046
    },
    {
      "epoch": 0.8494962216624685,
      "grad_norm": 0.11110463738441467,
      "learning_rate": 6.7446036056939775e-06,
      "loss": 0.1683,
      "step": 4047
    },
    {
      "epoch": 0.8497061293031066,
      "grad_norm": 0.12355831265449524,
      "learning_rate": 6.726236657246504e-06,
      "loss": 0.1418,
      "step": 4048
    },
    {
      "epoch": 0.8499160369437447,
      "grad_norm": 0.131812185049057,
      "learning_rate": 6.707892947798422e-06,
      "loss": 0.1572,
      "step": 4049
    },
    {
      "epoch": 0.8501259445843828,
      "grad_norm": 0.14924898743629456,
      "learning_rate": 6.689572487200718e-06,
      "loss": 0.1597,
      "step": 4050
    },
    {
      "epoch": 0.850335852225021,
      "grad_norm": 0.10449478030204773,
      "learning_rate": 6.671275285291911e-06,
      "loss": 0.1476,
      "step": 4051
    },
    {
      "epoch": 0.8505457598656591,
      "grad_norm": 0.11552565544843674,
      "learning_rate": 6.653001351897991e-06,
      "loss": 0.1545,
      "step": 4052
    },
    {
      "epoch": 0.8507556675062973,
      "grad_norm": 0.11661684513092041,
      "learning_rate": 6.634750696832514e-06,
      "loss": 0.1701,
      "step": 4053
    },
    {
      "epoch": 0.8509655751469354,
      "grad_norm": 0.10988535732030869,
      "learning_rate": 6.616523329896474e-06,
      "loss": 0.1623,
      "step": 4054
    },
    {
      "epoch": 0.8511754827875735,
      "grad_norm": 0.1000858023762703,
      "learning_rate": 6.598319260878405e-06,
      "loss": 0.1403,
      "step": 4055
    },
    {
      "epoch": 0.8513853904282116,
      "grad_norm": 0.13062427937984467,
      "learning_rate": 6.580138499554267e-06,
      "loss": 0.1551,
      "step": 4056
    },
    {
      "epoch": 0.8515952980688497,
      "grad_norm": 0.11242105811834335,
      "learning_rate": 6.561981055687577e-06,
      "loss": 0.1615,
      "step": 4057
    },
    {
      "epoch": 0.8518052057094878,
      "grad_norm": 0.11550282686948776,
      "learning_rate": 6.543846939029296e-06,
      "loss": 0.1584,
      "step": 4058
    },
    {
      "epoch": 0.8520151133501259,
      "grad_norm": 0.12232505530118942,
      "learning_rate": 6.525736159317852e-06,
      "loss": 0.1653,
      "step": 4059
    },
    {
      "epoch": 0.852225020990764,
      "grad_norm": 0.11430735141038895,
      "learning_rate": 6.5076487262791415e-06,
      "loss": 0.1486,
      "step": 4060
    },
    {
      "epoch": 0.8524349286314021,
      "grad_norm": 0.11951231956481934,
      "learning_rate": 6.489584649626534e-06,
      "loss": 0.1535,
      "step": 4061
    },
    {
      "epoch": 0.8526448362720404,
      "grad_norm": 0.10858254134654999,
      "learning_rate": 6.4715439390608644e-06,
      "loss": 0.1517,
      "step": 4062
    },
    {
      "epoch": 0.8528547439126785,
      "grad_norm": 0.09623479098081589,
      "learning_rate": 6.453526604270394e-06,
      "loss": 0.1449,
      "step": 4063
    },
    {
      "epoch": 0.8530646515533166,
      "grad_norm": 0.14456307888031006,
      "learning_rate": 6.435532654930826e-06,
      "loss": 0.1549,
      "step": 4064
    },
    {
      "epoch": 0.8532745591939547,
      "grad_norm": 0.1130475103855133,
      "learning_rate": 6.417562100705354e-06,
      "loss": 0.1563,
      "step": 4065
    },
    {
      "epoch": 0.8534844668345928,
      "grad_norm": 0.11575406044721603,
      "learning_rate": 6.399614951244559e-06,
      "loss": 0.1458,
      "step": 4066
    },
    {
      "epoch": 0.8536943744752309,
      "grad_norm": 0.11347532272338867,
      "learning_rate": 6.381691216186475e-06,
      "loss": 0.1507,
      "step": 4067
    },
    {
      "epoch": 0.853904282115869,
      "grad_norm": 0.1290491372346878,
      "learning_rate": 6.363790905156558e-06,
      "loss": 0.1645,
      "step": 4068
    },
    {
      "epoch": 0.8541141897565071,
      "grad_norm": 0.10183484852313995,
      "learning_rate": 6.34591402776768e-06,
      "loss": 0.1316,
      "step": 4069
    },
    {
      "epoch": 0.8543240973971452,
      "grad_norm": 0.14135852456092834,
      "learning_rate": 6.328060593620133e-06,
      "loss": 0.171,
      "step": 4070
    },
    {
      "epoch": 0.8545340050377834,
      "grad_norm": 0.11686300486326218,
      "learning_rate": 6.310230612301621e-06,
      "loss": 0.16,
      "step": 4071
    },
    {
      "epoch": 0.8547439126784215,
      "grad_norm": 0.13613487780094147,
      "learning_rate": 6.29242409338725e-06,
      "loss": 0.1665,
      "step": 4072
    },
    {
      "epoch": 0.8549538203190596,
      "grad_norm": 0.1191660612821579,
      "learning_rate": 6.274641046439517e-06,
      "loss": 0.1693,
      "step": 4073
    },
    {
      "epoch": 0.8551637279596978,
      "grad_norm": 0.106855608522892,
      "learning_rate": 6.256881481008348e-06,
      "loss": 0.1415,
      "step": 4074
    },
    {
      "epoch": 0.8553736356003359,
      "grad_norm": 0.11627862602472305,
      "learning_rate": 6.239145406631031e-06,
      "loss": 0.1502,
      "step": 4075
    },
    {
      "epoch": 0.855583543240974,
      "grad_norm": 0.16022130846977234,
      "learning_rate": 6.221432832832208e-06,
      "loss": 0.1395,
      "step": 4076
    },
    {
      "epoch": 0.8557934508816121,
      "grad_norm": 0.11979082226753235,
      "learning_rate": 6.203743769123982e-06,
      "loss": 0.1652,
      "step": 4077
    },
    {
      "epoch": 0.8560033585222502,
      "grad_norm": 0.12621894478797913,
      "learning_rate": 6.186078225005759e-06,
      "loss": 0.1387,
      "step": 4078
    },
    {
      "epoch": 0.8562132661628883,
      "grad_norm": 0.117207370698452,
      "learning_rate": 6.168436209964346e-06,
      "loss": 0.1399,
      "step": 4079
    },
    {
      "epoch": 0.8564231738035264,
      "grad_norm": 0.13264822959899902,
      "learning_rate": 6.150817733473907e-06,
      "loss": 0.1677,
      "step": 4080
    },
    {
      "epoch": 0.8566330814441646,
      "grad_norm": 0.1263989955186844,
      "learning_rate": 6.1332228049959525e-06,
      "loss": 0.1549,
      "step": 4081
    },
    {
      "epoch": 0.8568429890848027,
      "grad_norm": 0.11813917756080627,
      "learning_rate": 6.115651433979402e-06,
      "loss": 0.1543,
      "step": 4082
    },
    {
      "epoch": 0.8570528967254408,
      "grad_norm": 0.13481532037258148,
      "learning_rate": 6.098103629860441e-06,
      "loss": 0.1667,
      "step": 4083
    },
    {
      "epoch": 0.8572628043660789,
      "grad_norm": 0.13173486292362213,
      "learning_rate": 6.080579402062659e-06,
      "loss": 0.1572,
      "step": 4084
    },
    {
      "epoch": 0.857472712006717,
      "grad_norm": 0.11830896884202957,
      "learning_rate": 6.063078759996954e-06,
      "loss": 0.1664,
      "step": 4085
    },
    {
      "epoch": 0.8576826196473551,
      "grad_norm": 0.11736118793487549,
      "learning_rate": 6.0456017130615886e-06,
      "loss": 0.1507,
      "step": 4086
    },
    {
      "epoch": 0.8578925272879933,
      "grad_norm": 0.1546947956085205,
      "learning_rate": 6.028148270642131e-06,
      "loss": 0.1592,
      "step": 4087
    },
    {
      "epoch": 0.8581024349286314,
      "grad_norm": 0.1077655777335167,
      "learning_rate": 6.0107184421114734e-06,
      "loss": 0.1524,
      "step": 4088
    },
    {
      "epoch": 0.8583123425692695,
      "grad_norm": 0.14029689133167267,
      "learning_rate": 5.9933122368298346e-06,
      "loss": 0.1617,
      "step": 4089
    },
    {
      "epoch": 0.8585222502099077,
      "grad_norm": 0.12839555740356445,
      "learning_rate": 5.975929664144747e-06,
      "loss": 0.1648,
      "step": 4090
    },
    {
      "epoch": 0.8587321578505458,
      "grad_norm": 0.11472969502210617,
      "learning_rate": 5.958570733391039e-06,
      "loss": 0.158,
      "step": 4091
    },
    {
      "epoch": 0.8589420654911839,
      "grad_norm": 0.12960311770439148,
      "learning_rate": 5.941235453890864e-06,
      "loss": 0.1501,
      "step": 4092
    },
    {
      "epoch": 0.859151973131822,
      "grad_norm": 0.12296982854604721,
      "learning_rate": 5.923923834953648e-06,
      "loss": 0.1516,
      "step": 4093
    },
    {
      "epoch": 0.8593618807724601,
      "grad_norm": 0.14688728749752045,
      "learning_rate": 5.90663588587615e-06,
      "loss": 0.1536,
      "step": 4094
    },
    {
      "epoch": 0.8595717884130982,
      "grad_norm": 0.09548702090978622,
      "learning_rate": 5.889371615942379e-06,
      "loss": 0.1465,
      "step": 4095
    },
    {
      "epoch": 0.8597816960537363,
      "grad_norm": 0.1445847600698471,
      "learning_rate": 5.87213103442365e-06,
      "loss": 0.1558,
      "step": 4096
    },
    {
      "epoch": 0.8599916036943744,
      "grad_norm": 0.14124633371829987,
      "learning_rate": 5.854914150578527e-06,
      "loss": 0.1678,
      "step": 4097
    },
    {
      "epoch": 0.8602015113350125,
      "grad_norm": 0.12228516489267349,
      "learning_rate": 5.837720973652894e-06,
      "loss": 0.1665,
      "step": 4098
    },
    {
      "epoch": 0.8604114189756508,
      "grad_norm": 0.1366337239742279,
      "learning_rate": 5.820551512879873e-06,
      "loss": 0.1619,
      "step": 4099
    },
    {
      "epoch": 0.8606213266162889,
      "grad_norm": 0.15276916325092316,
      "learning_rate": 5.803405777479854e-06,
      "loss": 0.1687,
      "step": 4100
    },
    {
      "epoch": 0.860831234256927,
      "grad_norm": 0.12131724506616592,
      "learning_rate": 5.78628377666049e-06,
      "loss": 0.1653,
      "step": 4101
    },
    {
      "epoch": 0.8610411418975651,
      "grad_norm": 0.1318751573562622,
      "learning_rate": 5.769185519616671e-06,
      "loss": 0.1502,
      "step": 4102
    },
    {
      "epoch": 0.8612510495382032,
      "grad_norm": 0.12935899198055267,
      "learning_rate": 5.752111015530593e-06,
      "loss": 0.1631,
      "step": 4103
    },
    {
      "epoch": 0.8614609571788413,
      "grad_norm": 0.11931544542312622,
      "learning_rate": 5.735060273571618e-06,
      "loss": 0.1315,
      "step": 4104
    },
    {
      "epoch": 0.8616708648194794,
      "grad_norm": 0.11439146846532822,
      "learning_rate": 5.71803330289638e-06,
      "loss": 0.1484,
      "step": 4105
    },
    {
      "epoch": 0.8618807724601175,
      "grad_norm": 0.1328824758529663,
      "learning_rate": 5.701030112648781e-06,
      "loss": 0.1601,
      "step": 4106
    },
    {
      "epoch": 0.8620906801007556,
      "grad_norm": 0.13809579610824585,
      "learning_rate": 5.684050711959904e-06,
      "loss": 0.1609,
      "step": 4107
    },
    {
      "epoch": 0.8623005877413937,
      "grad_norm": 0.13090141117572784,
      "learning_rate": 5.667095109948078e-06,
      "loss": 0.1469,
      "step": 4108
    },
    {
      "epoch": 0.862510495382032,
      "grad_norm": 0.1246684342622757,
      "learning_rate": 5.650163315718854e-06,
      "loss": 0.1417,
      "step": 4109
    },
    {
      "epoch": 0.8627204030226701,
      "grad_norm": 0.12435910850763321,
      "learning_rate": 5.633255338364979e-06,
      "loss": 0.1511,
      "step": 4110
    },
    {
      "epoch": 0.8629303106633082,
      "grad_norm": 0.12332982569932938,
      "learning_rate": 5.6163711869664345e-06,
      "loss": 0.1438,
      "step": 4111
    },
    {
      "epoch": 0.8631402183039463,
      "grad_norm": 0.12709768116474152,
      "learning_rate": 5.599510870590391e-06,
      "loss": 0.1544,
      "step": 4112
    },
    {
      "epoch": 0.8633501259445844,
      "grad_norm": 0.12950602173805237,
      "learning_rate": 5.5826743982912265e-06,
      "loss": 0.1315,
      "step": 4113
    },
    {
      "epoch": 0.8635600335852225,
      "grad_norm": 0.12958264350891113,
      "learning_rate": 5.565861779110499e-06,
      "loss": 0.1665,
      "step": 4114
    },
    {
      "epoch": 0.8637699412258606,
      "grad_norm": 0.11284958571195602,
      "learning_rate": 5.549073022076989e-06,
      "loss": 0.1583,
      "step": 4115
    },
    {
      "epoch": 0.8639798488664987,
      "grad_norm": 0.12838315963745117,
      "learning_rate": 5.532308136206632e-06,
      "loss": 0.1599,
      "step": 4116
    },
    {
      "epoch": 0.8641897565071368,
      "grad_norm": 0.11664266884326935,
      "learning_rate": 5.515567130502558e-06,
      "loss": 0.1719,
      "step": 4117
    },
    {
      "epoch": 0.864399664147775,
      "grad_norm": 0.10369646549224854,
      "learning_rate": 5.498850013955065e-06,
      "loss": 0.1469,
      "step": 4118
    },
    {
      "epoch": 0.8646095717884131,
      "grad_norm": 0.11366793513298035,
      "learning_rate": 5.482156795541632e-06,
      "loss": 0.134,
      "step": 4119
    },
    {
      "epoch": 0.8648194794290512,
      "grad_norm": 0.11349675804376602,
      "learning_rate": 5.465487484226889e-06,
      "loss": 0.154,
      "step": 4120
    },
    {
      "epoch": 0.8650293870696893,
      "grad_norm": 0.12319415807723999,
      "learning_rate": 5.448842088962647e-06,
      "loss": 0.1545,
      "step": 4121
    },
    {
      "epoch": 0.8652392947103275,
      "grad_norm": 0.11100887507200241,
      "learning_rate": 5.4322206186878495e-06,
      "loss": 0.1574,
      "step": 4122
    },
    {
      "epoch": 0.8654492023509656,
      "grad_norm": 0.10524474829435349,
      "learning_rate": 5.415623082328625e-06,
      "loss": 0.1418,
      "step": 4123
    },
    {
      "epoch": 0.8656591099916037,
      "grad_norm": 0.1143570989370346,
      "learning_rate": 5.399049488798225e-06,
      "loss": 0.164,
      "step": 4124
    },
    {
      "epoch": 0.8658690176322418,
      "grad_norm": 0.12815353274345398,
      "learning_rate": 5.382499846997041e-06,
      "loss": 0.1611,
      "step": 4125
    },
    {
      "epoch": 0.8660789252728799,
      "grad_norm": 0.11535675823688507,
      "learning_rate": 5.365974165812599e-06,
      "loss": 0.1447,
      "step": 4126
    },
    {
      "epoch": 0.8662888329135181,
      "grad_norm": 0.13475079834461212,
      "learning_rate": 5.349472454119586e-06,
      "loss": 0.1869,
      "step": 4127
    },
    {
      "epoch": 0.8664987405541562,
      "grad_norm": 0.1347225308418274,
      "learning_rate": 5.332994720779794e-06,
      "loss": 0.158,
      "step": 4128
    },
    {
      "epoch": 0.8667086481947943,
      "grad_norm": 0.12452715635299683,
      "learning_rate": 5.316540974642137e-06,
      "loss": 0.1515,
      "step": 4129
    },
    {
      "epoch": 0.8669185558354324,
      "grad_norm": 0.13022229075431824,
      "learning_rate": 5.300111224542648e-06,
      "loss": 0.1605,
      "step": 4130
    },
    {
      "epoch": 0.8671284634760705,
      "grad_norm": 0.1267905980348587,
      "learning_rate": 5.2837054793045016e-06,
      "loss": 0.1723,
      "step": 4131
    },
    {
      "epoch": 0.8673383711167086,
      "grad_norm": 0.13651971518993378,
      "learning_rate": 5.2673237477379286e-06,
      "loss": 0.1691,
      "step": 4132
    },
    {
      "epoch": 0.8675482787573467,
      "grad_norm": 0.12502793967723846,
      "learning_rate": 5.250966038640315e-06,
      "loss": 0.1622,
      "step": 4133
    },
    {
      "epoch": 0.8677581863979849,
      "grad_norm": 0.12554888427257538,
      "learning_rate": 5.234632360796099e-06,
      "loss": 0.1553,
      "step": 4134
    },
    {
      "epoch": 0.867968094038623,
      "grad_norm": 0.11070720851421356,
      "learning_rate": 5.218322722976865e-06,
      "loss": 0.1463,
      "step": 4135
    },
    {
      "epoch": 0.8681780016792612,
      "grad_norm": 0.15806786715984344,
      "learning_rate": 5.202037133941251e-06,
      "loss": 0.1615,
      "step": 4136
    },
    {
