{
  "best_metric": 0.08259893208742142,
  "best_model_checkpoint": "arabic-embedding-model-only-pairs-v2/checkpoint-49115",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 49115,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0025450473378804844,
      "grad_norm": 17.451679229736328,
      "learning_rate": 1.0179153094462541e-07,
      "loss": 0.3458,
      "step": 25
    },
    {
      "epoch": 0.005090094675760969,
      "grad_norm": 8.519526481628418,
      "learning_rate": 2.0358306188925083e-07,
      "loss": 0.3213,
      "step": 50
    },
    {
      "epoch": 0.0076351420136414536,
      "grad_norm": 14.82270622253418,
      "learning_rate": 3.053745928338762e-07,
      "loss": 0.3414,
      "step": 75
    },
    {
      "epoch": 0.010180189351521938,
      "grad_norm": 20.397518157958984,
      "learning_rate": 4.0716612377850166e-07,
      "loss": 0.3622,
      "step": 100
    },
    {
      "epoch": 0.012725236689402423,
      "grad_norm": 14.741048812866211,
      "learning_rate": 5.089576547231271e-07,
      "loss": 0.3607,
      "step": 125
    },
    {
      "epoch": 0.015270284027282907,
      "grad_norm": 16.42844581604004,
      "learning_rate": 6.107491856677524e-07,
      "loss": 0.3514,
      "step": 150
    },
    {
      "epoch": 0.017815331365163393,
      "grad_norm": 4.865312099456787,
      "learning_rate": 7.12540716612378e-07,
      "loss": 0.2282,
      "step": 175
    },
    {
      "epoch": 0.020360378703043875,
      "grad_norm": 14.216117858886719,
      "learning_rate": 8.143322475570033e-07,
      "loss": 0.2781,
      "step": 200
    },
    {
      "epoch": 0.02290542604092436,
      "grad_norm": 11.410103797912598,
      "learning_rate": 9.161237785016288e-07,
      "loss": 0.3206,
      "step": 225
    },
    {
      "epoch": 0.025450473378804846,
      "grad_norm": 13.930344581604004,
      "learning_rate": 1.0179153094462542e-06,
      "loss": 0.3592,
      "step": 250
    },
    {
      "epoch": 0.027995520716685332,
      "grad_norm": 9.695392608642578,
      "learning_rate": 1.1197068403908795e-06,
      "loss": 0.2837,
      "step": 275
    },
    {
      "epoch": 0.030540568054565814,
      "grad_norm": 19.564462661743164,
      "learning_rate": 1.2214983713355049e-06,
      "loss": 0.2227,
      "step": 300
    },
    {
      "epoch": 0.033085615392446296,
      "grad_norm": 6.403764247894287,
      "learning_rate": 1.3232899022801304e-06,
      "loss": 0.3072,
      "step": 325
    },
    {
      "epoch": 0.035630662730326786,
      "grad_norm": 4.398595333099365,
      "learning_rate": 1.425081433224756e-06,
      "loss": 0.3487,
      "step": 350
    },
    {
      "epoch": 0.03817571006820727,
      "grad_norm": 14.698935508728027,
      "learning_rate": 1.5268729641693813e-06,
      "loss": 0.2439,
      "step": 375
    },
    {
      "epoch": 0.04072075740608775,
      "grad_norm": 4.241194248199463,
      "learning_rate": 1.6286644951140066e-06,
      "loss": 0.291,
      "step": 400
    },
    {
      "epoch": 0.04326580474396824,
      "grad_norm": 12.753772735595703,
      "learning_rate": 1.730456026058632e-06,
      "loss": 0.2325,
      "step": 425
    },
    {
      "epoch": 0.04581085208184872,
      "grad_norm": 8.159408569335938,
      "learning_rate": 1.8322475570032575e-06,
      "loss": 0.3068,
      "step": 450
    },
    {
      "epoch": 0.048355899419729204,
      "grad_norm": 7.215402126312256,
      "learning_rate": 1.934039087947883e-06,
      "loss": 0.3268,
      "step": 475
    },
    {
      "epoch": 0.05090094675760969,
      "grad_norm": 13.441118240356445,
      "learning_rate": 2.0358306188925084e-06,
      "loss": 0.2973,
      "step": 500
    },
    {
      "epoch": 0.053445994095490175,
      "grad_norm": 4.826434135437012,
      "learning_rate": 2.137622149837134e-06,
      "loss": 0.3116,
      "step": 525
    },
    {
      "epoch": 0.055991041433370664,
      "grad_norm": 9.862593650817871,
      "learning_rate": 2.239413680781759e-06,
      "loss": 0.2543,
      "step": 550
    },
    {
      "epoch": 0.058536088771251146,
      "grad_norm": 7.720365524291992,
      "learning_rate": 2.3412052117263846e-06,
      "loss": 0.2747,
      "step": 575
    },
    {
      "epoch": 0.06108113610913163,
      "grad_norm": 10.59146785736084,
      "learning_rate": 2.4429967426710097e-06,
      "loss": 0.2809,
      "step": 600
    },
    {
      "epoch": 0.06362618344701211,
      "grad_norm": 9.026193618774414,
      "learning_rate": 2.5447882736156353e-06,
      "loss": 0.3561,
      "step": 625
    },
    {
      "epoch": 0.06617123078489259,
      "grad_norm": 11.523633003234863,
      "learning_rate": 2.6425081433224757e-06,
      "loss": 0.2683,
      "step": 650
    },
    {
      "epoch": 0.06871627812277309,
      "grad_norm": 13.002677917480469,
      "learning_rate": 2.7442996742671013e-06,
      "loss": 0.249,
      "step": 675
    },
    {
      "epoch": 0.07126132546065357,
      "grad_norm": 12.167911529541016,
      "learning_rate": 2.846091205211727e-06,
      "loss": 0.2482,
      "step": 700
    },
    {
      "epoch": 0.07380637279853405,
      "grad_norm": 7.002504348754883,
      "learning_rate": 2.9478827361563524e-06,
      "loss": 0.3015,
      "step": 725
    },
    {
      "epoch": 0.07635142013641454,
      "grad_norm": 14.07034683227539,
      "learning_rate": 3.0496742671009775e-06,
      "loss": 0.2422,
      "step": 750
    },
    {
      "epoch": 0.07889646747429502,
      "grad_norm": 9.051650047302246,
      "learning_rate": 3.1514657980456026e-06,
      "loss": 0.259,
      "step": 775
    },
    {
      "epoch": 0.0814415148121755,
      "grad_norm": 7.5605692863464355,
      "learning_rate": 3.253257328990228e-06,
      "loss": 0.2565,
      "step": 800
    },
    {
      "epoch": 0.083986562150056,
      "grad_norm": 7.689245223999023,
      "learning_rate": 3.3550488599348537e-06,
      "loss": 0.254,
      "step": 825
    },
    {
      "epoch": 0.08653160948793648,
      "grad_norm": 10.289682388305664,
      "learning_rate": 3.456840390879479e-06,
      "loss": 0.2366,
      "step": 850
    },
    {
      "epoch": 0.08907665682581696,
      "grad_norm": 14.462525367736816,
      "learning_rate": 3.5586319218241044e-06,
      "loss": 0.2805,
      "step": 875
    },
    {
      "epoch": 0.09162170416369744,
      "grad_norm": 12.019689559936523,
      "learning_rate": 3.66042345276873e-06,
      "loss": 0.2881,
      "step": 900
    },
    {
      "epoch": 0.09416675150157792,
      "grad_norm": 6.248305320739746,
      "learning_rate": 3.7622149837133555e-06,
      "loss": 0.2424,
      "step": 925
    },
    {
      "epoch": 0.09671179883945841,
      "grad_norm": 8.205031394958496,
      "learning_rate": 3.864006514657981e-06,
      "loss": 0.2342,
      "step": 950
    },
    {
      "epoch": 0.0992568461773389,
      "grad_norm": 10.721994400024414,
      "learning_rate": 3.965798045602606e-06,
      "loss": 0.2036,
      "step": 975
    },
    {
      "epoch": 0.10180189351521939,
      "grad_norm": 10.379674911499023,
      "learning_rate": 4.067589576547232e-06,
      "loss": 0.2432,
      "step": 1000
    },
    {
      "epoch": 0.10434694085309987,
      "grad_norm": 11.975798606872559,
      "learning_rate": 4.169381107491857e-06,
      "loss": 0.2782,
      "step": 1025
    },
    {
      "epoch": 0.10689198819098035,
      "grad_norm": 9.442329406738281,
      "learning_rate": 4.271172638436483e-06,
      "loss": 0.2303,
      "step": 1050
    },
    {
      "epoch": 0.10943703552886083,
      "grad_norm": 11.444186210632324,
      "learning_rate": 4.372964169381108e-06,
      "loss": 0.2418,
      "step": 1075
    },
    {
      "epoch": 0.11198208286674133,
      "grad_norm": 7.217404365539551,
      "learning_rate": 4.474755700325733e-06,
      "loss": 0.1989,
      "step": 1100
    },
    {
      "epoch": 0.11452713020462181,
      "grad_norm": 13.797436714172363,
      "learning_rate": 4.5765472312703586e-06,
      "loss": 0.2189,
      "step": 1125
    },
    {
      "epoch": 0.11707217754250229,
      "grad_norm": 7.028829097747803,
      "learning_rate": 4.678338762214984e-06,
      "loss": 0.2295,
      "step": 1150
    },
    {
      "epoch": 0.11961722488038277,
      "grad_norm": 8.137258529663086,
      "learning_rate": 4.78013029315961e-06,
      "loss": 0.1583,
      "step": 1175
    },
    {
      "epoch": 0.12216227221826326,
      "grad_norm": 10.967634201049805,
      "learning_rate": 4.881921824104235e-06,
      "loss": 0.2247,
      "step": 1200
    },
    {
      "epoch": 0.12470731955614374,
      "grad_norm": 4.914444446563721,
      "learning_rate": 4.98371335504886e-06,
      "loss": 0.2127,
      "step": 1225
    },
    {
      "epoch": 0.12725236689402422,
      "grad_norm": 6.940595626831055,
      "learning_rate": 5.0855048859934855e-06,
      "loss": 0.2244,
      "step": 1250
    },
    {
      "epoch": 0.12979741423190472,
      "grad_norm": 11.294093132019043,
      "learning_rate": 5.187296416938111e-06,
      "loss": 0.2218,
      "step": 1275
    },
    {
      "epoch": 0.13234246156978519,
      "grad_norm": 9.178437232971191,
      "learning_rate": 5.2890879478827366e-06,
      "loss": 0.2489,
      "step": 1300
    },
    {
      "epoch": 0.13488750890766568,
      "grad_norm": 7.749022006988525,
      "learning_rate": 5.390879478827362e-06,
      "loss": 0.1778,
      "step": 1325
    },
    {
      "epoch": 0.13743255624554618,
      "grad_norm": 32.12667465209961,
      "learning_rate": 5.492671009771987e-06,
      "loss": 0.2243,
      "step": 1350
    },
    {
      "epoch": 0.13997760358342665,
      "grad_norm": 8.55077075958252,
      "learning_rate": 5.594462540716613e-06,
      "loss": 0.2063,
      "step": 1375
    },
    {
      "epoch": 0.14252265092130714,
      "grad_norm": 15.310531616210938,
      "learning_rate": 5.696254071661238e-06,
      "loss": 0.2187,
      "step": 1400
    },
    {
      "epoch": 0.1450676982591876,
      "grad_norm": 11.116787910461426,
      "learning_rate": 5.798045602605864e-06,
      "loss": 0.204,
      "step": 1425
    },
    {
      "epoch": 0.1476127455970681,
      "grad_norm": 6.040707588195801,
      "learning_rate": 5.899837133550489e-06,
      "loss": 0.2039,
      "step": 1450
    },
    {
      "epoch": 0.1501577929349486,
      "grad_norm": 10.210320472717285,
      "learning_rate": 6.0016286644951145e-06,
      "loss": 0.2186,
      "step": 1475
    },
    {
      "epoch": 0.15270284027282907,
      "grad_norm": 11.306329727172852,
      "learning_rate": 6.10342019543974e-06,
      "loss": 0.2352,
      "step": 1500
    },
    {
      "epoch": 0.15524788761070957,
      "grad_norm": 8.576631546020508,
      "learning_rate": 6.205211726384366e-06,
      "loss": 0.2578,
      "step": 1525
    },
    {
      "epoch": 0.15779293494859004,
      "grad_norm": 16.613567352294922,
      "learning_rate": 6.307003257328991e-06,
      "loss": 0.2047,
      "step": 1550
    },
    {
      "epoch": 0.16033798228647053,
      "grad_norm": 6.0128936767578125,
      "learning_rate": 6.408794788273616e-06,
      "loss": 0.1833,
      "step": 1575
    },
    {
      "epoch": 0.162883029624351,
      "grad_norm": 11.588747024536133,
      "learning_rate": 6.5105863192182414e-06,
      "loss": 0.2564,
      "step": 1600
    },
    {
      "epoch": 0.1654280769622315,
      "grad_norm": 14.360913276672363,
      "learning_rate": 6.612377850162867e-06,
      "loss": 0.2021,
      "step": 1625
    },
    {
      "epoch": 0.167973124300112,
      "grad_norm": 7.827113628387451,
      "learning_rate": 6.7141693811074925e-06,
      "loss": 0.1901,
      "step": 1650
    },
    {
      "epoch": 0.17051817163799246,
      "grad_norm": 6.506629467010498,
      "learning_rate": 6.815960912052117e-06,
      "loss": 0.2183,
      "step": 1675
    },
    {
      "epoch": 0.17306321897587296,
      "grad_norm": 9.81983470916748,
      "learning_rate": 6.917752442996744e-06,
      "loss": 0.1661,
      "step": 1700
    },
    {
      "epoch": 0.17560826631375342,
      "grad_norm": 15.31557559967041,
      "learning_rate": 7.019543973941368e-06,
      "loss": 0.1797,
      "step": 1725
    },
    {
      "epoch": 0.17815331365163392,
      "grad_norm": 13.521235466003418,
      "learning_rate": 7.121335504885995e-06,
      "loss": 0.2105,
      "step": 1750
    },
    {
      "epoch": 0.18069836098951442,
      "grad_norm": 6.115505695343018,
      "learning_rate": 7.223127035830619e-06,
      "loss": 0.2214,
      "step": 1775
    },
    {
      "epoch": 0.18324340832739489,
      "grad_norm": 11.252988815307617,
      "learning_rate": 7.324918566775245e-06,
      "loss": 0.2475,
      "step": 1800
    },
    {
      "epoch": 0.18578845566527538,
      "grad_norm": 20.72943878173828,
      "learning_rate": 7.4267100977198705e-06,
      "loss": 0.2395,
      "step": 1825
    },
    {
      "epoch": 0.18833350300315585,
      "grad_norm": 12.99436092376709,
      "learning_rate": 7.528501628664495e-06,
      "loss": 0.1911,
      "step": 1850
    },
    {
      "epoch": 0.19087855034103635,
      "grad_norm": 3.998025417327881,
      "learning_rate": 7.630293159609122e-06,
      "loss": 0.1588,
      "step": 1875
    },
    {
      "epoch": 0.19342359767891681,
      "grad_norm": 6.960880279541016,
      "learning_rate": 7.732084690553745e-06,
      "loss": 0.2358,
      "step": 1900
    },
    {
      "epoch": 0.1959686450167973,
      "grad_norm": 7.048129558563232,
      "learning_rate": 7.833876221498373e-06,
      "loss": 0.1773,
      "step": 1925
    },
    {
      "epoch": 0.1985136923546778,
      "grad_norm": 35.43663787841797,
      "learning_rate": 7.935667752442997e-06,
      "loss": 0.2347,
      "step": 1950
    },
    {
      "epoch": 0.20105873969255827,
      "grad_norm": 4.078209400177002,
      "learning_rate": 8.037459283387624e-06,
      "loss": 0.188,
      "step": 1975
    },
    {
      "epoch": 0.20360378703043877,
      "grad_norm": 7.5787458419799805,
      "learning_rate": 8.139250814332248e-06,
      "loss": 0.1925,
      "step": 2000
    },
    {
      "epoch": 0.20614883436831924,
      "grad_norm": 5.975426197052002,
      "learning_rate": 8.241042345276873e-06,
      "loss": 0.191,
      "step": 2025
    },
    {
      "epoch": 0.20869388170619974,
      "grad_norm": 18.67757797241211,
      "learning_rate": 8.342833876221499e-06,
      "loss": 0.2032,
      "step": 2050
    },
    {
      "epoch": 0.21123892904408023,
      "grad_norm": 7.5380449295043945,
      "learning_rate": 8.444625407166124e-06,
      "loss": 0.1787,
      "step": 2075
    },
    {
      "epoch": 0.2137839763819607,
      "grad_norm": 11.195531845092773,
      "learning_rate": 8.54641693811075e-06,
      "loss": 0.2327,
      "step": 2100
    },
    {
      "epoch": 0.2163290237198412,
      "grad_norm": 3.4572274684906006,
      "learning_rate": 8.648208469055375e-06,
      "loss": 0.196,
      "step": 2125
    },
    {
      "epoch": 0.21887407105772166,
      "grad_norm": 8.280584335327148,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.1945,
      "step": 2150
    },
    {
      "epoch": 0.22141911839560216,
      "grad_norm": 7.217907905578613,
      "learning_rate": 8.851791530944625e-06,
      "loss": 0.1918,
      "step": 2175
    },
    {
      "epoch": 0.22396416573348266,
      "grad_norm": 2.9696857929229736,
      "learning_rate": 8.953583061889252e-06,
      "loss": 0.2011,
      "step": 2200
    },
    {
      "epoch": 0.22650921307136312,
      "grad_norm": 13.338550567626953,
      "learning_rate": 9.055374592833876e-06,
      "loss": 0.1711,
      "step": 2225
    },
    {
      "epoch": 0.22905426040924362,
      "grad_norm": 11.651854515075684,
      "learning_rate": 9.157166123778503e-06,
      "loss": 0.2687,
      "step": 2250
    },
    {
      "epoch": 0.2315993077471241,
      "grad_norm": 1.1671559810638428,
      "learning_rate": 9.258957654723127e-06,
      "loss": 0.1998,
      "step": 2275
    },
    {
      "epoch": 0.23414435508500459,
      "grad_norm": 8.616079330444336,
      "learning_rate": 9.360749185667754e-06,
      "loss": 0.1604,
      "step": 2300
    },
    {
      "epoch": 0.23668940242288505,
      "grad_norm": 4.227112770080566,
      "learning_rate": 9.462540716612378e-06,
      "loss": 0.1871,
      "step": 2325
    },
    {
      "epoch": 0.23923444976076555,
      "grad_norm": 7.659778594970703,
      "learning_rate": 9.564332247557004e-06,
      "loss": 0.273,
      "step": 2350
    },
    {
      "epoch": 0.24177949709864605,
      "grad_norm": 5.232609272003174,
      "learning_rate": 9.66612377850163e-06,
      "loss": 0.2405,
      "step": 2375
    },
    {
      "epoch": 0.24432454443652651,
      "grad_norm": 8.198054313659668,
      "learning_rate": 9.767915309446255e-06,
      "loss": 0.2212,
      "step": 2400
    },
    {
      "epoch": 0.246869591774407,
      "grad_norm": 10.666227340698242,
      "learning_rate": 9.86970684039088e-06,
      "loss": 0.2293,
      "step": 2425
    },
    {
      "epoch": 0.24941463911228748,
      "grad_norm": 12.714786529541016,
      "learning_rate": 9.971498371335506e-06,
      "loss": 0.1887,
      "step": 2450
    },
    {
      "epoch": 0.251959686450168,
      "grad_norm": 9.544090270996094,
      "learning_rate": 1.007328990228013e-05,
      "loss": 0.233,
      "step": 2475
    },
    {
      "epoch": 0.25450473378804844,
      "grad_norm": 9.07693862915039,
      "learning_rate": 1.0175081433224757e-05,
      "loss": 0.1761,
      "step": 2500
    },
    {
      "epoch": 0.25704978112592897,
      "grad_norm": 6.585052490234375,
      "learning_rate": 1.0276872964169382e-05,
      "loss": 0.2103,
      "step": 2525
    },
    {
      "epoch": 0.25959482846380943,
      "grad_norm": 10.828688621520996,
      "learning_rate": 1.0378664495114008e-05,
      "loss": 0.2042,
      "step": 2550
    },
    {
      "epoch": 0.2621398758016899,
      "grad_norm": 4.312686920166016,
      "learning_rate": 1.0480456026058632e-05,
      "loss": 0.1831,
      "step": 2575
    },
    {
      "epoch": 0.26468492313957037,
      "grad_norm": 7.761200428009033,
      "learning_rate": 1.0582247557003257e-05,
      "loss": 0.2006,
      "step": 2600
    },
    {
      "epoch": 0.2672299704774509,
      "grad_norm": 12.342726707458496,
      "learning_rate": 1.0684039087947885e-05,
      "loss": 0.1801,
      "step": 2625
    },
    {
      "epoch": 0.26977501781533136,
      "grad_norm": 12.449806213378906,
      "learning_rate": 1.078583061889251e-05,
      "loss": 0.1625,
      "step": 2650
    },
    {
      "epoch": 0.27232006515321183,
      "grad_norm": 1.6762300729751587,
      "learning_rate": 1.0887622149837134e-05,
      "loss": 0.1741,
      "step": 2675
    },
    {
      "epoch": 0.27486511249109236,
      "grad_norm": 13.451739311218262,
      "learning_rate": 1.098941368078176e-05,
      "loss": 0.2158,
      "step": 2700
    },
    {
      "epoch": 0.2774101598289728,
      "grad_norm": 5.619725227355957,
      "learning_rate": 1.1091205211726385e-05,
      "loss": 0.2246,
      "step": 2725
    },
    {
      "epoch": 0.2799552071668533,
      "grad_norm": 3.5796515941619873,
      "learning_rate": 1.1192996742671012e-05,
      "loss": 0.1823,
      "step": 2750
    },
    {
      "epoch": 0.28250025450473376,
      "grad_norm": 6.106029033660889,
      "learning_rate": 1.1294788273615636e-05,
      "loss": 0.1484,
      "step": 2775
    },
    {
      "epoch": 0.2850453018426143,
      "grad_norm": 10.085928916931152,
      "learning_rate": 1.1396579804560262e-05,
      "loss": 0.1743,
      "step": 2800
    },
    {
      "epoch": 0.28759034918049475,
      "grad_norm": 5.781569480895996,
      "learning_rate": 1.1498371335504887e-05,
      "loss": 0.2208,
      "step": 2825
    },
    {
      "epoch": 0.2901353965183752,
      "grad_norm": 9.282120704650879,
      "learning_rate": 1.1600162866449511e-05,
      "loss": 0.1441,
      "step": 2850
    },
    {
      "epoch": 0.29268044385625575,
      "grad_norm": 21.168132781982422,
      "learning_rate": 1.1701954397394137e-05,
      "loss": 0.1998,
      "step": 2875
    },
    {
      "epoch": 0.2952254911941362,
      "grad_norm": 7.404329776763916,
      "learning_rate": 1.1803745928338764e-05,
      "loss": 0.2111,
      "step": 2900
    },
    {
      "epoch": 0.2977705385320167,
      "grad_norm": 11.75802230834961,
      "learning_rate": 1.190553745928339e-05,
      "loss": 0.1657,
      "step": 2925
    },
    {
      "epoch": 0.3003155858698972,
      "grad_norm": 7.526225566864014,
      "learning_rate": 1.2007328990228013e-05,
      "loss": 0.2189,
      "step": 2950
    },
    {
      "epoch": 0.3028606332077777,
      "grad_norm": 5.306471347808838,
      "learning_rate": 1.2109120521172639e-05,
      "loss": 0.203,
      "step": 2975
    },
    {
      "epoch": 0.30540568054565814,
      "grad_norm": 2.9105184078216553,
      "learning_rate": 1.2210912052117264e-05,
      "loss": 0.1789,
      "step": 3000
    },
    {
      "epoch": 0.3079507278835386,
      "grad_norm": 0.9518977403640747,
      "learning_rate": 1.2312703583061892e-05,
      "loss": 0.1435,
      "step": 3025
    },
    {
      "epoch": 0.31049577522141913,
      "grad_norm": 1.7667123079299927,
      "learning_rate": 1.2414495114006516e-05,
      "loss": 0.1727,
      "step": 3050
    },
    {
      "epoch": 0.3130408225592996,
      "grad_norm": 9.456025123596191,
      "learning_rate": 1.2516286644951141e-05,
      "loss": 0.2045,
      "step": 3075
    },
    {
      "epoch": 0.31558586989718007,
      "grad_norm": 8.54640007019043,
      "learning_rate": 1.2618078175895767e-05,
      "loss": 0.2143,
      "step": 3100
    },
    {
      "epoch": 0.3181309172350606,
      "grad_norm": 4.005561828613281,
      "learning_rate": 1.271986970684039e-05,
      "loss": 0.1971,
      "step": 3125
    },
    {
      "epoch": 0.32067596457294106,
      "grad_norm": 4.398591995239258,
      "learning_rate": 1.2821661237785018e-05,
      "loss": 0.2121,
      "step": 3150
    },
    {
      "epoch": 0.32322101191082153,
      "grad_norm": 8.242807388305664,
      "learning_rate": 1.2923452768729643e-05,
      "loss": 0.183,
      "step": 3175
    },
    {
      "epoch": 0.325766059248702,
      "grad_norm": 7.716986656188965,
      "learning_rate": 1.3025244299674269e-05,
      "loss": 0.2471,
      "step": 3200
    },
    {
      "epoch": 0.3283111065865825,
      "grad_norm": 2.0308592319488525,
      "learning_rate": 1.3127035830618893e-05,
      "loss": 0.1714,
      "step": 3225
    },
    {
      "epoch": 0.330856153924463,
      "grad_norm": 5.050072193145752,
      "learning_rate": 1.3228827361563518e-05,
      "loss": 0.1689,
      "step": 3250
    },
    {
      "epoch": 0.33340120126234346,
      "grad_norm": 8.50062370300293,
      "learning_rate": 1.3330618892508145e-05,
      "loss": 0.1817,
      "step": 3275
    },
    {
      "epoch": 0.335946248600224,
      "grad_norm": 12.829822540283203,
      "learning_rate": 1.3432410423452771e-05,
      "loss": 0.1616,
      "step": 3300
    },
    {
      "epoch": 0.33849129593810445,
      "grad_norm": 6.215601921081543,
      "learning_rate": 1.3534201954397395e-05,
      "loss": 0.152,
      "step": 3325
    },
    {
      "epoch": 0.3410363432759849,
      "grad_norm": 9.163058280944824,
      "learning_rate": 1.363599348534202e-05,
      "loss": 0.1619,
      "step": 3350
    },
    {
      "epoch": 0.34358139061386544,
      "grad_norm": 6.206748008728027,
      "learning_rate": 1.3737785016286646e-05,
      "loss": 0.1881,
      "step": 3375
    },
    {
      "epoch": 0.3461264379517459,
      "grad_norm": 10.556598663330078,
      "learning_rate": 1.383957654723127e-05,
      "loss": 0.1974,
      "step": 3400
    },
    {
      "epoch": 0.3486714852896264,
      "grad_norm": 5.033000946044922,
      "learning_rate": 1.3941368078175897e-05,
      "loss": 0.1609,
      "step": 3425
    },