      "epoch": 0.8683879093198993,
      "grad_norm": 0.11458860337734222,
      "learning_rate": 5.185775602434989e-06,
      "loss": 0.16,
      "step": 4137
    },
    {
      "epoch": 0.8685978169605374,
      "grad_norm": 0.12466058880090714,
      "learning_rate": 5.169538137190893e-06,
      "loss": 0.15,
      "step": 4138
    },
    {
      "epoch": 0.8688077246011755,
      "grad_norm": 0.11807677894830704,
      "learning_rate": 5.1533247469288555e-06,
      "loss": 0.1686,
      "step": 4139
    },
    {
      "epoch": 0.8690176322418136,
      "grad_norm": 0.1190905049443245,
      "learning_rate": 5.137135440355834e-06,
      "loss": 0.1494,
      "step": 4140
    },
    {
      "epoch": 0.8692275398824517,
      "grad_norm": 0.12761694192886353,
      "learning_rate": 5.120970226165861e-06,
      "loss": 0.1424,
      "step": 4141
    },
    {
      "epoch": 0.8694374475230898,
      "grad_norm": 0.13342903554439545,
      "learning_rate": 5.104829113040016e-06,
      "loss": 0.1581,
      "step": 4142
    },
    {
      "epoch": 0.8696473551637279,
      "grad_norm": 0.10999980568885803,
      "learning_rate": 5.088712109646449e-06,
      "loss": 0.1511,
      "step": 4143
    },
    {
      "epoch": 0.869857262804366,
      "grad_norm": 0.12726296484470367,
      "learning_rate": 5.072619224640363e-06,
      "loss": 0.1562,
      "step": 4144
    },
    {
      "epoch": 0.8700671704450041,
      "grad_norm": 0.11894024908542633,
      "learning_rate": 5.056550466664018e-06,
      "loss": 0.1649,
      "step": 4145
    },
    {
      "epoch": 0.8702770780856424,
      "grad_norm": 0.11655131727457047,
      "learning_rate": 5.040505844346671e-06,
      "loss": 0.1537,
      "step": 4146
    },
    {
      "epoch": 0.8704869857262805,
      "grad_norm": 0.14043211936950684,
      "learning_rate": 5.0244853663046765e-06,
      "loss": 0.15,
      "step": 4147
    },
    {
      "epoch": 0.8706968933669186,
      "grad_norm": 0.12577585875988007,
      "learning_rate": 5.00848904114139e-06,
      "loss": 0.1608,
      "step": 4148
    },
    {
      "epoch": 0.8709068010075567,
      "grad_norm": 0.1366935670375824,
      "learning_rate": 4.9925168774472065e-06,
      "loss": 0.162,
      "step": 4149
    },
    {
      "epoch": 0.8711167086481948,
      "grad_norm": 0.139494389295578,
      "learning_rate": 4.976568883799537e-06,
      "loss": 0.1458,
      "step": 4150
    },
    {
      "epoch": 0.8713266162888329,
      "grad_norm": 0.1221657320857048,
      "learning_rate": 4.960645068762809e-06,
      "loss": 0.1635,
      "step": 4151
    },
    {
      "epoch": 0.871536523929471,
      "grad_norm": 0.10618814080953598,
      "learning_rate": 4.9447454408885095e-06,
      "loss": 0.1581,
      "step": 4152
    },
    {
      "epoch": 0.8717464315701091,
      "grad_norm": 0.13304735720157623,
      "learning_rate": 4.9288700087150694e-06,
      "loss": 0.1416,
      "step": 4153
    },
    {
      "epoch": 0.8719563392107472,
      "grad_norm": 0.1176249235868454,
      "learning_rate": 4.913018780767969e-06,
      "loss": 0.1449,
      "step": 4154
    },
    {
      "epoch": 0.8721662468513854,
      "grad_norm": 0.11962371319532394,
      "learning_rate": 4.897191765559667e-06,
      "loss": 0.1606,
      "step": 4155
    },
    {
      "epoch": 0.8723761544920235,
      "grad_norm": 0.12428487837314606,
      "learning_rate": 4.881388971589656e-06,
      "loss": 0.156,
      "step": 4156
    },
    {
      "epoch": 0.8725860621326617,
      "grad_norm": 0.18144725263118744,
      "learning_rate": 4.865610407344384e-06,
      "loss": 0.1437,
      "step": 4157
    },
    {
      "epoch": 0.8727959697732998,
      "grad_norm": 0.14240662753582,
      "learning_rate": 4.8498560812973045e-06,
      "loss": 0.1378,
      "step": 4158
    },
    {
      "epoch": 0.8730058774139379,
      "grad_norm": 0.13271981477737427,
      "learning_rate": 4.834126001908845e-06,
      "loss": 0.1613,
      "step": 4159
    },
    {
      "epoch": 0.873215785054576,
      "grad_norm": 0.119957834482193,
      "learning_rate": 4.818420177626421e-06,
      "loss": 0.1441,
      "step": 4160
    },
    {
      "epoch": 0.8734256926952141,
      "grad_norm": 0.13145653903484344,
      "learning_rate": 4.802738616884417e-06,
      "loss": 0.1557,
      "step": 4161
    },
    {
      "epoch": 0.8736356003358522,
      "grad_norm": 0.13927792012691498,
      "learning_rate": 4.787081328104193e-06,
      "loss": 0.1488,
      "step": 4162
    },
    {
      "epoch": 0.8738455079764903,
      "grad_norm": 0.1195240393280983,
      "learning_rate": 4.771448319694061e-06,
      "loss": 0.1436,
      "step": 4163
    },
    {
      "epoch": 0.8740554156171285,
      "grad_norm": 0.11769021302461624,
      "learning_rate": 4.755839600049317e-06,
      "loss": 0.1546,
      "step": 4164
    },
    {
      "epoch": 0.8742653232577666,
      "grad_norm": 0.13738517463207245,
      "learning_rate": 4.7402551775522e-06,
      "loss": 0.1571,
      "step": 4165
    },
    {
      "epoch": 0.8744752308984047,
      "grad_norm": 0.13803741335868835,
      "learning_rate": 4.724695060571888e-06,
      "loss": 0.1429,
      "step": 4166
    },
    {
      "epoch": 0.8746851385390428,
      "grad_norm": 0.12481208890676498,
      "learning_rate": 4.7091592574645184e-06,
      "loss": 0.1498,
      "step": 4167
    },
    {
      "epoch": 0.874895046179681,
      "grad_norm": 0.10957111418247223,
      "learning_rate": 4.6936477765731905e-06,
      "loss": 0.1439,
      "step": 4168
    },
    {
      "epoch": 0.875104953820319,
      "grad_norm": 0.11496666073799133,
      "learning_rate": 4.6781606262279066e-06,
      "loss": 0.1538,
      "step": 4169
    },
    {
      "epoch": 0.8753148614609572,
      "grad_norm": 0.1191176176071167,
      "learning_rate": 4.662697814745631e-06,
      "loss": 0.1374,
      "step": 4170
    },
    {
      "epoch": 0.8755247691015953,
      "grad_norm": 0.14217902719974518,
      "learning_rate": 4.647259350430233e-06,
      "loss": 0.1665,
      "step": 4171
    },
    {
      "epoch": 0.8757346767422334,
      "grad_norm": 0.12614914774894714,
      "learning_rate": 4.631845241572552e-06,
      "loss": 0.1713,
      "step": 4172
    },
    {
      "epoch": 0.8759445843828715,
      "grad_norm": 0.14050200581550598,
      "learning_rate": 4.616455496450279e-06,
      "loss": 0.1798,
      "step": 4173
    },
    {
      "epoch": 0.8761544920235097,
      "grad_norm": 0.10861799865961075,
      "learning_rate": 4.601090123328078e-06,
      "loss": 0.1637,
      "step": 4174
    },
    {
      "epoch": 0.8763643996641478,
      "grad_norm": 0.12950311601161957,
      "learning_rate": 4.585749130457495e-06,
      "loss": 0.1709,
      "step": 4175
    },
    {
      "epoch": 0.8765743073047859,
      "grad_norm": 0.13461849093437195,
      "learning_rate": 4.570432526077006e-06,
      "loss": 0.157,
      "step": 4176
    },
    {
      "epoch": 0.876784214945424,
      "grad_norm": 0.1331222951412201,
      "learning_rate": 4.555140318411971e-06,
      "loss": 0.1447,
      "step": 4177
    },
    {
      "epoch": 0.8769941225860621,
      "grad_norm": 0.12269078940153122,
      "learning_rate": 4.539872515674659e-06,
      "loss": 0.1413,
      "step": 4178
    },
    {
      "epoch": 0.8772040302267002,
      "grad_norm": 0.11593735963106155,
      "learning_rate": 4.524629126064223e-06,
      "loss": 0.1412,
      "step": 4179
    },
    {
      "epoch": 0.8774139378673383,
      "grad_norm": 0.12780681252479553,
      "learning_rate": 4.509410157766714e-06,
      "loss": 0.1514,
      "step": 4180
    },
    {
      "epoch": 0.8776238455079765,
      "grad_norm": 0.12802265584468842,
      "learning_rate": 4.494215618955061e-06,
      "loss": 0.1486,
      "step": 4181
    },
    {
      "epoch": 0.8778337531486146,
      "grad_norm": 0.11089033633470535,
      "learning_rate": 4.479045517789088e-06,
      "loss": 0.156,
      "step": 4182
    },
    {
      "epoch": 0.8780436607892528,
      "grad_norm": 0.15606744587421417,
      "learning_rate": 4.463899862415483e-06,
      "loss": 0.1572,
      "step": 4183
    },
    {
      "epoch": 0.8782535684298909,
      "grad_norm": 0.11626232415437698,
      "learning_rate": 4.448778660967795e-06,
      "loss": 0.1608,
      "step": 4184
    },
    {
      "epoch": 0.878463476070529,
      "grad_norm": 0.13428574800491333,
      "learning_rate": 4.433681921566474e-06,
      "loss": 0.1445,
      "step": 4185
    },
    {
      "epoch": 0.8786733837111671,
      "grad_norm": 0.11833200603723526,
      "learning_rate": 4.418609652318817e-06,
      "loss": 0.1372,
      "step": 4186
    },
    {
      "epoch": 0.8788832913518052,
      "grad_norm": 0.12063129991292953,
      "learning_rate": 4.4035618613189444e-06,
      "loss": 0.1723,
      "step": 4187
    },
    {
      "epoch": 0.8790931989924433,
      "grad_norm": 0.11937091499567032,
      "learning_rate": 4.388538556647897e-06,
      "loss": 0.135,
      "step": 4188
    },
    {
      "epoch": 0.8793031066330814,
      "grad_norm": 0.1174740195274353,
      "learning_rate": 4.3735397463735175e-06,
      "loss": 0.1317,
      "step": 4189
    },
    {
      "epoch": 0.8795130142737195,
      "grad_norm": 0.1225159540772438,
      "learning_rate": 4.358565438550516e-06,
      "loss": 0.162,
      "step": 4190
    },
    {
      "epoch": 0.8797229219143576,
      "grad_norm": 0.11938232183456421,
      "learning_rate": 4.343615641220433e-06,
      "loss": 0.1599,
      "step": 4191
    },
    {
      "epoch": 0.8799328295549959,
      "grad_norm": 0.10696382075548172,
      "learning_rate": 4.328690362411647e-06,
      "loss": 0.148,
      "step": 4192
    },
    {
      "epoch": 0.880142737195634,
      "grad_norm": 0.15508660674095154,
      "learning_rate": 4.313789610139396e-06,
      "loss": 0.1693,
      "step": 4193
    },
    {
      "epoch": 0.8803526448362721,
      "grad_norm": 0.12078120559453964,
      "learning_rate": 4.298913392405702e-06,
      "loss": 0.1331,
      "step": 4194
    },
    {
      "epoch": 0.8805625524769102,
      "grad_norm": 0.1250450164079666,
      "learning_rate": 4.284061717199433e-06,
      "loss": 0.1589,
      "step": 4195
    },
    {
      "epoch": 0.8807724601175483,
      "grad_norm": 0.13361555337905884,
      "learning_rate": 4.26923459249628e-06,
      "loss": 0.1607,
      "step": 4196
    },
    {
      "epoch": 0.8809823677581864,
      "grad_norm": 0.13139186799526215,
      "learning_rate": 4.254432026258759e-06,
      "loss": 0.1457,
      "step": 4197
    },
    {
      "epoch": 0.8811922753988245,
      "grad_norm": 0.11095529049634933,
      "learning_rate": 4.23965402643618e-06,
      "loss": 0.1487,
      "step": 4198
    },
    {
      "epoch": 0.8814021830394626,
      "grad_norm": 0.11550076305866241,
      "learning_rate": 4.224900600964665e-06,
      "loss": 0.1418,
      "step": 4199
    },
    {
      "epoch": 0.8816120906801007,
      "grad_norm": 0.11938626319169998,
      "learning_rate": 4.210171757767145e-06,
      "loss": 0.1607,
      "step": 4200
    },
    {
      "epoch": 0.8818219983207388,
      "grad_norm": 0.11363077908754349,
      "learning_rate": 4.1954675047533365e-06,
      "loss": 0.1495,
      "step": 4201
    },
    {
      "epoch": 0.882031905961377,
      "grad_norm": 0.1299435794353485,
      "learning_rate": 4.180787849819767e-06,
      "loss": 0.1479,
      "step": 4202
    },
    {
      "epoch": 0.8822418136020151,
      "grad_norm": 0.11774297058582306,
      "learning_rate": 4.166132800849743e-06,
      "loss": 0.1397,
      "step": 4203
    },
    {
      "epoch": 0.8824517212426533,
      "grad_norm": 0.11965102702379227,
      "learning_rate": 4.151502365713356e-06,
      "loss": 0.157,
      "step": 4204
    },
    {
      "epoch": 0.8826616288832914,
      "grad_norm": 0.1356746107339859,
      "learning_rate": 4.136896552267494e-06,
      "loss": 0.1574,
      "step": 4205
    },
    {
      "epoch": 0.8828715365239295,
      "grad_norm": 0.12477898597717285,
      "learning_rate": 4.122315368355812e-06,
      "loss": 0.1287,
      "step": 4206
    },
    {
      "epoch": 0.8830814441645676,
      "grad_norm": 0.12956759333610535,
      "learning_rate": 4.107758821808738e-06,
      "loss": 0.1412,
      "step": 4207
    },
    {
      "epoch": 0.8832913518052057,
      "grad_norm": 0.12473218142986298,
      "learning_rate": 4.093226920443449e-06,
      "loss": 0.1597,
      "step": 4208
    },
    {
      "epoch": 0.8835012594458438,
      "grad_norm": 0.12124751508235931,
      "learning_rate": 4.0787196720639355e-06,
      "loss": 0.1483,
      "step": 4209
    },
    {
      "epoch": 0.8837111670864819,
      "grad_norm": 0.12192367017269135,
      "learning_rate": 4.064237084460915e-06,
      "loss": 0.1628,
      "step": 4210
    },
    {
      "epoch": 0.8839210747271201,
      "grad_norm": 0.113349050283432,
      "learning_rate": 4.04977916541186e-06,
      "loss": 0.137,
      "step": 4211
    },
    {
      "epoch": 0.8841309823677582,
      "grad_norm": 0.12009799480438232,
      "learning_rate": 4.035345922681e-06,
      "loss": 0.1451,
      "step": 4212
    },
    {
      "epoch": 0.8843408900083963,
      "grad_norm": 0.13234446942806244,
      "learning_rate": 4.020937364019328e-06,
      "loss": 0.1671,
      "step": 4213
    },
    {
      "epoch": 0.8845507976490344,
      "grad_norm": 0.11510760337114334,
      "learning_rate": 4.006553497164578e-06,
      "loss": 0.144,
      "step": 4214
    },
    {
      "epoch": 0.8847607052896725,
      "grad_norm": 0.1241830363869667,
      "learning_rate": 3.992194329841192e-06,
      "loss": 0.1509,
      "step": 4215
    },
    {
      "epoch": 0.8849706129303107,
      "grad_norm": 0.12140922993421555,
      "learning_rate": 3.977859869760375e-06,
      "loss": 0.1665,
      "step": 4216
    },
    {
      "epoch": 0.8851805205709488,
      "grad_norm": 0.11366656422615051,
      "learning_rate": 3.963550124620069e-06,
      "loss": 0.1619,
      "step": 4217
    },
    {
      "epoch": 0.8853904282115869,
      "grad_norm": 0.1194479912519455,
      "learning_rate": 3.949265102104932e-06,
      "loss": 0.1564,
      "step": 4218
    },
    {
      "epoch": 0.885600335852225,
      "grad_norm": 0.11060012876987457,
      "learning_rate": 3.935004809886344e-06,
      "loss": 0.1634,
      "step": 4219
    },
    {
      "epoch": 0.8858102434928632,
      "grad_norm": 0.12809467315673828,
      "learning_rate": 3.920769255622409e-06,
      "loss": 0.1583,
      "step": 4220
    },
    {
      "epoch": 0.8860201511335013,
      "grad_norm": 0.12603220343589783,
      "learning_rate": 3.906558446957947e-06,
      "loss": 0.1392,
      "step": 4221
    },
    {
      "epoch": 0.8862300587741394,
      "grad_norm": 0.19363883137702942,
      "learning_rate": 3.892372391524479e-06,
      "loss": 0.1511,
      "step": 4222
    },
    {
      "epoch": 0.8864399664147775,
      "grad_norm": 0.12570086121559143,
      "learning_rate": 3.878211096940254e-06,
      "loss": 0.1607,
      "step": 4223
    },
    {
      "epoch": 0.8866498740554156,
      "grad_norm": 0.17181172966957092,
      "learning_rate": 3.8640745708101996e-06,
      "loss": 0.1587,
      "step": 4224
    },
    {
      "epoch": 0.8868597816960537,
      "grad_norm": 0.11868228018283844,
      "learning_rate": 3.849962820725949e-06,
      "loss": 0.163,
      "step": 4225
    },
    {
      "epoch": 0.8870696893366918,
      "grad_norm": 0.12187950313091278,
      "learning_rate": 3.835875854265847e-06,
      "loss": 0.1495,
      "step": 4226
    },
    {
      "epoch": 0.8872795969773299,
      "grad_norm": 0.12567535042762756,
      "learning_rate": 3.8218136789949175e-06,
      "loss": 0.1573,
      "step": 4227
    },
    {
      "epoch": 0.887489504617968,
      "grad_norm": 0.1315799057483673,
      "learning_rate": 3.8077763024648606e-06,
      "loss": 0.1643,
      "step": 4228
    },
    {
      "epoch": 0.8876994122586063,
      "grad_norm": 0.11614145338535309,
      "learning_rate": 3.793763732214073e-06,
      "loss": 0.1656,
      "step": 4229
    },
    {
      "epoch": 0.8879093198992444,
      "grad_norm": 0.13281457126140594,
      "learning_rate": 3.779775975767619e-06,
      "loss": 0.1591,
      "step": 4230
    },
    {
      "epoch": 0.8881192275398825,
      "grad_norm": 0.10211368650197983,
      "learning_rate": 3.7658130406372503e-06,
      "loss": 0.1517,
      "step": 4231
    },
    {
      "epoch": 0.8883291351805206,
      "grad_norm": 0.10663280636072159,
      "learning_rate": 3.751874934321381e-06,
      "loss": 0.1539,
      "step": 4232
    },
    {
      "epoch": 0.8885390428211587,
      "grad_norm": 0.12947557866573334,
      "learning_rate": 3.7379616643050842e-06,
      "loss": 0.157,
      "step": 4233
    },
    {
      "epoch": 0.8887489504617968,
      "grad_norm": 0.10921616107225418,
      "learning_rate": 3.7240732380601162e-06,
      "loss": 0.1516,
      "step": 4234
    },
    {
      "epoch": 0.8889588581024349,
      "grad_norm": 0.1422072798013687,
      "learning_rate": 3.7102096630448835e-06,
      "loss": 0.1566,
      "step": 4235
    },
    {
      "epoch": 0.889168765743073,
      "grad_norm": 0.19969822466373444,
      "learning_rate": 3.6963709467044194e-06,
      "loss": 0.1541,
      "step": 4236
    },
    {
      "epoch": 0.8893786733837111,
      "grad_norm": 0.11462516337633133,
      "learning_rate": 3.6825570964704403e-06,
      "loss": 0.1422,
      "step": 4237
    },
    {
      "epoch": 0.8895885810243492,
      "grad_norm": 0.13612455129623413,
      "learning_rate": 3.6687681197613124e-06,
      "loss": 0.1509,
      "step": 4238
    },
    {
      "epoch": 0.8897984886649875,
      "grad_norm": 0.1351628452539444,
      "learning_rate": 3.655004023982017e-06,
      "loss": 0.1414,
      "step": 4239
    },
    {
      "epoch": 0.8900083963056256,
      "grad_norm": 0.13770541548728943,
      "learning_rate": 3.6412648165241923e-06,
      "loss": 0.1422,
      "step": 4240
    },
    {
      "epoch": 0.8902183039462637,
      "grad_norm": 0.13861404359340668,
      "learning_rate": 3.627550504766103e-06,
      "loss": 0.1543,
      "step": 4241
    },
    {
      "epoch": 0.8904282115869018,
      "grad_norm": 0.13922984898090363,
      "learning_rate": 3.6138610960726625e-06,
      "loss": 0.1676,
      "step": 4242
    },
    {
      "epoch": 0.8906381192275399,
      "grad_norm": 0.13794945180416107,
      "learning_rate": 3.60019659779538e-06,
      "loss": 0.1644,
      "step": 4243
    },
    {
      "epoch": 0.890848026868178,
      "grad_norm": 0.11094696819782257,
      "learning_rate": 3.586557017272413e-06,
      "loss": 0.1504,
      "step": 4244
    },
    {
      "epoch": 0.8910579345088161,
      "grad_norm": 0.12312065064907074,
      "learning_rate": 3.5729423618285074e-06,
      "loss": 0.1493,
      "step": 4245
    },
    {
      "epoch": 0.8912678421494542,
      "grad_norm": 0.12740106880664825,
      "learning_rate": 3.559352638775071e-06,
      "loss": 0.1575,
      "step": 4246
    },
    {
      "epoch": 0.8914777497900923,
      "grad_norm": 0.11770659685134888,
      "learning_rate": 3.545787855410082e-06,
      "loss": 0.152,
      "step": 4247
    },
    {
      "epoch": 0.8916876574307305,
      "grad_norm": 0.1185218021273613,
      "learning_rate": 3.5322480190181473e-06,
      "loss": 0.1632,
      "step": 4248
    },
    {
      "epoch": 0.8918975650713686,
      "grad_norm": 0.12321362644433975,
      "learning_rate": 3.5187331368704557e-06,
      "loss": 0.1471,
      "step": 4249
    },
    {
      "epoch": 0.8921074727120067,
      "grad_norm": 0.11762972176074982,
      "learning_rate": 3.505243216224818e-06,
      "loss": 0.1572,
      "step": 4250
    },
    {
      "epoch": 0.8923173803526449,
      "grad_norm": 0.10275141894817352,
      "learning_rate": 3.4917782643256237e-06,
      "loss": 0.1376,
      "step": 4251
    },
    {
      "epoch": 0.892527287993283,
      "grad_norm": 0.13671152293682098,
      "learning_rate": 3.478338288403865e-06,
      "loss": 0.1674,
      "step": 4252
    },
    {
      "epoch": 0.8927371956339211,
      "grad_norm": 0.13780957460403442,
      "learning_rate": 3.464923295677103e-06,
      "loss": 0.1413,
      "step": 4253
    },
    {
      "epoch": 0.8929471032745592,
      "grad_norm": 0.130636528134346,
      "learning_rate": 3.451533293349507e-06,
      "loss": 0.1513,
      "step": 4254
    },
    {
      "epoch": 0.8931570109151973,
      "grad_norm": 0.10833965986967087,
      "learning_rate": 3.438168288611826e-06,
      "loss": 0.1485,
      "step": 4255
    },
    {
      "epoch": 0.8933669185558354,
      "grad_norm": 0.12039763480424881,
      "learning_rate": 3.4248282886413517e-06,
      "loss": 0.1308,
      "step": 4256
    },
    {
      "epoch": 0.8935768261964736,
      "grad_norm": 0.12278316169977188,
      "learning_rate": 3.411513300601965e-06,
      "loss": 0.1373,
      "step": 4257
    },
    {
      "epoch": 0.8937867338371117,
      "grad_norm": 0.1298147439956665,
      "learning_rate": 3.39822333164414e-06,
      "loss": 0.1633,
      "step": 4258
    },
    {
      "epoch": 0.8939966414777498,
      "grad_norm": 0.11132081598043442,
      "learning_rate": 3.3849583889048853e-06,
      "loss": 0.1608,
      "step": 4259
    },
    {
      "epoch": 0.8942065491183879,
      "grad_norm": 0.12436234951019287,
      "learning_rate": 3.371718479507774e-06,
      "loss": 0.1603,
      "step": 4260
    },
    {
      "epoch": 0.894416456759026,
      "grad_norm": 0.12160570919513702,
      "learning_rate": 3.358503610562941e-06,
      "loss": 0.1515,
      "step": 4261
    },
    {
      "epoch": 0.8946263643996641,
      "grad_norm": 0.10866683721542358,
      "learning_rate": 3.3453137891670704e-06,
      "loss": 0.1364,
      "step": 4262
    },
    {
      "epoch": 0.8948362720403022,
      "grad_norm": 0.11041587591171265,
      "learning_rate": 3.3321490224034024e-06,
      "loss": 0.1435,
      "step": 4263
    },
    {
      "epoch": 0.8950461796809404,
      "grad_norm": 0.1161423921585083,
      "learning_rate": 3.3190093173417203e-06,
      "loss": 0.1515,
      "step": 4264
    },
    {
      "epoch": 0.8952560873215785,
      "grad_norm": 0.130842387676239,
      "learning_rate": 3.3058946810383406e-06,
      "loss": 0.161,
      "step": 4265
    },
    {
      "epoch": 0.8954659949622166,
      "grad_norm": 0.15605497360229492,
      "learning_rate": 3.2928051205361134e-06,
      "loss": 0.133,
      "step": 4266
    },
    {
      "epoch": 0.8956759026028548,
      "grad_norm": 0.10918981581926346,
      "learning_rate": 3.2797406428644483e-06,
      "loss": 0.1506,
      "step": 4267
    },
    {
      "epoch": 0.8958858102434929,
      "grad_norm": 0.11886347085237503,
      "learning_rate": 3.266701255039267e-06,
      "loss": 0.1506,
      "step": 4268
    },
    {
      "epoch": 0.896095717884131,
      "grad_norm": 0.12514930963516235,
      "learning_rate": 3.2536869640630074e-06,
      "loss": 0.1512,
      "step": 4269
    },
    {
      "epoch": 0.8963056255247691,
      "grad_norm": 0.14683309197425842,
      "learning_rate": 3.2406977769246505e-06,
      "loss": 0.1536,
      "step": 4270
    },
    {
      "epoch": 0.8965155331654072,
      "grad_norm": 0.13536719977855682,
      "learning_rate": 3.227733700599683e-06,
      "loss": 0.162,
      "step": 4271
    },
    {
      "epoch": 0.8967254408060453,
      "grad_norm": 0.11566172540187836,
      "learning_rate": 3.214794742050109e-06,
      "loss": 0.1523,
      "step": 4272
    },
    {
      "epoch": 0.8969353484466834,
      "grad_norm": 0.13411453366279602,
      "learning_rate": 3.201880908224447e-06,
      "loss": 0.1476,
      "step": 4273
    },
    {
      "epoch": 0.8971452560873215,
      "grad_norm": 0.1259935200214386,
      "learning_rate": 3.1889922060577115e-06,
      "loss": 0.1567,
      "step": 4274
    },
    {
      "epoch": 0.8973551637279596,
      "grad_norm": 0.10426774621009827,
      "learning_rate": 3.176128642471443e-06,
      "loss": 0.1411,
      "step": 4275
    },
    {
      "epoch": 0.8975650713685979,
      "grad_norm": 0.11820261180400848,
      "learning_rate": 3.163290224373672e-06,
      "loss": 0.1548,
      "step": 4276
    },
    {
      "epoch": 0.897774979009236,
      "grad_norm": 0.12722282111644745,
      "learning_rate": 3.15047695865891e-06,
      "loss": 0.1411,
      "step": 4277
    },
    {
      "epoch": 0.8979848866498741,
      "grad_norm": 0.11995641887187958,
      "learning_rate": 3.137688852208165e-06,
      "loss": 0.1528,
      "step": 4278
    },
    {
      "epoch": 0.8981947942905122,
      "grad_norm": 0.1154097393155098,
      "learning_rate": 3.124925911888971e-06,
      "loss": 0.1593,
      "step": 4279
    },
    {
      "epoch": 0.8984047019311503,
      "grad_norm": 0.15159837901592255,
      "learning_rate": 3.1121881445552947e-06,
      "loss": 0.1753,
      "step": 4280
    },
    {
      "epoch": 0.8986146095717884,
      "grad_norm": 0.11547110974788666,
      "learning_rate": 3.099475557047621e-06,
      "loss": 0.1609,
      "step": 4281
    },
    {
      "epoch": 0.8988245172124265,
      "grad_norm": 0.13358214497566223,
      "learning_rate": 3.0867881561928826e-06,
      "loss": 0.1488,
      "step": 4282
    },
    {
      "epoch": 0.8990344248530646,
      "grad_norm": 0.14517351984977722,
      "learning_rate": 3.0741259488045293e-06,
      "loss": 0.1549,
      "step": 4283
    },
    {
      "epoch": 0.8992443324937027,
      "grad_norm": 0.14740052819252014,
      "learning_rate": 3.061488941682439e-06,
      "loss": 0.1639,
      "step": 4284
    },
    {
      "epoch": 0.8994542401343409,
      "grad_norm": 0.11952229589223862,
      "learning_rate": 3.0488771416129767e-06,
      "loss": 0.1698,
      "step": 4285
    },
    {
      "epoch": 0.899664147774979,
      "grad_norm": 0.12098808586597443,
      "learning_rate": 3.0362905553689538e-06,
      "loss": 0.162,
      "step": 4286
    },
    {
      "epoch": 0.8998740554156172,
      "grad_norm": 0.11084499210119247,
      "learning_rate": 3.0237291897096788e-06,
      "loss": 0.1532,
      "step": 4287
    },
    {
      "epoch": 0.9000839630562553,
      "grad_norm": 0.12836486101150513,
      "learning_rate": 3.0111930513808785e-06,
      "loss": 0.1651,
      "step": 4288
    },
    {
      "epoch": 0.9002938706968934,
      "grad_norm": 0.14764492213726044,
      "learning_rate": 2.998682147114751e-06,
      "loss": 0.178,
      "step": 4289
    },
    {
      "epoch": 0.9005037783375315,
      "grad_norm": 0.1267375499010086,
      "learning_rate": 2.986196483629933e-06,
      "loss": 0.14,
      "step": 4290
    },
    {
      "epoch": 0.9007136859781696,
      "grad_norm": 0.1228199303150177,
      "learning_rate": 2.9737360676315106e-06,
      "loss": 0.1696,
      "step": 4291
    },
    {
      "epoch": 0.9009235936188077,
      "grad_norm": 0.1166803389787674,
      "learning_rate": 2.961300905811015e-06,
      "loss": 0.1489,
      "step": 4292
    },
    {
      "epoch": 0.9011335012594458,
      "grad_norm": 0.14002828299999237,
      "learning_rate": 2.94889100484641e-06,
      "loss": 0.168,
      "step": 4293
    },
    {
      "epoch": 0.9013434089000839,
      "grad_norm": 0.14958305656909943,
      "learning_rate": 2.9365063714020866e-06,
      "loss": 0.1609,
      "step": 4294
    },
    {
      "epoch": 0.9015533165407221,
      "grad_norm": 0.1365419179201126,
      "learning_rate": 2.924147012128897e-06,
      "loss": 0.1389,
      "step": 4295
    },
    {
      "epoch": 0.9017632241813602,
      "grad_norm": 0.1232651099562645,
      "learning_rate": 2.9118129336640874e-06,
      "loss": 0.1676,
      "step": 4296
    },
    {
      "epoch": 0.9019731318219983,
      "grad_norm": 0.10802242159843445,
      "learning_rate": 2.8995041426313486e-06,
      "loss": 0.1411,
      "step": 4297
    },
    {
      "epoch": 0.9021830394626364,
      "grad_norm": 0.10859952867031097,
      "learning_rate": 2.887220645640759e-06,
      "loss": 0.1423,
      "step": 4298
    },
    {
      "epoch": 0.9023929471032746,
      "grad_norm": 0.12386050075292587,
      "learning_rate": 2.8749624492888595e-06,
      "loss": 0.1526,
      "step": 4299
    },
    {
      "epoch": 0.9026028547439127,
      "grad_norm": 0.11284506320953369,
      "learning_rate": 2.862729560158578e-06,
      "loss": 0.1698,
      "step": 4300
    },
    {
      "epoch": 0.9028127623845508,
      "grad_norm": 0.13383999466896057,
      "learning_rate": 2.8505219848192488e-06,
      "loss": 0.1741,
      "step": 4301
    },
    {
      "epoch": 0.9030226700251889,
      "grad_norm": 0.09913118928670883,
      "learning_rate": 2.838339729826617e-06,
      "loss": 0.1575,
      "step": 4302
    },
    {
      "epoch": 0.903232577665827,
      "grad_norm": 0.14457812905311584,
      "learning_rate": 2.8261828017228375e-06,
      "loss": 0.1596,
      "step": 4303
    },
    {
      "epoch": 0.9034424853064652,
      "grad_norm": 0.12830603122711182,
      "learning_rate": 2.8140512070364665e-06,
      "loss": 0.1578,
      "step": 4304
    },
    {
      "epoch": 0.9036523929471033,
      "grad_norm": 0.11501707136631012,
      "learning_rate": 2.801944952282437e-06,
      "loss": 0.1354,
      "step": 4305
    },
    {
      "epoch": 0.9038623005877414,
      "grad_norm": 0.1554587483406067,
      "learning_rate": 2.7898640439620815e-06,
      "loss": 0.1448,
      "step": 4306
    },
    {
      "epoch": 0.9040722082283795,
      "grad_norm": 0.10302325338125229,
      "learning_rate": 2.7778084885631217e-06,
      "loss": 0.1709,
      "step": 4307
    },
    {
      "epoch": 0.9042821158690176,
      "grad_norm": 0.11777830123901367,
      "learning_rate": 2.7657782925596786e-06,
      "loss": 0.1542,
      "step": 4308
    },
    {
      "epoch": 0.9044920235096557,
      "grad_norm": 0.12227735668420792,
      "learning_rate": 2.7537734624122402e-06,
      "loss": 0.1687,
      "step": 4309
    },
    {
      "epoch": 0.9047019311502938,
      "grad_norm": 0.12648968398571014,
      "learning_rate": 2.741794004567666e-06,
      "loss": 0.1573,
      "step": 4310
    },
    {
      "epoch": 0.904911838790932,
      "grad_norm": 0.12902848422527313,
      "learning_rate": 2.7298399254592045e-06,
      "loss": 0.1565,
      "step": 4311
    },
    {
      "epoch": 0.9051217464315701,
      "grad_norm": 0.11057978123426437,
      "learning_rate": 2.7179112315064703e-06,
      "loss": 0.1469,
      "step": 4312
    },
    {
      "epoch": 0.9053316540722083,
      "grad_norm": 0.13053567707538605,
      "learning_rate": 2.7060079291154394e-06,
      "loss": 0.1511,
      "step": 4313
    },
    {
      "epoch": 0.9055415617128464,
      "grad_norm": 0.13333049416542053,
      "learning_rate": 2.6941300246784597e-06,
      "loss": 0.1542,
      "step": 4314
    },
    {
      "epoch": 0.9057514693534845,
      "grad_norm": 0.12608638405799866,
      "learning_rate": 2.6822775245742336e-06,
      "loss": 0.1535,
      "step": 4315
    },
    {
      "epoch": 0.9059613769941226,
      "grad_norm": 0.12417828291654587,
      "learning_rate": 2.6704504351678315e-06,
      "loss": 0.1688,
      "step": 4316
    },
    {
      "epoch": 0.9061712846347607,