    {
      "epoch": 0.35121653262750685,
      "grad_norm": 9.810371398925781,
      "learning_rate": 1.4043159609120523e-05,
      "loss": 0.185,
      "step": 3450
    },
    {
      "epoch": 0.3537615799653874,
      "grad_norm": 2.4697887897491455,
      "learning_rate": 1.4144951140065148e-05,
      "loss": 0.1459,
      "step": 3475
    },
    {
      "epoch": 0.35630662730326784,
      "grad_norm": 7.219648838043213,
      "learning_rate": 1.4246742671009772e-05,
      "loss": 0.1896,
      "step": 3500
    },
    {
      "epoch": 0.3588516746411483,
      "grad_norm": 5.063100337982178,
      "learning_rate": 1.4348534201954398e-05,
      "loss": 0.1944,
      "step": 3525
    },
    {
      "epoch": 0.36139672197902883,
      "grad_norm": 4.737133502960205,
      "learning_rate": 1.4450325732899025e-05,
      "loss": 0.2766,
      "step": 3550
    },
    {
      "epoch": 0.3639417693169093,
      "grad_norm": 12.325284004211426,
      "learning_rate": 1.455211726384365e-05,
      "loss": 0.166,
      "step": 3575
    },
    {
      "epoch": 0.36648681665478977,
      "grad_norm": 5.75470495223999,
      "learning_rate": 1.4653908794788274e-05,
      "loss": 0.1889,
      "step": 3600
    },
    {
      "epoch": 0.36903186399267024,
      "grad_norm": 0.9477280378341675,
      "learning_rate": 1.47557003257329e-05,
      "loss": 0.1586,
      "step": 3625
    },
    {
      "epoch": 0.37157691133055076,
      "grad_norm": 12.12957763671875,
      "learning_rate": 1.4857491856677525e-05,
      "loss": 0.1832,
      "step": 3650
    },
    {
      "epoch": 0.37412195866843123,
      "grad_norm": 4.511873245239258,
      "learning_rate": 1.4959283387622153e-05,
      "loss": 0.15,
      "step": 3675
    },
    {
      "epoch": 0.3766670060063117,
      "grad_norm": 2.4949402809143066,
      "learning_rate": 1.5061074918566776e-05,
      "loss": 0.2203,
      "step": 3700
    },
    {
      "epoch": 0.3792120533441922,
      "grad_norm": 6.411753177642822,
      "learning_rate": 1.5162866449511402e-05,
      "loss": 0.1409,
      "step": 3725
    },
    {
      "epoch": 0.3817571006820727,
      "grad_norm": 7.92652702331543,
      "learning_rate": 1.5264657980456027e-05,
      "loss": 0.157,
      "step": 3750
    },
    {
      "epoch": 0.38430214801995316,
      "grad_norm": 6.347306728363037,
      "learning_rate": 1.536644951140065e-05,
      "loss": 0.2132,
      "step": 3775
    },
    {
      "epoch": 0.38684719535783363,
      "grad_norm": 4.59040641784668,
      "learning_rate": 1.546824104234528e-05,
      "loss": 0.1653,
      "step": 3800
    },
    {
      "epoch": 0.38939224269571415,
      "grad_norm": 6.297945022583008,
      "learning_rate": 1.5570032573289902e-05,
      "loss": 0.2345,
      "step": 3825
    },
    {
      "epoch": 0.3919372900335946,
      "grad_norm": 6.927093982696533,
      "learning_rate": 1.567182410423453e-05,
      "loss": 0.171,
      "step": 3850
    },
    {
      "epoch": 0.3944823373714751,
      "grad_norm": 3.081691265106201,
      "learning_rate": 1.5773615635179154e-05,
      "loss": 0.1914,
      "step": 3875
    },
    {
      "epoch": 0.3970273847093556,
      "grad_norm": 8.08444595336914,
      "learning_rate": 1.587540716612378e-05,
      "loss": 0.1674,
      "step": 3900
    },
    {
      "epoch": 0.3995724320472361,
      "grad_norm": 4.390102386474609,
      "learning_rate": 1.5977198697068405e-05,
      "loss": 0.2075,
      "step": 3925
    },
    {
      "epoch": 0.40211747938511655,
      "grad_norm": 4.687044620513916,
      "learning_rate": 1.6078990228013032e-05,
      "loss": 0.1706,
      "step": 3950
    },
    {
      "epoch": 0.4046625267229971,
      "grad_norm": 6.264878749847412,
      "learning_rate": 1.6180781758957656e-05,
      "loss": 0.1797,
      "step": 3975
    },
    {
      "epoch": 0.40720757406087754,
      "grad_norm": 6.146754264831543,
      "learning_rate": 1.6282573289902283e-05,
      "loss": 0.2039,
      "step": 4000
    },
    {
      "epoch": 0.409752621398758,
      "grad_norm": 4.093338966369629,
      "learning_rate": 1.6384364820846907e-05,
      "loss": 0.1719,
      "step": 4025
    },
    {
      "epoch": 0.4122976687366385,
      "grad_norm": 4.742345809936523,
      "learning_rate": 1.648615635179153e-05,
      "loss": 0.1802,
      "step": 4050
    },
    {
      "epoch": 0.414842716074519,
      "grad_norm": 7.697132110595703,
      "learning_rate": 1.6587947882736158e-05,
      "loss": 0.1495,
      "step": 4075
    },
    {
      "epoch": 0.41738776341239947,
      "grad_norm": 9.093936920166016,
      "learning_rate": 1.6689739413680782e-05,
      "loss": 0.1765,
      "step": 4100
    },
    {
      "epoch": 0.41993281075027994,
      "grad_norm": 2.59102725982666,
      "learning_rate": 1.679153094462541e-05,
      "loss": 0.2024,
      "step": 4125
    },
    {
      "epoch": 0.42247785808816046,
      "grad_norm": 14.66810131072998,
      "learning_rate": 1.6893322475570033e-05,
      "loss": 0.2391,
      "step": 4150
    },
    {
      "epoch": 0.42502290542604093,
      "grad_norm": 9.222769737243652,
      "learning_rate": 1.699511400651466e-05,
      "loss": 0.1881,
      "step": 4175
    },
    {
      "epoch": 0.4275679527639214,
      "grad_norm": 5.003434658050537,
      "learning_rate": 1.7096905537459284e-05,
      "loss": 0.1537,
      "step": 4200
    },
    {
      "epoch": 0.43011300010180187,
      "grad_norm": 5.4720139503479,
      "learning_rate": 1.719869706840391e-05,
      "loss": 0.1573,
      "step": 4225
    },
    {
      "epoch": 0.4326580474396824,
      "grad_norm": 4.233233451843262,
      "learning_rate": 1.7300488599348535e-05,
      "loss": 0.1619,
      "step": 4250
    },
    {
      "epoch": 0.43520309477756286,
      "grad_norm": 5.573329925537109,
      "learning_rate": 1.7402280130293162e-05,
      "loss": 0.1827,
      "step": 4275
    },
    {
      "epoch": 0.43774814211544333,
      "grad_norm": 7.554567813873291,
      "learning_rate": 1.7504071661237786e-05,
      "loss": 0.1722,
      "step": 4300
    },
    {
      "epoch": 0.44029318945332385,
      "grad_norm": 8.24718952178955,
      "learning_rate": 1.760586319218241e-05,
      "loss": 0.138,
      "step": 4325
    },
    {
      "epoch": 0.4428382367912043,
      "grad_norm": 8.050121307373047,
      "learning_rate": 1.7707654723127037e-05,
      "loss": 0.1826,
      "step": 4350
    },
    {
      "epoch": 0.4453832841290848,
      "grad_norm": 2.46492338180542,
      "learning_rate": 1.7809446254071664e-05,
      "loss": 0.1447,
      "step": 4375
    },
    {
      "epoch": 0.4479283314669653,
      "grad_norm": 8.03940486907959,
      "learning_rate": 1.791123778501629e-05,
      "loss": 0.1745,
      "step": 4400
    },
    {
      "epoch": 0.4504733788048458,
      "grad_norm": 4.312441825866699,
      "learning_rate": 1.8013029315960912e-05,
      "loss": 0.1308,
      "step": 4425
    },
    {
      "epoch": 0.45301842614272625,
      "grad_norm": 8.30965805053711,
      "learning_rate": 1.811482084690554e-05,
      "loss": 0.2061,
      "step": 4450
    },
    {
      "epoch": 0.4555634734806067,
      "grad_norm": 4.426910400390625,
      "learning_rate": 1.8216612377850163e-05,
      "loss": 0.1735,
      "step": 4475
    },
    {
      "epoch": 0.45810852081848724,
      "grad_norm": 12.524816513061523,
      "learning_rate": 1.831840390879479e-05,
      "loss": 0.1937,
      "step": 4500
    },
    {
      "epoch": 0.4606535681563677,
      "grad_norm": 7.2023797035217285,
      "learning_rate": 1.8420195439739414e-05,
      "loss": 0.1794,
      "step": 4525
    },
    {
      "epoch": 0.4631986154942482,
      "grad_norm": 5.721170425415039,
      "learning_rate": 1.852198697068404e-05,
      "loss": 0.2055,
      "step": 4550
    },
    {
      "epoch": 0.4657436628321287,
      "grad_norm": 6.1515350341796875,
      "learning_rate": 1.8623778501628665e-05,
      "loss": 0.1835,
      "step": 4575
    },
    {
      "epoch": 0.46828871017000917,
      "grad_norm": 6.489041805267334,
      "learning_rate": 1.8725570032573293e-05,
      "loss": 0.1597,
      "step": 4600
    },
    {
      "epoch": 0.47083375750788964,
      "grad_norm": 6.8791093826293945,
      "learning_rate": 1.8827361563517917e-05,
      "loss": 0.1608,
      "step": 4625
    },
    {
      "epoch": 0.4733788048457701,
      "grad_norm": 3.147822141647339,
      "learning_rate": 1.8929153094462544e-05,
      "loss": 0.1653,
      "step": 4650
    },
    {
      "epoch": 0.47592385218365063,
      "grad_norm": 1.5910171270370483,
      "learning_rate": 1.9030944625407168e-05,
      "loss": 0.1526,
      "step": 4675
    },
    {
      "epoch": 0.4784688995215311,
      "grad_norm": 1.3465055227279663,
      "learning_rate": 1.9128664495114007e-05,
      "loss": 0.1507,
      "step": 4700
    },
    {
      "epoch": 0.48101394685941157,
      "grad_norm": 7.712373733520508,
      "learning_rate": 1.923045602605863e-05,
      "loss": 0.1571,
      "step": 4725
    },
    {
      "epoch": 0.4835589941972921,
      "grad_norm": 9.888788223266602,
      "learning_rate": 1.933224755700326e-05,
      "loss": 0.2194,
      "step": 4750
    },
    {
      "epoch": 0.48610404153517256,
      "grad_norm": 7.199795722961426,
      "learning_rate": 1.9434039087947886e-05,
      "loss": 0.1781,
      "step": 4775
    },
    {
      "epoch": 0.48864908887305303,
      "grad_norm": 12.238067626953125,
      "learning_rate": 1.953583061889251e-05,
      "loss": 0.1745,
      "step": 4800
    },
    {
      "epoch": 0.4911941362109335,
      "grad_norm": 1.1686809062957764,
      "learning_rate": 1.9637622149837133e-05,
      "loss": 0.1348,
      "step": 4825
    },
    {
      "epoch": 0.493739183548814,
      "grad_norm": 6.479466915130615,
      "learning_rate": 1.973941368078176e-05,
      "loss": 0.2164,
      "step": 4850
    },
    {
      "epoch": 0.4962842308866945,
      "grad_norm": 6.764782905578613,
      "learning_rate": 1.9841205211726388e-05,
      "loss": 0.1724,
      "step": 4875
    },
    {
      "epoch": 0.49882927822457496,
      "grad_norm": 2.8050501346588135,
      "learning_rate": 1.994299674267101e-05,
      "loss": 0.1563,
      "step": 4900
    },
    {
      "epoch": 0.5013743255624554,
      "grad_norm": 7.177031517028809,
      "learning_rate": 1.9995022962242382e-05,
      "loss": 0.2448,
      "step": 4925
    },
    {
      "epoch": 0.503919372900336,
      "grad_norm": 12.320595741271973,
      "learning_rate": 1.998371151279325e-05,
      "loss": 0.1796,
      "step": 4950
    },
    {
      "epoch": 0.5064644202382165,
      "grad_norm": 7.938055038452148,
      "learning_rate": 1.997240006334412e-05,
      "loss": 0.1541,
      "step": 4975
    },
    {
      "epoch": 0.5090094675760969,
      "grad_norm": 4.647416591644287,
      "learning_rate": 1.9961088613894987e-05,
      "loss": 0.1906,
      "step": 5000
    },
    {
      "epoch": 0.5115545149139774,
      "grad_norm": 4.05308723449707,
      "learning_rate": 1.9949777164445853e-05,
      "loss": 0.1822,
      "step": 5025
    },
    {
      "epoch": 0.5140995622518579,
      "grad_norm": 3.471247434616089,
      "learning_rate": 1.9938465714996723e-05,
      "loss": 0.1778,
      "step": 5050
    },
    {
      "epoch": 0.5166446095897383,
      "grad_norm": 8.766022682189941,
      "learning_rate": 1.992715426554759e-05,
      "loss": 0.1741,
      "step": 5075
    },
    {
      "epoch": 0.5191896569276189,
      "grad_norm": 7.638113021850586,
      "learning_rate": 1.9915842816098455e-05,
      "loss": 0.2008,
      "step": 5100
    },
    {
      "epoch": 0.5217347042654993,
      "grad_norm": 2.3895423412323,
      "learning_rate": 1.9904531366649325e-05,
      "loss": 0.1857,
      "step": 5125
    },
    {
      "epoch": 0.5242797516033798,
      "grad_norm": 9.395495414733887,
      "learning_rate": 1.9893219917200194e-05,
      "loss": 0.1966,
      "step": 5150
    },
    {
      "epoch": 0.5268247989412603,
      "grad_norm": 7.309564590454102,
      "learning_rate": 1.988190846775106e-05,
      "loss": 0.1868,
      "step": 5175
    },
    {
      "epoch": 0.5293698462791407,
      "grad_norm": 2.8665430545806885,
      "learning_rate": 1.9870597018301927e-05,
      "loss": 0.1607,
      "step": 5200
    },
    {
      "epoch": 0.5319148936170213,
      "grad_norm": 4.512245178222656,
      "learning_rate": 1.9859285568852796e-05,
      "loss": 0.1793,
      "step": 5225
    },
    {
      "epoch": 0.5344599409549018,
      "grad_norm": 5.876766681671143,
      "learning_rate": 1.9847974119403662e-05,
      "loss": 0.173,
      "step": 5250
    },
    {
      "epoch": 0.5370049882927822,
      "grad_norm": 13.551549911499023,
      "learning_rate": 1.983666266995453e-05,
      "loss": 0.1564,
      "step": 5275
    },
    {
      "epoch": 0.5395500356306627,
      "grad_norm": 8.268135070800781,
      "learning_rate": 1.9825351220505398e-05,
      "loss": 0.1681,
      "step": 5300
    },
    {
      "epoch": 0.5420950829685433,
      "grad_norm": 4.490458011627197,
      "learning_rate": 1.9814039771056264e-05,
      "loss": 0.2074,
      "step": 5325
    },
    {
      "epoch": 0.5446401303064237,
      "grad_norm": 7.859119415283203,
      "learning_rate": 1.9802728321607134e-05,
      "loss": 0.208,
      "step": 5350
    },
    {
      "epoch": 0.5471851776443042,
      "grad_norm": 3.8342528343200684,
      "learning_rate": 1.9791416872158e-05,
      "loss": 0.1809,
      "step": 5375
    },
    {
      "epoch": 0.5497302249821847,
      "grad_norm": 6.6303181648254395,
      "learning_rate": 1.978010542270887e-05,
      "loss": 0.1548,
      "step": 5400
    },
    {
      "epoch": 0.5522752723200651,
      "grad_norm": 5.547021865844727,
      "learning_rate": 1.9768793973259735e-05,
      "loss": 0.1755,
      "step": 5425
    },
    {
      "epoch": 0.5548203196579456,
      "grad_norm": 5.783038139343262,
      "learning_rate": 1.97574825238106e-05,
      "loss": 0.1891,
      "step": 5450
    },
    {
      "epoch": 0.5573653669958262,
      "grad_norm": 4.4773359298706055,
      "learning_rate": 1.974617107436147e-05,
      "loss": 0.1569,
      "step": 5475
    },
    {
      "epoch": 0.5599104143337066,
      "grad_norm": 7.39841890335083,
      "learning_rate": 1.9734859624912337e-05,
      "loss": 0.1599,
      "step": 5500
    },
    {
      "epoch": 0.5624554616715871,
      "grad_norm": 7.901022434234619,
      "learning_rate": 1.9723548175463207e-05,
      "loss": 0.1919,
      "step": 5525
    },
    {
      "epoch": 0.5650005090094675,
      "grad_norm": 5.0281596183776855,
      "learning_rate": 1.9712236726014073e-05,
      "loss": 0.1901,
      "step": 5550
    },
    {
      "epoch": 0.567545556347348,
      "grad_norm": 5.846480369567871,
      "learning_rate": 1.9700925276564943e-05,
      "loss": 0.1736,
      "step": 5575
    },
    {
      "epoch": 0.5700906036852286,
      "grad_norm": 5.52054500579834,
      "learning_rate": 1.968961382711581e-05,
      "loss": 0.1477,
      "step": 5600
    },
    {
      "epoch": 0.572635651023109,
      "grad_norm": 8.760751724243164,
      "learning_rate": 1.9678302377666675e-05,
      "loss": 0.189,
      "step": 5625
    },
    {
      "epoch": 0.5751806983609895,
      "grad_norm": 5.988430500030518,
      "learning_rate": 1.9666990928217544e-05,
      "loss": 0.2132,
      "step": 5650
    },
    {
      "epoch": 0.57772574569887,
      "grad_norm": 5.293154716491699,
      "learning_rate": 1.965567947876841e-05,
      "loss": 0.1636,
      "step": 5675
    },
    {
      "epoch": 0.5802707930367504,
      "grad_norm": 6.205268859863281,
      "learning_rate": 1.964436802931928e-05,
      "loss": 0.1592,
      "step": 5700
    },
    {
      "epoch": 0.582815840374631,
      "grad_norm": 3.520357370376587,
      "learning_rate": 1.9633056579870146e-05,
      "loss": 0.1607,
      "step": 5725
    },
    {
      "epoch": 0.5853608877125115,
      "grad_norm": 9.744399070739746,
      "learning_rate": 1.9621745130421012e-05,
      "loss": 0.1978,
      "step": 5750
    },
    {
      "epoch": 0.5879059350503919,
      "grad_norm": 7.006613254547119,
      "learning_rate": 1.9610433680971882e-05,
      "loss": 0.1954,
      "step": 5775
    },
    {
      "epoch": 0.5904509823882724,
      "grad_norm": 2.8386662006378174,
      "learning_rate": 1.9599122231522748e-05,
      "loss": 0.1457,
      "step": 5800
    },
    {
      "epoch": 0.592996029726153,
      "grad_norm": 9.212706565856934,
      "learning_rate": 1.9587810782073618e-05,
      "loss": 0.1229,
      "step": 5825
    },
    {
      "epoch": 0.5955410770640334,
      "grad_norm": 2.414585828781128,
      "learning_rate": 1.9576499332624484e-05,
      "loss": 0.1644,
      "step": 5850
    },
    {
      "epoch": 0.5980861244019139,
      "grad_norm": 9.854455947875977,
      "learning_rate": 1.9565187883175353e-05,
      "loss": 0.1804,
      "step": 5875
    },
    {
      "epoch": 0.6006311717397944,
      "grad_norm": 5.590005874633789,
      "learning_rate": 1.955387643372622e-05,
      "loss": 0.1697,
      "step": 5900
    },
    {
      "epoch": 0.6031762190776748,
      "grad_norm": 3.120067834854126,
      "learning_rate": 1.9542564984277086e-05,
      "loss": 0.1828,
      "step": 5925
    },
    {
      "epoch": 0.6057212664155553,
      "grad_norm": 11.758387565612793,
      "learning_rate": 1.9531253534827955e-05,
      "loss": 0.1476,
      "step": 5950
    },
    {
      "epoch": 0.6082663137534358,
      "grad_norm": 3.0814547538757324,
      "learning_rate": 1.951994208537882e-05,
      "loss": 0.1765,
      "step": 5975
    },
    {
      "epoch": 0.6108113610913163,
      "grad_norm": 5.264235496520996,
      "learning_rate": 1.950863063592969e-05,
      "loss": 0.1937,
      "step": 6000
    },
    {
      "epoch": 0.6133564084291968,
      "grad_norm": 7.090575695037842,
      "learning_rate": 1.9497319186480557e-05,
      "loss": 0.1737,
      "step": 6025
    },
    {
      "epoch": 0.6159014557670772,
      "grad_norm": 1.6197588443756104,
      "learning_rate": 1.9486007737031426e-05,
      "loss": 0.1403,
      "step": 6050
    },
    {
      "epoch": 0.6184465031049577,
      "grad_norm": 6.1466240882873535,
      "learning_rate": 1.9474696287582293e-05,
      "loss": 0.1869,
      "step": 6075
    },
    {
      "epoch": 0.6209915504428383,
      "grad_norm": 3.4034574031829834,
      "learning_rate": 1.946338483813316e-05,
      "loss": 0.17,
      "step": 6100
    },
    {
      "epoch": 0.6235365977807187,
      "grad_norm": 6.991599082946777,
      "learning_rate": 1.945207338868403e-05,
      "loss": 0.1534,
      "step": 6125
    },
    {
      "epoch": 0.6260816451185992,
      "grad_norm": 4.051377773284912,
      "learning_rate": 1.9440761939234894e-05,
      "loss": 0.1778,
      "step": 6150
    },
    {
      "epoch": 0.6286266924564797,
      "grad_norm": 5.443165302276611,
      "learning_rate": 1.942945048978576e-05,
      "loss": 0.1741,
      "step": 6175
    },
    {
      "epoch": 0.6311717397943601,
      "grad_norm": 5.732553482055664,
      "learning_rate": 1.941813904033663e-05,
      "loss": 0.1622,
      "step": 6200
    },
    {
      "epoch": 0.6337167871322407,
      "grad_norm": 6.2127180099487305,
      "learning_rate": 1.94068275908875e-05,
      "loss": 0.2172,
      "step": 6225
    },
    {
      "epoch": 0.6362618344701212,
      "grad_norm": 5.593441963195801,
      "learning_rate": 1.9395516141438366e-05,
      "loss": 0.1618,
      "step": 6250
    },
    {
      "epoch": 0.6388068818080016,
      "grad_norm": 4.706175804138184,
      "learning_rate": 1.9384204691989232e-05,
      "loss": 0.1973,
      "step": 6275
    },
    {
      "epoch": 0.6413519291458821,
      "grad_norm": 5.412227630615234,
      "learning_rate": 1.93728932425401e-05,
      "loss": 0.1896,
      "step": 6300
    },
    {
      "epoch": 0.6438969764837627,
      "grad_norm": 4.554426193237305,
      "learning_rate": 1.9361581793090968e-05,
      "loss": 0.1766,
      "step": 6325
    },
    {
      "epoch": 0.6464420238216431,
      "grad_norm": 5.6457839012146,
      "learning_rate": 1.9350270343641834e-05,
      "loss": 0.1692,
      "step": 6350
    },
    {
      "epoch": 0.6489870711595236,
      "grad_norm": 5.07610559463501,
      "learning_rate": 1.9338958894192703e-05,
      "loss": 0.1687,
      "step": 6375
    },
    {
      "epoch": 0.651532118497404,
      "grad_norm": 8.431342124938965,
      "learning_rate": 1.9327647444743573e-05,
      "loss": 0.1635,
      "step": 6400
    },
    {
      "epoch": 0.6540771658352845,
      "grad_norm": 2.4447405338287354,
      "learning_rate": 1.931633599529444e-05,
      "loss": 0.1625,
      "step": 6425
    },
    {
      "epoch": 0.656622213173165,
      "grad_norm": 5.066686153411865,
      "learning_rate": 1.9305024545845305e-05,
      "loss": 0.1532,
      "step": 6450
    },
    {
      "epoch": 0.6591672605110455,
      "grad_norm": 5.495453357696533,
      "learning_rate": 1.9293713096396175e-05,
      "loss": 0.1668,
      "step": 6475
    },
    {
      "epoch": 0.661712307848926,
      "grad_norm": 12.891332626342773,
      "learning_rate": 1.928240164694704e-05,
      "loss": 0.1869,
      "step": 6500
    },
    {
      "epoch": 0.6642573551868065,
      "grad_norm": 4.865134239196777,
      "learning_rate": 1.9271090197497907e-05,
      "loss": 0.1891,
      "step": 6525
    },
    {
      "epoch": 0.6668024025246869,
      "grad_norm": 7.412440776824951,
      "learning_rate": 1.9259778748048777e-05,
      "loss": 0.1914,
      "step": 6550
    },
    {
      "epoch": 0.6693474498625674,
      "grad_norm": 4.157654285430908,
      "learning_rate": 1.9248467298599646e-05,
      "loss": 0.1951,
      "step": 6575
    },
    {
      "epoch": 0.671892497200448,
      "grad_norm": 3.6485140323638916,
      "learning_rate": 1.9237155849150512e-05,
      "loss": 0.1453,
      "step": 6600
    },
    {
      "epoch": 0.6744375445383284,
      "grad_norm": 6.457579612731934,
      "learning_rate": 1.922584439970138e-05,
      "loss": 0.1671,
      "step": 6625
    },
    {
      "epoch": 0.6769825918762089,
      "grad_norm": 4.9921650886535645,
      "learning_rate": 1.9214532950252248e-05,
      "loss": 0.1527,
      "step": 6650
    },
    {
      "epoch": 0.6795276392140894,
      "grad_norm": 11.607213973999023,
      "learning_rate": 1.9203221500803114e-05,
      "loss": 0.1945,
      "step": 6675
    },
    {
      "epoch": 0.6820726865519698,
      "grad_norm": 1.1529160737991333,
      "learning_rate": 1.919191005135398e-05,
      "loss": 0.1575,
      "step": 6700
    },
    {
      "epoch": 0.6846177338898504,
      "grad_norm": 3.0068256855010986,
      "learning_rate": 1.918059860190485e-05,
      "loss": 0.1633,
      "step": 6725
    },
    {
      "epoch": 0.6871627812277309,
      "grad_norm": 4.913267135620117,
      "learning_rate": 1.916928715245572e-05,
      "loss": 0.1672,
      "step": 6750
    },
    {
      "epoch": 0.6897078285656113,
      "grad_norm": 6.968467712402344,
      "learning_rate": 1.9157975703006582e-05,
      "loss": 0.1768,
      "step": 6775
    },
    {
      "epoch": 0.6922528759034918,
      "grad_norm": 3.2676682472229004,
      "learning_rate": 1.914711671153542e-05,
      "loss": 0.1957,
      "step": 6800
    },
    {
      "epoch": 0.6947979232413722,
      "grad_norm": 2.3184611797332764,
      "learning_rate": 1.9135805262086286e-05,
      "loss": 0.1427,
      "step": 6825
    },
    {
      "epoch": 0.6973429705792528,
      "grad_norm": 7.692404270172119,
      "learning_rate": 1.9124493812637152e-05,
      "loss": 0.1333,
      "step": 6850
    },
    {
      "epoch": 0.6998880179171333,
      "grad_norm": 10.22700309753418,
      "learning_rate": 1.9113182363188022e-05,
      "loss": 0.1457,
      "step": 6875
    },
    {
      "epoch": 0.7024330652550137,
      "grad_norm": 6.087644577026367,