      "grad_norm": 0.15067745745182037,
      "learning_rate": 2.658648762810667e-06,
      "loss": 0.1694,
      "step": 4317
    },
    {
      "epoch": 0.9063811922753988,
      "grad_norm": 0.1259569376707077,
      "learning_rate": 2.6468725138405148e-06,
      "loss": 0.1548,
      "step": 4318
    },
    {
      "epoch": 0.9065910999160369,
      "grad_norm": 0.129777729511261,
      "learning_rate": 2.6351216945814717e-06,
      "loss": 0.1444,
      "step": 4319
    },
    {
      "epoch": 0.906801007556675,
      "grad_norm": 0.11254890263080597,
      "learning_rate": 2.623396311344012e-06,
      "loss": 0.1666,
      "step": 4320
    },
    {
      "epoch": 0.9070109151973131,
      "grad_norm": 0.11656245589256287,
      "learning_rate": 2.611696370424932e-06,
      "loss": 0.1517,
      "step": 4321
    },
    {
      "epoch": 0.9072208228379512,
      "grad_norm": 0.1227530986070633,
      "learning_rate": 2.6000218781073615e-06,
      "loss": 0.1488,
      "step": 4322
    },
    {
      "epoch": 0.9074307304785895,
      "grad_norm": 0.14218954741954803,
      "learning_rate": 2.5883728406607634e-06,
      "loss": 0.1356,
      "step": 4323
    },
    {
      "epoch": 0.9076406381192276,
      "grad_norm": 0.13774141669273376,
      "learning_rate": 2.576749264340955e-06,
      "loss": 0.1685,
      "step": 4324
    },
    {
      "epoch": 0.9078505457598657,
      "grad_norm": 0.12770342826843262,
      "learning_rate": 2.5651511553900663e-06,
      "loss": 0.1664,
      "step": 4325
    },
    {
      "epoch": 0.9080604534005038,
      "grad_norm": 0.12919233739376068,
      "learning_rate": 2.5535785200365203e-06,
      "loss": 0.1515,
      "step": 4326
    },
    {
      "epoch": 0.9082703610411419,
      "grad_norm": 0.13036112487316132,
      "learning_rate": 2.5420313644950957e-06,
      "loss": 0.155,
      "step": 4327
    },
    {
      "epoch": 0.90848026868178,
      "grad_norm": 0.12815465033054352,
      "learning_rate": 2.5305096949668872e-06,
      "loss": 0.1449,
      "step": 4328
    },
    {
      "epoch": 0.9086901763224181,
      "grad_norm": 0.1162593811750412,
      "learning_rate": 2.5190135176392902e-06,
      "loss": 0.1529,
      "step": 4329
    },
    {
      "epoch": 0.9089000839630562,
      "grad_norm": 0.11832818388938904,
      "learning_rate": 2.507542838686022e-06,
      "loss": 0.161,
      "step": 4330
    },
    {
      "epoch": 0.9091099916036943,
      "grad_norm": 0.1431998759508133,
      "learning_rate": 2.496097664267083e-06,
      "loss": 0.1733,
      "step": 4331
    },
    {
      "epoch": 0.9093198992443325,
      "grad_norm": 0.1280035823583603,
      "learning_rate": 2.484678000528806e-06,
      "loss": 0.1432,
      "step": 4332
    },
    {
      "epoch": 0.9095298068849706,
      "grad_norm": 0.13008855283260345,
      "learning_rate": 2.473283853603808e-06,
      "loss": 0.1487,
      "step": 4333
    },
    {
      "epoch": 0.9097397145256088,
      "grad_norm": 0.12129084020853043,
      "learning_rate": 2.461915229611006e-06,
      "loss": 0.1565,
      "step": 4334
    },
    {
      "epoch": 0.9099496221662469,
      "grad_norm": 0.13218483328819275,
      "learning_rate": 2.450572134655604e-06,
      "loss": 0.1539,
      "step": 4335
    },
    {
      "epoch": 0.910159529806885,
      "grad_norm": 0.12636138498783112,
      "learning_rate": 2.4392545748291197e-06,
      "loss": 0.1521,
      "step": 4336
    },
    {
      "epoch": 0.9103694374475231,
      "grad_norm": 0.13400456309318542,
      "learning_rate": 2.4279625562093234e-06,
      "loss": 0.1657,
      "step": 4337
    },
    {
      "epoch": 0.9105793450881612,
      "grad_norm": 0.11741022765636444,
      "learning_rate": 2.4166960848603037e-06,
      "loss": 0.1578,
      "step": 4338
    },
    {
      "epoch": 0.9107892527287993,
      "grad_norm": 0.10905701667070389,
      "learning_rate": 2.4054551668323986e-06,
      "loss": 0.1492,
      "step": 4339
    },
    {
      "epoch": 0.9109991603694374,
      "grad_norm": 0.1346195489168167,
      "learning_rate": 2.394239808162241e-06,
      "loss": 0.1383,
      "step": 4340
    },
    {
      "epoch": 0.9112090680100756,
      "grad_norm": 0.13491378724575043,
      "learning_rate": 2.38305001487274e-06,
      "loss": 0.1678,
      "step": 4341
    },
    {
      "epoch": 0.9114189756507137,
      "grad_norm": 0.12403254956007004,
      "learning_rate": 2.371885792973072e-06,
      "loss": 0.154,
      "step": 4342
    },
    {
      "epoch": 0.9116288832913518,
      "grad_norm": 0.11544390022754669,
      "learning_rate": 2.3607471484586697e-06,
      "loss": 0.1582,
      "step": 4343
    },
    {
      "epoch": 0.9118387909319899,
      "grad_norm": 0.11051946133375168,
      "learning_rate": 2.3496340873112368e-06,
      "loss": 0.1465,
      "step": 4344
    },
    {
      "epoch": 0.912048698572628,
      "grad_norm": 0.12439233809709549,
      "learning_rate": 2.3385466154987655e-06,
      "loss": 0.1463,
      "step": 4345
    },
    {
      "epoch": 0.9122586062132662,
      "grad_norm": 0.10868985950946808,
      "learning_rate": 2.327484738975455e-06,
      "loss": 0.1615,
      "step": 4346
    },
    {
      "epoch": 0.9124685138539043,
      "grad_norm": 0.11184863746166229,
      "learning_rate": 2.3164484636817973e-06,
      "loss": 0.1413,
      "step": 4347
    },
    {
      "epoch": 0.9126784214945424,
      "grad_norm": 0.13447201251983643,
      "learning_rate": 2.305437795544513e-06,
      "loss": 0.1658,
      "step": 4348
    },
    {
      "epoch": 0.9128883291351805,
      "grad_norm": 0.11348041892051697,
      "learning_rate": 2.294452740476599e-06,
      "loss": 0.1634,
      "step": 4349
    },
    {
      "epoch": 0.9130982367758187,
      "grad_norm": 0.11615937948226929,
      "learning_rate": 2.2834933043772766e-06,
      "loss": 0.1795,
      "step": 4350
    },
    {
      "epoch": 0.9133081444164568,
      "grad_norm": 0.1186215952038765,
      "learning_rate": 2.2725594931320037e-06,
      "loss": 0.1595,
      "step": 4351
    },
    {
      "epoch": 0.9135180520570949,
      "grad_norm": 0.11456868052482605,
      "learning_rate": 2.2616513126125004e-06,
      "loss": 0.1648,
      "step": 4352
    },
    {
      "epoch": 0.913727959697733,
      "grad_norm": 0.10546039789915085,
      "learning_rate": 2.2507687686766975e-06,
      "loss": 0.1625,
      "step": 4353
    },
    {
      "epoch": 0.9139378673383711,
      "grad_norm": 0.13526315987110138,
      "learning_rate": 2.239911867168776e-06,
      "loss": 0.1654,
      "step": 4354
    },
    {
      "epoch": 0.9141477749790092,
      "grad_norm": 0.1123739704489708,
      "learning_rate": 2.2290806139191377e-06,
      "loss": 0.1558,
      "step": 4355
    },
    {
      "epoch": 0.9143576826196473,
      "grad_norm": 0.1591699868440628,
      "learning_rate": 2.2182750147444033e-06,
      "loss": 0.1424,
      "step": 4356
    },
    {
      "epoch": 0.9145675902602854,
      "grad_norm": 0.11519675701856613,
      "learning_rate": 2.2074950754474465e-06,
      "loss": 0.1592,
      "step": 4357
    },
    {
      "epoch": 0.9147774979009236,
      "grad_norm": 0.1345817595720291,
      "learning_rate": 2.1967408018173264e-06,
      "loss": 0.1672,
      "step": 4358
    },
    {
      "epoch": 0.9149874055415617,
      "grad_norm": 0.12809444963932037,
      "learning_rate": 2.1860121996293504e-06,
      "loss": 0.167,
      "step": 4359
    },
    {
      "epoch": 0.9151973131821999,
      "grad_norm": 0.11888294667005539,
      "learning_rate": 2.1753092746449933e-06,
      "loss": 0.1632,
      "step": 4360
    },
    {
      "epoch": 0.915407220822838,
      "grad_norm": 0.1209503635764122,
      "learning_rate": 2.1646320326119908e-06,
      "loss": 0.1538,
      "step": 4361
    },
    {
      "epoch": 0.9156171284634761,
      "grad_norm": 0.1321111023426056,
      "learning_rate": 2.153980479264267e-06,
      "loss": 0.1638,
      "step": 4362
    },
    {
      "epoch": 0.9158270361041142,
      "grad_norm": 0.1377587914466858,
      "learning_rate": 2.143354620321947e-06,
      "loss": 0.1631,
      "step": 4363
    },
    {
      "epoch": 0.9160369437447523,
      "grad_norm": 0.13363386690616608,
      "learning_rate": 2.132754461491343e-06,
      "loss": 0.1479,
      "step": 4364
    },
    {
      "epoch": 0.9162468513853904,
      "grad_norm": 0.12408062070608139,
      "learning_rate": 2.122180008465008e-06,
      "loss": 0.1529,
      "step": 4365
    },
    {
      "epoch": 0.9164567590260285,
      "grad_norm": 0.1353202611207962,
      "learning_rate": 2.11163126692166e-06,
      "loss": 0.1424,
      "step": 4366
    },
    {
      "epoch": 0.9166666666666666,
      "grad_norm": 0.12677113711833954,
      "learning_rate": 2.1011082425262007e-06,
      "loss": 0.1504,
      "step": 4367
    },
    {
      "epoch": 0.9168765743073047,
      "grad_norm": 0.11069560050964355,
      "learning_rate": 2.090610940929738e-06,
      "loss": 0.1541,
      "step": 4368
    },
    {
      "epoch": 0.917086481947943,
      "grad_norm": 0.11136747896671295,
      "learning_rate": 2.080139367769568e-06,
      "loss": 0.1599,
      "step": 4369
    },
    {
      "epoch": 0.9172963895885811,
      "grad_norm": 0.10797339677810669,
      "learning_rate": 2.0696935286691697e-06,
      "loss": 0.154,
      "step": 4370
    },
    {
      "epoch": 0.9175062972292192,
      "grad_norm": 0.1661231368780136,
      "learning_rate": 2.0592734292381844e-06,
      "loss": 0.1599,
      "step": 4371
    },
    {
      "epoch": 0.9177162048698573,
      "grad_norm": 0.11923778802156448,
      "learning_rate": 2.048879075072452e-06,
      "loss": 0.1685,
      "step": 4372
    },
    {
      "epoch": 0.9179261125104954,
      "grad_norm": 0.12219996005296707,
      "learning_rate": 2.038510471753979e-06,
      "loss": 0.1592,
      "step": 4373
    },
    {
      "epoch": 0.9181360201511335,
      "grad_norm": 0.1292019933462143,
      "learning_rate": 2.028167624850935e-06,
      "loss": 0.1608,
      "step": 4374
    },
    {
      "epoch": 0.9183459277917716,
      "grad_norm": 0.11530636996030807,
      "learning_rate": 2.0178505399176704e-06,
      "loss": 0.1483,
      "step": 4375
    },
    {
      "epoch": 0.9185558354324097,
      "grad_norm": 0.11238705366849899,
      "learning_rate": 2.007559222494687e-06,
      "loss": 0.1477,
      "step": 4376
    },
    {
      "epoch": 0.9187657430730478,
      "grad_norm": 0.11351519078016281,
      "learning_rate": 1.9972936781086693e-06,
      "loss": 0.1593,
      "step": 4377
    },
    {
      "epoch": 0.918975650713686,
      "grad_norm": 0.1325710266828537,
      "learning_rate": 1.987053912272446e-06,
      "loss": 0.142,
      "step": 4378
    },
    {
      "epoch": 0.9191855583543241,
      "grad_norm": 0.11840063333511353,
      "learning_rate": 1.976839930485008e-06,
      "loss": 0.1433,
      "step": 4379
    },
    {
      "epoch": 0.9193954659949622,
      "grad_norm": 0.114000603556633,
      "learning_rate": 1.9666517382314897e-06,
      "loss": 0.1491,
      "step": 4380
    },
    {
      "epoch": 0.9196053736356004,
      "grad_norm": 0.10446514934301376,
      "learning_rate": 1.956489340983181e-06,
      "loss": 0.1572,
      "step": 4381
    },
    {
      "epoch": 0.9198152812762385,
      "grad_norm": 0.11224140971899033,
      "learning_rate": 1.9463527441975325e-06,
      "loss": 0.1389,
      "step": 4382
    },
    {
      "epoch": 0.9200251889168766,
      "grad_norm": 0.13361458480358124,
      "learning_rate": 1.9362419533181187e-06,
      "loss": 0.1628,
      "step": 4383
    },
    {
      "epoch": 0.9202350965575147,
      "grad_norm": 0.11145322024822235,
      "learning_rate": 1.9261569737746675e-06,
      "loss": 0.1533,
      "step": 4384
    },
    {
      "epoch": 0.9204450041981528,
      "grad_norm": 0.09477204829454422,
      "learning_rate": 1.916097810983042e-06,
      "loss": 0.1589,
      "step": 4385
    },
    {
      "epoch": 0.9206549118387909,
      "grad_norm": 0.12341588735580444,
      "learning_rate": 1.906064470345248e-06,
      "loss": 0.1664,
      "step": 4386
    },
    {
      "epoch": 0.920864819479429,
      "grad_norm": 0.11354007571935654,
      "learning_rate": 1.8960569572494203e-06,
      "loss": 0.151,
      "step": 4387
    },
    {
      "epoch": 0.9210747271200672,
      "grad_norm": 0.10435278713703156,
      "learning_rate": 1.8860752770698155e-06,
      "loss": 0.1529,
      "step": 4388
    },
    {
      "epoch": 0.9212846347607053,
      "grad_norm": 0.11676231026649475,
      "learning_rate": 1.8761194351668176e-06,
      "loss": 0.1622,
      "step": 4389
    },
    {
      "epoch": 0.9214945424013434,
      "grad_norm": 0.11899908632040024,
      "learning_rate": 1.8661894368869505e-06,
      "loss": 0.1299,
      "step": 4390
    },
    {
      "epoch": 0.9217044500419815,
      "grad_norm": 0.12129631638526917,
      "learning_rate": 1.8562852875628534e-06,
      "loss": 0.1408,
      "step": 4391
    },
    {
      "epoch": 0.9219143576826196,
      "grad_norm": 0.12856750190258026,
      "learning_rate": 1.846406992513272e-06,
      "loss": 0.1555,
      "step": 4392
    },
    {
      "epoch": 0.9221242653232578,
      "grad_norm": 0.11216092109680176,
      "learning_rate": 1.8365545570430798e-06,
      "loss": 0.1479,
      "step": 4393
    },
    {
      "epoch": 0.9223341729638959,
      "grad_norm": 0.11643592268228531,
      "learning_rate": 1.8267279864432662e-06,
      "loss": 0.137,
      "step": 4394
    },
    {
      "epoch": 0.922544080604534,
      "grad_norm": 0.11467355489730835,
      "learning_rate": 1.8169272859909158e-06,
      "loss": 0.1541,
      "step": 4395
    },
    {
      "epoch": 0.9227539882451721,
      "grad_norm": 0.11284631490707397,
      "learning_rate": 1.8071524609492295e-06,
      "loss": 0.1328,
      "step": 4396
    },
    {
      "epoch": 0.9229638958858103,
      "grad_norm": 0.12459743767976761,
      "learning_rate": 1.797403516567503e-06,
      "loss": 0.1555,
      "step": 4397
    },
    {
      "epoch": 0.9231738035264484,
      "grad_norm": 0.13417114317417145,
      "learning_rate": 1.7876804580811645e-06,
      "loss": 0.1484,
      "step": 4398
    },
    {
      "epoch": 0.9233837111670865,
      "grad_norm": 0.11709386855363846,
      "learning_rate": 1.7779832907117045e-06,
      "loss": 0.1446,
      "step": 4399
    },
    {
      "epoch": 0.9235936188077246,
      "grad_norm": 0.14955921471118927,
      "learning_rate": 1.7683120196667235e-06,
      "loss": 0.185,
      "step": 4400
    },
    {
      "epoch": 0.9238035264483627,
      "grad_norm": 0.12120398879051208,
      "learning_rate": 1.7586666501399118e-06,
      "loss": 0.1655,
      "step": 4401
    },
    {
      "epoch": 0.9240134340890008,
      "grad_norm": 0.12226814031600952,
      "learning_rate": 1.7490471873110647e-06,
      "loss": 0.1701,
      "step": 4402
    },
    {
      "epoch": 0.9242233417296389,
      "grad_norm": 0.11423861980438232,
      "learning_rate": 1.7394536363460445e-06,
      "loss": 0.1466,
      "step": 4403
    },
    {
      "epoch": 0.924433249370277,
      "grad_norm": 0.14197257161140442,
      "learning_rate": 1.7298860023968078e-06,
      "loss": 0.1527,
      "step": 4404
    },
    {
      "epoch": 0.9246431570109152,
      "grad_norm": 0.13661980628967285,
      "learning_rate": 1.720344290601389e-06,
      "loss": 0.158,
      "step": 4405
    },
    {
      "epoch": 0.9248530646515534,
      "grad_norm": 0.1328696757555008,
      "learning_rate": 1.7108285060839169e-06,
      "loss": 0.1617,
      "step": 4406
    },
    {
      "epoch": 0.9250629722921915,