      "learning_rate": 1.9101870913738888e-05,
      "loss": 0.1821,
      "step": 6900
    },
    {
      "epoch": 0.7049781125928942,
      "grad_norm": 4.5725836753845215,
      "learning_rate": 1.9090559464289754e-05,
      "loss": 0.1419,
      "step": 6925
    },
    {
      "epoch": 0.7075231599307747,
      "grad_norm": 5.20031213760376,
      "learning_rate": 1.9079248014840624e-05,
      "loss": 0.1687,
      "step": 6950
    },
    {
      "epoch": 0.7100682072686552,
      "grad_norm": 5.029293060302734,
      "learning_rate": 1.9067936565391493e-05,
      "loss": 0.2075,
      "step": 6975
    },
    {
      "epoch": 0.7126132546065357,
      "grad_norm": 6.754817962646484,
      "learning_rate": 1.905662511594236e-05,
      "loss": 0.1931,
      "step": 7000
    },
    {
      "epoch": 0.7151583019444162,
      "grad_norm": 0.8655216693878174,
      "learning_rate": 1.9045313666493225e-05,
      "loss": 0.1519,
      "step": 7025
    },
    {
      "epoch": 0.7177033492822966,
      "grad_norm": 6.249109268188477,
      "learning_rate": 1.9034002217044095e-05,
      "loss": 0.135,
      "step": 7050
    },
    {
      "epoch": 0.7202483966201771,
      "grad_norm": 2.2959837913513184,
      "learning_rate": 1.902269076759496e-05,
      "loss": 0.1467,
      "step": 7075
    },
    {
      "epoch": 0.7227934439580577,
      "grad_norm": 7.267690658569336,
      "learning_rate": 1.9011379318145827e-05,
      "loss": 0.2012,
      "step": 7100
    },
    {
      "epoch": 0.7253384912959381,
      "grad_norm": 3.546318292617798,
      "learning_rate": 1.9000067868696697e-05,
      "loss": 0.1632,
      "step": 7125
    },
    {
      "epoch": 0.7278835386338186,
      "grad_norm": 7.313410758972168,
      "learning_rate": 1.8988756419247566e-05,
      "loss": 0.1581,
      "step": 7150
    },
    {
      "epoch": 0.730428585971699,
      "grad_norm": 5.693080902099609,
      "learning_rate": 1.8977444969798432e-05,
      "loss": 0.1336,
      "step": 7175
    },
    {
      "epoch": 0.7329736333095795,
      "grad_norm": 5.479151248931885,
      "learning_rate": 1.89661335203493e-05,
      "loss": 0.1555,
      "step": 7200
    },
    {
      "epoch": 0.7355186806474601,
      "grad_norm": 6.8671345710754395,
      "learning_rate": 1.8954822070900168e-05,
      "loss": 0.1901,
      "step": 7225
    },
    {
      "epoch": 0.7380637279853405,
      "grad_norm": 5.530952453613281,
      "learning_rate": 1.8943510621451034e-05,
      "loss": 0.1602,
      "step": 7250
    },
    {
      "epoch": 0.740608775323221,
      "grad_norm": 8.927380561828613,
      "learning_rate": 1.89321991720019e-05,
      "loss": 0.1987,
      "step": 7275
    },
    {
      "epoch": 0.7431538226611015,
      "grad_norm": 7.032819747924805,
      "learning_rate": 1.892088772255277e-05,
      "loss": 0.1757,
      "step": 7300
    },
    {
      "epoch": 0.7456988699989819,
      "grad_norm": 1.044941782951355,
      "learning_rate": 1.890957627310364e-05,
      "loss": 0.1622,
      "step": 7325
    },
    {
      "epoch": 0.7482439173368625,
      "grad_norm": 4.259761810302734,
      "learning_rate": 1.8898264823654506e-05,
      "loss": 0.1402,
      "step": 7350
    },
    {
      "epoch": 0.750788964674743,
      "grad_norm": 6.318974494934082,
      "learning_rate": 1.8886953374205372e-05,
      "loss": 0.1463,
      "step": 7375
    },
    {
      "epoch": 0.7533340120126234,
      "grad_norm": 9.901176452636719,
      "learning_rate": 1.887564192475624e-05,
      "loss": 0.1683,
      "step": 7400
    },
    {
      "epoch": 0.7558790593505039,
      "grad_norm": 5.005266189575195,
      "learning_rate": 1.8864330475307108e-05,
      "loss": 0.1813,
      "step": 7425
    },
    {
      "epoch": 0.7584241066883844,
      "grad_norm": 7.308044910430908,
      "learning_rate": 1.8853019025857974e-05,
      "loss": 0.137,
      "step": 7450
    },
    {
      "epoch": 0.7609691540262649,
      "grad_norm": 7.757829666137695,
      "learning_rate": 1.8841707576408843e-05,
      "loss": 0.1778,
      "step": 7475
    },
    {
      "epoch": 0.7635142013641454,
      "grad_norm": 8.0330228805542,
      "learning_rate": 1.883039612695971e-05,
      "loss": 0.1876,
      "step": 7500
    },
    {
      "epoch": 0.7660592487020259,
      "grad_norm": 11.038161277770996,
      "learning_rate": 1.881908467751058e-05,
      "loss": 0.151,
      "step": 7525
    },
    {
      "epoch": 0.7686042960399063,
      "grad_norm": 7.047212600708008,
      "learning_rate": 1.8807773228061445e-05,
      "loss": 0.1672,
      "step": 7550
    },
    {
      "epoch": 0.7711493433777868,
      "grad_norm": 2.1965415477752686,
      "learning_rate": 1.8796461778612315e-05,
      "loss": 0.119,
      "step": 7575
    },
    {
      "epoch": 0.7736943907156673,
      "grad_norm": 5.676666259765625,
      "learning_rate": 1.878515032916318e-05,
      "loss": 0.1732,
      "step": 7600
    },
    {
      "epoch": 0.7762394380535478,
      "grad_norm": 5.984830856323242,
      "learning_rate": 1.8773838879714047e-05,
      "loss": 0.1328,
      "step": 7625
    },
    {
      "epoch": 0.7787844853914283,
      "grad_norm": 5.381333827972412,
      "learning_rate": 1.8762527430264916e-05,
      "loss": 0.1857,
      "step": 7650
    },
    {
      "epoch": 0.7813295327293087,
      "grad_norm": 6.377498149871826,
      "learning_rate": 1.8751215980815783e-05,
      "loss": 0.1954,
      "step": 7675
    },
    {
      "epoch": 0.7838745800671892,
      "grad_norm": 7.693028926849365,
      "learning_rate": 1.8739904531366652e-05,
      "loss": 0.1599,
      "step": 7700
    },
    {
      "epoch": 0.7864196274050698,
      "grad_norm": 5.4145426750183105,
      "learning_rate": 1.8728593081917518e-05,
      "loss": 0.1917,
      "step": 7725
    },
    {
      "epoch": 0.7889646747429502,
      "grad_norm": 4.35727596282959,
      "learning_rate": 1.8717281632468388e-05,
      "loss": 0.1735,
      "step": 7750
    },
    {
      "epoch": 0.7915097220808307,
      "grad_norm": 8.166193008422852,
      "learning_rate": 1.8705970183019254e-05,
      "loss": 0.158,
      "step": 7775
    },
    {
      "epoch": 0.7940547694187112,
      "grad_norm": 2.884567975997925,
      "learning_rate": 1.8695111191548085e-05,
      "loss": 0.1422,
      "step": 7800
    },
    {
      "epoch": 0.7965998167565916,
      "grad_norm": 7.425231456756592,
      "learning_rate": 1.8683799742098954e-05,
      "loss": 0.1841,
      "step": 7825
    },
    {
      "epoch": 0.7991448640944722,
      "grad_norm": 5.505438804626465,
      "learning_rate": 1.867248829264982e-05,
      "loss": 0.1473,
      "step": 7850
    },
    {
      "epoch": 0.8016899114323527,
      "grad_norm": 8.657942771911621,
      "learning_rate": 1.866117684320069e-05,
      "loss": 0.1494,
      "step": 7875
    },
    {
      "epoch": 0.8042349587702331,
      "grad_norm": 10.716988563537598,
      "learning_rate": 1.8649865393751556e-05,
      "loss": 0.1657,
      "step": 7900
    },
    {
      "epoch": 0.8067800061081136,
      "grad_norm": 5.407134056091309,
      "learning_rate": 1.8638553944302426e-05,
      "loss": 0.1637,
      "step": 7925
    },
    {
      "epoch": 0.8093250534459941,
      "grad_norm": 3.173985719680786,
      "learning_rate": 1.8627242494853292e-05,
      "loss": 0.1618,
      "step": 7950
    },
    {
      "epoch": 0.8118701007838746,
      "grad_norm": 7.4745330810546875,
      "learning_rate": 1.8615931045404158e-05,
      "loss": 0.1896,
      "step": 7975
    },
    {
      "epoch": 0.8144151481217551,
      "grad_norm": 4.304909706115723,
      "learning_rate": 1.8604619595955028e-05,
      "loss": 0.1893,
      "step": 8000
    },
    {
      "epoch": 0.8169601954596355,
      "grad_norm": 7.029309272766113,
      "learning_rate": 1.8593308146505894e-05,
      "loss": 0.2058,
      "step": 8025
    },
    {
      "epoch": 0.819505242797516,
      "grad_norm": 7.313731670379639,
      "learning_rate": 1.8581996697056763e-05,
      "loss": 0.1796,
      "step": 8050
    },
    {
      "epoch": 0.8220502901353965,
      "grad_norm": 5.254171371459961,
      "learning_rate": 1.857068524760763e-05,
      "loss": 0.1906,
      "step": 8075
    },
    {
      "epoch": 0.824595337473277,
      "grad_norm": 4.730832099914551,
      "learning_rate": 1.85593737981585e-05,
      "loss": 0.1581,
      "step": 8100
    },
    {
      "epoch": 0.8271403848111575,
      "grad_norm": 7.461226463317871,
      "learning_rate": 1.8548062348709365e-05,
      "loss": 0.1871,
      "step": 8125
    },
    {
      "epoch": 0.829685432149038,
      "grad_norm": 3.6966590881347656,
      "learning_rate": 1.853675089926023e-05,
      "loss": 0.1231,
      "step": 8150
    },
    {
      "epoch": 0.8322304794869184,
      "grad_norm": 9.96469783782959,
      "learning_rate": 1.85254394498111e-05,
      "loss": 0.2116,
      "step": 8175
    },
    {
      "epoch": 0.8347755268247989,
      "grad_norm": 6.719648361206055,
      "learning_rate": 1.8514128000361967e-05,
      "loss": 0.1809,
      "step": 8200
    },
    {
      "epoch": 0.8373205741626795,
      "grad_norm": 2.1537251472473145,
      "learning_rate": 1.8502816550912837e-05,
      "loss": 0.1456,
      "step": 8225
    },
    {
      "epoch": 0.8398656215005599,
      "grad_norm": 7.317068576812744,
      "learning_rate": 1.8491505101463703e-05,
      "loss": 0.1526,
      "step": 8250
    },
    {
      "epoch": 0.8424106688384404,
      "grad_norm": 11.354976654052734,
      "learning_rate": 1.8480193652014572e-05,
      "loss": 0.1698,
      "step": 8275
    },
    {
      "epoch": 0.8449557161763209,
      "grad_norm": 4.932126045227051,
      "learning_rate": 1.846888220256544e-05,
      "loss": 0.1893,
      "step": 8300
    },
    {
      "epoch": 0.8475007635142013,
      "grad_norm": 7.772342681884766,
      "learning_rate": 1.8457570753116305e-05,
      "loss": 0.1755,
      "step": 8325
    },
    {
      "epoch": 0.8500458108520819,
      "grad_norm": 6.9974541664123535,
      "learning_rate": 1.8446259303667174e-05,
      "loss": 0.1748,
      "step": 8350
    },
    {
      "epoch": 0.8525908581899624,
      "grad_norm": 3.3724286556243896,
      "learning_rate": 1.843494785421804e-05,
      "loss": 0.172,
      "step": 8375
    },
    {
      "epoch": 0.8551359055278428,
      "grad_norm": 3.287820339202881,
      "learning_rate": 1.8423636404768906e-05,
      "loss": 0.1222,
      "step": 8400
    },
    {
      "epoch": 0.8576809528657233,
      "grad_norm": 4.461840629577637,
      "learning_rate": 1.8412324955319776e-05,
      "loss": 0.1604,
      "step": 8425
    },
    {
      "epoch": 0.8602260002036037,
      "grad_norm": 10.443443298339844,
      "learning_rate": 1.8401013505870646e-05,
      "loss": 0.1862,
      "step": 8450
    },
    {
      "epoch": 0.8627710475414843,
      "grad_norm": 6.476747035980225,
      "learning_rate": 1.838970205642151e-05,
      "loss": 0.1848,
      "step": 8475
    },
    {
      "epoch": 0.8653160948793648,
      "grad_norm": 8.956707000732422,
      "learning_rate": 1.8378390606972378e-05,
      "loss": 0.1445,
      "step": 8500
    },
    {
      "epoch": 0.8678611422172452,
      "grad_norm": 2.2105791568756104,
      "learning_rate": 1.8367079157523247e-05,
      "loss": 0.1686,
      "step": 8525
    },
    {
      "epoch": 0.8704061895551257,
      "grad_norm": 10.493624687194824,
      "learning_rate": 1.8355767708074114e-05,
      "loss": 0.1835,
      "step": 8550
    },
    {
      "epoch": 0.8729512368930062,
      "grad_norm": 9.583802223205566,
      "learning_rate": 1.834445625862498e-05,
      "loss": 0.1464,
      "step": 8575
    },
    {
      "epoch": 0.8754962842308867,
      "grad_norm": 10.556769371032715,
      "learning_rate": 1.833314480917585e-05,
      "loss": 0.187,
      "step": 8600
    },
    {
      "epoch": 0.8780413315687672,
      "grad_norm": 5.583988666534424,
      "learning_rate": 1.832183335972672e-05,
      "loss": 0.1457,
      "step": 8625
    },
    {
      "epoch": 0.8805863789066477,
      "grad_norm": 3.9103634357452393,
      "learning_rate": 1.8310521910277585e-05,
      "loss": 0.1535,
      "step": 8650
    },
    {
      "epoch": 0.8831314262445281,
      "grad_norm": 5.067739009857178,
      "learning_rate": 1.829921046082845e-05,
      "loss": 0.1457,
      "step": 8675
    },
    {
      "epoch": 0.8856764735824086,
      "grad_norm": 4.853885650634766,
      "learning_rate": 1.828789901137932e-05,
      "loss": 0.1417,
      "step": 8700
    },
    {
      "epoch": 0.8882215209202892,
      "grad_norm": 9.625697135925293,
      "learning_rate": 1.8276587561930187e-05,
      "loss": 0.1428,
      "step": 8725
    },
    {
      "epoch": 0.8907665682581696,
      "grad_norm": 7.169886112213135,
      "learning_rate": 1.8265276112481053e-05,
      "loss": 0.1764,
      "step": 8750
    },
    {
      "epoch": 0.8933116155960501,
      "grad_norm": 3.6491191387176514,
      "learning_rate": 1.8253964663031922e-05,
      "loss": 0.1454,
      "step": 8775
    },
    {
      "epoch": 0.8958566629339306,
      "grad_norm": 6.77921724319458,
      "learning_rate": 1.8242653213582792e-05,
      "loss": 0.2102,
      "step": 8800
    },
    {
      "epoch": 0.898401710271811,
      "grad_norm": 5.925548076629639,
      "learning_rate": 1.8231341764133658e-05,
      "loss": 0.1703,
      "step": 8825
    },
    {
      "epoch": 0.9009467576096916,
      "grad_norm": 5.68595552444458,
      "learning_rate": 1.8220030314684524e-05,
      "loss": 0.1503,
      "step": 8850
    },
    {
      "epoch": 0.903491804947572,
      "grad_norm": 4.654073238372803,
      "learning_rate": 1.8208718865235394e-05,
      "loss": 0.1741,
      "step": 8875
    },
    {
      "epoch": 0.9060368522854525,
      "grad_norm": 4.704840660095215,
      "learning_rate": 1.819740741578626e-05,
      "loss": 0.1608,
      "step": 8900
    },
    {
      "epoch": 0.908581899623333,
      "grad_norm": 4.562308311462402,
      "learning_rate": 1.8186095966337126e-05,
      "loss": 0.1881,
      "step": 8925
    },
    {
      "epoch": 0.9111269469612134,
      "grad_norm": 7.123320579528809,
      "learning_rate": 1.8174784516887996e-05,
      "loss": 0.1739,
      "step": 8950
    },
    {
      "epoch": 0.913671994299094,
      "grad_norm": 1.0297778844833374,
      "learning_rate": 1.8163473067438865e-05,
      "loss": 0.1314,
      "step": 8975
    },
    {
      "epoch": 0.9162170416369745,
      "grad_norm": 2.9744787216186523,
      "learning_rate": 1.815216161798973e-05,
      "loss": 0.1516,
      "step": 9000
    },
    {
      "epoch": 0.9187620889748549,
      "grad_norm": 7.400439262390137,
      "learning_rate": 1.8140850168540597e-05,
      "loss": 0.1486,
      "step": 9025
    },
    {
      "epoch": 0.9213071363127354,
      "grad_norm": 7.018895149230957,
      "learning_rate": 1.8129538719091467e-05,
      "loss": 0.1222,
      "step": 9050
    },
    {
      "epoch": 0.9238521836506159,
      "grad_norm": 5.089198589324951,
      "learning_rate": 1.8118227269642333e-05,
      "loss": 0.2115,
      "step": 9075
    },
    {
      "epoch": 0.9263972309884964,
      "grad_norm": 6.323294162750244,
      "learning_rate": 1.81069158201932e-05,
      "loss": 0.1607,
      "step": 9100
    },
    {
      "epoch": 0.9289422783263769,
      "grad_norm": 3.201341390609741,
      "learning_rate": 1.809560437074407e-05,
      "loss": 0.1259,
      "step": 9125
    },
    {
      "epoch": 0.9314873256642574,
      "grad_norm": 6.361789703369141,
      "learning_rate": 1.808429292129494e-05,
      "loss": 0.1885,
      "step": 9150
    },
    {
      "epoch": 0.9340323730021378,
      "grad_norm": 4.8685383796691895,
      "learning_rate": 1.8072981471845805e-05,
      "loss": 0.1819,
      "step": 9175
    },
    {
      "epoch": 0.9365774203400183,
      "grad_norm": 6.192000865936279,
      "learning_rate": 1.806167002239667e-05,
      "loss": 0.1732,
      "step": 9200
    },
    {
      "epoch": 0.9391224676778988,
      "grad_norm": 7.19197940826416,
      "learning_rate": 1.805035857294754e-05,
      "loss": 0.1635,
      "step": 9225
    },
    {
      "epoch": 0.9416675150157793,
      "grad_norm": 8.539782524108887,
      "learning_rate": 1.8039047123498406e-05,
      "loss": 0.1896,
      "step": 9250
    },
    {
      "epoch": 0.9442125623536598,
      "grad_norm": 5.8749494552612305,
      "learning_rate": 1.8027735674049273e-05,
      "loss": 0.1402,
      "step": 9275
    },
    {
      "epoch": 0.9467576096915402,
      "grad_norm": 1.8965274095535278,
      "learning_rate": 1.8016424224600142e-05,
      "loss": 0.1247,
      "step": 9300
    },
    {
      "epoch": 0.9493026570294207,
      "grad_norm": 0.6471710801124573,
      "learning_rate": 1.800511277515101e-05,
      "loss": 0.1337,
      "step": 9325
    },
    {
      "epoch": 0.9518477043673013,
      "grad_norm": 9.45545768737793,
      "learning_rate": 1.7993801325701878e-05,
      "loss": 0.1447,
      "step": 9350
    },
    {
      "epoch": 0.9543927517051817,
      "grad_norm": 3.9917619228363037,
      "learning_rate": 1.7982489876252744e-05,
      "loss": 0.1978,
      "step": 9375
    },
    {
      "epoch": 0.9569377990430622,
      "grad_norm": 5.922330379486084,
      "learning_rate": 1.7971178426803613e-05,
      "loss": 0.1773,
      "step": 9400
    },
    {
      "epoch": 0.9594828463809427,
      "grad_norm": 7.029813289642334,
      "learning_rate": 1.795986697735448e-05,
      "loss": 0.185,
      "step": 9425
    },
    {
      "epoch": 0.9620278937188231,
      "grad_norm": 2.9291632175445557,
      "learning_rate": 1.7948555527905346e-05,
      "loss": 0.1906,
      "step": 9450
    },
    {
      "epoch": 0.9645729410567037,
      "grad_norm": 2.9473659992218018,
      "learning_rate": 1.7937244078456215e-05,
      "loss": 0.1606,
      "step": 9475
    },
    {
      "epoch": 0.9671179883945842,
      "grad_norm": 7.865957260131836,
      "learning_rate": 1.7925932629007085e-05,
      "loss": 0.1815,
      "step": 9500
    },
    {
      "epoch": 0.9696630357324646,
      "grad_norm": 4.849057197570801,
      "learning_rate": 1.791462117955795e-05,
      "loss": 0.1642,
      "step": 9525
    },
    {
      "epoch": 0.9722080830703451,
      "grad_norm": 6.303313255310059,
      "learning_rate": 1.7903309730108817e-05,
      "loss": 0.1785,
      "step": 9550
    },
    {
      "epoch": 0.9747531304082256,
      "grad_norm": 2.399278402328491,
      "learning_rate": 1.7891998280659687e-05,
      "loss": 0.1578,
      "step": 9575
    },
    {
      "epoch": 0.9772981777461061,
      "grad_norm": 7.749370574951172,
      "learning_rate": 1.7880686831210553e-05,
      "loss": 0.1341,
      "step": 9600
    },
    {
      "epoch": 0.9798432250839866,
      "grad_norm": 4.824361801147461,
      "learning_rate": 1.786937538176142e-05,
      "loss": 0.1381,
      "step": 9625
    },
    {
      "epoch": 0.982388272421867,
      "grad_norm": 4.130845069885254,
      "learning_rate": 1.785806393231229e-05,
      "loss": 0.1397,
      "step": 9650
    },
    {
      "epoch": 0.9849333197597475,
      "grad_norm": 3.9672348499298096,
      "learning_rate": 1.7846752482863158e-05,
      "loss": 0.1547,
      "step": 9675
    },
    {
      "epoch": 0.987478367097628,
      "grad_norm": 4.505708694458008,
      "learning_rate": 1.7835441033414024e-05,
      "loss": 0.1801,
      "step": 9700
    },
    {
      "epoch": 0.9900234144355085,
      "grad_norm": 5.979536533355713,
      "learning_rate": 1.782412958396489e-05,
      "loss": 0.1419,
      "step": 9725
    },
    {
      "epoch": 0.992568461773389,
      "grad_norm": 3.7445387840270996,
      "learning_rate": 1.781281813451576e-05,
      "loss": 0.1665,
      "step": 9750
    },
    {
      "epoch": 0.9951135091112695,
      "grad_norm": 3.4875290393829346,
      "learning_rate": 1.7801506685066626e-05,
      "loss": 0.1535,
      "step": 9775
    },
    {
      "epoch": 0.9976585564491499,
      "grad_norm": 3.284198760986328,
      "learning_rate": 1.7790195235617492e-05,
      "loss": 0.1482,
      "step": 9800
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.10157867521047592,
      "eval_runtime": 7.0586,
      "eval_samples_per_second": 964.495,
      "eval_steps_per_second": 15.159,
      "step": 9823
    },
    {
      "epoch": 1.0002036037870305,
      "grad_norm": 2.1478254795074463,
      "learning_rate": 1.7778883786168362e-05,
      "loss": 0.0948,
      "step": 9825
    },
    {
      "epoch": 1.0027486511249109,
      "grad_norm": 5.228854656219482,
      "learning_rate": 1.7767572336719228e-05,
      "loss": 0.1329,
      "step": 9850
    },
    {
      "epoch": 1.0052936984627914,
      "grad_norm": 1.1476223468780518,
      "learning_rate": 1.7756260887270097e-05,
      "loss": 0.1203,
      "step": 9875
    },
    {
      "epoch": 1.007838745800672,
      "grad_norm": 5.739945411682129,
      "learning_rate": 1.7744949437820964e-05,
      "loss": 0.1418,
      "step": 9900
    },
    {
      "epoch": 1.0103837931385524,
      "grad_norm": 2.3966896533966064,
      "learning_rate": 1.7733637988371833e-05,
      "loss": 0.1122,
      "step": 9925
    },
    {
      "epoch": 1.012928840476433,
      "grad_norm": 4.959926128387451,
      "learning_rate": 1.77223265389227e-05,
      "loss": 0.1192,
      "step": 9950
    },
    {
      "epoch": 1.0154738878143132,
      "grad_norm": 5.601362705230713,
      "learning_rate": 1.7711015089473565e-05,
      "loss": 0.0988,
      "step": 9975
    },
    {
      "epoch": 1.0180189351521938,
      "grad_norm": 3.998955249786377,
      "learning_rate": 1.7699703640024435e-05,
      "loss": 0.1295,
      "step": 10000
    },
    {
      "epoch": 1.0205639824900743,
      "grad_norm": 4.754059314727783,
      "learning_rate": 1.76883921905753e-05,
      "loss": 0.1348,
      "step": 10025
    },
    {
      "epoch": 1.0231090298279548,
      "grad_norm": 4.0749735832214355,
      "learning_rate": 1.767708074112617e-05,
      "loss": 0.1333,
      "step": 10050
    },
    {
      "epoch": 1.0256540771658353,
      "grad_norm": 10.547972679138184,
      "learning_rate": 1.7665769291677037e-05,
      "loss": 0.1693,
      "step": 10075
    },
    {
      "epoch": 1.0281991245037159,
      "grad_norm": 6.241606712341309,
      "learning_rate": 1.7654457842227906e-05,
      "loss": 0.0902,
      "step": 10100
    },
    {
      "epoch": 1.0307441718415962,
      "grad_norm": 2.9035451412200928,
      "learning_rate": 1.7643146392778773e-05,
      "loss": 0.1196,
      "step": 10125
    },
    {
      "epoch": 1.0332892191794767,
      "grad_norm": 5.422848224639893,
      "learning_rate": 1.763183494332964e-05,
      "loss": 0.1624,
      "step": 10150
    },
    {
      "epoch": 1.0358342665173572,
      "grad_norm": 4.132188320159912,
      "learning_rate": 1.7620523493880508e-05,
      "loss": 0.1308,
      "step": 10175
    },
    {
      "epoch": 1.0383793138552377,
      "grad_norm": 5.723974704742432,
      "learning_rate": 1.7609212044431374e-05,
      "loss": 0.1226,
      "step": 10200
    },
    {
      "epoch": 1.0409243611931183,
      "grad_norm": 2.8668813705444336,
      "learning_rate": 1.7597900594982244e-05,
      "loss": 0.1049,
      "step": 10225
    },
    {
      "epoch": 1.0434694085309988,
      "grad_norm": 6.671713829040527,
      "learning_rate": 1.758658914553311e-05,
      "loss": 0.1377,
      "step": 10250
    },
    {
      "epoch": 1.046014455868879,
      "grad_norm": 1.478005290031433,
      "learning_rate": 1.7575277696083976e-05,
      "loss": 0.1183,
      "step": 10275
    },
    {
      "epoch": 1.0485595032067596,
      "grad_norm": 7.64422607421875,
      "learning_rate": 1.7563966246634846e-05,
      "loss": 0.1262,
      "step": 10300
    },
    {
      "epoch": 1.0511045505446401,
      "grad_norm": 9.314327239990234,
      "learning_rate": 1.7552654797185712e-05,
      "loss": 0.1159,
      "step": 10325
    },
    {
      "epoch": 1.0536495978825207,
      "grad_norm": 7.101593017578125,
      "learning_rate": 1.754134334773658e-05,
      "loss": 0.1545,
      "step": 10350
    },
    {
      "epoch": 1.0561946452204012,
      "grad_norm": 3.6233043670654297,
      "learning_rate": 1.7530031898287448e-05,
      "loss": 0.1602,
      "step": 10375
    },
    {
      "epoch": 1.0587396925582815,
      "grad_norm": 0.4728439450263977,
      "learning_rate": 1.7518720448838317e-05,
      "loss": 0.1373,
      "step": 10400
    },
    {
      "epoch": 1.061284739896162,
      "grad_norm": 2.2750091552734375,
      "learning_rate": 1.7507408999389183e-05,
      "loss": 0.1503,
      "step": 10425
    },
    {
      "epoch": 1.0638297872340425,
      "grad_norm": 6.410482406616211,
      "learning_rate": 1.749609754994005e-05,
      "loss": 0.1787,
      "step": 10450
    },
    {
      "epoch": 1.066374834571923,
      "grad_norm": 5.287363052368164,
      "learning_rate": 1.748478610049092e-05,
      "loss": 0.1642,
      "step": 10475
    },
    {
      "epoch": 1.0689198819098036,
      "grad_norm": 4.545390605926514,
      "learning_rate": 1.7473474651041785e-05,
      "loss": 0.1128,
      "step": 10500
    },
    {
      "epoch": 1.071464929247684,
      "grad_norm": 4.9489288330078125,
      "learning_rate": 1.7462163201592655e-05,
      "loss": 0.1206,
      "step": 10525
    },
    {
      "epoch": 1.0740099765855644,
      "grad_norm": 3.488626003265381,
      "learning_rate": 1.745085175214352e-05,
      "loss": 0.1176,
      "step": 10550
    },
    {
      "epoch": 1.076555023923445,
      "grad_norm": 5.163557529449463,
      "learning_rate": 1.743954030269439e-05,
      "loss": 0.1531,
      "step": 10575
    },
    {
      "epoch": 1.0791000712613255,
      "grad_norm": 7.763261795043945,
      "learning_rate": 1.7428228853245256e-05,
      "loss": 0.1371,
      "step": 10600
    },
    {
      "epoch": 1.081645118599206,
      "grad_norm": 3.1402170658111572,
      "learning_rate": 1.7416917403796123e-05,
      "loss": 0.1588,
      "step": 10625
    },
    {
      "epoch": 1.0841901659370865,
      "grad_norm": 3.7544312477111816,
      "learning_rate": 1.7405605954346992e-05,
      "loss": 0.1445,
      "step": 10650
    },
    {
      "epoch": 1.086735213274967,
      "grad_norm": 6.183403968811035,
      "learning_rate": 1.739429450489786e-05,
      "loss": 0.1113,
      "step": 10675
    },
    {
      "epoch": 1.0892802606128473,
      "grad_norm": 6.552886962890625,
      "learning_rate": 1.7382983055448724e-05,
      "loss": 0.1347,
      "step": 10700
    },
    {
      "epoch": 1.0918253079507279,
      "grad_norm": 3.3716797828674316,
      "learning_rate": 1.7371671605999594e-05,
      "loss": 0.1546,
      "step": 10725
    },
    {
      "epoch": 1.0943703552886084,
      "grad_norm": 4.622589111328125,
      "learning_rate": 1.7360360156550464e-05,
      "loss": 0.1687,
      "step": 10750
    },
    {
      "epoch": 1.096915402626489,
      "grad_norm": 4.305492401123047,
      "learning_rate": 1.734904870710133e-05,
      "loss": 0.1446,
      "step": 10775
    },
    {
      "epoch": 1.0994604499643694,
      "grad_norm": 6.439985275268555,
      "learning_rate": 1.7337737257652196e-05,
      "loss": 0.1295,
      "step": 10800
    },
    {
      "epoch": 1.1020054973022497,
      "grad_norm": 3.2125208377838135,
      "learning_rate": 1.7326425808203065e-05,
      "loss": 0.1218,
      "step": 10825
    },
    {
      "epoch": 1.1045505446401302,
      "grad_norm": 6.812816143035889,
      "learning_rate": 1.731511435875393e-05,
      "loss": 0.1201,
      "step": 10850
    },
    {
      "epoch": 1.1070955919780108,
      "grad_norm": 9.083202362060547,
      "learning_rate": 1.7303802909304798e-05,
      "loss": 0.1475,
      "step": 10875
    },
    {
      "epoch": 1.1096406393158913,
      "grad_norm": 9.970014572143555,
      "learning_rate": 1.7292491459855667e-05,
      "loss": 0.1278,
      "step": 10900
    },
    {
      "epoch": 1.1121856866537718,
      "grad_norm": 6.782029151916504,
      "learning_rate": 1.7281180010406537e-05,
      "loss": 0.1655,
      "step": 10925
    },
    {
      "epoch": 1.1147307339916523,
      "grad_norm": 2.6370034217834473,
      "learning_rate": 1.7269868560957403e-05,
      "loss": 0.134,
      "step": 10950
    },
    {
      "epoch": 1.1172757813295326,
      "grad_norm": 10.335969924926758,
      "learning_rate": 1.725855711150827e-05,
      "loss": 0.1174,
      "step": 10975
    },
    {
      "epoch": 1.1198208286674132,
      "grad_norm": 7.169666290283203,
      "learning_rate": 1.724724566205914e-05,
      "loss": 0.1559,
      "step": 11000
    },
    {
      "epoch": 1.1223658760052937,
      "grad_norm": 3.894927740097046,
      "learning_rate": 1.7235934212610005e-05,
      "loss": 0.1276,
      "step": 11025
    },
    {
      "epoch": 1.1249109233431742,
      "grad_norm": 8.561812400817871,
      "learning_rate": 1.722462276316087e-05,
      "loss": 0.1026,
      "step": 11050
    },
    {
      "epoch": 1.1274559706810547,
      "grad_norm": 7.845750331878662,
      "learning_rate": 1.721331131371174e-05,
      "loss": 0.1386,
      "step": 11075
    },
    {
      "epoch": 1.1300010180189353,
      "grad_norm": 3.596691370010376,
      "learning_rate": 1.720199986426261e-05,
      "loss": 0.1558,
      "step": 11100
    },
    {
      "epoch": 1.1325460653568156,
      "grad_norm": 2.8355915546417236,
      "learning_rate": 1.7190688414813476e-05,
      "loss": 0.1428,
      "step": 11125
    },
    {
      "epoch": 1.135091112694696,
      "grad_norm": 3.2784042358398438,
      "learning_rate": 1.7179376965364342e-05,
      "loss": 0.1838,
      "step": 11150
    },
    {
      "epoch": 1.1376361600325766,
      "grad_norm": 5.514401912689209,
      "learning_rate": 1.7168065515915212e-05,
      "loss": 0.0894,
      "step": 11175
    },
    {
      "epoch": 1.1401812073704571,
      "grad_norm": 6.449738502502441,
      "learning_rate": 1.7156754066466078e-05,
      "loss": 0.125,
      "step": 11200
    },
    {
      "epoch": 1.1427262547083377,
      "grad_norm": 3.2359158992767334,
      "learning_rate": 1.7145442617016944e-05,
      "loss": 0.1509,
      "step": 11225
    },
    {
      "epoch": 1.145271302046218,
      "grad_norm": 4.851259708404541,
      "learning_rate": 1.7134131167567814e-05,
      "loss": 0.1359,
      "step": 11250
    },
    {
      "epoch": 1.1478163493840985,
      "grad_norm": 3.3549108505249023,
      "learning_rate": 1.7122819718118683e-05,
      "loss": 0.1086,
      "step": 11275
    },
    {
      "epoch": 1.150361396721979,
      "grad_norm": 7.090845584869385,
      "learning_rate": 1.711150826866955e-05,
      "loss": 0.1361,
      "step": 11300
    },
    {
      "epoch": 1.1529064440598595,
      "grad_norm": 0.37024664878845215,
      "learning_rate": 1.7100196819220416e-05,
      "loss": 0.0929,
      "step": 11325
    },
    {
      "epoch": 1.15545149139774,
      "grad_norm": 2.443056583404541,
      "learning_rate": 1.7088885369771285e-05,
      "loss": 0.1291,
      "step": 11350
    },
    {
      "epoch": 1.1579965387356204,
      "grad_norm": 2.6655914783477783,
      "learning_rate": 1.707757392032215e-05,
      "loss": 0.0942,
      "step": 11375
    },
    {
      "epoch": 1.1605415860735009,
      "grad_norm": 4.226424217224121,
      "learning_rate": 1.7066262470873017e-05,
      "loss": 0.1658,
      "step": 11400
    },
    {
      "epoch": 1.1630866334113814,
      "grad_norm": 7.379846096038818,
      "learning_rate": 1.7054951021423887e-05,
      "loss": 0.1238,
      "step": 11425
    },
    {
      "epoch": 1.165631680749262,
      "grad_norm": 4.336475849151611,
      "learning_rate": 1.7043639571974756e-05,
      "loss": 0.143,
      "step": 11450
    },
    {
      "epoch": 1.1681767280871425,
      "grad_norm": 7.550657272338867,
      "learning_rate": 1.7032328122525623e-05,
      "loss": 0.1232,
      "step": 11475
    },
    {
      "epoch": 1.170721775425023,
      "grad_norm": 2.325849771499634,
      "learning_rate": 1.702101667307649e-05,
      "loss": 0.1253,
      "step": 11500
    },
    {
      "epoch": 1.1732668227629035,
      "grad_norm": 0.20483657717704773,
      "learning_rate": 1.7009705223627358e-05,
      "loss": 0.1415,
      "step": 11525
    },
    {
      "epoch": 1.1758118701007838,
      "grad_norm": 6.844511032104492,
      "learning_rate": 1.6998393774178224e-05,
      "loss": 0.1451,
      "step": 11550
    },
    {
      "epoch": 1.1783569174386643,
      "grad_norm": 2.30995512008667,
      "learning_rate": 1.698708232472909e-05,
      "loss": 0.1371,
      "step": 11575
    },
    {
      "epoch": 1.1809019647765449,
      "grad_norm": 5.756036758422852,
      "learning_rate": 1.697577087527996e-05,
      "loss": 0.1172,
      "step": 11600
    },
    {
      "epoch": 1.1834470121144254,
      "grad_norm": 4.841132164001465,
      "learning_rate": 1.696445942583083e-05,
      "loss": 0.1409,
      "step": 11625
    },
    {
      "epoch": 1.185992059452306,
      "grad_norm": 4.341500759124756,
      "learning_rate": 1.6953147976381696e-05,
      "loss": 0.1197,
      "step": 11650
    },
    {
      "epoch": 1.1885371067901862,
      "grad_norm": 3.437917947769165,
      "learning_rate": 1.6941836526932562e-05,
      "loss": 0.1384,
      "step": 11675
    },
    {
      "epoch": 1.1910821541280667,
      "grad_norm": 0.36744824051856995,
      "learning_rate": 1.693052507748343e-05,
      "loss": 0.1053,
      "step": 11700
    },
    {
      "epoch": 1.1936272014659473,
      "grad_norm": 6.063575744628906,
      "learning_rate": 1.6919213628034298e-05,
      "loss": 0.1402,
      "step": 11725
    },
    {
      "epoch": 1.1961722488038278,
      "grad_norm": 3.604062795639038,
      "learning_rate": 1.6907902178585164e-05,
      "loss": 0.1094,
      "step": 11750
    },
    {
      "epoch": 1.1987172961417083,
      "grad_norm": 6.017195701599121,
      "learning_rate": 1.6896590729136033e-05,
      "loss": 0.1301,
      "step": 11775
    },
    {
      "epoch": 1.2012623434795886,
      "grad_norm": 2.256840229034424,
      "learning_rate": 1.6885279279686903e-05,
      "loss": 0.1302,
      "step": 11800
    },
    {
      "epoch": 1.2038073908174691,
      "grad_norm": 2.5422046184539795,
      "learning_rate": 1.687396783023777e-05,
      "loss": 0.1111,
      "step": 11825
    },
    {
      "epoch": 1.2063524381553496,
      "grad_norm": 1.03132963180542,
      "learning_rate": 1.6862656380788635e-05,
      "loss": 0.1131,
      "step": 11850
    },
    {
      "epoch": 1.2088974854932302,
      "grad_norm": 3.2958521842956543,
      "learning_rate": 1.6851344931339505e-05,
      "loss": 0.1303,
      "step": 11875
    },
    {
      "epoch": 1.2114425328311107,
      "grad_norm": 4.987472057342529,
      "learning_rate": 1.684003348189037e-05,
      "loss": 0.1298,
      "step": 11900
    },
    {
      "epoch": 1.2139875801689912,
      "grad_norm": 2.972264289855957,
      "learning_rate": 1.6828722032441237e-05,
      "loss": 0.1426,
      "step": 11925
    },
    {
      "epoch": 1.2165326275068717,
      "grad_norm": 1.8931164741516113,
      "learning_rate": 1.6817410582992107e-05,
      "loss": 0.1437,
      "step": 11950
    },
    {
      "epoch": 1.219077674844752,
      "grad_norm": 4.350847244262695,
      "learning_rate": 1.6806099133542976e-05,
      "loss": 0.1301,
      "step": 11975
    },
    {
      "epoch": 1.2216227221826326,
      "grad_norm": 3.05570387840271,
      "learning_rate": 1.6794787684093842e-05,
      "loss": 0.1209,
      "step": 12000
    },
    {
      "epoch": 1.224167769520513,
      "grad_norm": 6.202553749084473,
      "learning_rate": 1.678347623464471e-05,
      "loss": 0.1203,
      "step": 12025
    },
    {
      "epoch": 1.2267128168583936,
      "grad_norm": 4.269680500030518,
      "learning_rate": 1.6772164785195578e-05,
      "loss": 0.1393,
      "step": 12050
    },
    {
      "epoch": 1.2292578641962741,
      "grad_norm": 13.57045841217041,
      "learning_rate": 1.6760853335746444e-05,
      "loss": 0.1442,
      "step": 12075
    },
    {
      "epoch": 1.2318029115341544,
      "grad_norm": 4.8651018142700195,
      "learning_rate": 1.674954188629731e-05,
      "loss": 0.1376,
      "step": 12100
    },
    {
      "epoch": 1.234347958872035,
      "grad_norm": 7.477112770080566,
      "learning_rate": 1.673823043684818e-05,
      "loss": 0.1278,
      "step": 12125
    },
    {
      "epoch": 1.2368930062099155,
      "grad_norm": 6.843944549560547,
      "learning_rate": 1.6726918987399046e-05,
      "loss": 0.1118,
      "step": 12150
    },
    {
      "epoch": 1.239438053547796,
      "grad_norm": 5.739787578582764,
      "learning_rate": 1.6715607537949915e-05,
      "loss": 0.1169,
      "step": 12175
    },
    {
      "epoch": 1.2419831008856765,
      "grad_norm": 2.3320653438568115,
      "learning_rate": 1.670429608850078e-05,
      "loss": 0.1234,
      "step": 12200
    },
    {
      "epoch": 1.2445281482235568,
      "grad_norm": 2.789673328399658,
      "learning_rate": 1.669298463905165e-05,
      "loss": 0.1482,
      "step": 12225
    },
    {
      "epoch": 1.2470731955614374,
      "grad_norm": 6.369720935821533,
      "learning_rate": 1.6681673189602517e-05,
      "loss": 0.1398,
      "step": 12250
    },
    {
      "epoch": 1.2496182428993179,
      "grad_norm": 5.785031318664551,
      "learning_rate": 1.6670361740153383e-05,
      "loss": 0.1433,
      "step": 12275
    },
    {
      "epoch": 1.2521632902371984,
      "grad_norm": 4.050633430480957,
      "learning_rate": 1.6659050290704253e-05,
      "loss": 0.1128,
      "step": 12300
    },
    {
      "epoch": 1.254708337575079,
      "grad_norm": 8.750215530395508,
      "learning_rate": 1.664773884125512e-05,
      "loss": 0.1636,
      "step": 12325
    },
    {
      "epoch": 1.2572533849129595,
      "grad_norm": 3.612732410430908,
      "learning_rate": 1.663642739180599e-05,
      "loss": 0.1722,
      "step": 12350
    },
    {
      "epoch": 1.25979843225084,
      "grad_norm": 5.242950439453125,
      "learning_rate": 1.6625115942356855e-05,
      "loss": 0.1421,
      "step": 12375
    },
    {
      "epoch": 1.2623434795887203,
      "grad_norm": 3.4691531658172607,
      "learning_rate": 1.6613804492907724e-05,
      "loss": 0.1413,
      "step": 12400
    },
    {
      "epoch": 1.2648885269266008,
      "grad_norm": 9.25007438659668,
      "learning_rate": 1.660249304345859e-05,
      "loss": 0.142,
      "step": 12425
    },
    {
      "epoch": 1.2674335742644813,
      "grad_norm": 4.163516044616699,
      "learning_rate": 1.6591181594009457e-05,
      "loss": 0.1528,
      "step": 12450
    },
    {
      "epoch": 1.2699786216023619,
      "grad_norm": 4.836366176605225,
      "learning_rate": 1.6579870144560326e-05,
      "loss": 0.13,
      "step": 12475
    },
    {
      "epoch": 1.2725236689402424,
      "grad_norm": 5.360639572143555,
      "learning_rate": 1.6568558695111192e-05,
      "loss": 0.1355,
      "step": 12500
    },
    {
      "epoch": 1.2750687162781227,
      "grad_norm": 3.893960952758789,
      "learning_rate": 1.6557247245662062e-05,
      "loss": 0.1,
      "step": 12525
    },
    {
      "epoch": 1.2776137636160032,
      "grad_norm": 5.7345991134643555,
      "learning_rate": 1.6545935796212928e-05,
      "loss": 0.1435,
      "step": 12550
    },
    {
      "epoch": 1.2801588109538837,
      "grad_norm": 8.154664993286133,
      "learning_rate": 1.6534624346763794e-05,
      "loss": 0.1381,
      "step": 12575
    },
    {
      "epoch": 1.2827038582917643,
      "grad_norm": 2.673351287841797,
      "learning_rate": 1.6523312897314664e-05,
      "loss": 0.127,
      "step": 12600
    },
    {
      "epoch": 1.2852489056296448,
      "grad_norm": 6.447815418243408,
      "learning_rate": 1.651200144786553e-05,
      "loss": 0.1345,
      "step": 12625
    },
    {
      "epoch": 1.287793952967525,
      "grad_norm": 6.898881435394287,
      "learning_rate": 1.65006899984164e-05,
      "loss": 0.1195,
      "step": 12650
    },
    {
      "epoch": 1.2903390003054056,
      "grad_norm": 6.427828311920166,
      "learning_rate": 1.6489378548967266e-05,
      "loss": 0.0898,
      "step": 12675
    },
    {
      "epoch": 1.2928840476432861,
      "grad_norm": 4.725465774536133,
      "learning_rate": 1.6478067099518135e-05,
      "loss": 0.1683,
      "step": 12700
    },
    {
      "epoch": 1.2954290949811667,
      "grad_norm": 5.63037633895874,
      "learning_rate": 1.6466755650069e-05,
      "loss": 0.1517,
      "step": 12725
    },
    {
      "epoch": 1.2979741423190472,
      "grad_norm": 4.656158447265625,
      "learning_rate": 1.6455444200619867e-05,
      "loss": 0.1627,
      "step": 12750
    },
    {
      "epoch": 1.3005191896569277,
      "grad_norm": 6.372010707855225,
      "learning_rate": 1.6444132751170737e-05,
      "loss": 0.1247,
      "step": 12775
    },
    {
      "epoch": 1.3030642369948082,
      "grad_norm": 5.822622299194336,
      "learning_rate": 1.6432821301721603e-05,
      "loss": 0.1421,
      "step": 12800
    },
    {
      "epoch": 1.3056092843326885,
      "grad_norm": 6.20810079574585,
      "learning_rate": 1.6421509852272473e-05,
      "loss": 0.1499,
      "step": 12825
    },
    {
      "epoch": 1.308154331670569,
      "grad_norm": 3.9495532512664795,
      "learning_rate": 1.641019840282334e-05,
      "loss": 0.1329,
      "step": 12850
    },
    {
      "epoch": 1.3106993790084496,
      "grad_norm": 4.609329700469971,
      "learning_rate": 1.639888695337421e-05,
      "loss": 0.1182,
      "step": 12875
    },
    {
      "epoch": 1.31324442634633,
      "grad_norm": 6.07358455657959,
      "learning_rate": 1.6387575503925075e-05,
      "loss": 0.1256,
      "step": 12900
    },
    {
      "epoch": 1.3157894736842106,
      "grad_norm": 1.8420575857162476,
      "learning_rate": 1.637626405447594e-05,
      "loss": 0.1173,
      "step": 12925
    },
    {
      "epoch": 1.318334521022091,
      "grad_norm": 7.779773235321045,
      "learning_rate": 1.636495260502681e-05,
      "loss": 0.1328,
      "step": 12950
    },
    {
      "epoch": 1.3208795683599714,
      "grad_norm": 8.295707702636719,
      "learning_rate": 1.6353641155577676e-05,
      "loss": 0.1315,
      "step": 12975
    },
    {
      "epoch": 1.323424615697852,
      "grad_norm": 3.329450845718384,
      "learning_rate": 1.6342329706128546e-05,
      "loss": 0.1224,
      "step": 13000
    },
    {
      "epoch": 1.3259696630357325,
      "grad_norm": 7.493209362030029,
      "learning_rate": 1.6331018256679412e-05,
      "loss": 0.1203,
      "step": 13025
    },
    {
      "epoch": 1.328514710373613,
      "grad_norm": 3.4829013347625732,
      "learning_rate": 1.631970680723028e-05,
      "loss": 0.1319,
      "step": 13050
    },
    {
      "epoch": 1.3310597577114933,
      "grad_norm": 4.998099327087402,
      "learning_rate": 1.6308395357781148e-05,
      "loss": 0.1272,
      "step": 13075
    },
    {
      "epoch": 1.3336048050493738,
      "grad_norm": 6.8269524574279785,
      "learning_rate": 1.6297083908332014e-05,
      "loss": 0.1759,
      "step": 13100
    },
    {
      "epoch": 1.3361498523872544,
      "grad_norm": 2.3060414791107178,
      "learning_rate": 1.6285772458882883e-05,
      "loss": 0.1174,
      "step": 13125
    },
    {
      "epoch": 1.338694899725135,
      "grad_norm": 2.708559274673462,
      "learning_rate": 1.627446100943375e-05,
      "loss": 0.1042,
      "step": 13150
    },
    {
      "epoch": 1.3412399470630154,
      "grad_norm": 5.088305950164795,
      "learning_rate": 1.6263149559984616e-05,
      "loss": 0.1399,
      "step": 13175
    },
    {
      "epoch": 1.343784994400896,
      "grad_norm": 3.7776496410369873,
      "learning_rate": 1.6251838110535485e-05,
      "loss": 0.1133,
      "step": 13200
    },
    {
      "epoch": 1.3463300417387765,
      "grad_norm": 1.5227245092391968,
      "learning_rate": 1.6240526661086355e-05,
      "loss": 0.1199,
      "step": 13225
    },
    {
      "epoch": 1.3488750890766568,
      "grad_norm": 3.7397494316101074,
      "learning_rate": 1.622921521163722e-05,
      "loss": 0.1066,
      "step": 13250
    },
    {
      "epoch": 1.3514201364145373,
      "grad_norm": 6.5920209884643555,
      "learning_rate": 1.6217903762188087e-05,
      "loss": 0.153,
      "step": 13275
    },
    {
      "epoch": 1.3539651837524178,
      "grad_norm": 8.203237533569336,
      "learning_rate": 1.6206592312738957e-05,
      "loss": 0.1267,
      "step": 13300
    },
    {
      "epoch": 1.3565102310902983,
      "grad_norm": 6.9769415855407715,
      "learning_rate": 1.6195280863289823e-05,
      "loss": 0.1227,
      "step": 13325
    },
    {
      "epoch": 1.3590552784281789,
      "grad_norm": 7.504185199737549,
      "learning_rate": 1.618396941384069e-05,
      "loss": 0.1384,
      "step": 13350
    },
    {
      "epoch": 1.3616003257660592,
      "grad_norm": 7.599664688110352,
      "learning_rate": 1.617265796439156e-05,
      "loss": 0.1268,
      "step": 13375
    },
    {
      "epoch": 1.3641453731039397,
      "grad_norm": 4.518153667449951,
      "learning_rate": 1.6161346514942428e-05,
      "loss": 0.1095,
      "step": 13400
    },
    {
      "epoch": 1.3666904204418202,
      "grad_norm": 5.297679901123047,
      "learning_rate": 1.6150035065493294e-05,
      "loss": 0.168,
      "step": 13425
    },
    {
      "epoch": 1.3692354677797007,
      "grad_norm": 5.273420810699463,
      "learning_rate": 1.613872361604416e-05,
      "loss": 0.1334,
      "step": 13450
    },
    {
      "epoch": 1.3717805151175813,
      "grad_norm": 3.286939859390259,
      "learning_rate": 1.612741216659503e-05,
      "loss": 0.085,
      "step": 13475
    },
    {
      "epoch": 1.3743255624554616,
      "grad_norm": 8.941014289855957,
      "learning_rate": 1.6116100717145896e-05,
      "loss": 0.135,
      "step": 13500
    },
    {
      "epoch": 1.376870609793342,
      "grad_norm": 4.475488185882568,
      "learning_rate": 1.6104789267696762e-05,
      "loss": 0.1172,
      "step": 13525
    },
    {
      "epoch": 1.3794156571312226,
      "grad_norm": 3.8119313716888428,
      "learning_rate": 1.6093477818247632e-05,
      "loss": 0.1344,
      "step": 13550
    },
    {
      "epoch": 1.3819607044691031,
      "grad_norm": 1.4924836158752441,
      "learning_rate": 1.60821663687985e-05,
      "loss": 0.1608,
      "step": 13575
    },
    {
      "epoch": 1.3845057518069837,
      "grad_norm": 5.886949062347412,
      "learning_rate": 1.6070854919349364e-05,
      "loss": 0.1453,
      "step": 13600
    },
    {
      "epoch": 1.3870507991448642,
      "grad_norm": 5.144103527069092,
      "learning_rate": 1.6059543469900234e-05,
      "loss": 0.1446,
      "step": 13625
    },
    {
      "epoch": 1.3895958464827447,
      "grad_norm": 7.620888710021973,
      "learning_rate": 1.6048232020451103e-05,
      "loss": 0.1304,
      "step": 13650
    },
    {
      "epoch": 1.392140893820625,
      "grad_norm": 5.124198913574219,
      "learning_rate": 1.603692057100197e-05,
      "loss": 0.1467,
      "step": 13675
    },
    {
      "epoch": 1.3946859411585055,
      "grad_norm": 0.9429985284805298,
      "learning_rate": 1.6025609121552835e-05,
      "loss": 0.0788,
      "step": 13700
    },
    {
      "epoch": 1.397230988496386,
      "grad_norm": 7.610143184661865,
      "learning_rate": 1.6014297672103705e-05,
      "loss": 0.1546,
      "step": 13725
    },
    {
      "epoch": 1.3997760358342666,
      "grad_norm": 3.9099016189575195,
      "learning_rate": 1.6002986222654574e-05,
      "loss": 0.0995,
      "step": 13750
    },
    {
      "epoch": 1.402321083172147,
      "grad_norm": 2.6341960430145264,
      "learning_rate": 1.5991674773205437e-05,
      "loss": 0.1381,