      "grad_norm": 0.1082594245672226,
      "learning_rate": 1.7013386539545761e-06,
      "loss": 0.1529,
      "step": 4407
    },
    {
      "epoch": 0.9252728799328296,
      "grad_norm": 0.12489262223243713,
      "learning_rate": 1.6918747393096456e-06,
      "loss": 0.1524,
      "step": 4408
    },
    {
      "epoch": 0.9254827875734677,
      "grad_norm": 0.1348544806241989,
      "learning_rate": 1.6824367672314433e-06,
      "loss": 0.1536,
      "step": 4409
    },
    {
      "epoch": 0.9256926952141058,
      "grad_norm": 0.12416580319404602,
      "learning_rate": 1.6730247427883872e-06,
      "loss": 0.1535,
      "step": 4410
    },
    {
      "epoch": 0.9259026028547439,
      "grad_norm": 0.12714757025241852,
      "learning_rate": 1.663638671034956e-06,
      "loss": 0.1826,
      "step": 4411
    },
    {
      "epoch": 0.926112510495382,
      "grad_norm": 0.13926753401756287,
      "learning_rate": 1.6542785570116736e-06,
      "loss": 0.1477,
      "step": 4412
    },
    {
      "epoch": 0.9263224181360201,
      "grad_norm": 0.11886262148618698,
      "learning_rate": 1.6449444057451414e-06,
      "loss": 0.1558,
      "step": 4413
    },
    {
      "epoch": 0.9265323257766582,
      "grad_norm": 0.11909741908311844,
      "learning_rate": 1.6356362222479993e-06,
      "loss": 0.1637,
      "step": 4414
    },
    {
      "epoch": 0.9267422334172963,
      "grad_norm": 0.1119212731719017,
      "learning_rate": 1.6263540115189823e-06,
      "loss": 0.1406,
      "step": 4415
    },
    {
      "epoch": 0.9269521410579346,
      "grad_norm": 0.11529149860143661,
      "learning_rate": 1.6170977785428254e-06,
      "loss": 0.1537,
      "step": 4416
    },
    {
      "epoch": 0.9271620486985727,
      "grad_norm": 0.10490969568490982,
      "learning_rate": 1.6078675282903354e-06,
      "loss": 0.1465,
      "step": 4417
    },
    {
      "epoch": 0.9273719563392108,
      "grad_norm": 0.11506490409374237,
      "learning_rate": 1.5986632657183865e-06,
      "loss": 0.166,
      "step": 4418
    },
    {
      "epoch": 0.9275818639798489,
      "grad_norm": 0.12108638882637024,
      "learning_rate": 1.5894849957698632e-06,
      "loss": 0.1598,
      "step": 4419
    },
    {
      "epoch": 0.927791771620487,
      "grad_norm": 0.10503645241260529,
      "learning_rate": 1.580332723373712e-06,
      "loss": 0.1662,
      "step": 4420
    },
    {
      "epoch": 0.9280016792611251,
      "grad_norm": 0.12522466480731964,
      "learning_rate": 1.5712064534449067e-06,
      "loss": 0.1527,
      "step": 4421
    },
    {
      "epoch": 0.9282115869017632,
      "grad_norm": 0.12102288752794266,
      "learning_rate": 1.5621061908844714e-06,
      "loss": 0.17,
      "step": 4422
    },
    {
      "epoch": 0.9284214945424013,
      "grad_norm": 0.12997739017009735,
      "learning_rate": 1.5530319405794525e-06,
      "loss": 0.1323,
      "step": 4423
    },
    {
      "epoch": 0.9286314021830394,
      "grad_norm": 0.12769199907779694,
      "learning_rate": 1.5439837074029239e-06,
      "loss": 0.1465,
      "step": 4424
    },
    {
      "epoch": 0.9288413098236776,
      "grad_norm": 0.10771086812019348,
      "learning_rate": 1.5349614962139936e-06,
      "loss": 0.155,
      "step": 4425
    },
    {
      "epoch": 0.9290512174643157,
      "grad_norm": 0.13490477204322815,
      "learning_rate": 1.525965311857802e-06,
      "loss": 0.1665,
      "step": 4426
    },
    {
      "epoch": 0.9292611251049538,
      "grad_norm": 0.11361309885978699,
      "learning_rate": 1.516995159165502e-06,
      "loss": 0.1578,
      "step": 4427
    },
    {
      "epoch": 0.929471032745592,
      "grad_norm": 0.1189640536904335,
      "learning_rate": 1.5080510429542672e-06,
      "loss": 0.1415,
      "step": 4428
    },
    {
      "epoch": 0.9296809403862301,
      "grad_norm": 0.11884358525276184,
      "learning_rate": 1.499132968027306e-06,
      "loss": 0.1617,
      "step": 4429
    },
    {
      "epoch": 0.9298908480268682,
      "grad_norm": 0.14828267693519592,
      "learning_rate": 1.4902409391738091e-06,
      "loss": 0.1596,
      "step": 4430
    },
    {
      "epoch": 0.9301007556675063,
      "grad_norm": 0.13310547173023224,
      "learning_rate": 1.4813749611690075e-06,
      "loss": 0.1535,
      "step": 4431
    },
    {
      "epoch": 0.9303106633081444,
      "grad_norm": 0.11745025217533112,
      "learning_rate": 1.472535038774142e-06,
      "loss": 0.1728,
      "step": 4432
    },
    {
      "epoch": 0.9305205709487825,
      "grad_norm": 0.11373081803321838,
      "learning_rate": 1.4637211767364378e-06,
      "loss": 0.1595,
      "step": 4433
    },
    {
      "epoch": 0.9307304785894207,
      "grad_norm": 0.1132584884762764,
      "learning_rate": 1.4549333797891472e-06,
      "loss": 0.1439,
      "step": 4434
    },
    {
      "epoch": 0.9309403862300588,
      "grad_norm": 0.11979074031114578,
      "learning_rate": 1.4461716526515234e-06,
      "loss": 0.1511,
      "step": 4435
    },
    {
      "epoch": 0.9311502938706969,
      "grad_norm": 0.13811197876930237,
      "learning_rate": 1.4374360000288134e-06,
      "loss": 0.15,
      "step": 4436
    },
    {
      "epoch": 0.931360201511335,
      "grad_norm": 0.10705514252185822,
      "learning_rate": 1.428726426612248e-06,
      "loss": 0.1465,
      "step": 4437
    },
    {
      "epoch": 0.9315701091519731,
      "grad_norm": 0.13573043048381805,
      "learning_rate": 1.4200429370790802e-06,
      "loss": 0.1302,
      "step": 4438
    },
    {
      "epoch": 0.9317800167926112,
      "grad_norm": 0.11842697113752365,
      "learning_rate": 1.4113855360925354e-06,
      "loss": 0.148,
      "step": 4439
    },
    {
      "epoch": 0.9319899244332494,
      "grad_norm": 0.11456546932458878,
      "learning_rate": 1.4027542283018448e-06,
      "loss": 0.1573,
      "step": 4440
    },
    {
      "epoch": 0.9321998320738875,
      "grad_norm": 0.1204330176115036,
      "learning_rate": 1.3941490183422113e-06,
      "loss": 0.1516,
      "step": 4441
    },
    {
      "epoch": 0.9324097397145256,
      "grad_norm": 0.126027449965477,
      "learning_rate": 1.385569910834833e-06,
      "loss": 0.1512,
      "step": 4442
    },
    {
      "epoch": 0.9326196473551638,
      "grad_norm": 0.1196598932147026,
      "learning_rate": 1.3770169103868802e-06,
      "loss": 0.165,
      "step": 4443
    },
    {
      "epoch": 0.9328295549958019,
      "grad_norm": 0.17786267399787903,
      "learning_rate": 1.368490021591512e-06,
      "loss": 0.1337,
      "step": 4444
    },
    {
      "epoch": 0.93303946263644,
      "grad_norm": 0.1896003633737564,
      "learning_rate": 1.3599892490278654e-06,
      "loss": 0.1614,
      "step": 4445
    },
    {
      "epoch": 0.9332493702770781,
      "grad_norm": 0.1261228770017624,
      "learning_rate": 1.3515145972610443e-06,
      "loss": 0.1517,
      "step": 4446
    },
    {
      "epoch": 0.9334592779177162,
      "grad_norm": 0.12102353572845459,
      "learning_rate": 1.343066070842136e-06,
      "loss": 0.1561,
      "step": 4447
    },
    {
      "epoch": 0.9336691855583543,
      "grad_norm": 0.12365438044071198,
      "learning_rate": 1.334643674308189e-06,
      "loss": 0.165,
      "step": 4448
    },
    {
      "epoch": 0.9338790931989924,
      "grad_norm": 0.1193314641714096,
      "learning_rate": 1.32624741218223e-06,
      "loss": 0.1686,
      "step": 4449
    },
    {
      "epoch": 0.9340890008396305,
      "grad_norm": 0.11541610211133957,
      "learning_rate": 1.3178772889732293e-06,
      "loss": 0.1624,
      "step": 4450
    },
    {
      "epoch": 0.9342989084802686,
      "grad_norm": 0.13764046132564545,
      "learning_rate": 1.309533309176142e-06,
      "loss": 0.1497,
      "step": 4451
    },
    {
      "epoch": 0.9345088161209067,
      "grad_norm": 0.1194557324051857,
      "learning_rate": 1.3012154772718777e-06,
      "loss": 0.1665,
      "step": 4452
    },
    {
      "epoch": 0.934718723761545,
      "grad_norm": 0.13757576048374176,
      "learning_rate": 1.2929237977273023e-06,
      "loss": 0.1688,
      "step": 4453
    },
    {
      "epoch": 0.9349286314021831,
      "grad_norm": 0.12633179128170013,
      "learning_rate": 1.2846582749952317e-06,
      "loss": 0.1358,
      "step": 4454
    },
    {
      "epoch": 0.9351385390428212,
      "grad_norm": 0.12188289314508438,
      "learning_rate": 1.2764189135144377e-06,
      "loss": 0.1544,
      "step": 4455
    },
    {
      "epoch": 0.9353484466834593,
      "grad_norm": 0.11736936867237091,
      "learning_rate": 1.268205717709664e-06,
      "loss": 0.1607,
      "step": 4456
    },
    {
      "epoch": 0.9355583543240974,
      "grad_norm": 0.12075451761484146,
      "learning_rate": 1.2600186919915712e-06,
      "loss": 0.1577,
      "step": 4457
    },
    {
      "epoch": 0.9357682619647355,
      "grad_norm": 0.12973789870738983,
      "learning_rate": 1.2518578407567706e-06,
      "loss": 0.1482,
      "step": 4458
    },
    {
      "epoch": 0.9359781696053736,
      "grad_norm": 0.12099910527467728,
      "learning_rate": 1.2437231683878393e-06,
      "loss": 0.1564,
      "step": 4459
    },
    {
      "epoch": 0.9361880772460117,
      "grad_norm": 0.13991937041282654,
      "learning_rate": 1.2356146792532775e-06,
      "loss": 0.1357,
      "step": 4460
    },
    {
      "epoch": 0.9363979848866498,
      "grad_norm": 0.13311533629894257,
      "learning_rate": 1.2275323777075297e-06,
      "loss": 0.1652,
      "step": 4461
    },
    {
      "epoch": 0.936607892527288,
      "grad_norm": 0.11790844053030014,
      "learning_rate": 1.2194762680909743e-06,
      "loss": 0.1722,
      "step": 4462
    },
    {
      "epoch": 0.9368178001679262,
      "grad_norm": 0.1021541953086853,
      "learning_rate": 1.2114463547299338e-06,
      "loss": 0.1554,
      "step": 4463
    },
    {
      "epoch": 0.9370277078085643,
      "grad_norm": 0.1211385428905487,
      "learning_rate": 1.2034426419366473e-06,
      "loss": 0.174,
      "step": 4464
    },
    {
      "epoch": 0.9372376154492024,
      "grad_norm": 0.16256774961948395,
      "learning_rate": 1.1954651340092936e-06,
      "loss": 0.1276,
      "step": 4465
    },
    {
      "epoch": 0.9374475230898405,
      "grad_norm": 0.12904593348503113,
      "learning_rate": 1.1875138352319737e-06,
      "loss": 0.1466,
      "step": 4466
    },
    {
      "epoch": 0.9376574307304786,
      "grad_norm": 0.12402849644422531,
      "learning_rate": 1.179588749874716e-06,
      "loss": 0.1295,
      "step": 4467
    },
    {
      "epoch": 0.9378673383711167,
      "grad_norm": 0.14143963158130646,
      "learning_rate": 1.171689882193483e-06,
      "loss": 0.165,
      "step": 4468
    },
    {
      "epoch": 0.9380772460117548,
      "grad_norm": 0.1427275836467743,
      "learning_rate": 1.163817236430137e-06,
      "loss": 0.1736,
      "step": 4469
    },
    {
      "epoch": 0.9382871536523929,
      "grad_norm": 0.12246612459421158,
      "learning_rate": 1.1559708168124793e-06,
      "loss": 0.1681,
      "step": 4470
    },
    {
      "epoch": 0.9384970612930311,
      "grad_norm": 0.11412835866212845,
      "learning_rate": 1.1481506275541953e-06,
      "loss": 0.1462,
      "step": 4471
    },
    {
      "epoch": 0.9387069689336692,
      "grad_norm": 0.13155506551265717,
      "learning_rate": 1.1403566728549253e-06,
      "loss": 0.1659,
      "step": 4472
    },
    {
      "epoch": 0.9389168765743073,
      "grad_norm": 0.11218184232711792,
      "learning_rate": 1.1325889569001935e-06,
      "loss": 0.1421,
      "step": 4473
    },
    {
      "epoch": 0.9391267842149454,
      "grad_norm": 0.13464675843715668,
      "learning_rate": 1.1248474838614465e-06,
      "loss": 0.1593,
      "step": 4474
    },
    {
      "epoch": 0.9393366918555835,
      "grad_norm": 0.11895395815372467,
      "learning_rate": 1.1171322578960197e-06,
      "loss": 0.1504,
      "step": 4475
    },
    {
      "epoch": 0.9395465994962217,
      "grad_norm": 0.12074901163578033,
      "learning_rate": 1.1094432831471768e-06,
      "loss": 0.1641,
      "step": 4476
    },
    {
      "epoch": 0.9397565071368598,
      "grad_norm": 0.1069439947605133,
      "learning_rate": 1.101780563744076e-06,
      "loss": 0.1479,
      "step": 4477
    },
    {
      "epoch": 0.9399664147774979,
      "grad_norm": 0.10945051163434982,
      "learning_rate": 1.0941441038017586e-06,
      "loss": 0.161,
      "step": 4478
    },
    {
      "epoch": 0.940176322418136,
      "grad_norm": 0.12555761635303497,
      "learning_rate": 1.0865339074211833e-06,
      "loss": 0.1463,
      "step": 4479
    },
    {
      "epoch": 0.9403862300587741,
      "grad_norm": 0.1251411885023117,
      "learning_rate": 1.0789499786892032e-06,
      "loss": 0.153,
      "step": 4480
    },
    {
      "epoch": 0.9405961376994123,
      "grad_norm": 0.10783831775188446,
      "learning_rate": 1.071392321678566e-06,
      "loss": 0.1574,
      "step": 4481
    },
    {
      "epoch": 0.9408060453400504,
      "grad_norm": 0.119156114757061,
      "learning_rate": 1.0638609404478916e-06,
      "loss": 0.1579,
      "step": 4482
    },
    {
      "epoch": 0.9410159529806885,
      "grad_norm": 0.12803539633750916,
      "learning_rate": 1.0563558390417172e-06,
      "loss": 0.1532,
      "step": 4483
    },
    {
      "epoch": 0.9412258606213266,
      "grad_norm": 0.11497396975755692,
      "learning_rate": 1.0488770214904465e-06,
      "loss": 0.1618,
      "step": 4484
    },
    {
      "epoch": 0.9414357682619647,
      "grad_norm": 0.15252766013145447,
      "learning_rate": 1.0414244918103777e-06,
      "loss": 0.1586,
      "step": 4485
    },
    {
      "epoch": 0.9416456759026028,
      "grad_norm": 0.11600207537412643,
      "learning_rate": 1.033998254003693e-06,
      "loss": 0.1373,
      "step": 4486
    },
    {
      "epoch": 0.941855583543241,
      "grad_norm": 0.11400733888149261,
      "learning_rate": 1.026598312058441e-06,
      "loss": 0.1475,
      "step": 4487
    },
    {
      "epoch": 0.9420654911838791,
      "grad_norm": 0.12781751155853271,
      "learning_rate": 1.0192246699485652e-06,
      "loss": 0.1657,
      "step": 4488
    },
    {
      "epoch": 0.9422753988245172,
      "grad_norm": 0.13421955704689026,
      "learning_rate": 1.0118773316338869e-06,
      "loss": 0.1624,
      "step": 4489
    },
    {
      "epoch": 0.9424853064651554,
      "grad_norm": 0.12550324201583862,
      "learning_rate": 1.004556301060089e-06,
      "loss": 0.1554,
      "step": 4490
    },
    {
      "epoch": 0.9426952141057935,
      "grad_norm": 0.13535472750663757,
      "learning_rate": 9.972615821587316e-07,
      "loss": 0.1354,
      "step": 4491
    },
    {
      "epoch": 0.9429051217464316,
      "grad_norm": 0.146240696310997,
      "learning_rate": 9.899931788472427e-07,
      "loss": 0.1489,
      "step": 4492
    },
    {
      "epoch": 0.9431150293870697,
      "grad_norm": 0.12862592935562134,
      "learning_rate": 9.827510950289275e-07,
      "loss": 0.1612,
      "step": 4493
    },
    {
      "epoch": 0.9433249370277078,
      "grad_norm": 0.11790810525417328,
      "learning_rate": 9.75535334592942e-07,
      "loss": 0.1415,
      "step": 4494
    },
    {
      "epoch": 0.9435348446683459,
      "grad_norm": 0.11090364307165146,
      "learning_rate": 9.683459014143248e-07,
      "loss": 0.1649,
      "step": 4495
    },
    {
      "epoch": 0.943744752308984,
      "grad_norm": 0.12426898628473282,
      "learning_rate": 9.611827993539547e-07,
      "loss": 0.1573,
      "step": 4496
    },
    {
      "epoch": 0.9439546599496221,
      "grad_norm": 0.15216581523418427,
      "learning_rate": 9.540460322585875e-07,
      "loss": 0.1518,
      "step": 4497
    },
    {
      "epoch": 0.9441645675902602,
      "grad_norm": 0.11171586066484451,
      "learning_rate": 9.469356039608357e-07,
      "loss": 0.1326,
      "step": 4498
    },
    {
      "epoch": 0.9443744752308985,
      "grad_norm": 0.12237754464149475,
      "learning_rate": 9.398515182791446e-07,
      "loss": 0.1544,
      "step": 4499
    },
    {
      "epoch": 0.9445843828715366,
      "grad_norm": 0.12762495875358582,
      "learning_rate": 9.327937790178487e-07,
      "loss": 0.1702,
      "step": 4500
    },
    {
      "epoch": 0.9447942905121747,
      "grad_norm": 0.12392231822013855,
      "learning_rate": 9.257623899671053e-07,
      "loss": 0.1591,
      "step": 4501
    },
    {
      "epoch": 0.9450041981528128,
      "grad_norm": 0.11078055948019028,
      "learning_rate": 9.187573549029327e-07,
      "loss": 0.1492,
      "step": 4502
    },
    {
      "epoch": 0.9452141057934509,
      "grad_norm": 0.14925430715084076,
      "learning_rate": 9.117786775871939e-07,
      "loss": 0.1325,
      "step": 4503
    },
    {
      "epoch": 0.945424013434089,
      "grad_norm": 0.12183661013841629,
      "learning_rate": 9.048263617675967e-07,
      "loss": 0.1605,
      "step": 4504
    },
    {
      "epoch": 0.9456339210747271,
      "grad_norm": 0.11003254354000092,
      "learning_rate": 8.97900411177699e-07,
      "loss": 0.1536,
      "step": 4505
    },
    {
      "epoch": 0.9458438287153652,
      "grad_norm": 0.11058861762285233,
      "learning_rate": 8.910008295368921e-07,
      "loss": 0.145,
      "step": 4506
    },
    {
      "epoch": 0.9460537363560033,
      "grad_norm": 0.12683282792568207,
      "learning_rate": 8.841276205504067e-07,
      "loss": 0.1614,
      "step": 4507
    },
    {
      "epoch": 0.9462636439966414,
      "grad_norm": 0.10946714878082275,
      "learning_rate": 8.772807879093126e-07,
      "loss": 0.1584,
      "step": 4508
    },
    {
      "epoch": 0.9464735516372796,
      "grad_norm": 0.11910434067249298,
      "learning_rate": 8.704603352905183e-07,
      "loss": 0.1635,
      "step": 4509
    },
    {
      "epoch": 0.9466834592779177,
      "grad_norm": 0.12066495418548584,
      "learning_rate": 8.636662663567608e-07,
      "loss": 0.1588,
      "step": 4510
    },
    {
      "epoch": 0.9468933669185559,
      "grad_norm": 0.12396088242530823,
      "learning_rate": 8.568985847566157e-07,
      "loss": 0.1564,
      "step": 4511
    },
    {
      "epoch": 0.947103274559194,
      "grad_norm": 0.12058629095554352,
      "learning_rate": 8.501572941244762e-07,
      "loss": 0.1484,
      "step": 4512
    },
    {
      "epoch": 0.9473131821998321,
      "grad_norm": 0.1340767741203308,
      "learning_rate": 8.434423980805684e-07,
      "loss": 0.1525,
      "step": 4513
    },
    {
      "epoch": 0.9475230898404702,
      "grad_norm": 0.14222848415374756,
      "learning_rate": 8.367539002309465e-07,
      "loss": 0.1531,
      "step": 4514
    },
    {
      "epoch": 0.9477329974811083,
      "grad_norm": 0.14000943303108215,
      "learning_rate": 8.300918041674932e-07,
      "loss": 0.1748,
      "step": 4515
    },
    {
      "epoch": 0.9479429051217464,
      "grad_norm": 0.1195470541715622,
      "learning_rate": 8.234561134678909e-07,
      "loss": 0.1471,
      "step": 4516
    },
    {
      "epoch": 0.9481528127623845,
      "grad_norm": 0.10464131087064743,
      "learning_rate": 8.168468316956723e-07,
      "loss": 0.1451,
      "step": 4517
    },
    {
      "epoch": 0.9483627204030227,
      "grad_norm": 0.11033995449542999,
      "learning_rate": 8.102639624001707e-07,
      "loss": 0.1417,
      "step": 4518
    },
    {
      "epoch": 0.9485726280436608,
      "grad_norm": 0.12074270844459534,
      "learning_rate": 8.037075091165358e-07,
      "loss": 0.1486,
      "step": 4519
    },
    {
      "epoch": 0.9487825356842989,
      "grad_norm": 0.1229119673371315,
      "learning_rate": 7.971774753657235e-07,
      "loss": 0.1597,
      "step": 4520
    },
    {
      "epoch": 0.948992443324937,
      "grad_norm": 0.12053476274013519,
      "learning_rate": 7.906738646545176e-07,
      "loss": 0.1463,
      "step": 4521
    },
    {
      "epoch": 0.9492023509655751,
      "grad_norm": 0.11490915715694427,
      "learning_rate": 7.841966804755129e-07,
      "loss": 0.1505,
      "step": 4522
    },
    {
      "epoch": 0.9494122586062133,
      "grad_norm": 0.1358073353767395,
      "learning_rate": 7.77745926307094e-07,
      "loss": 0.1653,
      "step": 4523
    },
    {
      "epoch": 0.9496221662468514,
      "grad_norm": 0.13225747644901276,
      "learning_rate": 7.713216056134731e-07,
      "loss": 0.1442,
      "step": 4524
    },
    {
      "epoch": 0.9498320738874895,
      "grad_norm": 0.1354922205209732,
      "learning_rate": 7.649237218446458e-07,
      "loss": 0.156,
      "step": 4525
    },
    {
      "epoch": 0.9500419815281276,
      "grad_norm": 0.11272266507148743,
      "learning_rate": 7.585522784364363e-07,
      "loss": 0.1669,
      "step": 4526
    },
    {
      "epoch": 0.9502518891687658,
      "grad_norm": 0.12645363807678223,
      "learning_rate": 7.522072788104406e-07,
      "loss": 0.1653,
      "step": 4527
    },
    {
      "epoch": 0.9504617968094039,
      "grad_norm": 0.12966369092464447,
      "learning_rate": 7.458887263740721e-07,
      "loss": 0.1421,
      "step": 4528
    },
    {
      "epoch": 0.950671704450042,
      "grad_norm": 0.1188059002161026,
      "learning_rate": 7.395966245205443e-07,
      "loss": 0.1632,
      "step": 4529
    },
    {
      "epoch": 0.9508816120906801,
      "grad_norm": 0.1147613450884819,
      "learning_rate": 7.333309766288599e-07,
      "loss": 0.1567,
      "step": 4530
    },
    {
      "epoch": 0.9510915197313182,
      "grad_norm": 0.12427303940057755,
      "learning_rate": 7.270917860638049e-07,
      "loss": 0.1619,
      "step": 4531
    },
    {
      "epoch": 0.9513014273719563,
      "grad_norm": 0.10449483245611191,
      "learning_rate": 7.20879056175977e-07,
      "loss": 0.1574,
      "step": 4532
    },
    {
      "epoch": 0.9515113350125944,
      "grad_norm": 0.10406437516212463,
      "learning_rate": 7.146927903017464e-07,
      "loss": 0.1531,
      "step": 4533
    },
    {
      "epoch": 0.9517212426532325,
      "grad_norm": 0.12593679130077362,
      "learning_rate": 7.085329917632888e-07,
      "loss": 0.1596,
      "step": 4534
    },
    {
      "epoch": 0.9519311502938707,
      "grad_norm": 0.15952709317207336,
      "learning_rate": 7.02399663868547e-07,
      "loss": 0.1507,
      "step": 4535
    },
    {
      "epoch": 0.9521410579345088,
      "grad_norm": 0.14857974648475647,
      "learning_rate": 6.962928099112643e-07,
      "loss": 0.1593,
      "step": 4536
    },
    {
      "epoch": 0.952350965575147,
      "grad_norm": 0.12613141536712646,
      "learning_rate": 6.902124331709503e-07,
      "loss": 0.1467,
      "step": 4537
    },
    {
      "epoch": 0.9525608732157851,
      "grad_norm": 0.12536998093128204,
      "learning_rate": 6.841585369129266e-07,
      "loss": 0.1616,
      "step": 4538
    },
    {
      "epoch": 0.9527707808564232,
      "grad_norm": 0.11849667876958847,
      "learning_rate": 6.781311243882593e-07,
      "loss": 0.1564,
      "step": 4539
    },
    {
      "epoch": 0.9529806884970613,
      "grad_norm": 0.13249583542346954,
      "learning_rate": 6.721301988338036e-07,
      "loss": 0.1581,
      "step": 4540
    },
    {
      "epoch": 0.9531905961376994,
      "grad_norm": 0.13716809451580048,
      "learning_rate": 6.661557634721982e-07,
      "loss": 0.1582,
      "step": 4541
    },
    {
      "epoch": 0.9534005037783375,
      "grad_norm": 0.11763902753591537,
      "learning_rate": 6.602078215118601e-07,
      "loss": 0.1544,
      "step": 4542
    },
    {
      "epoch": 0.9536104114189756,
      "grad_norm": 0.11533381044864655,
      "learning_rate": 6.542863761469565e-07,
      "loss": 0.1462,
      "step": 4543
    },
    {
      "epoch": 0.9538203190596137,
      "grad_norm": 0.10741002857685089,
      "learning_rate": 6.483914305574434e-07,
      "loss": 0.141,
      "step": 4544
    },
    {
      "epoch": 0.9540302267002518,
      "grad_norm": 0.11406449973583221,
      "learning_rate": 6.425229879090444e-07,
      "loss": 0.165,
      "step": 4545
    },
    {
      "epoch": 0.9542401343408901,
      "grad_norm": 0.13373495638370514,
      "learning_rate": 6.366810513532495e-07,
      "loss": 0.1569,
      "step": 4546
    },
    {
      "epoch": 0.9544500419815282,
      "grad_norm": 0.10696794092655182,
      "learning_rate": 6.308656240272992e-07,
      "loss": 0.1377,
      "step": 4547
    },
    {
      "epoch": 0.9546599496221663,
      "grad_norm": 0.12216789275407791,
      "learning_rate": 6.250767090542231e-07,
      "loss": 0.1471,
      "step": 4548
    },
    {
      "epoch": 0.9548698572628044,
      "grad_norm": 0.1208495944738388,
      "learning_rate": 6.193143095427956e-07,
      "loss": 0.1516,
      "step": 4549
    },
    {
      "epoch": 0.9550797649034425,
      "grad_norm": 0.11360879987478256,
      "learning_rate": 6.135784285875579e-07,
      "loss": 0.1607,
      "step": 4550
    },
    {
      "epoch": 0.9552896725440806,
      "grad_norm": 0.11046485602855682,
      "learning_rate": 6.078690692688127e-07,
      "loss": 0.1551,
      "step": 4551
    },
    {
      "epoch": 0.9554995801847187,
      "grad_norm": 0.12528584897518158,
      "learning_rate": 6.02186234652613e-07,
      "loss": 0.1537,
      "step": 4552
    },
    {
      "epoch": 0.9557094878253568,
      "grad_norm": 0.12984655797481537,
      "learning_rate": 5.965299277907677e-07,
      "loss": 0.1542,
      "step": 4553
    },
    {
      "epoch": 0.9559193954659949,
      "grad_norm": 0.1632155030965805,
      "learning_rate": 5.909001517208468e-07,
      "loss": 0.1564,
      "step": 4554
    },
    {
      "epoch": 0.9561293031066331,
      "grad_norm": 0.11997831612825394,
      "learning_rate": 5.852969094661709e-07,
      "loss": 0.1497,
      "step": 4555
    },
    {
      "epoch": 0.9563392107472712,
      "grad_norm": 0.11431669443845749,
      "learning_rate": 5.797202040358052e-07,
      "loss": 0.1433,
      "step": 4556
    },
    {
      "epoch": 0.9565491183879093,
      "grad_norm": 0.12266053259372711,
      "learning_rate": 5.741700384245596e-07,
      "loss": 0.1592,
      "step": 4557
    },
    {
      "epoch": 0.9567590260285475,
      "grad_norm": 0.09542907774448395,
      "learning_rate": 5.686464156130167e-07,
      "loss": 0.1376,
      "step": 4558
    },
    {
      "epoch": 0.9569689336691856,
      "grad_norm": 0.10986118763685226,
      "learning_rate": 5.631493385674814e-07,
      "loss": 0.173,
      "step": 4559
    },
    {
      "epoch": 0.9571788413098237,
      "grad_norm": 0.12995390594005585,
      "learning_rate": 5.576788102400144e-07,
      "loss": 0.1373,
      "step": 4560
    },
    {
      "epoch": 0.9573887489504618,
      "grad_norm": 0.11493270099163055,
      "learning_rate": 5.522348335683935e-07,
      "loss": 0.1528,
      "step": 4561
    },
    {
      "epoch": 0.9575986565910999,
      "grad_norm": 0.1076839342713356,
      "learning_rate": 5.468174114761859e-07,
      "loss": 0.1406,
      "step": 4562
    },
    {
      "epoch": 0.957808564231738,
      "grad_norm": 0.15728503465652466,
      "learning_rate": 5.414265468726531e-07,
      "loss": 0.1595,
      "step": 4563
    },
    {
      "epoch": 0.9580184718723762,
      "grad_norm": 0.12800243496894836,
      "learning_rate": 5.360622426528183e-07,
      "loss": 0.1685,
      "step": 4564
    },
    {
      "epoch": 0.9582283795130143,
      "grad_norm": 0.11527343839406967,
      "learning_rate": 5.307245016974382e-07,
      "loss": 0.1527,
      "step": 4565
    },
    {
      "epoch": 0.9584382871536524,
      "grad_norm": 0.12676075100898743,
      "learning_rate": 5.254133268729921e-07,
      "loss": 0.161,
      "step": 4566
    },
    {
      "epoch": 0.9586481947942905,
      "grad_norm": 0.13688981533050537,
      "learning_rate": 5.201287210317151e-07,
      "loss": 0.1551,
      "step": 4567
    },
    {
      "epoch": 0.9588581024349286,
      "grad_norm": 0.11639519780874252,
      "learning_rate": 5.148706870115539e-07,
      "loss": 0.1414,
      "step": 4568
    },
    {
      "epoch": 0.9590680100755667,
      "grad_norm": 0.11846988648176193,
      "learning_rate": 5.096392276361883e-07,
      "loss": 0.1552,
      "step": 4569
    },
    {
      "epoch": 0.9592779177162049,
      "grad_norm": 0.1394091099500656,
      "learning_rate": 5.044343457150436e-07,
      "loss": 0.1479,
      "step": 4570
    },
    {
      "epoch": 0.959487825356843,
      "grad_norm": 0.11145713180303574,
      "learning_rate": 4.992560440432503e-07,
      "loss": 0.1563,
      "step": 4571
    },
    {
      "epoch": 0.9596977329974811,
      "grad_norm": 0.11287238448858261,
      "learning_rate": 4.941043254016786e-07,
      "loss": 0.16,
      "step": 4572
    },
    {
      "epoch": 0.9599076406381192,
      "grad_norm": 0.11632929742336273,
      "learning_rate": 4.889791925569209e-07,
      "loss": 0.1571,
      "step": 4573
    },
    {
      "epoch": 0.9601175482787574,
      "grad_norm": 0.12415865063667297,
      "learning_rate": 4.838806482612867e-07,
      "loss": 0.1635,
      "step": 4574
    },
    {
      "epoch": 0.9603274559193955,
      "grad_norm": 0.1393875777721405,
      "learning_rate": 4.788086952528137e-07,
      "loss": 0.166,
      "step": 4575
    },
    {
      "epoch": 0.9605373635600336,
      "grad_norm": 0.11887677013874054,
      "learning_rate": 4.737633362552507e-07,
      "loss": 0.1635,
      "step": 4576
    },
    {
      "epoch": 0.9607472712006717,
      "grad_norm": 0.13373823463916779,
      "learning_rate": 4.6874457397808045e-07,
      "loss": 0.1603,
      "step": 4577
    },
    {
      "epoch": 0.9609571788413098,
      "grad_norm": 0.13354790210723877,
      "learning_rate": 4.637524111164804e-07,
      "loss": 0.1722,
      "step": 4578
    },
    {
      "epoch": 0.9611670864819479,
      "grad_norm": 0.11867116391658783,
      "learning_rate": 4.5878685035136706e-07,
      "loss": 0.1515,
      "step": 4579
    },
    {
      "epoch": 0.961376994122586,
      "grad_norm": 0.12191498279571533,
      "learning_rate": 4.5384789434935735e-07,
      "loss": 0.1505,
      "step": 4580
    },
    {
      "epoch": 0.9615869017632241,
      "grad_norm": 0.10336653143167496,
      "learning_rate": 4.489355457627853e-07,
      "loss": 0.1638,
      "step": 4581
    },
    {
      "epoch": 0.9617968094038623,
      "grad_norm": 0.12683366239070892,
      "learning_rate": 4.440498072296906e-07,
      "loss": 0.1631,
      "step": 4582
    },
    {
      "epoch": 0.9620067170445005,
      "grad_norm": 0.12422499805688858,
      "learning_rate": 4.391906813738245e-07,
      "loss": 0.1663,
      "step": 4583
    },
    {
      "epoch": 0.9622166246851386,
      "grad_norm": 0.12688611447811127,
      "learning_rate": 4.343581708046496e-07,
      "loss": 0.148,
      "step": 4584
    },
    {
      "epoch": 0.9624265323257767,
      "grad_norm": 0.12651990354061127,
      "learning_rate": 4.2955227811734e-07,
      "loss": 0.1581,
      "step": 4585
    },
    {
      "epoch": 0.9626364399664148,
      "grad_norm": 0.14679253101348877,
      "learning_rate": 4.24773005892759e-07,
      "loss": 0.1606,
      "step": 4586
    },
    {
      "epoch": 0.9628463476070529,