      "step": 13775
    },
    {
      "epoch": 1.4048661305100274,
      "grad_norm": 1.491227149963379,
      "learning_rate": 1.5980815781734275e-05,
      "loss": 0.1514,
      "step": 13800
    },
    {
      "epoch": 1.407411177847908,
      "grad_norm": 3.6851260662078857,
      "learning_rate": 1.596950433228514e-05,
      "loss": 0.1678,
      "step": 13825
    },
    {
      "epoch": 1.4099562251857884,
      "grad_norm": 6.391075134277344,
      "learning_rate": 1.5958192882836007e-05,
      "loss": 0.1329,
      "step": 13850
    },
    {
      "epoch": 1.412501272523669,
      "grad_norm": 4.474203109741211,
      "learning_rate": 1.5946881433386877e-05,
      "loss": 0.1137,
      "step": 13875
    },
    {
      "epoch": 1.4150463198615495,
      "grad_norm": 0.19339196383953094,
      "learning_rate": 1.5935569983937743e-05,
      "loss": 0.1353,
      "step": 13900
    },
    {
      "epoch": 1.4175913671994298,
      "grad_norm": 3.198152780532837,
      "learning_rate": 1.592425853448861e-05,
      "loss": 0.1347,
      "step": 13925
    },
    {
      "epoch": 1.4201364145373103,
      "grad_norm": 2.9210307598114014,
      "learning_rate": 1.591294708503948e-05,
      "loss": 0.0968,
      "step": 13950
    },
    {
      "epoch": 1.4226814618751908,
      "grad_norm": 0.791731059551239,
      "learning_rate": 1.5901635635590348e-05,
      "loss": 0.1162,
      "step": 13975
    },
    {
      "epoch": 1.4252265092130714,
      "grad_norm": 7.955835342407227,
      "learning_rate": 1.5890324186141214e-05,
      "loss": 0.1213,
      "step": 14000
    },
    {
      "epoch": 1.427771556550952,
      "grad_norm": 8.638495445251465,
      "learning_rate": 1.587901273669208e-05,
      "loss": 0.1339,
      "step": 14025
    },
    {
      "epoch": 1.4303166038888322,
      "grad_norm": 4.877995014190674,
      "learning_rate": 1.586770128724295e-05,
      "loss": 0.1296,
      "step": 14050
    },
    {
      "epoch": 1.432861651226713,
      "grad_norm": 10.070806503295898,
      "learning_rate": 1.5856389837793816e-05,
      "loss": 0.1447,
      "step": 14075
    },
    {
      "epoch": 1.4354066985645932,
      "grad_norm": 3.1885571479797363,
      "learning_rate": 1.5845078388344682e-05,
      "loss": 0.1256,
      "step": 14100
    },
    {
      "epoch": 1.4379517459024738,
      "grad_norm": 9.940485954284668,
      "learning_rate": 1.5833766938895552e-05,
      "loss": 0.1325,
      "step": 14125
    },
    {
      "epoch": 1.4404967932403543,
      "grad_norm": 9.942497253417969,
      "learning_rate": 1.582245548944642e-05,
      "loss": 0.1258,
      "step": 14150
    },
    {
      "epoch": 1.4430418405782348,
      "grad_norm": 1.4657670259475708,
      "learning_rate": 1.5811144039997288e-05,
      "loss": 0.1295,
      "step": 14175
    },
    {
      "epoch": 1.4455868879161153,
      "grad_norm": 1.3000506162643433,
      "learning_rate": 1.5799832590548154e-05,
      "loss": 0.1533,
      "step": 14200
    },
    {
      "epoch": 1.4481319352539956,
      "grad_norm": 3.836527109146118,
      "learning_rate": 1.5788521141099023e-05,
      "loss": 0.1188,
      "step": 14225
    },
    {
      "epoch": 1.4506769825918762,
      "grad_norm": 7.592291355133057,
      "learning_rate": 1.577720969164989e-05,
      "loss": 0.1273,
      "step": 14250
    },
    {
      "epoch": 1.4532220299297567,
      "grad_norm": 0.6994871497154236,
      "learning_rate": 1.5765898242200756e-05,
      "loss": 0.1291,
      "step": 14275
    },
    {
      "epoch": 1.4557670772676372,
      "grad_norm": 5.725295066833496,
      "learning_rate": 1.5754586792751625e-05,
      "loss": 0.1534,
      "step": 14300
    },
    {
      "epoch": 1.4583121246055177,
      "grad_norm": 8.091775894165039,
      "learning_rate": 1.574327534330249e-05,
      "loss": 0.1256,
      "step": 14325
    },
    {
      "epoch": 1.460857171943398,
      "grad_norm": 1.0600472688674927,
      "learning_rate": 1.573196389385336e-05,
      "loss": 0.1317,
      "step": 14350
    },
    {
      "epoch": 1.4634022192812786,
      "grad_norm": 4.549612522125244,
      "learning_rate": 1.5720652444404227e-05,
      "loss": 0.1729,
      "step": 14375
    },
    {
      "epoch": 1.465947266619159,
      "grad_norm": 3.8479483127593994,
      "learning_rate": 1.5709340994955096e-05,
      "loss": 0.1498,
      "step": 14400
    },
    {
      "epoch": 1.4684923139570396,
      "grad_norm": 6.539849281311035,
      "learning_rate": 1.5698029545505963e-05,
      "loss": 0.113,
      "step": 14425
    },
    {
      "epoch": 1.4710373612949201,
      "grad_norm": 3.222187042236328,
      "learning_rate": 1.568671809605683e-05,
      "loss": 0.1169,
      "step": 14450
    },
    {
      "epoch": 1.4735824086328004,
      "grad_norm": 6.104904651641846,
      "learning_rate": 1.56754066466077e-05,
      "loss": 0.1429,
      "step": 14475
    },
    {
      "epoch": 1.4761274559706812,
      "grad_norm": 2.730999708175659,
      "learning_rate": 1.5664095197158564e-05,
      "loss": 0.1178,
      "step": 14500
    },
    {
      "epoch": 1.4786725033085615,
      "grad_norm": 6.132184982299805,
      "learning_rate": 1.5652783747709434e-05,
      "loss": 0.1333,
      "step": 14525
    },
    {
      "epoch": 1.481217550646442,
      "grad_norm": 2.02093505859375,
      "learning_rate": 1.56414722982603e-05,
      "loss": 0.1247,
      "step": 14550
    },
    {
      "epoch": 1.4837625979843225,
      "grad_norm": 5.066205978393555,
      "learning_rate": 1.563016084881117e-05,
      "loss": 0.08,
      "step": 14575
    },
    {
      "epoch": 1.486307645322203,
      "grad_norm": 2.487128973007202,
      "learning_rate": 1.5618849399362036e-05,
      "loss": 0.107,
      "step": 14600
    },
    {
      "epoch": 1.4888526926600836,
      "grad_norm": 4.138731479644775,
      "learning_rate": 1.5607537949912902e-05,
      "loss": 0.126,
      "step": 14625
    },
    {
      "epoch": 1.4913977399979639,
      "grad_norm": 3.9390079975128174,
      "learning_rate": 1.559622650046377e-05,
      "loss": 0.1206,
      "step": 14650
    },
    {
      "epoch": 1.4939427873358444,
      "grad_norm": 5.015639305114746,
      "learning_rate": 1.5584915051014638e-05,
      "loss": 0.1353,
      "step": 14675
    },
    {
      "epoch": 1.496487834673725,
      "grad_norm": 6.541854381561279,
      "learning_rate": 1.5573603601565507e-05,
      "loss": 0.1401,
      "step": 14700
    },
    {
      "epoch": 1.4990328820116054,
      "grad_norm": 8.357683181762695,
      "learning_rate": 1.5562292152116373e-05,
      "loss": 0.1244,
      "step": 14725
    },
    {
      "epoch": 1.501577929349486,
      "grad_norm": 0.8040742874145508,
      "learning_rate": 1.5550980702667243e-05,
      "loss": 0.127,
      "step": 14750
    },
    {
      "epoch": 1.5041229766873663,
      "grad_norm": 4.00300931930542,
      "learning_rate": 1.553966925321811e-05,
      "loss": 0.1174,
      "step": 14775
    },
    {
      "epoch": 1.506668024025247,
      "grad_norm": 4.621161460876465,
      "learning_rate": 1.5528357803768975e-05,
      "loss": 0.1155,
      "step": 14800
    },
    {
      "epoch": 1.5092130713631273,
      "grad_norm": 4.954023838043213,
      "learning_rate": 1.5517046354319845e-05,
      "loss": 0.1682,
      "step": 14825
    },
    {
      "epoch": 1.5117581187010078,
      "grad_norm": 1.7072348594665527,
      "learning_rate": 1.550573490487071e-05,
      "loss": 0.1423,
      "step": 14850
    },
    {
      "epoch": 1.5143031660388884,
      "grad_norm": 6.226248264312744,
      "learning_rate": 1.549442345542158e-05,
      "loss": 0.1285,
      "step": 14875
    },
    {
      "epoch": 1.5168482133767687,
      "grad_norm": 6.3383259773254395,
      "learning_rate": 1.5483112005972447e-05,
      "loss": 0.108,
      "step": 14900
    },
    {
      "epoch": 1.5193932607146494,
      "grad_norm": 4.042302131652832,
      "learning_rate": 1.5471800556523313e-05,
      "loss": 0.1085,
      "step": 14925
    },
    {
      "epoch": 1.5219383080525297,
      "grad_norm": 2.5637173652648926,
      "learning_rate": 1.5460489107074182e-05,
      "loss": 0.1202,
      "step": 14950
    },
    {
      "epoch": 1.5244833553904102,
      "grad_norm": 2.83852219581604,
      "learning_rate": 1.544917765762505e-05,
      "loss": 0.1306,
      "step": 14975
    },
    {
      "epoch": 1.5270284027282908,
      "grad_norm": 5.585648059844971,
      "learning_rate": 1.5437866208175918e-05,
      "loss": 0.1391,
      "step": 15000
    },
    {
      "epoch": 1.529573450066171,
      "grad_norm": 4.606128692626953,
      "learning_rate": 1.5426554758726784e-05,
      "loss": 0.1483,
      "step": 15025
    },
    {
      "epoch": 1.5321184974040518,
      "grad_norm": 9.550539016723633,
      "learning_rate": 1.5415243309277654e-05,
      "loss": 0.1047,
      "step": 15050
    },
    {
      "epoch": 1.5346635447419321,
      "grad_norm": 4.530235767364502,
      "learning_rate": 1.540393185982852e-05,
      "loss": 0.1181,
      "step": 15075
    },
    {
      "epoch": 1.5372085920798126,
      "grad_norm": 4.647919654846191,
      "learning_rate": 1.5392620410379386e-05,
      "loss": 0.118,
      "step": 15100
    },
    {
      "epoch": 1.5397536394176932,
      "grad_norm": 2.899007797241211,
      "learning_rate": 1.5381308960930256e-05,
      "loss": 0.1248,
      "step": 15125
    },
    {
      "epoch": 1.5422986867555737,
      "grad_norm": 11.52777099609375,
      "learning_rate": 1.536999751148112e-05,
      "loss": 0.1114,
      "step": 15150
    },
    {
      "epoch": 1.5448437340934542,
      "grad_norm": 1.5432101488113403,
      "learning_rate": 1.535868606203199e-05,
      "loss": 0.1162,
      "step": 15175
    },
    {
      "epoch": 1.5473887814313345,
      "grad_norm": 6.4070658683776855,
      "learning_rate": 1.5347374612582857e-05,
      "loss": 0.1618,
      "step": 15200
    },
    {
      "epoch": 1.5499338287692153,
      "grad_norm": 9.226142883300781,
      "learning_rate": 1.5336063163133727e-05,
      "loss": 0.1624,
      "step": 15225
    },
    {
      "epoch": 1.5524788761070956,
      "grad_norm": 1.0666420459747314,
      "learning_rate": 1.5324751713684593e-05,
      "loss": 0.1109,
      "step": 15250
    },
    {
      "epoch": 1.555023923444976,
      "grad_norm": 4.738593578338623,
      "learning_rate": 1.531344026423546e-05,
      "loss": 0.1333,
      "step": 15275
    },
    {
      "epoch": 1.5575689707828566,
      "grad_norm": 1.8429763317108154,
      "learning_rate": 1.530212881478633e-05,
      "loss": 0.1386,
      "step": 15300
    },
    {
      "epoch": 1.560114018120737,
      "grad_norm": 7.202968120574951,
      "learning_rate": 1.5290817365337195e-05,
      "loss": 0.1179,
      "step": 15325
    },
    {
      "epoch": 1.5626590654586177,
      "grad_norm": 9.482736587524414,
      "learning_rate": 1.527950591588806e-05,
      "loss": 0.16,
      "step": 15350
    },
    {
      "epoch": 1.565204112796498,
      "grad_norm": 3.1656908988952637,
      "learning_rate": 1.526819446643893e-05,
      "loss": 0.1053,
      "step": 15375
    },
    {
      "epoch": 1.5677491601343785,
      "grad_norm": 7.062704563140869,
      "learning_rate": 1.5256883016989798e-05,
      "loss": 0.1121,
      "step": 15400
    },
    {
      "epoch": 1.570294207472259,
      "grad_norm": 3.370267629623413,
      "learning_rate": 1.5245571567540666e-05,
      "loss": 0.112,
      "step": 15425
    },
    {
      "epoch": 1.5728392548101393,
      "grad_norm": 5.777838230133057,
      "learning_rate": 1.5234260118091532e-05,
      "loss": 0.1433,
      "step": 15450
    },
    {
      "epoch": 1.57538430214802,
      "grad_norm": 2.517376184463501,
      "learning_rate": 1.52229486686424e-05,
      "loss": 0.1068,
      "step": 15475
    },
    {
      "epoch": 1.5779293494859004,
      "grad_norm": 4.369190216064453,
      "learning_rate": 1.521163721919327e-05,
      "loss": 0.151,
      "step": 15500
    },
    {
      "epoch": 1.5804743968237809,
      "grad_norm": 1.2526782751083374,
      "learning_rate": 1.5200325769744136e-05,
      "loss": 0.1137,
      "step": 15525
    },
    {
      "epoch": 1.5830194441616614,
      "grad_norm": 3.6932199001312256,
      "learning_rate": 1.5189014320295004e-05,
      "loss": 0.1045,
      "step": 15550
    },
    {
      "epoch": 1.585564491499542,
      "grad_norm": 0.895078182220459,
      "learning_rate": 1.5177702870845872e-05,
      "loss": 0.1014,
      "step": 15575
    },
    {
      "epoch": 1.5881095388374225,
      "grad_norm": 4.923023223876953,
      "learning_rate": 1.516639142139674e-05,
      "loss": 0.1102,
      "step": 15600
    },
    {
      "epoch": 1.5906545861753028,
      "grad_norm": 6.089389801025391,
      "learning_rate": 1.5155079971947606e-05,
      "loss": 0.1446,
      "step": 15625
    },
    {
      "epoch": 1.5931996335131835,
      "grad_norm": 3.304971933364868,
      "learning_rate": 1.5143768522498474e-05,
      "loss": 0.098,
      "step": 15650
    },
    {
      "epoch": 1.5957446808510638,
      "grad_norm": 8.50531005859375,
      "learning_rate": 1.5132457073049343e-05,
      "loss": 0.1312,
      "step": 15675
    },
    {
      "epoch": 1.5982897281889443,
      "grad_norm": 2.5584232807159424,
      "learning_rate": 1.512114562360021e-05,
      "loss": 0.108,
      "step": 15700
    },
    {
      "epoch": 1.6008347755268248,
      "grad_norm": 7.9185590744018555,
      "learning_rate": 1.5109834174151077e-05,
      "loss": 0.1321,
      "step": 15725
    },
    {
      "epoch": 1.6033798228647052,
      "grad_norm": 4.034428119659424,
      "learning_rate": 1.5098522724701945e-05,
      "loss": 0.1454,
      "step": 15750
    },
    {
      "epoch": 1.605924870202586,
      "grad_norm": 0.8627036809921265,
      "learning_rate": 1.5087211275252811e-05,
      "loss": 0.1119,
      "step": 15775
    },
    {
      "epoch": 1.6084699175404662,
      "grad_norm": 5.200289726257324,
      "learning_rate": 1.5075899825803679e-05,
      "loss": 0.1152,
      "step": 15800
    },
    {
      "epoch": 1.6110149648783467,
      "grad_norm": 7.20810079574585,
      "learning_rate": 1.5064588376354547e-05,
      "loss": 0.0935,
      "step": 15825
    },
    {
      "epoch": 1.6135600122162272,
      "grad_norm": 1.9433012008666992,
      "learning_rate": 1.5053276926905416e-05,
      "loss": 0.1604,
      "step": 15850
    },
    {
      "epoch": 1.6161050595541075,
      "grad_norm": 3.855752944946289,
      "learning_rate": 1.5042417935434249e-05,
      "loss": 0.1198,
      "step": 15875
    },
    {
      "epoch": 1.6186501068919883,
      "grad_norm": 4.6323652267456055,
      "learning_rate": 1.5031106485985117e-05,
      "loss": 0.1409,
      "step": 15900
    },
    {
      "epoch": 1.6211951542298686,
      "grad_norm": 2.9197826385498047,
      "learning_rate": 1.5019795036535983e-05,
      "loss": 0.0972,
      "step": 15925
    },
    {
      "epoch": 1.6237402015677491,
      "grad_norm": 4.8417158126831055,
      "learning_rate": 1.500848358708685e-05,
      "loss": 0.1041,
      "step": 15950
    },
    {
      "epoch": 1.6262852489056296,
      "grad_norm": 0.9690869450569153,
      "learning_rate": 1.4997172137637719e-05,
      "loss": 0.1215,
      "step": 15975
    },
    {
      "epoch": 1.6288302962435102,
      "grad_norm": 5.960648536682129,
      "learning_rate": 1.4985860688188585e-05,
      "loss": 0.1569,
      "step": 16000
    },
    {
      "epoch": 1.6313753435813907,
      "grad_norm": 9.493753433227539,
      "learning_rate": 1.4974549238739453e-05,
      "loss": 0.1169,
      "step": 16025
    },
    {
      "epoch": 1.633920390919271,
      "grad_norm": 2.922004461288452,
      "learning_rate": 1.4963237789290322e-05,
      "loss": 0.112,
      "step": 16050
    },
    {
      "epoch": 1.6364654382571517,
      "grad_norm": 7.904358863830566,
      "learning_rate": 1.495192633984119e-05,
      "loss": 0.1441,
      "step": 16075
    },
    {
      "epoch": 1.639010485595032,
      "grad_norm": 0.4537775218486786,
      "learning_rate": 1.4940614890392056e-05,
      "loss": 0.0963,
      "step": 16100
    },
    {
      "epoch": 1.6415555329329126,
      "grad_norm": 8.233470916748047,
      "learning_rate": 1.4929303440942924e-05,
      "loss": 0.1392,
      "step": 16125
    },
    {
      "epoch": 1.644100580270793,
      "grad_norm": 10.062689781188965,
      "learning_rate": 1.4917991991493792e-05,
      "loss": 0.1349,
      "step": 16150
    },
    {
      "epoch": 1.6466456276086734,
      "grad_norm": 7.0017218589782715,
      "learning_rate": 1.4906680542044658e-05,
      "loss": 0.1563,
      "step": 16175
    },
    {
      "epoch": 1.6491906749465541,
      "grad_norm": 1.1416449546813965,
      "learning_rate": 1.4895369092595526e-05,
      "loss": 0.1232,
      "step": 16200
    },
    {
      "epoch": 1.6517357222844344,
      "grad_norm": 1.434586763381958,
      "learning_rate": 1.4884057643146395e-05,
      "loss": 0.1227,
      "step": 16225
    },
    {
      "epoch": 1.654280769622315,
      "grad_norm": 4.737888336181641,
      "learning_rate": 1.487274619369726e-05,
      "loss": 0.0974,
      "step": 16250
    },
    {
      "epoch": 1.6568258169601955,
      "grad_norm": 7.9679718017578125,
      "learning_rate": 1.486143474424813e-05,
      "loss": 0.1188,
      "step": 16275
    },
    {
      "epoch": 1.6593708642980758,
      "grad_norm": 3.9078972339630127,
      "learning_rate": 1.4850123294798997e-05,
      "loss": 0.0956,
      "step": 16300
    },
    {
      "epoch": 1.6619159116359565,
      "grad_norm": 0.9742254614830017,
      "learning_rate": 1.4838811845349865e-05,
      "loss": 0.1065,
      "step": 16325
    },
    {
      "epoch": 1.6644609589738368,
      "grad_norm": 8.059602737426758,
      "learning_rate": 1.4827500395900731e-05,
      "loss": 0.1578,
      "step": 16350
    },
    {
      "epoch": 1.6670060063117174,
      "grad_norm": 2.904923677444458,
      "learning_rate": 1.4816188946451599e-05,
      "loss": 0.0992,
      "step": 16375
    },
    {
      "epoch": 1.6695510536495979,
      "grad_norm": 3.145519733428955,
      "learning_rate": 1.4804877497002469e-05,
      "loss": 0.119,
      "step": 16400
    },
    {
      "epoch": 1.6720961009874784,
      "grad_norm": 3.9910943508148193,
      "learning_rate": 1.4793566047553333e-05,
      "loss": 0.1349,
      "step": 16425
    },
    {
      "epoch": 1.674641148325359,
      "grad_norm": 6.027825355529785,
      "learning_rate": 1.4782254598104203e-05,
      "loss": 0.1392,
      "step": 16450
    },
    {
      "epoch": 1.6771861956632392,
      "grad_norm": 2.852295398712158,
      "learning_rate": 1.477094314865507e-05,
      "loss": 0.1405,
      "step": 16475
    },
    {
      "epoch": 1.67973124300112,
      "grad_norm": 3.4339046478271484,
      "learning_rate": 1.4759631699205938e-05,
      "loss": 0.1204,
      "step": 16500
    },
    {
      "epoch": 1.6822762903390003,
      "grad_norm": 9.087562561035156,
      "learning_rate": 1.4748320249756804e-05,
      "loss": 0.1483,
      "step": 16525
    },
    {
      "epoch": 1.6848213376768808,
      "grad_norm": 8.998224258422852,
      "learning_rate": 1.4737008800307672e-05,
      "loss": 0.1192,
      "step": 16550
    },
    {
      "epoch": 1.6873663850147613,
      "grad_norm": 4.188343048095703,
      "learning_rate": 1.4725697350858542e-05,
      "loss": 0.1044,
      "step": 16575
    },
    {
      "epoch": 1.6899114323526416,
      "grad_norm": 0.9458564519882202,
      "learning_rate": 1.4714385901409406e-05,
      "loss": 0.1418,
      "step": 16600
    },
    {
      "epoch": 1.6924564796905224,
      "grad_norm": 7.491810321807861,
      "learning_rate": 1.4703074451960276e-05,
      "loss": 0.1299,
      "step": 16625
    },
    {
      "epoch": 1.6950015270284027,
      "grad_norm": 8.30079460144043,
      "learning_rate": 1.4691763002511144e-05,
      "loss": 0.1225,
      "step": 16650
    },
    {
      "epoch": 1.6975465743662832,
      "grad_norm": 4.6412272453308105,
      "learning_rate": 1.468045155306201e-05,
      "loss": 0.1487,
      "step": 16675
    },
    {
      "epoch": 1.7000916217041637,
      "grad_norm": 0.662234365940094,
      "learning_rate": 1.4669140103612878e-05,
      "loss": 0.0988,
      "step": 16700
    },
    {
      "epoch": 1.702636669042044,
      "grad_norm": 0.9764494895935059,
      "learning_rate": 1.4657828654163745e-05,
      "loss": 0.1137,
      "step": 16725
    },
    {
      "epoch": 1.7051817163799248,
      "grad_norm": 3.0529532432556152,
      "learning_rate": 1.4646517204714615e-05,
      "loss": 0.104,
      "step": 16750
    },
    {
      "epoch": 1.707726763717805,
      "grad_norm": 2.24320387840271,
      "learning_rate": 1.463520575526548e-05,
      "loss": 0.1038,
      "step": 16775
    },
    {
      "epoch": 1.7102718110556856,
      "grad_norm": 12.847587585449219,
      "learning_rate": 1.4623894305816349e-05,
      "loss": 0.1447,
      "step": 16800
    },
    {
      "epoch": 1.7128168583935661,
      "grad_norm": 2.5556368827819824,
      "learning_rate": 1.4612582856367217e-05,
      "loss": 0.1125,
      "step": 16825
    },
    {
      "epoch": 1.7153619057314466,
      "grad_norm": 1.6241698265075684,
      "learning_rate": 1.4601271406918083e-05,
      "loss": 0.0814,
      "step": 16850
    },
    {
      "epoch": 1.7179069530693272,
      "grad_norm": 7.694450855255127,
      "learning_rate": 1.4589959957468951e-05,
      "loss": 0.1414,
      "step": 16875
    },
    {
      "epoch": 1.7204520004072075,
      "grad_norm": 3.1575894355773926,
      "learning_rate": 1.4578648508019819e-05,
      "loss": 0.1473,
      "step": 16900
    },
    {
      "epoch": 1.7229970477450882,
      "grad_norm": 8.672683715820312,
      "learning_rate": 1.4567337058570688e-05,
      "loss": 0.1017,
      "step": 16925
    },
    {
      "epoch": 1.7255420950829685,
      "grad_norm": 6.179168701171875,
      "learning_rate": 1.4556025609121553e-05,
      "loss": 0.1454,
      "step": 16950
    },
    {
      "epoch": 1.728087142420849,
      "grad_norm": 4.374868392944336,
      "learning_rate": 1.4544714159672422e-05,
      "loss": 0.117,
      "step": 16975
    },
    {
      "epoch": 1.7306321897587296,
      "grad_norm": 3.160280227661133,
      "learning_rate": 1.453340271022329e-05,
      "loss": 0.0905,
      "step": 17000
    },
    {
      "epoch": 1.7331772370966099,
      "grad_norm": 3.6998531818389893,
      "learning_rate": 1.4522091260774156e-05,
      "loss": 0.1017,
      "step": 17025
    },
    {
      "epoch": 1.7357222844344906,
      "grad_norm": 5.593207359313965,
      "learning_rate": 1.4510779811325024e-05,
      "loss": 0.1295,
      "step": 17050
    },
    {
      "epoch": 1.738267331772371,
      "grad_norm": 6.101248264312744,
      "learning_rate": 1.4499468361875892e-05,
      "loss": 0.1057,
      "step": 17075
    },
    {
      "epoch": 1.7408123791102514,
      "grad_norm": 2.3513197898864746,
      "learning_rate": 1.4488156912426758e-05,
      "loss": 0.1048,
      "step": 17100
    },
    {
      "epoch": 1.743357426448132,
      "grad_norm": 5.067135810852051,
      "learning_rate": 1.4476845462977626e-05,
      "loss": 0.1083,
      "step": 17125
    },
    {
      "epoch": 1.7459024737860123,
      "grad_norm": 7.471678256988525,
      "learning_rate": 1.4465534013528495e-05,
      "loss": 0.1393,
      "step": 17150
    },
    {
      "epoch": 1.748447521123893,
      "grad_norm": 6.90990686416626,
      "learning_rate": 1.4454222564079363e-05,
      "loss": 0.1097,
      "step": 17175
    },
    {
      "epoch": 1.7509925684617733,
      "grad_norm": 6.522189617156982,
      "learning_rate": 1.444291111463023e-05,
      "loss": 0.1335,
      "step": 17200
    },
    {
      "epoch": 1.7535376157996538,
      "grad_norm": 3.7323899269104004,
      "learning_rate": 1.4431599665181097e-05,