      "grad_norm": 0.13088642060756683,
      "learning_rate": 4.2002035669749806e-07,
      "loss": 0.1519,
      "step": 4587
    },
    {
      "epoch": 0.963056255247691,
      "grad_norm": 0.12559828162193298,
      "learning_rate": 4.152943330838377e-07,
      "loss": 0.1526,
      "step": 4588
    },
    {
      "epoch": 0.9632661628883291,
      "grad_norm": 0.10229134559631348,
      "learning_rate": 4.105949375897478e-07,
      "loss": 0.1408,
      "step": 4589
    },
    {
      "epoch": 0.9634760705289672,
      "grad_norm": 0.12710246443748474,
      "learning_rate": 4.059221727389151e-07,
      "loss": 0.1634,
      "step": 4590
    },
    {
      "epoch": 0.9636859781696053,
      "grad_norm": 0.1332884132862091,
      "learning_rate": 4.012760410407324e-07,
      "loss": 0.1715,
      "step": 4591
    },
    {
      "epoch": 0.9638958858102435,
      "grad_norm": 0.11965722590684891,
      "learning_rate": 3.9665654499026486e-07,
      "loss": 0.1553,
      "step": 4592
    },
    {
      "epoch": 0.9641057934508817,
      "grad_norm": 0.12464660406112671,
      "learning_rate": 3.920636870682948e-07,
      "loss": 0.1404,
      "step": 4593
    },
    {
      "epoch": 0.9643157010915198,
      "grad_norm": 0.11357168853282928,
      "learning_rate": 3.874974697412881e-07,
      "loss": 0.132,
      "step": 4594
    },
    {
      "epoch": 0.9645256087321579,
      "grad_norm": 0.12431029975414276,
      "learning_rate": 3.8295789546141656e-07,
      "loss": 0.1411,
      "step": 4595
    },
    {
      "epoch": 0.964735516372796,
      "grad_norm": 0.12060980498790741,
      "learning_rate": 3.7844496666651906e-07,
      "loss": 0.1745,
      "step": 4596
    },
    {
      "epoch": 0.9649454240134341,
      "grad_norm": 0.12177833914756775,
      "learning_rate": 3.7395868578015693e-07,
      "loss": 0.1497,
      "step": 4597
    },
    {
      "epoch": 0.9651553316540722,
      "grad_norm": 0.12002568691968918,
      "learning_rate": 3.694990552115585e-07,
      "loss": 0.1559,
      "step": 4598
    },
    {
      "epoch": 0.9653652392947103,
      "grad_norm": 0.11855290830135345,
      "learning_rate": 3.650660773556469e-07,
      "loss": 0.1494,
      "step": 4599
    },
    {
      "epoch": 0.9655751469353484,
      "grad_norm": 0.12736105918884277,
      "learning_rate": 3.6065975459303456e-07,
      "loss": 0.1301,
      "step": 4600
    },
    {
      "epoch": 0.9657850545759865,
      "grad_norm": 0.12847037613391876,
      "learning_rate": 3.5628008929001753e-07,
      "loss": 0.1367,
      "step": 4601
    },
    {
      "epoch": 0.9659949622166247,
      "grad_norm": 0.11419855803251266,
      "learning_rate": 3.5192708379856997e-07,
      "loss": 0.1432,
      "step": 4602
    },
    {
      "epoch": 0.9662048698572628,
      "grad_norm": 0.11686286330223083,
      "learning_rate": 3.4760074045636637e-07,
      "loss": 0.1561,
      "step": 4603
    },
    {
      "epoch": 0.966414777497901,
      "grad_norm": 0.12151585519313812,
      "learning_rate": 3.433010615867427e-07,
      "loss": 0.1552,
      "step": 4604
    },
    {
      "epoch": 0.966624685138539,
      "grad_norm": 0.14519096910953522,
      "learning_rate": 3.3902804949872967e-07,
      "loss": 0.1587,
      "step": 4605
    },
    {
      "epoch": 0.9668345927791772,
      "grad_norm": 0.12270315736532211,
      "learning_rate": 3.347817064870307e-07,
      "loss": 0.1582,
      "step": 4606
    },
    {
      "epoch": 0.9670445004198153,
      "grad_norm": 0.12619507312774658,
      "learning_rate": 3.3056203483202705e-07,
      "loss": 0.1546,
      "step": 4607
    },
    {
      "epoch": 0.9672544080604534,
      "grad_norm": 0.15550176799297333,
      "learning_rate": 3.2636903679978405e-07,
      "loss": 0.1434,
      "step": 4608
    },
    {
      "epoch": 0.9674643157010915,
      "grad_norm": 0.14259426295757294,
      "learning_rate": 3.222027146420337e-07,
      "loss": 0.1554,
      "step": 4609
    },
    {
      "epoch": 0.9676742233417296,
      "grad_norm": 0.12078682333230972,
      "learning_rate": 3.1806307059618625e-07,
      "loss": 0.1171,
      "step": 4610
    },
    {
      "epoch": 0.9678841309823678,
      "grad_norm": 0.11571485549211502,
      "learning_rate": 3.1395010688532454e-07,
      "loss": 0.153,
      "step": 4611
    },
    {
      "epoch": 0.9680940386230059,
      "grad_norm": 0.15199881792068481,
      "learning_rate": 3.0986382571820406e-07,
      "loss": 0.1633,
      "step": 4612
    },
    {
      "epoch": 0.968303946263644,
      "grad_norm": 0.12040861696004868,
      "learning_rate": 3.0580422928925824e-07,
      "loss": 0.166,
      "step": 4613
    },
    {
      "epoch": 0.9685138539042821,
      "grad_norm": 0.14831073582172394,
      "learning_rate": 3.01771319778571e-07,
      "loss": 0.1626,
      "step": 4614
    },
    {
      "epoch": 0.9687237615449202,
      "grad_norm": 0.12354908138513565,
      "learning_rate": 2.977650993519099e-07,
      "loss": 0.1603,
      "step": 4615
    },
    {
      "epoch": 0.9689336691855583,
      "grad_norm": 0.1218198910355568,
      "learning_rate": 2.937855701607206e-07,
      "loss": 0.1749,
      "step": 4616
    },
    {
      "epoch": 0.9691435768261965,
      "grad_norm": 0.1256335824728012,
      "learning_rate": 2.898327343420826e-07,
      "loss": 0.1417,
      "step": 4617
    },
    {
      "epoch": 0.9693534844668346,
      "grad_norm": 0.3487498164176941,
      "learning_rate": 2.8590659401876996e-07,
      "loss": 0.1604,
      "step": 4618
    },
    {
      "epoch": 0.9695633921074727,
      "grad_norm": 0.100483737885952,
      "learning_rate": 2.8200715129920175e-07,
      "loss": 0.1562,
      "step": 4619
    },
    {
      "epoch": 0.9697732997481109,
      "grad_norm": 0.13053487241268158,
      "learning_rate": 2.7813440827747504e-07,
      "loss": 0.1552,
      "step": 4620
    },
    {
      "epoch": 0.969983207388749,
      "grad_norm": 0.1330847293138504,
      "learning_rate": 2.7428836703333737e-07,
      "loss": 0.1602,
      "step": 4621
    },
    {
      "epoch": 0.9701931150293871,
      "grad_norm": 0.13313066959381104,
      "learning_rate": 2.704690296322032e-07,
      "loss": 0.1574,
      "step": 4622
    },
    {
      "epoch": 0.9704030226700252,
      "grad_norm": 0.11992690712213516,
      "learning_rate": 2.66676398125143e-07,
      "loss": 0.1513,
      "step": 4623
    },
    {
      "epoch": 0.9706129303106633,
      "grad_norm": 0.1217782199382782,
      "learning_rate": 2.629104745488886e-07,
      "loss": 0.1639,
      "step": 4624
    },
    {
      "epoch": 0.9708228379513014,
      "grad_norm": 0.13410820066928864,
      "learning_rate": 2.5917126092582234e-07,
      "loss": 0.1624,
      "step": 4625
    },
    {
      "epoch": 0.9710327455919395,
      "grad_norm": 0.1345527172088623,
      "learning_rate": 2.5545875926398786e-07,
      "loss": 0.14,
      "step": 4626
    },
    {
      "epoch": 0.9712426532325776,
      "grad_norm": 0.11989013850688934,
      "learning_rate": 2.5177297155709046e-07,
      "loss": 0.1574,
      "step": 4627
    },
    {
      "epoch": 0.9714525608732157,
      "grad_norm": 0.12046922743320465,
      "learning_rate": 2.481138997844745e-07,
      "loss": 0.1605,
      "step": 4628
    },
    {
      "epoch": 0.9716624685138538,
      "grad_norm": 0.12484660744667053,
      "learning_rate": 2.44481545911146e-07,
      "loss": 0.1545,
      "step": 4629
    },
    {
      "epoch": 0.9718723761544921,
      "grad_norm": 0.10974811762571335,
      "learning_rate": 2.4087591188776683e-07,
      "loss": 0.1552,
      "step": 4630
    },
    {
      "epoch": 0.9720822837951302,
      "grad_norm": 0.13422971963882446,
      "learning_rate": 2.3729699965063802e-07,
      "loss": 0.1461,
      "step": 4631
    },
    {
      "epoch": 0.9722921914357683,
      "grad_norm": 0.1330391764640808,
      "learning_rate": 2.3374481112172775e-07,
      "loss": 0.1688,
      "step": 4632
    },
    {
      "epoch": 0.9725020990764064,
      "grad_norm": 0.13906224071979523,
      "learning_rate": 2.3021934820862678e-07,
      "loss": 0.1665,
      "step": 4633
    },
    {
      "epoch": 0.9727120067170445,
      "grad_norm": 0.1329282969236374,
      "learning_rate": 2.2672061280460399e-07,
      "loss": 0.1581,
      "step": 4634
    },
    {
      "epoch": 0.9729219143576826,
      "grad_norm": 0.12213360518217087,
      "learning_rate": 2.2324860678855087e-07,
      "loss": 0.1712,
      "step": 4635
    },
    {
      "epoch": 0.9731318219983207,
      "grad_norm": 0.1564663201570511,
      "learning_rate": 2.1980333202502036e-07,
      "loss": 0.1634,
      "step": 4636
    },
    {
      "epoch": 0.9733417296389588,
      "grad_norm": 0.13800211250782013,
      "learning_rate": 2.1638479036419912e-07,
      "loss": 0.1504,
      "step": 4637
    },
    {
      "epoch": 0.9735516372795969,
      "grad_norm": 0.11233645677566528,
      "learning_rate": 2.129929836419242e-07,
      "loss": 0.1623,
      "step": 4638
    },
    {
      "epoch": 0.9737615449202351,
      "grad_norm": 0.13837553560733795,
      "learning_rate": 2.0962791367967195e-07,
      "loss": 0.1611,
      "step": 4639
    },
    {
      "epoch": 0.9739714525608733,
      "grad_norm": 0.11701783537864685,
      "learning_rate": 2.0628958228456341e-07,
      "loss": 0.1553,
      "step": 4640
    },
    {
      "epoch": 0.9741813602015114,
      "grad_norm": 0.13130126893520355,
      "learning_rate": 2.02977991249359e-07,
      "loss": 0.1379,
      "step": 4641
    },
    {
      "epoch": 0.9743912678421495,
      "grad_norm": 0.12259750068187714,
      "learning_rate": 1.996931423524584e-07,
      "loss": 0.1514,
      "step": 4642
    },
    {
      "epoch": 0.9746011754827876,
      "grad_norm": 0.14832593500614166,
      "learning_rate": 1.9643503735789493e-07,
      "loss": 0.1533,
      "step": 4643
    },
    {
      "epoch": 0.9748110831234257,
      "grad_norm": 0.12299138307571411,
      "learning_rate": 1.9320367801535232e-07,
      "loss": 0.1528,
      "step": 4644
    },
    {
      "epoch": 0.9750209907640638,
      "grad_norm": 0.11201336234807968,
      "learning_rate": 1.8999906606014806e-07,
      "loss": 0.1555,
      "step": 4645
    },
    {
      "epoch": 0.9752308984047019,
      "grad_norm": 0.11352569609880447,
      "learning_rate": 1.8682120321322217e-07,
      "loss": 0.1502,
      "step": 4646
    },
    {
      "epoch": 0.97544080604534,
      "grad_norm": 0.12598967552185059,
      "learning_rate": 1.8367009118115953e-07,
      "loss": 0.138,
      "step": 4647
    },
    {
      "epoch": 0.9756507136859782,
      "grad_norm": 0.10122178494930267,
      "learning_rate": 1.8054573165618426e-07,
      "loss": 0.1397,
      "step": 4648
    },
    {
      "epoch": 0.9758606213266163,
      "grad_norm": 0.15723147988319397,
      "learning_rate": 1.7744812631615425e-07,
      "loss": 0.1501,
      "step": 4649
    },
    {
      "epoch": 0.9760705289672544,
      "grad_norm": 0.13425561785697937,
      "learning_rate": 1.743772768245444e-07,
      "loss": 0.1795,
      "step": 4650
    },
    {
      "epoch": 0.9762804366078925,
      "grad_norm": 0.11695228517055511,
      "learning_rate": 1.7133318483047445e-07,
      "loss": 0.1559,
      "step": 4651
    },
    {
      "epoch": 0.9764903442485307,
      "grad_norm": 0.1349976360797882,
      "learning_rate": 1.6831585196869226e-07,
      "loss": 0.1651,
      "step": 4652
    },
    {
      "epoch": 0.9767002518891688,
      "grad_norm": 0.12021506577730179,
      "learning_rate": 1.6532527985957392e-07,
      "loss": 0.1567,
      "step": 4653
    },
    {
      "epoch": 0.9769101595298069,
      "grad_norm": 0.11645930260419846,
      "learning_rate": 1.6236147010912362e-07,
      "loss": 0.1484,
      "step": 4654
    },
    {
      "epoch": 0.977120067170445,
      "grad_norm": 0.12351713329553604,
      "learning_rate": 1.5942442430897376e-07,
      "loss": 0.1496,
      "step": 4655
    },
    {
      "epoch": 0.9773299748110831,
      "grad_norm": 0.11955785006284714,
      "learning_rate": 1.5651414403639043e-07,
      "loss": 0.155,
      "step": 4656
    },
    {
      "epoch": 0.9775398824517213,
      "grad_norm": 0.12057064473628998,
      "learning_rate": 1.5363063085425678e-07,
      "loss": 0.1698,
      "step": 4657
    },
    {
      "epoch": 0.9777497900923594,
      "grad_norm": 0.1271531581878662,
      "learning_rate": 1.507738863110897e-07,
      "loss": 0.1419,
      "step": 4658
    },
    {
      "epoch": 0.9779596977329975,
      "grad_norm": 0.1251571774482727,
      "learning_rate": 1.4794391194101754e-07,
      "loss": 0.1495,
      "step": 4659
    },
    {
      "epoch": 0.9781696053736356,
      "grad_norm": 0.14511661231517792,
      "learning_rate": 1.4514070926380795e-07,
      "loss": 0.1397,
      "step": 4660
    },
    {
      "epoch": 0.9783795130142737,
      "grad_norm": 0.10586293786764145,
      "learning_rate": 1.4236427978484013e-07,
      "loss": 0.1662,
      "step": 4661
    },
    {
      "epoch": 0.9785894206549118,
      "grad_norm": 0.12456677854061127,
      "learning_rate": 1.396146249951269e-07,
      "loss": 0.1655,
      "step": 4662
    },
    {
      "epoch": 0.9787993282955499,
      "grad_norm": 0.1349252313375473,
      "learning_rate": 1.3689174637129264e-07,
      "loss": 0.1753,
      "step": 4663
    },
    {
      "epoch": 0.979009235936188,
      "grad_norm": 0.11700873076915741,
      "learning_rate": 1.341956453755844e-07,
      "loss": 0.1436,
      "step": 4664
    },
    {
      "epoch": 0.9792191435768262,
      "grad_norm": 0.1258639246225357,
      "learning_rate": 1.3152632345586613e-07,
      "loss": 0.1508,
      "step": 4665
    },
    {
      "epoch": 0.9794290512174643,
      "grad_norm": 0.12187707424163818,
      "learning_rate": 1.2888378204563012e-07,
      "loss": 0.1591,
      "step": 4666
    },
    {
      "epoch": 0.9796389588581025,
      "grad_norm": 0.1171417310833931,
      "learning_rate": 1.2626802256398561e-07,
      "loss": 0.1586,
      "step": 4667
    },
    {
      "epoch": 0.9798488664987406,
      "grad_norm": 0.12227719277143478,
      "learning_rate": 1.236790464156423e-07,
      "loss": 0.1579,
      "step": 4668
    },
    {
      "epoch": 0.9800587741393787,
      "grad_norm": 0.12303393334150314,
      "learning_rate": 1.2111685499094915e-07,
      "loss": 0.1452,
      "step": 4669
    },
    {
      "epoch": 0.9802686817800168,
      "grad_norm": 0.12497860938310623,
      "learning_rate": 1.1858144966586104e-07,
      "loss": 0.174,
      "step": 4670
    },
    {
      "epoch": 0.9804785894206549,
      "grad_norm": 0.11376795172691345,
      "learning_rate": 1.160728318019444e-07,
      "loss": 0.1686,
      "step": 4671
    },
    {
      "epoch": 0.980688497061293,
      "grad_norm": 0.12611518800258636,
      "learning_rate": 1.1359100274638824e-07,
      "loss": 0.1358,
      "step": 4672
    },
    {
      "epoch": 0.9808984047019311,
      "grad_norm": 0.13577139377593994,
      "learning_rate": 1.1113596383199309e-07,
      "loss": 0.157,
      "step": 4673
    },
    {
      "epoch": 0.9811083123425692,
      "grad_norm": 0.13323377072811127,
      "learning_rate": 1.0870771637715993e-07,
      "loss": 0.1636,
      "step": 4674
    },
    {
      "epoch": 0.9813182199832073,
      "grad_norm": 0.11575714498758316,
      "learning_rate": 1.0630626168592894e-07,
      "loss": 0.1697,
      "step": 4675
    },
    {
      "epoch": 0.9815281276238456,
      "grad_norm": 0.11939375847578049,
      "learning_rate": 1.039316010479241e-07,
      "loss": 0.151,
      "step": 4676
    },
    {
      "epoch": 0.9817380352644837,