      "loss": 0.1182,
      "step": 17225
    },
    {
      "epoch": 1.7560826631375344,
      "grad_norm": 5.302699089050293,
      "learning_rate": 1.4420288215731965e-05,
      "loss": 0.1103,
      "step": 17250
    },
    {
      "epoch": 1.7586277104754149,
      "grad_norm": 0.8149896860122681,
      "learning_rate": 1.4408976766282831e-05,
      "loss": 0.1399,
      "step": 17275
    },
    {
      "epoch": 1.7611727578132954,
      "grad_norm": 5.488561153411865,
      "learning_rate": 1.43976653168337e-05,
      "loss": 0.1057,
      "step": 17300
    },
    {
      "epoch": 1.7637178051511757,
      "grad_norm": 6.034390926361084,
      "learning_rate": 1.4386353867384569e-05,
      "loss": 0.1112,
      "step": 17325
    },
    {
      "epoch": 1.7662628524890565,
      "grad_norm": 4.187770366668701,
      "learning_rate": 1.4375042417935437e-05,
      "loss": 0.1282,
      "step": 17350
    },
    {
      "epoch": 1.7688078998269368,
      "grad_norm": 5.549367427825928,
      "learning_rate": 1.4363730968486303e-05,
      "loss": 0.1056,
      "step": 17375
    },
    {
      "epoch": 1.7713529471648173,
      "grad_norm": 1.8145831823349,
      "learning_rate": 1.435241951903717e-05,
      "loss": 0.0953,
      "step": 17400
    },
    {
      "epoch": 1.7738979945026978,
      "grad_norm": 2.8839643001556396,
      "learning_rate": 1.4341108069588038e-05,
      "loss": 0.1086,
      "step": 17425
    },
    {
      "epoch": 1.776443041840578,
      "grad_norm": 0.7977824807167053,
      "learning_rate": 1.4329796620138905e-05,
      "loss": 0.1209,
      "step": 17450
    },
    {
      "epoch": 1.7789880891784589,
      "grad_norm": 8.126140594482422,
      "learning_rate": 1.4318485170689772e-05,
      "loss": 0.1062,
      "step": 17475
    },
    {
      "epoch": 1.7815331365163392,
      "grad_norm": 4.229276657104492,
      "learning_rate": 1.4307173721240642e-05,
      "loss": 0.1279,
      "step": 17500
    },
    {
      "epoch": 1.7840781838542197,
      "grad_norm": 4.933651924133301,
      "learning_rate": 1.4295862271791508e-05,
      "loss": 0.1199,
      "step": 17525
    },
    {
      "epoch": 1.7866232311921002,
      "grad_norm": 4.240781784057617,
      "learning_rate": 1.4284550822342376e-05,
      "loss": 0.1176,
      "step": 17550
    },
    {
      "epoch": 1.7891682785299805,
      "grad_norm": 2.898059368133545,
      "learning_rate": 1.4273239372893244e-05,
      "loss": 0.0957,
      "step": 17575
    },
    {
      "epoch": 1.7917133258678613,
      "grad_norm": 3.8518853187561035,
      "learning_rate": 1.4261927923444112e-05,
      "loss": 0.0978,
      "step": 17600
    },
    {
      "epoch": 1.7942583732057416,
      "grad_norm": 3.217885732650757,
      "learning_rate": 1.4250616473994978e-05,
      "loss": 0.1244,
      "step": 17625
    },
    {
      "epoch": 1.796803420543622,
      "grad_norm": 2.9937503337860107,
      "learning_rate": 1.4239305024545846e-05,
      "loss": 0.1163,
      "step": 17650
    },
    {
      "epoch": 1.7993484678815026,
      "grad_norm": 3.574305534362793,
      "learning_rate": 1.4227993575096715e-05,
      "loss": 0.1512,
      "step": 17675
    },
    {
      "epoch": 1.8018935152193831,
      "grad_norm": 10.027637481689453,
      "learning_rate": 1.4216682125647581e-05,
      "loss": 0.1373,
      "step": 17700
    },
    {
      "epoch": 1.8044385625572636,
      "grad_norm": 1.1961842775344849,
      "learning_rate": 1.4205370676198449e-05,
      "loss": 0.1315,
      "step": 17725
    },
    {
      "epoch": 1.806983609895144,
      "grad_norm": 3.1464712619781494,
      "learning_rate": 1.4194059226749317e-05,
      "loss": 0.1205,
      "step": 17750
    },
    {
      "epoch": 1.8095286572330247,
      "grad_norm": 5.239190578460693,
      "learning_rate": 1.4182747777300185e-05,
      "loss": 0.1095,
      "step": 17775
    },
    {
      "epoch": 1.812073704570905,
      "grad_norm": 6.392675399780273,
      "learning_rate": 1.4171436327851051e-05,
      "loss": 0.1442,
      "step": 17800
    },
    {
      "epoch": 1.8146187519087855,
      "grad_norm": 5.46378755569458,
      "learning_rate": 1.4160124878401919e-05,
      "loss": 0.1126,
      "step": 17825
    },
    {
      "epoch": 1.817163799246666,
      "grad_norm": 3.494110107421875,
      "learning_rate": 1.4148813428952788e-05,
      "loss": 0.1166,
      "step": 17850
    },
    {
      "epoch": 1.8197088465845463,
      "grad_norm": 4.842367649078369,
      "learning_rate": 1.4137501979503655e-05,
      "loss": 0.1038,
      "step": 17875
    },
    {
      "epoch": 1.822253893922427,
      "grad_norm": 3.657884359359741,
      "learning_rate": 1.4126190530054522e-05,
      "loss": 0.1669,
      "step": 17900
    },
    {
      "epoch": 1.8247989412603074,
      "grad_norm": 3.146895408630371,
      "learning_rate": 1.411487908060539e-05,
      "loss": 0.1639,
      "step": 17925
    },
    {
      "epoch": 1.827343988598188,
      "grad_norm": 4.50389289855957,
      "learning_rate": 1.4103567631156258e-05,
      "loss": 0.1279,
      "step": 17950
    },
    {
      "epoch": 1.8298890359360684,
      "grad_norm": 7.898961067199707,
      "learning_rate": 1.4092256181707124e-05,
      "loss": 0.1571,
      "step": 17975
    },
    {
      "epoch": 1.8324340832739487,
      "grad_norm": 0.7812437415122986,
      "learning_rate": 1.4080944732257992e-05,
      "loss": 0.1268,
      "step": 18000
    },
    {
      "epoch": 1.8349791306118295,
      "grad_norm": 5.2326579093933105,
      "learning_rate": 1.4069633282808862e-05,
      "loss": 0.1148,
      "step": 18025
    },
    {
      "epoch": 1.8375241779497098,
      "grad_norm": 3.574772357940674,
      "learning_rate": 1.4058321833359728e-05,
      "loss": 0.1484,
      "step": 18050
    },
    {
      "epoch": 1.8400692252875903,
      "grad_norm": 5.390368461608887,
      "learning_rate": 1.4047010383910596e-05,
      "loss": 0.1509,
      "step": 18075
    },
    {
      "epoch": 1.8426142726254708,
      "grad_norm": 3.923863410949707,
      "learning_rate": 1.4035698934461463e-05,
      "loss": 0.1158,
      "step": 18100
    },
    {
      "epoch": 1.8451593199633514,
      "grad_norm": 1.9413028955459595,
      "learning_rate": 1.402438748501233e-05,
      "loss": 0.1092,
      "step": 18125
    },
    {
      "epoch": 1.8477043673012319,
      "grad_norm": 4.675200462341309,
      "learning_rate": 1.4013528493541164e-05,
      "loss": 0.0942,
      "step": 18150
    },
    {
      "epoch": 1.8502494146391122,
      "grad_norm": 4.470402717590332,
      "learning_rate": 1.400221704409203e-05,
      "loss": 0.1146,
      "step": 18175
    },
    {
      "epoch": 1.852794461976993,
      "grad_norm": 3.648837089538574,
      "learning_rate": 1.3990905594642898e-05,
      "loss": 0.1386,
      "step": 18200
    },
    {
      "epoch": 1.8553395093148732,
      "grad_norm": 3.77079701423645,
      "learning_rate": 1.3979594145193767e-05,
      "loss": 0.1316,
      "step": 18225
    },
    {
      "epoch": 1.8578845566527538,
      "grad_norm": 4.982840538024902,
      "learning_rate": 1.3968282695744635e-05,
      "loss": 0.13,
      "step": 18250
    },
    {
      "epoch": 1.8604296039906343,
      "grad_norm": 8.14816665649414,
      "learning_rate": 1.3956971246295501e-05,
      "loss": 0.0977,
      "step": 18275
    },
    {
      "epoch": 1.8629746513285146,
      "grad_norm": 2.803504467010498,
      "learning_rate": 1.394565979684637e-05,
      "loss": 0.1599,
      "step": 18300
    },
    {
      "epoch": 1.8655196986663953,
      "grad_norm": 3.785940408706665,
      "learning_rate": 1.3934348347397237e-05,
      "loss": 0.094,
      "step": 18325
    },
    {
      "epoch": 1.8680647460042756,
      "grad_norm": 0.8537298440933228,
      "learning_rate": 1.3923036897948103e-05,
      "loss": 0.0993,
      "step": 18350
    },
    {
      "epoch": 1.8706097933421562,
      "grad_norm": 3.6978354454040527,
      "learning_rate": 1.3911725448498971e-05,
      "loss": 0.1357,
      "step": 18375
    },
    {
      "epoch": 1.8731548406800367,
      "grad_norm": 2.4232470989227295,
      "learning_rate": 1.390041399904984e-05,
      "loss": 0.109,
      "step": 18400
    },
    {
      "epoch": 1.875699888017917,
      "grad_norm": 6.006959915161133,
      "learning_rate": 1.3889102549600705e-05,
      "loss": 0.1133,
      "step": 18425
    },
    {
      "epoch": 1.8782449353557977,
      "grad_norm": 3.3002946376800537,
      "learning_rate": 1.3877791100151575e-05,
      "loss": 0.1429,
      "step": 18450
    },
    {
      "epoch": 1.880789982693678,
      "grad_norm": 3.789641857147217,
      "learning_rate": 1.3866479650702443e-05,
      "loss": 0.1547,
      "step": 18475
    },
    {
      "epoch": 1.8833350300315586,
      "grad_norm": 5.670425891876221,
      "learning_rate": 1.385516820125331e-05,
      "loss": 0.1347,
      "step": 18500
    },
    {
      "epoch": 1.885880077369439,
      "grad_norm": 1.6732641458511353,
      "learning_rate": 1.3843856751804177e-05,
      "loss": 0.0954,
      "step": 18525
    },
    {
      "epoch": 1.8884251247073196,
      "grad_norm": 0.9046832919120789,
      "learning_rate": 1.3832545302355044e-05,
      "loss": 0.1444,
      "step": 18550
    },
    {
      "epoch": 1.8909701720452001,
      "grad_norm": 6.483386516571045,
      "learning_rate": 1.3821233852905914e-05,
      "loss": 0.1367,
      "step": 18575
    },
    {
      "epoch": 1.8935152193830804,
      "grad_norm": 4.16316556930542,
      "learning_rate": 1.3809922403456778e-05,
      "loss": 0.1292,
      "step": 18600
    },
    {
      "epoch": 1.8960602667209612,
      "grad_norm": 8.144868850708008,
      "learning_rate": 1.3798610954007648e-05,
      "loss": 0.0942,
      "step": 18625
    },
    {
      "epoch": 1.8986053140588415,
      "grad_norm": 5.008410453796387,
      "learning_rate": 1.3787299504558516e-05,
      "loss": 0.1263,
      "step": 18650
    },
    {
      "epoch": 1.901150361396722,
      "grad_norm": 5.292370796203613,
      "learning_rate": 1.3775988055109384e-05,
      "loss": 0.1237,
      "step": 18675
    },
    {
      "epoch": 1.9036954087346025,
      "grad_norm": 2.5285484790802,
      "learning_rate": 1.376467660566025e-05,
      "loss": 0.1126,
      "step": 18700
    },
    {
      "epoch": 1.9062404560724828,
      "grad_norm": 2.1306982040405273,
      "learning_rate": 1.3753365156211118e-05,
      "loss": 0.112,
      "step": 18725
    },
    {
      "epoch": 1.9087855034103636,
      "grad_norm": 0.5022045969963074,
      "learning_rate": 1.3742053706761987e-05,
      "loss": 0.1209,
      "step": 18750
    },
    {
      "epoch": 1.9113305507482439,
      "grad_norm": 3.427983045578003,
      "learning_rate": 1.3730742257312852e-05,
      "loss": 0.103,
      "step": 18775
    },
    {
      "epoch": 1.9138755980861244,
      "grad_norm": 3.986631393432617,
      "learning_rate": 1.3719430807863721e-05,
      "loss": 0.0819,
      "step": 18800
    },
    {
      "epoch": 1.916420645424005,
      "grad_norm": 3.5819921493530273,
      "learning_rate": 1.3708119358414589e-05,
      "loss": 0.1688,
      "step": 18825
    },
    {
      "epoch": 1.9189656927618852,
      "grad_norm": 9.21711540222168,
      "learning_rate": 1.3696807908965455e-05,
      "loss": 0.1095,
      "step": 18850
    },
    {
      "epoch": 1.921510740099766,
      "grad_norm": 4.607287883758545,
      "learning_rate": 1.3685496459516323e-05,
      "loss": 0.1251,
      "step": 18875
    },
    {
      "epoch": 1.9240557874376463,
      "grad_norm": 3.5619211196899414,
      "learning_rate": 1.367418501006719e-05,
      "loss": 0.1163,
      "step": 18900
    },
    {
      "epoch": 1.9266008347755268,
      "grad_norm": 4.004523754119873,
      "learning_rate": 1.366287356061806e-05,
      "loss": 0.1449,
      "step": 18925
    },
    {
      "epoch": 1.9291458821134073,
      "grad_norm": 6.862831115722656,
      "learning_rate": 1.3651562111168925e-05,
      "loss": 0.1331,
      "step": 18950
    },
    {
      "epoch": 1.9316909294512878,
      "grad_norm": 6.602372646331787,
      "learning_rate": 1.3640250661719794e-05,
      "loss": 0.1345,
      "step": 18975
    },
    {
      "epoch": 1.9342359767891684,
      "grad_norm": 4.988379955291748,
      "learning_rate": 1.3628939212270662e-05,
      "loss": 0.1184,
      "step": 19000
    },
    {
      "epoch": 1.9367810241270487,
      "grad_norm": 7.112786293029785,
      "learning_rate": 1.3617627762821528e-05,
      "loss": 0.1287,
      "step": 19025
    },
    {
      "epoch": 1.9393260714649294,
      "grad_norm": 3.288667917251587,
      "learning_rate": 1.3606316313372396e-05,
      "loss": 0.1298,
      "step": 19050
    },
    {
      "epoch": 1.9418711188028097,
      "grad_norm": 2.2869608402252197,
      "learning_rate": 1.3595004863923264e-05,
      "loss": 0.0996,
      "step": 19075
    },
    {
      "epoch": 1.9444161661406902,
      "grad_norm": 4.9739532470703125,
      "learning_rate": 1.3583693414474134e-05,
      "loss": 0.125,
      "step": 19100
    },
    {
      "epoch": 1.9469612134785708,
      "grad_norm": 7.579037189483643,
      "learning_rate": 1.3572381965024998e-05,
      "loss": 0.1243,
      "step": 19125
    },
    {
      "epoch": 1.949506260816451,
      "grad_norm": 3.219343662261963,
      "learning_rate": 1.3561070515575868e-05,
      "loss": 0.1329,
      "step": 19150
    },
    {
      "epoch": 1.9520513081543318,
      "grad_norm": 3.9998912811279297,
      "learning_rate": 1.3549759066126735e-05,
      "loss": 0.1067,
      "step": 19175
    },
    {
      "epoch": 1.9545963554922121,
      "grad_norm": 1.2954130172729492,
      "learning_rate": 1.3538447616677602e-05,
      "loss": 0.1176,
      "step": 19200
    },
    {
      "epoch": 1.9571414028300926,
      "grad_norm": 3.8018393516540527,
      "learning_rate": 1.352713616722847e-05,
      "loss": 0.1318,
      "step": 19225
    },
    {
      "epoch": 1.9596864501679732,
      "grad_norm": 4.415460586547852,
      "learning_rate": 1.3515824717779337e-05,
      "loss": 0.1273,
      "step": 19250
    },
    {
      "epoch": 1.9622314975058535,
      "grad_norm": 0.9868888854980469,
      "learning_rate": 1.3504513268330207e-05,
      "loss": 0.1303,
      "step": 19275
    },
    {
      "epoch": 1.9647765448437342,
      "grad_norm": 11.183004379272461,
      "learning_rate": 1.3493201818881071e-05,
      "loss": 0.1699,
      "step": 19300
    },
    {
      "epoch": 1.9673215921816145,
      "grad_norm": 8.36978816986084,
      "learning_rate": 1.348189036943194e-05,
      "loss": 0.1163,
      "step": 19325
    },
    {
      "epoch": 1.969866639519495,
      "grad_norm": 2.3566489219665527,
      "learning_rate": 1.3470578919982809e-05,
      "loss": 0.1186,
      "step": 19350
    },
    {
      "epoch": 1.9724116868573756,
      "grad_norm": 4.07636833190918,
      "learning_rate": 1.3459267470533675e-05,
      "loss": 0.1196,
      "step": 19375
    },
    {
      "epoch": 1.974956734195256,
      "grad_norm": 1.6372345685958862,
      "learning_rate": 1.3447956021084543e-05,
      "loss": 0.1028,
      "step": 19400
    },
    {
      "epoch": 1.9775017815331366,
      "grad_norm": 1.4987009763717651,
      "learning_rate": 1.343664457163541e-05,
      "loss": 0.089,
      "step": 19425
    },
    {
      "epoch": 1.980046828871017,
      "grad_norm": 2.3132851123809814,
      "learning_rate": 1.3425333122186277e-05,
      "loss": 0.1074,
      "step": 19450
    },
    {
      "epoch": 1.9825918762088974,
      "grad_norm": 1.2867132425308228,
      "learning_rate": 1.3414021672737144e-05,
      "loss": 0.1352,
      "step": 19475
    },
    {
      "epoch": 1.985136923546778,
      "grad_norm": 2.911659002304077,
      "learning_rate": 1.3402710223288014e-05,
      "loss": 0.1128,
      "step": 19500
    },
    {
      "epoch": 1.9876819708846585,
      "grad_norm": 3.1398777961730957,
      "learning_rate": 1.3391398773838882e-05,
      "loss": 0.1311,
      "step": 19525
    },
    {
      "epoch": 1.990227018222539,
      "grad_norm": 3.4940083026885986,
      "learning_rate": 1.3380087324389748e-05,
      "loss": 0.0797,
      "step": 19550
    },
    {
      "epoch": 1.9927720655604193,
      "grad_norm": 5.973903656005859,
      "learning_rate": 1.3368775874940616e-05,
      "loss": 0.0901,
      "step": 19575
    },
    {
      "epoch": 1.9953171128983,
      "grad_norm": 6.650625705718994,
      "learning_rate": 1.3357464425491484e-05,
      "loss": 0.1026,
      "step": 19600
    },
    {
      "epoch": 1.9978621602361804,
      "grad_norm": 2.1521973609924316,
      "learning_rate": 1.334615297604235e-05,
      "loss": 0.1444,
      "step": 19625
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.09097875654697418,
      "eval_runtime": 7.4187,
      "eval_samples_per_second": 917.686,
      "eval_steps_per_second": 14.423,
      "step": 19646
    },
    {
      "epoch": 2.000407207574061,
      "grad_norm": 5.6037187576293945,
      "learning_rate": 1.3334841526593218e-05,
      "loss": 0.0984,
      "step": 19650
    },
    {
      "epoch": 2.0029522549119414,
      "grad_norm": 1.090952754020691,
      "learning_rate": 1.3323530077144087e-05,
      "loss": 0.0879,
      "step": 19675
    },
    {
      "epoch": 2.0054973022498217,
      "grad_norm": 7.369726181030273,
      "learning_rate": 1.3312218627694955e-05,
      "loss": 0.1064,
      "step": 19700
    },
    {
      "epoch": 2.0080423495877024,
      "grad_norm": 0.6371269822120667,
      "learning_rate": 1.3300907178245821e-05,
      "loss": 0.1009,
      "step": 19725
    },
    {
      "epoch": 2.0105873969255827,
      "grad_norm": 1.3027335405349731,
      "learning_rate": 1.3289595728796689e-05,
      "loss": 0.1037,
      "step": 19750
    },
    {
      "epoch": 2.0131324442634635,
      "grad_norm": 1.485375165939331,
      "learning_rate": 1.3278284279347557e-05,
      "loss": 0.0781,
      "step": 19775
    },
    {
      "epoch": 2.015677491601344,
      "grad_norm": 3.413872718811035,
      "learning_rate": 1.3266972829898423e-05,
      "loss": 0.105,
      "step": 19800
    },
    {
      "epoch": 2.018222538939224,
      "grad_norm": 3.911144971847534,
      "learning_rate": 1.3255661380449291e-05,
      "loss": 0.1214,
      "step": 19825
    },
    {
      "epoch": 2.020767586277105,
      "grad_norm": 9.321831703186035,
      "learning_rate": 1.324434993100016e-05,
      "loss": 0.0892,
      "step": 19850
    },
    {
      "epoch": 2.023312633614985,
      "grad_norm": 5.0478081703186035,
      "learning_rate": 1.3233038481551027e-05,
      "loss": 0.0776,
      "step": 19875
    },
    {
      "epoch": 2.025857680952866,
      "grad_norm": 4.059922695159912,
      "learning_rate": 1.3221727032101894e-05,
      "loss": 0.1019,
      "step": 19900
    },
    {
      "epoch": 2.028402728290746,
      "grad_norm": 2.502460241317749,
      "learning_rate": 1.3210415582652762e-05,
      "loss": 0.1305,
      "step": 19925
    },
    {
      "epoch": 2.0309477756286265,
      "grad_norm": 8.318452835083008,
      "learning_rate": 1.319910413320363e-05,
      "loss": 0.0912,
      "step": 19950
    },
    {
      "epoch": 2.0334928229665072,
      "grad_norm": 0.444395512342453,
      "learning_rate": 1.3188245141732463e-05,
      "loss": 0.0858,
      "step": 19975
    },
    {
      "epoch": 2.0360378703043875,
      "grad_norm": 5.8228349685668945,
      "learning_rate": 1.317693369228333e-05,
      "loss": 0.1094,
      "step": 20000
    },
    {
      "epoch": 2.0385829176422683,
      "grad_norm": 6.549695014953613,
      "learning_rate": 1.3165622242834197e-05,
      "loss": 0.0803,
      "step": 20025
    },
    {
      "epoch": 2.0411279649801486,
      "grad_norm": 4.000819683074951,
      "learning_rate": 1.3154310793385066e-05,
      "loss": 0.0938,
      "step": 20050
    },
    {
      "epoch": 2.0436730123180293,
      "grad_norm": 1.8325241804122925,
      "learning_rate": 1.3142999343935934e-05,
      "loss": 0.0798,
      "step": 20075
    },
    {
      "epoch": 2.0462180596559096,
      "grad_norm": 7.206612586975098,
      "learning_rate": 1.31316878944868e-05,
      "loss": 0.0883,
      "step": 20100
    },
    {
      "epoch": 2.04876310699379,
      "grad_norm": 3.23553729057312,
      "learning_rate": 1.3120376445037668e-05,
      "loss": 0.114,
      "step": 20125
    },
    {
      "epoch": 2.0513081543316707,
      "grad_norm": 0.41115373373031616,
      "learning_rate": 1.3109064995588536e-05,
      "loss": 0.1187,
      "step": 20150
    },
    {
      "epoch": 2.053853201669551,
      "grad_norm": 1.5195305347442627,
      "learning_rate": 1.3097753546139402e-05,
      "loss": 0.1014,
      "step": 20175
    },
    {
      "epoch": 2.0563982490074317,
      "grad_norm": 4.764736652374268,
      "learning_rate": 1.308644209669027e-05,
      "loss": 0.1099,
      "step": 20200
    },
    {
      "epoch": 2.058943296345312,
      "grad_norm": 0.8784139156341553,
      "learning_rate": 1.307513064724114e-05,
      "loss": 0.1045,
      "step": 20225
    },
    {
      "epoch": 2.0614883436831923,
      "grad_norm": 0.10267910361289978,
      "learning_rate": 1.3063819197792007e-05,
      "loss": 0.1031,
      "step": 20250
    },
    {
      "epoch": 2.064033391021073,
      "grad_norm": 0.8482717871665955,
      "learning_rate": 1.3052507748342874e-05,
      "loss": 0.0902,
      "step": 20275
    },
    {
      "epoch": 2.0665784383589534,
      "grad_norm": 4.93229341506958,
      "learning_rate": 1.3041196298893741e-05,
      "loss": 0.0859,
      "step": 20300
    },
    {
      "epoch": 2.069123485696834,
      "grad_norm": 0.9881433248519897,
      "learning_rate": 1.302988484944461e-05,
      "loss": 0.0931,
      "step": 20325
    },
    {
      "epoch": 2.0716685330347144,
      "grad_norm": 0.9810713529586792,
      "learning_rate": 1.3018573399995475e-05,
      "loss": 0.0846,
      "step": 20350
    },
    {
      "epoch": 2.0742135803725947,
      "grad_norm": 7.200399398803711,
      "learning_rate": 1.3007261950546343e-05,
      "loss": 0.0842,
      "step": 20375
    },
    {
      "epoch": 2.0767586277104755,
      "grad_norm": 6.763378143310547,
      "learning_rate": 1.2995950501097213e-05,
      "loss": 0.1141,
      "step": 20400
    },
    {
      "epoch": 2.079303675048356,
      "grad_norm": 1.3273606300354004,
      "learning_rate": 1.298463905164808e-05,
      "loss": 0.0947,
      "step": 20425
    },
    {
      "epoch": 2.0818487223862365,
      "grad_norm": 1.6942657232284546,
      "learning_rate": 1.2973327602198947e-05,
      "loss": 0.1171,
      "step": 20450
    },
    {
      "epoch": 2.084393769724117,
      "grad_norm": 3.745817184448242,
      "learning_rate": 1.2962016152749815e-05,
      "loss": 0.0943,
      "step": 20475
    },
    {
      "epoch": 2.0869388170619976,
      "grad_norm": 2.1933345794677734,
      "learning_rate": 1.2950704703300682e-05,
      "loss": 0.0748,
      "step": 20500
    },
    {
      "epoch": 2.089483864399878,
      "grad_norm": 0.5361920595169067,
      "learning_rate": 1.2939393253851549e-05,
      "loss": 0.1071,
      "step": 20525
    },
    {
      "epoch": 2.092028911737758,
      "grad_norm": 2.4719314575195312,
      "learning_rate": 1.2928081804402416e-05,
      "loss": 0.0777,
      "step": 20550
    },
    {
      "epoch": 2.094573959075639,
      "grad_norm": 4.1660990715026855,
      "learning_rate": 1.2916770354953286e-05,
      "loss": 0.0773,
      "step": 20575
    },
    {
      "epoch": 2.0971190064135192,
      "grad_norm": 4.774954319000244,
      "learning_rate": 1.290545890550415e-05,
      "loss": 0.1215,
      "step": 20600
    },
    {
      "epoch": 2.0996640537514,
      "grad_norm": 6.706242561340332,
      "learning_rate": 1.289414745605502e-05,
      "loss": 0.1075,
      "step": 20625
    },
    {
      "epoch": 2.1022091010892803,
      "grad_norm": 4.085606575012207,
      "learning_rate": 1.2882836006605888e-05,
      "loss": 0.1012,
      "step": 20650
    },
    {
      "epoch": 2.1047541484271606,
      "grad_norm": 5.016777515411377,
      "learning_rate": 1.2871524557156756e-05,
      "loss": 0.1105,
      "step": 20675
    },
    {
      "epoch": 2.1072991957650413,
      "grad_norm": 6.600816249847412,
      "learning_rate": 1.2860213107707622e-05,
      "loss": 0.1018,
      "step": 20700
    },
    {
      "epoch": 2.1098442431029216,
      "grad_norm": 2.2283174991607666,
      "learning_rate": 1.284890165825849e-05,
      "loss": 0.0926,
      "step": 20725
    },
    {
      "epoch": 2.1123892904408024,
      "grad_norm": 5.262929916381836,
      "learning_rate": 1.283759020880936e-05,
      "loss": 0.1175,
      "step": 20750
    },
    {
      "epoch": 2.1149343377786827,
      "grad_norm": 1.850588083267212,
      "learning_rate": 1.2826278759360224e-05,
      "loss": 0.0949,
      "step": 20775
    },
    {
      "epoch": 2.117479385116563,
      "grad_norm": 3.330935478210449,
      "learning_rate": 1.2814967309911093e-05,
      "loss": 0.1297,
      "step": 20800
    },
    {
      "epoch": 2.1200244324544437,
      "grad_norm": 2.4614529609680176,
      "learning_rate": 1.2803655860461961e-05,
      "loss": 0.0803,
      "step": 20825
    },
    {
      "epoch": 2.122569479792324,
      "grad_norm": 4.3904337882995605,
      "learning_rate": 1.2792344411012829e-05,
      "loss": 0.0997,
      "step": 20850
    },
    {
      "epoch": 2.1251145271302048,
      "grad_norm": 1.6169228553771973,
      "learning_rate": 1.2781032961563695e-05,
      "loss": 0.1115,
      "step": 20875
    },
    {
      "epoch": 2.127659574468085,
      "grad_norm": 3.5835373401641846,
      "learning_rate": 1.2769721512114563e-05,
      "loss": 0.084,
      "step": 20900
    },
    {
      "epoch": 2.130204621805966,
      "grad_norm": 8.013703346252441,
      "learning_rate": 1.2758410062665432e-05,
      "loss": 0.0928,
      "step": 20925
    },
    {
      "epoch": 2.132749669143846,
      "grad_norm": 6.659077167510986,
      "learning_rate": 1.2747098613216297e-05,
      "loss": 0.1014,
      "step": 20950
    },
    {
      "epoch": 2.1352947164817264,
      "grad_norm": 2.2046802043914795,
      "learning_rate": 1.2735787163767166e-05,
      "loss": 0.1198,
      "step": 20975
    },
    {
      "epoch": 2.137839763819607,
      "grad_norm": 3.954470157623291,
      "learning_rate": 1.2724475714318034e-05,
      "loss": 0.1347,
      "step": 21000
    },
    {
      "epoch": 2.1403848111574875,
      "grad_norm": 1.1410102844238281,
      "learning_rate": 1.2713164264868902e-05,
      "loss": 0.0831,
      "step": 21025
    },
    {
      "epoch": 2.142929858495368,
      "grad_norm": 5.025262355804443,
      "learning_rate": 1.2701852815419768e-05,
      "loss": 0.0995,
      "step": 21050
    },
    {
      "epoch": 2.1454749058332485,
      "grad_norm": 4.646005630493164,
      "learning_rate": 1.2690541365970636e-05,
      "loss": 0.1139,
      "step": 21075
    },
    {
      "epoch": 2.148019953171129,
      "grad_norm": 5.704071521759033,
      "learning_rate": 1.2679229916521506e-05,
      "loss": 0.0884,
      "step": 21100
    },
    {
      "epoch": 2.1505650005090096,
      "grad_norm": 2.248732566833496,
      "learning_rate": 1.266791846707237e-05,
      "loss": 0.0799,
      "step": 21125
    },
    {
      "epoch": 2.15311004784689,
      "grad_norm": 2.9786360263824463,
      "learning_rate": 1.265660701762324e-05,
      "loss": 0.0761,
      "step": 21150
    },
    {
      "epoch": 2.1556550951847706,
      "grad_norm": 4.9691948890686035,
      "learning_rate": 1.2645295568174108e-05,
      "loss": 0.0813,
      "step": 21175
    },
    {
      "epoch": 2.158200142522651,
      "grad_norm": 3.8157265186309814,
      "learning_rate": 1.2633984118724974e-05,
      "loss": 0.0996,
      "step": 21200
    },
    {
      "epoch": 2.160745189860531,
      "grad_norm": 3.571915864944458,
      "learning_rate": 1.2622672669275842e-05,
      "loss": 0.0908,
      "step": 21225
    },
    {
      "epoch": 2.163290237198412,
      "grad_norm": 3.316852331161499,
      "learning_rate": 1.261136121982671e-05,
      "loss": 0.1066,
      "step": 21250
    },
    {
      "epoch": 2.1658352845362923,
      "grad_norm": 2.9165329933166504,
      "learning_rate": 1.2600049770377579e-05,
      "loss": 0.1119,
      "step": 21275
    },
    {
      "epoch": 2.168380331874173,
      "grad_norm": 17.929237365722656,
      "learning_rate": 1.2588738320928443e-05,
      "loss": 0.1363,
      "step": 21300
    },
    {
      "epoch": 2.1709253792120533,
      "grad_norm": 10.439239501953125,
      "learning_rate": 1.2577426871479313e-05,
      "loss": 0.0573,
      "step": 21325
    },
    {
      "epoch": 2.173470426549934,
      "grad_norm": 4.958606243133545,
      "learning_rate": 1.256611542203018e-05,
      "loss": 0.1083,
      "step": 21350
    },
    {
      "epoch": 2.1760154738878144,
      "grad_norm": 0.9944233298301697,
      "learning_rate": 1.2554803972581047e-05,
      "loss": 0.0806,
      "step": 21375
    },
    {
      "epoch": 2.1785605212256947,
      "grad_norm": 0.8803888559341431,
      "learning_rate": 1.2543492523131915e-05,
      "loss": 0.1165,
      "step": 21400
    },
    {
      "epoch": 2.1811055685635754,
      "grad_norm": 1.1710690259933472,
      "learning_rate": 1.2532181073682783e-05,
      "loss": 0.0919,
      "step": 21425
    },
    {
      "epoch": 2.1836506159014557,
      "grad_norm": 4.190867900848389,
      "learning_rate": 1.2520869624233652e-05,
      "loss": 0.1119,
      "step": 21450
    },
    {
      "epoch": 2.1861956632393365,
      "grad_norm": 5.893057346343994,
      "learning_rate": 1.2509558174784517e-05,
      "loss": 0.097,
      "step": 21475
    },
    {
      "epoch": 2.1887407105772168,
      "grad_norm": 3.533395767211914,
      "learning_rate": 1.2498246725335386e-05,
      "loss": 0.086,
      "step": 21500
    },
    {
      "epoch": 2.191285757915097,
      "grad_norm": 2.7442116737365723,
      "learning_rate": 1.2486935275886254e-05,
      "loss": 0.1254,
      "step": 21525
    },
    {
      "epoch": 2.193830805252978,
      "grad_norm": 4.103741645812988,
      "learning_rate": 1.247562382643712e-05,
      "loss": 0.1215,
      "step": 21550
    },
    {
      "epoch": 2.196375852590858,
      "grad_norm": 2.0932254791259766,
      "learning_rate": 1.2464312376987988e-05,
      "loss": 0.1174,
      "step": 21575
    },
    {
      "epoch": 2.198920899928739,
      "grad_norm": 5.704500198364258,
      "learning_rate": 1.2453000927538856e-05,
      "loss": 0.1348,
      "step": 21600
    },
    {
      "epoch": 2.201465947266619,
      "grad_norm": 4.199222087860107,
      "learning_rate": 1.2441689478089722e-05,
      "loss": 0.11,
      "step": 21625
    },
    {
      "epoch": 2.2040109946044995,
      "grad_norm": 3.764247417449951,
      "learning_rate": 1.243037802864059e-05,
      "loss": 0.1224,
      "step": 21650
    },
    {
      "epoch": 2.20655604194238,
      "grad_norm": 1.0659103393554688,
      "learning_rate": 1.241906657919146e-05,
      "loss": 0.1164,
      "step": 21675
    },
    {
      "epoch": 2.2091010892802605,
      "grad_norm": 3.8435420989990234,
      "learning_rate": 1.2407755129742327e-05,
      "loss": 0.1127,
      "step": 21700
    },
    {
      "epoch": 2.2116461366181412,
      "grad_norm": 2.6357994079589844,
      "learning_rate": 1.2396443680293193e-05,
      "loss": 0.0713,
      "step": 21725
    },
    {
      "epoch": 2.2141911839560215,
      "grad_norm": 5.551704406738281,
      "learning_rate": 1.2385132230844061e-05,
      "loss": 0.0999,
      "step": 21750
    },
    {
      "epoch": 2.2167362312939023,
      "grad_norm": 3.5522947311401367,
      "learning_rate": 1.2373820781394929e-05,
      "loss": 0.0822,
      "step": 21775
    },
    {
      "epoch": 2.2192812786317826,
      "grad_norm": 6.4888739585876465,
      "learning_rate": 1.2362509331945795e-05,
      "loss": 0.1165,
      "step": 21800
    },
    {
      "epoch": 2.221826325969663,
      "grad_norm": 2.415229082107544,
      "learning_rate": 1.2351197882496663e-05,
      "loss": 0.0943,
      "step": 21825
    },
    {
      "epoch": 2.2243713733075436,
      "grad_norm": 2.813169479370117,
      "learning_rate": 1.2339886433047533e-05,
      "loss": 0.0936,
      "step": 21850
    },
    {
      "epoch": 2.226916420645424,
      "grad_norm": 5.718323707580566,
      "learning_rate": 1.23285749835984e-05,
      "loss": 0.1375,
      "step": 21875
    },
    {
      "epoch": 2.2294614679833047,
      "grad_norm": 6.90181827545166,
      "learning_rate": 1.2317263534149267e-05,
      "loss": 0.0972,
      "step": 21900
    },
    {
      "epoch": 2.232006515321185,
      "grad_norm": 1.21304452419281,
      "learning_rate": 1.2305952084700134e-05,
      "loss": 0.1095,
      "step": 21925
    },
    {
      "epoch": 2.2345515626590653,
      "grad_norm": 9.322980880737305,
      "learning_rate": 1.2294640635251002e-05,
      "loss": 0.0997,
      "step": 21950
    },
    {
      "epoch": 2.237096609996946,
      "grad_norm": 4.291175842285156,
      "learning_rate": 1.2283329185801868e-05,
      "loss": 0.0908,
      "step": 21975
    },
    {
      "epoch": 2.2396416573348263,
      "grad_norm": 3.542860746383667,
      "learning_rate": 1.2272017736352736e-05,
      "loss": 0.0871,
      "step": 22000
    },
    {
      "epoch": 2.242186704672707,
      "grad_norm": 6.1715312004089355,
      "learning_rate": 1.2260706286903606e-05,
      "loss": 0.1264,
      "step": 22025
    },
    {
      "epoch": 2.2447317520105874,
      "grad_norm": 2.7312450408935547,
      "learning_rate": 1.2249394837454472e-05,
      "loss": 0.112,
      "step": 22050
    },
    {
      "epoch": 2.2472767993484677,
      "grad_norm": 4.341641902923584,
      "learning_rate": 1.223808338800534e-05,
      "loss": 0.0914,
      "step": 22075
    },
    {
      "epoch": 2.2498218466863484,
      "grad_norm": 2.093118667602539,
      "learning_rate": 1.2226771938556208e-05,
      "loss": 0.0638,
      "step": 22100
    },
    {
      "epoch": 2.2523668940242287,
      "grad_norm": 2.3410089015960693,
      "learning_rate": 1.2215460489107075e-05,
      "loss": 0.0941,
      "step": 22125
    },
    {
      "epoch": 2.2549119413621095,
      "grad_norm": 3.056443452835083,
      "learning_rate": 1.2204149039657942e-05,
      "loss": 0.0983,
      "step": 22150
    },
    {
      "epoch": 2.25745698869999,
      "grad_norm": 1.6339013576507568,
      "learning_rate": 1.2192837590208811e-05,
      "loss": 0.1036,
      "step": 22175
    },
    {
      "epoch": 2.2600020360378705,
      "grad_norm": 8.467519760131836,
      "learning_rate": 1.2181526140759679e-05,
      "loss": 0.0898,
      "step": 22200
    },
    {
      "epoch": 2.262547083375751,
      "grad_norm": 2.0889439582824707,
      "learning_rate": 1.2170214691310545e-05,
      "loss": 0.0984,
      "step": 22225
    },
    {
      "epoch": 2.265092130713631,
      "grad_norm": 0.9072188138961792,
      "learning_rate": 1.215935569983938e-05,
      "loss": 0.1118,
      "step": 22250
    },
    {
      "epoch": 2.267637178051512,
      "grad_norm": 4.564201354980469,
      "learning_rate": 1.2148044250390246e-05,
      "loss": 0.0865,
      "step": 22275
    },
    {
      "epoch": 2.270182225389392,
      "grad_norm": 3.500279188156128,
      "learning_rate": 1.2136732800941113e-05,
      "loss": 0.1191,
      "step": 22300
    },
    {
      "epoch": 2.2727272727272725,
      "grad_norm": 4.1375579833984375,
      "learning_rate": 1.2125421351491981e-05,
      "loss": 0.0984,
      "step": 22325
    },
    {
      "epoch": 2.2752723200651532,
      "grad_norm": 3.598430871963501,
      "learning_rate": 1.211410990204285e-05,
      "loss": 0.0829,
      "step": 22350
    },
    {
      "epoch": 2.2778173674030335,
      "grad_norm": 6.143867015838623,
      "learning_rate": 1.2102798452593715e-05,
      "loss": 0.1363,
      "step": 22375
    },
    {
      "epoch": 2.2803624147409143,
      "grad_norm": 2.5650177001953125,
      "learning_rate": 1.2091487003144585e-05,
      "loss": 0.0876,
      "step": 22400
    },
    {
      "epoch": 2.2829074620787946,
      "grad_norm": 6.063427448272705,
      "learning_rate": 1.2080175553695453e-05,
      "loss": 0.0932,
      "step": 22425
    },
    {
      "epoch": 2.2854525094166753,
      "grad_norm": 4.583015441894531,
      "learning_rate": 1.2068864104246319e-05,
      "loss": 0.095,
      "step": 22450
    },
    {
      "epoch": 2.2879975567545556,
      "grad_norm": 6.4546332359313965,
      "learning_rate": 1.2057552654797187e-05,
      "loss": 0.0957,
      "step": 22475
    },
    {
      "epoch": 2.290542604092436,
      "grad_norm": 3.2949342727661133,
      "learning_rate": 1.2046241205348055e-05,
      "loss": 0.0999,
      "step": 22500
    },
    {
      "epoch": 2.2930876514303167,
      "grad_norm": 8.154585838317871,
      "learning_rate": 1.203492975589892e-05,
      "loss": 0.1071,
      "step": 22525
    },
    {
      "epoch": 2.295632698768197,
      "grad_norm": 3.5857834815979004,
      "learning_rate": 1.2023618306449789e-05,
      "loss": 0.1223,
      "step": 22550
    },
    {
      "epoch": 2.2981777461060777,
      "grad_norm": 1.3705384731292725,
      "learning_rate": 1.2012306857000658e-05,
      "loss": 0.0748,
      "step": 22575
    },
    {
      "epoch": 2.300722793443958,
      "grad_norm": 2.1247451305389404,
      "learning_rate": 1.2000995407551526e-05,
      "loss": 0.0866,
      "step": 22600
    },
    {
      "epoch": 2.3032678407818388,
      "grad_norm": 3.3343911170959473,
      "learning_rate": 1.1989683958102392e-05,
      "loss": 0.1079,
      "step": 22625
    },
    {
      "epoch": 2.305812888119719,
      "grad_norm": 4.785033702850342,
      "learning_rate": 1.197837250865326e-05,
      "loss": 0.0928,
      "step": 22650
    },
    {
      "epoch": 2.3083579354575994,
      "grad_norm": 0.6399935483932495,
      "learning_rate": 1.1967061059204128e-05,
      "loss": 0.1176,
      "step": 22675
    },
    {
      "epoch": 2.31090298279548,
      "grad_norm": 6.403359413146973,
      "learning_rate": 1.1955749609754994e-05,
      "loss": 0.1455,
      "step": 22700
    },
    {
      "epoch": 2.3134480301333604,
      "grad_norm": 1.8424525260925293,
      "learning_rate": 1.1944438160305862e-05,
      "loss": 0.0922,
      "step": 22725
    },
    {
      "epoch": 2.3159930774712407,
      "grad_norm": 4.637752056121826,
      "learning_rate": 1.1933126710856731e-05,
      "loss": 0.1412,
      "step": 22750
    },
    {
      "epoch": 2.3185381248091215,
      "grad_norm": 2.118675708770752,
      "learning_rate": 1.19218152614076e-05,
      "loss": 0.0885,
      "step": 22775
    },
    {
      "epoch": 2.3210831721470018,
      "grad_norm": 2.0802664756774902,
      "learning_rate": 1.1910503811958465e-05,
      "loss": 0.084,
      "step": 22800
    },
    {
      "epoch": 2.3236282194848825,
      "grad_norm": 3.8109188079833984,
      "learning_rate": 1.1899192362509333e-05,
      "loss": 0.1206,
      "step": 22825
    },
    {
      "epoch": 2.326173266822763,
      "grad_norm": 2.5389983654022217,
      "learning_rate": 1.1887880913060201e-05,
      "loss": 0.1152,
      "step": 22850
    },
    {
      "epoch": 2.3287183141606436,
      "grad_norm": 2.187694787979126,
      "learning_rate": 1.1876569463611067e-05,
      "loss": 0.0903,
      "step": 22875
    },
    {
      "epoch": 2.331263361498524,
      "grad_norm": 7.514569282531738,
      "learning_rate": 1.1865258014161935e-05,
      "loss": 0.1087,
      "step": 22900
    },
    {
      "epoch": 2.333808408836404,
      "grad_norm": 4.942701816558838,
      "learning_rate": 1.1853946564712805e-05,
      "loss": 0.0602,
      "step": 22925
    },
    {
      "epoch": 2.336353456174285,
      "grad_norm": 2.9110524654388428,
      "learning_rate": 1.1842635115263669e-05,
      "loss": 0.0941,
      "step": 22950
    },
    {
      "epoch": 2.338898503512165,
      "grad_norm": 2.718231678009033,
      "learning_rate": 1.1831323665814539e-05,
      "loss": 0.0776,
      "step": 22975
    },
    {
      "epoch": 2.341443550850046,
      "grad_norm": 2.2280826568603516,
      "learning_rate": 1.1820012216365406e-05,
      "loss": 0.0844,
      "step": 23000
    },
    {
      "epoch": 2.3439885981879263,
      "grad_norm": 2.983081340789795,
      "learning_rate": 1.1808700766916274e-05,
      "loss": 0.0959,
      "step": 23025
    },
    {
      "epoch": 2.346533645525807,
      "grad_norm": 1.6823519468307495,
      "learning_rate": 1.179738931746714e-05,
      "loss": 0.0882,
      "step": 23050
    },
    {
      "epoch": 2.3490786928636873,
      "grad_norm": 4.962369918823242,
      "learning_rate": 1.1786077868018008e-05,
      "loss": 0.0826,
      "step": 23075
    },
    {
      "epoch": 2.3516237402015676,
      "grad_norm": 4.208407402038574,
      "learning_rate": 1.1774766418568878e-05,
      "loss": 0.0792,
      "step": 23100
    },
    {
      "epoch": 2.3541687875394484,
      "grad_norm": 2.5069408416748047,
      "learning_rate": 1.1763454969119742e-05,
      "loss": 0.0889,
      "step": 23125
    },
    {
      "epoch": 2.3567138348773287,
      "grad_norm": 2.7248406410217285,
      "learning_rate": 1.1752143519670612e-05,
      "loss": 0.0914,
      "step": 23150
    },
    {
      "epoch": 2.359258882215209,
      "grad_norm": 6.141697883605957,
      "learning_rate": 1.174083207022148e-05,
      "loss": 0.0866,
      "step": 23175
    },
    {
      "epoch": 2.3618039295530897,
      "grad_norm": 7.505110740661621,
      "learning_rate": 1.1729520620772347e-05,
      "loss": 0.0787,
      "step": 23200
    },
    {
      "epoch": 2.36434897689097,
      "grad_norm": 4.82844877243042,
      "learning_rate": 1.1718209171323214e-05,
      "loss": 0.1134,
      "step": 23225
    },
    {
      "epoch": 2.3668940242288508,
      "grad_norm": 3.9105417728424072,
      "learning_rate": 1.1706897721874081e-05,
      "loss": 0.1057,
      "step": 23250
    },
    {
      "epoch": 2.369439071566731,
      "grad_norm": 2.952836513519287,
      "learning_rate": 1.1695586272424951e-05,
      "loss": 0.1038,
      "step": 23275
    },
    {
      "epoch": 2.371984118904612,
      "grad_norm": 2.2733404636383057,
      "learning_rate": 1.1684274822975815e-05,
      "loss": 0.107,
      "step": 23300
    },
    {
      "epoch": 2.374529166242492,
      "grad_norm": 4.268314838409424,
      "learning_rate": 1.1672963373526685e-05,
      "loss": 0.1015,
      "step": 23325
    },
    {
      "epoch": 2.3770742135803724,
      "grad_norm": 1.9675124883651733,
      "learning_rate": 1.1661651924077553e-05,
      "loss": 0.0873,
      "step": 23350
    },
    {
      "epoch": 2.379619260918253,
      "grad_norm": 2.5415396690368652,
      "learning_rate": 1.1650340474628419e-05,
      "loss": 0.087,
      "step": 23375
    },
    {
      "epoch": 2.3821643082561335,
      "grad_norm": 3.148134231567383,
      "learning_rate": 1.1639029025179287e-05,
      "loss": 0.1126,
      "step": 23400
    },
    {
      "epoch": 2.384709355594014,
      "grad_norm": 0.9491432309150696,
      "learning_rate": 1.1627717575730155e-05,
      "loss": 0.0685,
      "step": 23425
    },
    {
      "epoch": 2.3872544029318945,
      "grad_norm": 4.692232131958008,
      "learning_rate": 1.1616406126281024e-05,
      "loss": 0.1078,
      "step": 23450
    },
    {
      "epoch": 2.3897994502697752,
      "grad_norm": 1.1305272579193115,
      "learning_rate": 1.1605094676831889e-05,
      "loss": 0.0767,
      "step": 23475
    },
    {
      "epoch": 2.3923444976076556,
      "grad_norm": 2.8438282012939453,
      "learning_rate": 1.1593783227382758e-05,
      "loss": 0.0982,
      "step": 23500
    },
    {
      "epoch": 2.394889544945536,
      "grad_norm": 4.794705867767334,
      "learning_rate": 1.1582471777933626e-05,
      "loss": 0.0794,
      "step": 23525
    },
    {
      "epoch": 2.3974345922834166,
      "grad_norm": 6.11577033996582,
      "learning_rate": 1.1571160328484492e-05,
      "loss": 0.1003,
      "step": 23550
    },
    {
      "epoch": 2.399979639621297,
      "grad_norm": 7.616250514984131,
      "learning_rate": 1.155984887903536e-05,
      "loss": 0.1272,
      "step": 23575
    },
    {
      "epoch": 2.402524686959177,
      "grad_norm": 1.4506219625473022,
      "learning_rate": 1.1548537429586228e-05,
      "loss": 0.1003,
      "step": 23600
    },
    {
      "epoch": 2.405069734297058,
      "grad_norm": 8.559371948242188,
      "learning_rate": 1.1537225980137097e-05,
      "loss": 0.1174,
      "step": 23625
    },
    {
      "epoch": 2.4076147816349383,
      "grad_norm": 0.7198318839073181,
      "learning_rate": 1.1525914530687964e-05,
      "loss": 0.1227,
      "step": 23650
    },
    {
      "epoch": 2.410159828972819,
      "grad_norm": 1.5881752967834473,
      "learning_rate": 1.1514603081238831e-05,
      "loss": 0.1009,
      "step": 23675
    },
    {
      "epoch": 2.4127048763106993,
      "grad_norm": 6.224447727203369,
      "learning_rate": 1.15032916317897e-05,
      "loss": 0.0939,
      "step": 23700
    },
    {
      "epoch": 2.41524992364858,
      "grad_norm": 2.6777849197387695,
      "learning_rate": 1.1491980182340565e-05,
      "loss": 0.0768,
      "step": 23725
    },
    {
      "epoch": 2.4177949709864603,
      "grad_norm": 4.088348388671875,
      "learning_rate": 1.1480668732891433e-05,
      "loss": 0.1155,
      "step": 23750
    },
    {
      "epoch": 2.4203400183243406,
      "grad_norm": 0.4692411422729492,
      "learning_rate": 1.1469357283442301e-05,
      "loss": 0.1223,
      "step": 23775
    },
    {
      "epoch": 2.4228850656622214,
      "grad_norm": 6.204223155975342,
      "learning_rate": 1.1458045833993167e-05,
      "loss": 0.1063,
      "step": 23800
    },
    {
      "epoch": 2.4254301130001017,
      "grad_norm": 6.6200032234191895,
      "learning_rate": 1.1446734384544037e-05,
      "loss": 0.0932,
      "step": 23825
    },
    {
      "epoch": 2.4279751603379824,
      "grad_norm": 4.673367023468018,
      "learning_rate": 1.1435422935094905e-05,
      "loss": 0.0962,
      "step": 23850
    },
    {
      "epoch": 2.4305202076758627,
      "grad_norm": 0.12596215307712555,
      "learning_rate": 1.1424111485645772e-05,
      "loss": 0.0822,
      "step": 23875
    },
    {
      "epoch": 2.4330652550137435,
      "grad_norm": 4.890219688415527,
      "learning_rate": 1.1412800036196639e-05,
      "loss": 0.086,
      "step": 23900
    },
    {
      "epoch": 2.435610302351624,
      "grad_norm": 7.528822422027588,
      "learning_rate": 1.1401488586747506e-05,
      "loss": 0.0967,
      "step": 23925
    },
    {
      "epoch": 2.438155349689504,
      "grad_norm": 4.572819232940674,
      "learning_rate": 1.1390177137298374e-05,
      "loss": 0.0703,
      "step": 23950
    },
    {
      "epoch": 2.440700397027385,
      "grad_norm": 1.6426405906677246,
      "learning_rate": 1.137886568784924e-05,
      "loss": 0.085,
      "step": 23975
    },
    {
      "epoch": 2.443245444365265,
      "grad_norm": 4.192370414733887,
      "learning_rate": 1.136755423840011e-05,
      "loss": 0.1188,
      "step": 24000
    },
    {
      "epoch": 2.4457904917031454,
      "grad_norm": 4.600924968719482,
      "learning_rate": 1.1356242788950978e-05,
      "loss": 0.1026,
      "step": 24025
    },
    {
      "epoch": 2.448335539041026,
      "grad_norm": 5.712591171264648,
      "learning_rate": 1.1344931339501846e-05,
      "loss": 0.079,
      "step": 24050
    },
    {
      "epoch": 2.4508805863789065,
      "grad_norm": 3.8217313289642334,
      "learning_rate": 1.1333619890052712e-05,
      "loss": 0.1202,
      "step": 24075
    },