      "grad_norm": 0.13515305519104004,
      "learning_rate": 1.0158373573839197e-07,
      "loss": 0.158,
      "step": 4677
    },
    {
      "epoch": 0.9819479429051218,
      "grad_norm": 0.12420187145471573,
      "learning_rate": 9.926266701820175e-08,
      "loss": 0.1578,
      "step": 4678
    },
    {
      "epoch": 0.9821578505457599,
      "grad_norm": 0.14188450574874878,
      "learning_rate": 9.696839613381193e-08,
      "loss": 0.1716,
      "step": 4679
    },
    {
      "epoch": 0.982367758186398,
      "grad_norm": 0.11995568871498108,
      "learning_rate": 9.470092431729805e-08,
      "loss": 0.147,
      "step": 4680
    },
    {
      "epoch": 0.9825776658270361,
      "grad_norm": 0.1103273257613182,
      "learning_rate": 9.246025278634163e-08,
      "loss": 0.1414,
      "step": 4681
    },
    {
      "epoch": 0.9827875734676742,
      "grad_norm": 0.15861214697360992,
      "learning_rate": 9.024638274424125e-08,
      "loss": 0.1489,
      "step": 4682
    },
    {
      "epoch": 0.9829974811083123,
      "grad_norm": 0.18855759501457214,
      "learning_rate": 8.805931537989586e-08,
      "loss": 0.1508,
      "step": 4683
    },
    {
      "epoch": 0.9832073887489504,
      "grad_norm": 0.11294414848089218,
      "learning_rate": 8.589905186780489e-08,
      "loss": 0.1566,
      "step": 4684
    },
    {
      "epoch": 0.9834172963895886,
      "grad_norm": 0.11468642950057983,
      "learning_rate": 8.376559336807922e-08,
      "loss": 0.1548,
      "step": 4685
    },
    {
      "epoch": 0.9836272040302267,
      "grad_norm": 0.14178474247455597,
      "learning_rate": 8.165894102644123e-08,
      "loss": 0.1494,
      "step": 4686
    },
    {
      "epoch": 0.9838371116708649,
      "grad_norm": 0.12208564579486847,
      "learning_rate": 7.957909597420821e-08,
      "loss": 0.1638,
      "step": 4687
    },
    {
      "epoch": 0.984047019311503,
      "grad_norm": 0.132912278175354,
      "learning_rate": 7.752605932830337e-08,
      "loss": 0.1541,
      "step": 4688
    },
    {
      "epoch": 0.9842569269521411,
      "grad_norm": 0.11923284828662872,
      "learning_rate": 7.549983219125035e-08,
      "loss": 0.1585,
      "step": 4689
    },
    {
      "epoch": 0.9844668345927792,
      "grad_norm": 0.12066918611526489,
      "learning_rate": 7.350041565118981e-08,
      "loss": 0.1498,
      "step": 4690
    },
    {
      "epoch": 0.9846767422334173,
      "grad_norm": 0.16000604629516602,
      "learning_rate": 7.152781078184622e-08,
      "loss": 0.1571,
      "step": 4691
    },
    {
      "epoch": 0.9848866498740554,
      "grad_norm": 0.11645068973302841,
      "learning_rate": 6.958201864254999e-08,
      "loss": 0.1472,
      "step": 4692
    },
    {
      "epoch": 0.9850965575146935,
      "grad_norm": 0.11970088630914688,
      "learning_rate": 6.766304027824855e-08,
      "loss": 0.1431,
      "step": 4693
    },
    {
      "epoch": 0.9853064651553316,
      "grad_norm": 0.11811143159866333,
      "learning_rate": 6.577087671946758e-08,
      "loss": 0.1406,
      "step": 4694
    },
    {
      "epoch": 0.9855163727959698,
      "grad_norm": 0.13158312439918518,
      "learning_rate": 6.390552898234425e-08,
      "loss": 0.1644,
      "step": 4695
    },
    {
      "epoch": 0.9857262804366079,
      "grad_norm": 0.11352595686912537,
      "learning_rate": 6.206699806861061e-08,
      "loss": 0.1587,
      "step": 4696
    },
    {
      "epoch": 0.985936188077246,
      "grad_norm": 0.11427728831768036,
      "learning_rate": 6.025528496560462e-08,
      "loss": 0.1509,
      "step": 4697
    },
    {
      "epoch": 0.9861460957178841,
      "grad_norm": 0.1264103204011917,
      "learning_rate": 5.847039064625359e-08,
      "loss": 0.1508,
      "step": 4698
    },
    {
      "epoch": 0.9863560033585222,
      "grad_norm": 0.11748611181974411,
      "learning_rate": 5.671231606909078e-08,
      "loss": 0.1546,
      "step": 4699
    },
    {
      "epoch": 0.9865659109991604,
      "grad_norm": 0.1317833513021469,
      "learning_rate": 5.498106217823873e-08,
      "loss": 0.1345,
      "step": 4700
    },
    {
      "epoch": 0.9867758186397985,
      "grad_norm": 0.12351711839437485,
      "learning_rate": 5.327662990342042e-08,
      "loss": 0.1615,
      "step": 4701
    },
    {
      "epoch": 0.9869857262804366,
      "grad_norm": 0.11820079386234283,
      "learning_rate": 5.159902015995366e-08,
      "loss": 0.144,
      "step": 4702
    },
    {
      "epoch": 0.9871956339210747,
      "grad_norm": 0.13129431009292603,
      "learning_rate": 4.994823384876224e-08,
      "loss": 0.1539,
      "step": 4703
    },
    {
      "epoch": 0.9874055415617129,
      "grad_norm": 0.12318190187215805,
      "learning_rate": 4.832427185634259e-08,
      "loss": 0.1536,
      "step": 4704
    },
    {
      "epoch": 0.987615449202351,
      "grad_norm": 0.12245703488588333,
      "learning_rate": 4.672713505480819e-08,
      "loss": 0.1579,
      "step": 4705
    },
    {
      "epoch": 0.9878253568429891,
      "grad_norm": 0.12133902311325073,
      "learning_rate": 4.5156824301856306e-08,
      "loss": 0.1676,
      "step": 4706
    },
    {
      "epoch": 0.9880352644836272,
      "grad_norm": 0.10715369135141373,
      "learning_rate": 4.361334044077903e-08,
      "loss": 0.1446,
      "step": 4707
    },
    {
      "epoch": 0.9882451721242653,
      "grad_norm": 0.12753887474536896,
      "learning_rate": 4.209668430046332e-08,
      "loss": 0.1637,
      "step": 4708
    },
    {
      "epoch": 0.9884550797649034,
      "grad_norm": 0.15687881410121918,
      "learning_rate": 4.0606856695385444e-08,
      "loss": 0.1553,
      "step": 4709
    },
    {
      "epoch": 0.9886649874055415,
      "grad_norm": 0.12655100226402283,
      "learning_rate": 3.914385842562207e-08,
      "loss": 0.1387,
      "step": 4710
    },
    {
      "epoch": 0.9888748950461796,
      "grad_norm": 0.11232425272464752,
      "learning_rate": 3.770769027683363e-08,
      "loss": 0.1517,
      "step": 4711
    },
    {
      "epoch": 0.9890848026868178,
      "grad_norm": 0.11365073174238205,
      "learning_rate": 3.629835302026985e-08,
      "loss": 0.1573,
      "step": 4712
    },
    {
      "epoch": 0.989294710327456,
      "grad_norm": 0.1190265566110611,
      "learning_rate": 3.491584741278642e-08,
      "loss": 0.1549,
      "step": 4713
    },
    {
      "epoch": 0.9895046179680941,
      "grad_norm": 0.12508736550807953,
      "learning_rate": 3.356017419681723e-08,
      "loss": 0.1497,
      "step": 4714
    },
    {
      "epoch": 0.9897145256087322,
      "grad_norm": 0.11955531686544418,
      "learning_rate": 3.223133410038548e-08,
      "loss": 0.1601,
      "step": 4715
    },
    {
      "epoch": 0.9899244332493703,
      "grad_norm": 0.0987209752202034,
      "learning_rate": 3.0929327837114774e-08,
      "loss": 0.1482,
      "step": 4716
    },
    {
      "epoch": 0.9901343408900084,
      "grad_norm": 0.13037124276161194,
      "learning_rate": 2.9654156106206922e-08,
      "loss": 0.1531,
      "step": 4717
    },
    {
      "epoch": 0.9903442485306465,
      "grad_norm": 0.1238628700375557,
      "learning_rate": 2.8405819592464135e-08,
      "loss": 0.1519,
      "step": 4718
    },
    {
      "epoch": 0.9905541561712846,
      "grad_norm": 0.14739884436130524,
      "learning_rate": 2.718431896626683e-08,
      "loss": 0.1736,
      "step": 4719
    },
    {
      "epoch": 0.9907640638119227,
      "grad_norm": 0.11886577308177948,
      "learning_rate": 2.5989654883590286e-08,
      "loss": 0.1669,
      "step": 4720
    },
    {
      "epoch": 0.9909739714525608,
      "grad_norm": 0.09993169456720352,
      "learning_rate": 2.482182798599353e-08,
      "loss": 0.1513,
      "step": 4721
    },
    {
      "epoch": 0.9911838790931989,
      "grad_norm": 0.13532714545726776,
      "learning_rate": 2.3680838900635995e-08,
      "loss": 0.1597,
      "step": 4722
    },
    {
      "epoch": 0.9913937867338372,
      "grad_norm": 0.13656656444072723,
      "learning_rate": 2.2566688240244216e-08,
      "loss": 0.1385,
      "step": 4723
    },
    {
      "epoch": 0.9916036943744753,
      "grad_norm": 0.11988739669322968,
      "learning_rate": 2.147937660314514e-08,
      "loss": 0.1531,
      "step": 4724
    },
    {
      "epoch": 0.9918136020151134,
      "grad_norm": 0.15171854197978973,
      "learning_rate": 2.0418904573249464e-08,
      "loss": 0.154,
      "step": 4725
    },
    {
      "epoch": 0.9920235096557515,
      "grad_norm": 0.10903916507959366,
      "learning_rate": 1.938527272005719e-08,
      "loss": 0.1532,
      "step": 4726
    },
    {
      "epoch": 0.9922334172963896,
      "grad_norm": 0.13691365718841553,
      "learning_rate": 1.8378481598657627e-08,
      "loss": 0.1564,
      "step": 4727
    },
    {
      "epoch": 0.9924433249370277,
      "grad_norm": 0.10917693376541138,
      "learning_rate": 1.739853174970718e-08,
      "loss": 0.157,
      "step": 4728
    },
    {
      "epoch": 0.9926532325776658,
      "grad_norm": 0.11434437334537506,
      "learning_rate": 1.644542369947377e-08,
      "loss": 0.1492,
      "step": 4729
    },
    {
      "epoch": 0.9928631402183039,
      "grad_norm": 0.13532006740570068,
      "learning_rate": 1.5519157959792417e-08,
      "loss": 0.1622,
      "step": 4730
    },
    {
      "epoch": 0.993073047858942,
      "grad_norm": 0.11654216051101685,
      "learning_rate": 1.4619735028087445e-08,
      "loss": 0.173,
      "step": 4731
    },
    {
      "epoch": 0.9932829554995802,
      "grad_norm": 0.13726460933685303,
      "learning_rate": 1.374715538737248e-08,
      "loss": 0.148,
      "step": 4732
    },
    {
      "epoch": 0.9934928631402183,
      "grad_norm": 0.12805287539958954,
      "learning_rate": 1.2901419506239354e-08,
      "loss": 0.1477,
      "step": 4733
    },
    {
      "epoch": 0.9937027707808564,
      "grad_norm": 0.12175727635622025,
      "learning_rate": 1.2082527838874757e-08,
      "loss": 0.1736,
      "step": 4734
    },
    {
      "epoch": 0.9939126784214946,
      "grad_norm": 0.14802756905555725,
      "learning_rate": 1.129048082503803e-08,
      "loss": 0.1662,
      "step": 4735
    },
    {
      "epoch": 0.9941225860621327,
      "grad_norm": 0.09921912103891373,
      "learning_rate": 1.0525278890072265e-08,
      "loss": 0.1419,
      "step": 4736
    },
    {
      "epoch": 0.9943324937027708,
      "grad_norm": 0.1295527070760727,
      "learning_rate": 9.786922444915415e-09,
      "loss": 0.1565,
      "step": 4737
    },
    {
      "epoch": 0.9945424013434089,
      "grad_norm": 0.13671144843101501,
      "learning_rate": 9.075411886078078e-09,
      "loss": 0.1452,
      "step": 4738
    },
    {
      "epoch": 0.994752308984047,
      "grad_norm": 0.1243344098329544,
      "learning_rate": 8.390747595654613e-09,
      "loss": 0.1305,
      "step": 4739
    },
    {
      "epoch": 0.9949622166246851,
      "grad_norm": 0.12893415987491608,
      "learning_rate": 7.732929941334233e-09,
      "loss": 0.1589,
      "step": 4740
    },
    {
      "epoch": 0.9951721242653233,
      "grad_norm": 0.1381278932094574,
      "learning_rate": 7.101959276373249e-09,
      "loss": 0.159,
      "step": 4741
    },
    {
      "epoch": 0.9953820319059614,
      "grad_norm": 0.11679843813180923,
      "learning_rate": 6.4978359396172806e-09,
      "loss": 0.168,
      "step": 4742
    },
    {
      "epoch": 0.9955919395465995,
      "grad_norm": 0.13081474602222443,
      "learning_rate": 5.9205602554957e-09,
      "loss": 0.1489,
      "step": 4743
    },
    {
      "epoch": 0.9958018471872376,
      "grad_norm": 0.13281650841236115,
      "learning_rate": 5.370132534021632e-09,
      "loss": 0.1624,
      "step": 4744
    },
    {
      "epoch": 0.9960117548278757,
      "grad_norm": 0.10714995861053467,
      "learning_rate": 4.8465530707808575e-09,
      "loss": 0.1534,
      "step": 4745
    },
    {
      "epoch": 0.9962216624685138,
      "grad_norm": 0.11529166996479034,
      "learning_rate": 4.349822146954008e-09,
      "loss": 0.1494,
      "step": 4746
    },
    {
      "epoch": 0.996431570109152,
      "grad_norm": 0.11040196567773819,
      "learning_rate": 3.879940029294371e-09,
      "loss": 0.1583,
      "step": 4747
    },
    {
      "epoch": 0.9966414777497901,
      "grad_norm": 0.10761739313602448,
      "learning_rate": 3.436906970138987e-09,
      "loss": 0.163,
      "step": 4748
    },
    {
      "epoch": 0.9968513853904282,
      "grad_norm": 0.13290688395500183,
      "learning_rate": 3.020723207408649e-09,
      "loss": 0.1525,
      "step": 4749
    },
    {
      "epoch": 0.9970612930310663,
      "grad_norm": 0.10621462017297745,
      "learning_rate": 2.631388964596804e-09,
      "loss": 0.1414,
      "step": 4750
    },
    {
      "epoch": 0.9972712006717045,
      "grad_norm": 0.12779180705547333,
      "learning_rate": 2.2689044507917534e-09,
      "loss": 0.164,
      "step": 4751
    },
    {
      "epoch": 0.9974811083123426,
      "grad_norm": 0.12876422703266144,
      "learning_rate": 1.9332698606544517e-09,
      "loss": 0.1419,
      "step": 4752
    },
    {
      "epoch": 0.9976910159529807,
      "grad_norm": 0.12758982181549072,
      "learning_rate": 1.6244853744296073e-09,
      "loss": 0.1386,
      "step": 4753
    },
    {
      "epoch": 0.9979009235936188,
      "grad_norm": 0.11979212611913681,
      "learning_rate": 1.3425511579401307e-09,
      "loss": 0.1541,
      "step": 4754
    },
    {
      "epoch": 0.9981108312342569,
      "grad_norm": 0.12721380591392517,
      "learning_rate": 1.0874673625926868e-09,
      "loss": 0.1404,
      "step": 4755
    },
    {
      "epoch": 0.998320738874895,
      "grad_norm": 0.12794804573059082,
      "learning_rate": 8.592341253665925e-10,
      "loss": 0.159,
      "step": 4756
    },
    {
      "epoch": 0.9985306465155331,
      "grad_norm": 0.11918237060308456,
      "learning_rate": 6.578515688360209e-10,
      "loss": 0.1439,
      "step": 4757
    },
    {
      "epoch": 0.9987405541561712,
      "grad_norm": 0.1348177194595337,
      "learning_rate": 4.833198011422457e-10,
      "loss": 0.171,
      "step": 4758
    },
    {
      "epoch": 0.9989504617968094,
      "grad_norm": 0.1429336965084076,
      "learning_rate": 3.3563891602139685e-10,
      "loss": 0.1625,
      "step": 4759
    },
    {
      "epoch": 0.9991603694374476,
      "grad_norm": 0.10812459141016006,
      "learning_rate": 2.1480899277115385e-10,
      "loss": 0.1494,
      "step": 4760
    },
    {
      "epoch": 0.9993702770780857,
      "grad_norm": 0.12720505893230438,
      "learning_rate": 1.2083009628405252e-10,
      "loss": 0.1366,
      "step": 4761
    },
    {
      "epoch": 0.9995801847187238,
      "grad_norm": 0.10869085788726807,
      "learning_rate": 5.3702277030831524e-11,
      "loss": 0.1306,
      "step": 4762
    },
    {
      "epoch": 0.9997900923593619,
      "grad_norm": 0.13745978474617004,
      "learning_rate": 1.3425571060432518e-11,
      "loss": 0.1508,
      "step": 4763
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1369509994983673,
      "learning_rate": 0.0,
      "loss": 0.1475,
      "step": 4764
    },
    {
      "epoch": 1.0,
      "step": 4764,
      "total_flos": 3.4097062376237957e+18,
      "train_loss": 0.17368370302859773,
      "train_runtime": 25648.7488,
      "train_samples_per_second": 5.943,
      "train_steps_per_second": 0.186
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 4764,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.4097062376237957e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}