    {
      "epoch": 2.4534256337167872,
      "grad_norm": 5.072776794433594,
      "learning_rate": 1.132230844060358e-05,
      "loss": 0.1105,
      "step": 24100
    },
    {
      "epoch": 2.4559706810546675,
      "grad_norm": 0.7305828928947449,
      "learning_rate": 1.1310996991154448e-05,
      "loss": 0.0713,
      "step": 24125
    },
    {
      "epoch": 2.4585157283925483,
      "grad_norm": 1.228698968887329,
      "learning_rate": 1.1299685541705314e-05,
      "loss": 0.0883,
      "step": 24150
    },
    {
      "epoch": 2.4610607757304286,
      "grad_norm": 4.841889381408691,
      "learning_rate": 1.1288374092256183e-05,
      "loss": 0.0943,
      "step": 24175
    },
    {
      "epoch": 2.463605823068309,
      "grad_norm": 5.082825660705566,
      "learning_rate": 1.1277062642807051e-05,
      "loss": 0.1042,
      "step": 24200
    },
    {
      "epoch": 2.4661508704061896,
      "grad_norm": 5.81834077835083,
      "learning_rate": 1.1265751193357919e-05,
      "loss": 0.0933,
      "step": 24225
    },
    {
      "epoch": 2.46869591774407,
      "grad_norm": 1.649214744567871,
      "learning_rate": 1.1254439743908785e-05,
      "loss": 0.1272,
      "step": 24250
    },
    {
      "epoch": 2.4712409650819507,
      "grad_norm": 3.9719619750976562,
      "learning_rate": 1.1243128294459653e-05,
      "loss": 0.0857,
      "step": 24275
    },
    {
      "epoch": 2.473786012419831,
      "grad_norm": 3.341063976287842,
      "learning_rate": 1.123181684501052e-05,
      "loss": 0.089,
      "step": 24300
    },
    {
      "epoch": 2.4763310597577117,
      "grad_norm": 5.203547477722168,
      "learning_rate": 1.1220505395561387e-05,
      "loss": 0.1298,
      "step": 24325
    },
    {
      "epoch": 2.478876107095592,
      "grad_norm": 8.336240768432617,
      "learning_rate": 1.1209193946112256e-05,
      "loss": 0.1025,
      "step": 24350
    },
    {
      "epoch": 2.4814211544334723,
      "grad_norm": 5.071165561676025,
      "learning_rate": 1.1197882496663124e-05,
      "loss": 0.1062,
      "step": 24375
    },
    {
      "epoch": 2.483966201771353,
      "grad_norm": 2.8873372077941895,
      "learning_rate": 1.118657104721399e-05,
      "loss": 0.0863,
      "step": 24400
    },
    {
      "epoch": 2.4865112491092334,
      "grad_norm": 0.5643141269683838,
      "learning_rate": 1.1175259597764858e-05,
      "loss": 0.0778,
      "step": 24425
    },
    {
      "epoch": 2.4890562964471137,
      "grad_norm": 5.715143203735352,
      "learning_rate": 1.1163948148315726e-05,
      "loss": 0.1379,
      "step": 24450
    },
    {
      "epoch": 2.4916013437849944,
      "grad_norm": 5.828744411468506,
      "learning_rate": 1.1152636698866594e-05,
      "loss": 0.1094,
      "step": 24475
    },
    {
      "epoch": 2.4941463911228747,
      "grad_norm": 5.1948933601379395,
      "learning_rate": 1.114132524941746e-05,
      "loss": 0.1035,
      "step": 24500
    },
    {
      "epoch": 2.4966914384607555,
      "grad_norm": 1.6300240755081177,
      "learning_rate": 1.113001379996833e-05,
      "loss": 0.0861,
      "step": 24525
    },
    {
      "epoch": 2.4992364857986358,
      "grad_norm": 3.659991979598999,
      "learning_rate": 1.1118702350519198e-05,
      "loss": 0.092,
      "step": 24550
    },
    {
      "epoch": 2.5017815331365165,
      "grad_norm": 4.651791095733643,
      "learning_rate": 1.1107390901070064e-05,
      "loss": 0.0907,
      "step": 24575
    },
    {
      "epoch": 2.504326580474397,
      "grad_norm": 2.967970609664917,
      "learning_rate": 1.1096079451620932e-05,
      "loss": 0.0933,
      "step": 24600
    },
    {
      "epoch": 2.506871627812277,
      "grad_norm": 1.0345920324325562,
      "learning_rate": 1.10847680021718e-05,
      "loss": 0.0955,
      "step": 24625
    },
    {
      "epoch": 2.509416675150158,
      "grad_norm": 0.5405799150466919,
      "learning_rate": 1.1073456552722667e-05,
      "loss": 0.1027,
      "step": 24650
    },
    {
      "epoch": 2.511961722488038,
      "grad_norm": 4.543354511260986,
      "learning_rate": 1.1062145103273533e-05,
      "loss": 0.1106,
      "step": 24675
    },
    {
      "epoch": 2.514506769825919,
      "grad_norm": 5.559842109680176,
      "learning_rate": 1.1050833653824403e-05,
      "loss": 0.1031,
      "step": 24700
    },
    {
      "epoch": 2.517051817163799,
      "grad_norm": 0.921740710735321,
      "learning_rate": 1.103952220437527e-05,
      "loss": 0.0935,
      "step": 24725
    },
    {
      "epoch": 2.51959686450168,
      "grad_norm": 7.651474475860596,
      "learning_rate": 1.1028210754926137e-05,
      "loss": 0.0732,
      "step": 24750
    },
    {
      "epoch": 2.5221419118395603,
      "grad_norm": 4.730767250061035,
      "learning_rate": 1.1016899305477005e-05,
      "loss": 0.0983,
      "step": 24775
    },
    {
      "epoch": 2.5246869591774406,
      "grad_norm": 6.074618339538574,
      "learning_rate": 1.1005587856027873e-05,
      "loss": 0.0886,
      "step": 24800
    },
    {
      "epoch": 2.5272320065153213,
      "grad_norm": 3.169323682785034,
      "learning_rate": 1.0994276406578739e-05,
      "loss": 0.0815,
      "step": 24825
    },
    {
      "epoch": 2.5297770538532016,
      "grad_norm": 3.6971890926361084,
      "learning_rate": 1.0982964957129607e-05,
      "loss": 0.1011,
      "step": 24850
    },
    {
      "epoch": 2.532322101191082,
      "grad_norm": 2.3330135345458984,
      "learning_rate": 1.0971653507680476e-05,
      "loss": 0.0796,
      "step": 24875
    },
    {
      "epoch": 2.5348671485289627,
      "grad_norm": 1.2037569284439087,
      "learning_rate": 1.0960342058231344e-05,
      "loss": 0.0883,
      "step": 24900
    },
    {
      "epoch": 2.537412195866843,
      "grad_norm": 0.8009597659111023,
      "learning_rate": 1.094903060878221e-05,
      "loss": 0.0981,
      "step": 24925
    },
    {
      "epoch": 2.5399572432047237,
      "grad_norm": 3.767286539077759,
      "learning_rate": 1.0937719159333078e-05,
      "loss": 0.1247,
      "step": 24950
    },
    {
      "epoch": 2.542502290542604,
      "grad_norm": 4.843240737915039,
      "learning_rate": 1.0926407709883946e-05,
      "loss": 0.1035,
      "step": 24975
    },
    {
      "epoch": 2.5450473378804848,
      "grad_norm": 0.4151710271835327,
      "learning_rate": 1.0915096260434812e-05,
      "loss": 0.115,
      "step": 25000
    },
    {
      "epoch": 2.547592385218365,
      "grad_norm": 3.2080724239349365,
      "learning_rate": 1.090378481098568e-05,
      "loss": 0.0916,
      "step": 25025
    },
    {
      "epoch": 2.5501374325562454,
      "grad_norm": 5.026012897491455,
      "learning_rate": 1.089247336153655e-05,
      "loss": 0.1199,
      "step": 25050
    },
    {
      "epoch": 2.552682479894126,
      "grad_norm": 2.0379691123962402,
      "learning_rate": 1.0881161912087417e-05,
      "loss": 0.1045,
      "step": 25075
    },
    {
      "epoch": 2.5552275272320064,
      "grad_norm": 2.8039698600769043,
      "learning_rate": 1.0869850462638283e-05,
      "loss": 0.1002,
      "step": 25100
    },
    {
      "epoch": 2.557772574569887,
      "grad_norm": 0.7228884696960449,
      "learning_rate": 1.0858539013189151e-05,
      "loss": 0.1184,
      "step": 25125
    },
    {
      "epoch": 2.5603176219077675,
      "grad_norm": 3.297924518585205,
      "learning_rate": 1.0847227563740019e-05,
      "loss": 0.077,
      "step": 25150
    },
    {
      "epoch": 2.562862669245648,
      "grad_norm": 0.8766332268714905,
      "learning_rate": 1.0835916114290885e-05,
      "loss": 0.0948,
      "step": 25175
    },
    {
      "epoch": 2.5654077165835285,
      "grad_norm": 6.345101833343506,
      "learning_rate": 1.0824604664841753e-05,
      "loss": 0.0991,
      "step": 25200
    },
    {
      "epoch": 2.567952763921409,
      "grad_norm": 2.1553688049316406,
      "learning_rate": 1.0813293215392623e-05,
      "loss": 0.1005,
      "step": 25225
    },
    {
      "epoch": 2.5704978112592896,
      "grad_norm": 5.529004096984863,
      "learning_rate": 1.0801981765943487e-05,
      "loss": 0.1139,
      "step": 25250
    },
    {
      "epoch": 2.57304285859717,
      "grad_norm": 6.425251007080078,
      "learning_rate": 1.0790670316494357e-05,
      "loss": 0.0919,
      "step": 25275
    },
    {
      "epoch": 2.57558790593505,
      "grad_norm": 4.08704137802124,
      "learning_rate": 1.0779358867045224e-05,
      "loss": 0.1483,
      "step": 25300
    },
    {
      "epoch": 2.578132953272931,
      "grad_norm": 5.872612476348877,
      "learning_rate": 1.0768047417596092e-05,
      "loss": 0.0965,
      "step": 25325
    },
    {
      "epoch": 2.580678000610811,
      "grad_norm": 3.455596923828125,
      "learning_rate": 1.0756735968146958e-05,
      "loss": 0.0925,
      "step": 25350
    },
    {
      "epoch": 2.583223047948692,
      "grad_norm": 3.5879950523376465,
      "learning_rate": 1.0745424518697826e-05,
      "loss": 0.0897,
      "step": 25375
    },
    {
      "epoch": 2.5857680952865723,
      "grad_norm": 1.831623911857605,
      "learning_rate": 1.0734113069248696e-05,
      "loss": 0.1098,
      "step": 25400
    },
    {
      "epoch": 2.588313142624453,
      "grad_norm": 3.8397438526153564,
      "learning_rate": 1.072280161979956e-05,
      "loss": 0.1025,
      "step": 25425
    },
    {
      "epoch": 2.5908581899623333,
      "grad_norm": 3.1397838592529297,
      "learning_rate": 1.071149017035043e-05,
      "loss": 0.0988,
      "step": 25450
    },
    {
      "epoch": 2.5934032373002136,
      "grad_norm": 0.3933914005756378,
      "learning_rate": 1.0700178720901298e-05,
      "loss": 0.0964,
      "step": 25475
    },
    {
      "epoch": 2.5959482846380943,
      "grad_norm": 4.579705238342285,
      "learning_rate": 1.0688867271452165e-05,
      "loss": 0.1126,
      "step": 25500
    },
    {
      "epoch": 2.5984933319759747,
      "grad_norm": 1.6875035762786865,
      "learning_rate": 1.0677555822003032e-05,
      "loss": 0.0902,
      "step": 25525
    },
    {
      "epoch": 2.6010383793138554,
      "grad_norm": 0.3768336772918701,
      "learning_rate": 1.06662443725539e-05,
      "loss": 0.1293,
      "step": 25550
    },
    {
      "epoch": 2.6035834266517357,
      "grad_norm": 3.3800220489501953,
      "learning_rate": 1.0654932923104769e-05,
      "loss": 0.1078,
      "step": 25575
    },
    {
      "epoch": 2.6061284739896164,
      "grad_norm": 3.3132853507995605,
      "learning_rate": 1.0643621473655633e-05,
      "loss": 0.1227,
      "step": 25600
    },
    {
      "epoch": 2.6086735213274967,
      "grad_norm": 3.5779366493225098,
      "learning_rate": 1.0632310024206503e-05,
      "loss": 0.0942,
      "step": 25625
    },
    {
      "epoch": 2.611218568665377,
      "grad_norm": 1.6557024717330933,
      "learning_rate": 1.0620998574757371e-05,
      "loss": 0.1184,
      "step": 25650
    },
    {
      "epoch": 2.613763616003258,
      "grad_norm": 2.5037038326263428,
      "learning_rate": 1.0609687125308239e-05,
      "loss": 0.1013,
      "step": 25675
    },
    {
      "epoch": 2.616308663341138,
      "grad_norm": 1.7757961750030518,
      "learning_rate": 1.0598375675859105e-05,
      "loss": 0.1082,
      "step": 25700
    },
    {
      "epoch": 2.6188537106790184,
      "grad_norm": 4.928213596343994,
      "learning_rate": 1.0587064226409973e-05,
      "loss": 0.0963,
      "step": 25725
    },
    {
      "epoch": 2.621398758016899,
      "grad_norm": 8.937736511230469,
      "learning_rate": 1.0575752776960842e-05,
      "loss": 0.1063,
      "step": 25750
    },
    {
      "epoch": 2.6239438053547794,
      "grad_norm": 2.6956536769866943,
      "learning_rate": 1.0564441327511707e-05,
      "loss": 0.0912,
      "step": 25775
    },
    {
      "epoch": 2.62648885269266,
      "grad_norm": 2.5104563236236572,
      "learning_rate": 1.0553129878062576e-05,
      "loss": 0.0929,
      "step": 25800
    },
    {
      "epoch": 2.6290339000305405,
      "grad_norm": 3.791407823562622,
      "learning_rate": 1.0541818428613444e-05,
      "loss": 0.1296,
      "step": 25825
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 1.7674590349197388,
      "learning_rate": 1.053050697916431e-05,
      "loss": 0.0797,
      "step": 25850
    },
    {
      "epoch": 2.6341239947063015,
      "grad_norm": 5.181350231170654,
      "learning_rate": 1.0519195529715178e-05,
      "loss": 0.0891,
      "step": 25875
    },
    {
      "epoch": 2.636669042044182,
      "grad_norm": 5.08585262298584,
      "learning_rate": 1.0507884080266046e-05,
      "loss": 0.101,
      "step": 25900
    },
    {
      "epoch": 2.6392140893820626,
      "grad_norm": 3.8872954845428467,
      "learning_rate": 1.0496572630816915e-05,
      "loss": 0.0651,
      "step": 25925
    },
    {
      "epoch": 2.641759136719943,
      "grad_norm": 2.1056861877441406,
      "learning_rate": 1.048526118136778e-05,
      "loss": 0.0778,
      "step": 25950
    },
    {
      "epoch": 2.6443041840578236,
      "grad_norm": 4.589896202087402,
      "learning_rate": 1.047394973191865e-05,
      "loss": 0.0976,
      "step": 25975
    },
    {
      "epoch": 2.646849231395704,
      "grad_norm": 1.1529279947280884,
      "learning_rate": 1.0462638282469517e-05,
      "loss": 0.0872,
      "step": 26000
    },
    {
      "epoch": 2.6493942787335847,
      "grad_norm": 8.964022636413574,
      "learning_rate": 1.0451326833020383e-05,
      "loss": 0.0974,
      "step": 26025
    },
    {
      "epoch": 2.651939326071465,
      "grad_norm": 0.959297776222229,
      "learning_rate": 1.0440015383571251e-05,
      "loss": 0.1183,
      "step": 26050
    },
    {
      "epoch": 2.6544843734093453,
      "grad_norm": 3.5030341148376465,
      "learning_rate": 1.0428703934122119e-05,
      "loss": 0.1122,
      "step": 26075
    },
    {
      "epoch": 2.657029420747226,
      "grad_norm": 3.678330659866333,
      "learning_rate": 1.0417392484672989e-05,
      "loss": 0.1087,
      "step": 26100
    },
    {
      "epoch": 2.6595744680851063,
      "grad_norm": 5.365817546844482,
      "learning_rate": 1.0406081035223853e-05,
      "loss": 0.1147,
      "step": 26125
    },
    {
      "epoch": 2.6621195154229866,
      "grad_norm": 5.787365436553955,
      "learning_rate": 1.0394769585774723e-05,
      "loss": 0.1113,
      "step": 26150
    },
    {
      "epoch": 2.6646645627608674,
      "grad_norm": 0.9475263953208923,
      "learning_rate": 1.038345813632559e-05,
      "loss": 0.0729,
      "step": 26175
    },
    {
      "epoch": 2.6672096100987477,
      "grad_norm": 2.679917573928833,
      "learning_rate": 1.0372146686876457e-05,
      "loss": 0.0869,
      "step": 26200
    },
    {
      "epoch": 2.6697546574366284,
      "grad_norm": 5.277957916259766,
      "learning_rate": 1.0360835237427325e-05,
      "loss": 0.1159,
      "step": 26225
    },
    {
      "epoch": 2.6722997047745087,
      "grad_norm": 2.8972973823547363,
      "learning_rate": 1.0349976245956157e-05,
      "loss": 0.0984,
      "step": 26250
    },
    {
      "epoch": 2.6748447521123895,
      "grad_norm": 7.060084819793701,
      "learning_rate": 1.0338664796507025e-05,
      "loss": 0.1173,
      "step": 26275
    },
    {
      "epoch": 2.67738979945027,
      "grad_norm": 3.6646735668182373,
      "learning_rate": 1.0327353347057893e-05,
      "loss": 0.0998,
      "step": 26300
    },
    {
      "epoch": 2.67993484678815,
      "grad_norm": 7.1751017570495605,
      "learning_rate": 1.0316041897608759e-05,
      "loss": 0.1257,
      "step": 26325
    },
    {
      "epoch": 2.682479894126031,
      "grad_norm": 4.124204158782959,
      "learning_rate": 1.0304730448159629e-05,
      "loss": 0.1221,
      "step": 26350
    },
    {
      "epoch": 2.685024941463911,
      "grad_norm": 2.890241861343384,
      "learning_rate": 1.0293418998710496e-05,
      "loss": 0.0779,
      "step": 26375
    },
    {
      "epoch": 2.687569988801792,
      "grad_norm": 3.6809606552124023,
      "learning_rate": 1.0282107549261364e-05,
      "loss": 0.1453,
      "step": 26400
    },
    {
      "epoch": 2.690115036139672,
      "grad_norm": 0.7839491963386536,
      "learning_rate": 1.027079609981223e-05,
      "loss": 0.1117,
      "step": 26425
    },
    {
      "epoch": 2.692660083477553,
      "grad_norm": 4.305838584899902,
      "learning_rate": 1.0259484650363098e-05,
      "loss": 0.1391,
      "step": 26450
    },
    {
      "epoch": 2.6952051308154332,
      "grad_norm": 5.504796504974365,
      "learning_rate": 1.0248173200913966e-05,
      "loss": 0.0844,
      "step": 26475
    },
    {
      "epoch": 2.6977501781533135,
      "grad_norm": 1.3983299732208252,
      "learning_rate": 1.0236861751464832e-05,
      "loss": 0.1011,
      "step": 26500
    },
    {
      "epoch": 2.7002952254911943,
      "grad_norm": 3.5293939113616943,
      "learning_rate": 1.0225550302015702e-05,
      "loss": 0.0949,
      "step": 26525
    },
    {
      "epoch": 2.7028402728290746,
      "grad_norm": 0.8638243675231934,
      "learning_rate": 1.021423885256657e-05,
      "loss": 0.0631,
      "step": 26550
    },
    {
      "epoch": 2.705385320166955,
      "grad_norm": 7.877193450927734,
      "learning_rate": 1.0202927403117436e-05,
      "loss": 0.1002,
      "step": 26575
    },
    {
      "epoch": 2.7079303675048356,
      "grad_norm": 4.632683753967285,
      "learning_rate": 1.0191615953668304e-05,
      "loss": 0.0856,
      "step": 26600
    },
    {
      "epoch": 2.710475414842716,
      "grad_norm": 2.3460853099823,
      "learning_rate": 1.0180304504219171e-05,
      "loss": 0.1046,
      "step": 26625
    },
    {
      "epoch": 2.7130204621805967,
      "grad_norm": 4.212901592254639,
      "learning_rate": 1.016899305477004e-05,
      "loss": 0.1039,
      "step": 26650
    },
    {
      "epoch": 2.715565509518477,
      "grad_norm": 5.075952053070068,
      "learning_rate": 1.0157681605320905e-05,
      "loss": 0.1168,
      "step": 26675
    },
    {
      "epoch": 2.7181105568563577,
      "grad_norm": 3.706630229949951,
      "learning_rate": 1.0146370155871775e-05,
      "loss": 0.1187,
      "step": 26700
    },
    {
      "epoch": 2.720655604194238,
      "grad_norm": 8.349642753601074,
      "learning_rate": 1.0135058706422643e-05,
      "loss": 0.1226,
      "step": 26725
    },
    {
      "epoch": 2.7232006515321183,
      "grad_norm": 7.56433629989624,
      "learning_rate": 1.0123747256973509e-05,
      "loss": 0.1012,
      "step": 26750
    },
    {
      "epoch": 2.725745698869999,
      "grad_norm": 5.832910060882568,
      "learning_rate": 1.0112435807524377e-05,
      "loss": 0.106,
      "step": 26775
    },
    {
      "epoch": 2.7282907462078794,
      "grad_norm": 7.008131504058838,
      "learning_rate": 1.0101124358075245e-05,
      "loss": 0.0892,
      "step": 26800
    },
    {
      "epoch": 2.73083579354576,
      "grad_norm": 4.23357629776001,
      "learning_rate": 1.0089812908626113e-05,
      "loss": 0.0899,
      "step": 26825
    },
    {
      "epoch": 2.7333808408836404,
      "grad_norm": 2.169849395751953,
      "learning_rate": 1.0078501459176979e-05,
      "loss": 0.105,
      "step": 26850
    },
    {
      "epoch": 2.735925888221521,
      "grad_norm": 4.076704025268555,
      "learning_rate": 1.0067190009727848e-05,
      "loss": 0.0766,
      "step": 26875
    },
    {
      "epoch": 2.7384709355594015,
      "grad_norm": 4.680348873138428,
      "learning_rate": 1.0055878560278716e-05,
      "loss": 0.0855,
      "step": 26900
    },
    {
      "epoch": 2.7410159828972818,
      "grad_norm": 2.6306278705596924,
      "learning_rate": 1.0044567110829582e-05,
      "loss": 0.127,
      "step": 26925
    },
    {
      "epoch": 2.7435610302351625,
      "grad_norm": 6.218433380126953,
      "learning_rate": 1.003325566138045e-05,
      "loss": 0.1265,
      "step": 26950
    },
    {
      "epoch": 2.746106077573043,
      "grad_norm": 3.8491039276123047,
      "learning_rate": 1.0021944211931318e-05,
      "loss": 0.1109,
      "step": 26975
    },
    {
      "epoch": 2.748651124910923,
      "grad_norm": 4.13882303237915,
      "learning_rate": 1.0010632762482184e-05,
      "loss": 0.1001,
      "step": 27000
    },
    {
      "epoch": 2.751196172248804,
      "grad_norm": 2.967843770980835,
      "learning_rate": 9.999321313033052e-06,
      "loss": 0.1056,
      "step": 27025
    },
    {
      "epoch": 2.753741219586684,
      "grad_norm": 1.7531013488769531,
      "learning_rate": 9.988009863583921e-06,
      "loss": 0.1146,
      "step": 27050
    },
    {
      "epoch": 2.756286266924565,
      "grad_norm": 3.6976428031921387,
      "learning_rate": 9.976698414134788e-06,
      "loss": 0.1213,
      "step": 27075
    },
    {
      "epoch": 2.758831314262445,
      "grad_norm": 4.366905212402344,
      "learning_rate": 9.965386964685655e-06,
      "loss": 0.084,
      "step": 27100
    },
    {
      "epoch": 2.761376361600326,
      "grad_norm": 0.365721732378006,
      "learning_rate": 9.954075515236523e-06,
      "loss": 0.0955,
      "step": 27125
    },
    {
      "epoch": 2.7639214089382063,
      "grad_norm": 3.8855741024017334,
      "learning_rate": 9.94276406578739e-06,
      "loss": 0.0983,
      "step": 27150
    },
    {
      "epoch": 2.7664664562760866,
      "grad_norm": 2.5487568378448486,
      "learning_rate": 9.931452616338259e-06,
      "loss": 0.0997,
      "step": 27175
    },
    {
      "epoch": 2.7690115036139673,
      "grad_norm": 2.598275661468506,
      "learning_rate": 9.920141166889125e-06,
      "loss": 0.1203,
      "step": 27200
    },
    {
      "epoch": 2.7715565509518476,
      "grad_norm": 7.30168342590332,
      "learning_rate": 9.908829717439995e-06,
      "loss": 0.1122,
      "step": 27225
    },
    {
      "epoch": 2.7741015982897284,
      "grad_norm": 6.6037468910217285,
      "learning_rate": 9.89751826799086e-06,
      "loss": 0.114,
      "step": 27250
    },
    {
      "epoch": 2.7766466456276087,
      "grad_norm": 3.6901676654815674,
      "learning_rate": 9.886206818541729e-06,
      "loss": 0.0957,
      "step": 27275
    },
    {
      "epoch": 2.7791916929654894,
      "grad_norm": 1.6863963603973389,
      "learning_rate": 9.874895369092597e-06,
      "loss": 0.1075,
      "step": 27300
    },
    {
      "epoch": 2.7817367403033697,
      "grad_norm": 2.468958854675293,
      "learning_rate": 9.863583919643464e-06,
      "loss": 0.0911,
      "step": 27325
    },
    {
      "epoch": 2.78428178764125,
      "grad_norm": 2.852034330368042,
      "learning_rate": 9.852272470194332e-06,
      "loss": 0.1005,
      "step": 27350
    },
    {
      "epoch": 2.7868268349791308,
      "grad_norm": 2.5854666233062744,
      "learning_rate": 9.840961020745198e-06,
      "loss": 0.0996,
      "step": 27375
    },
    {
      "epoch": 2.789371882317011,
      "grad_norm": 6.25841760635376,
      "learning_rate": 9.829649571296068e-06,
      "loss": 0.1035,
      "step": 27400
    },
    {
      "epoch": 2.7919169296548914,
      "grad_norm": 6.368521690368652,
      "learning_rate": 9.818338121846934e-06,
      "loss": 0.1205,
      "step": 27425
    },
    {
      "epoch": 2.794461976992772,
      "grad_norm": 1.1734464168548584,
      "learning_rate": 9.807026672397802e-06,
      "loss": 0.1068,
      "step": 27450
    },
    {
      "epoch": 2.7970070243306524,
      "grad_norm": 5.702926158905029,
      "learning_rate": 9.79571522294867e-06,
      "loss": 0.0723,
      "step": 27475
    },
    {
      "epoch": 2.799552071668533,
      "grad_norm": 5.801497459411621,
      "learning_rate": 9.784403773499538e-06,
      "loss": 0.0965,
      "step": 27500
    },
    {
      "epoch": 2.8020971190064135,
      "grad_norm": 3.9957456588745117,
      "learning_rate": 9.773092324050405e-06,
      "loss": 0.0719,
      "step": 27525
    },
    {
      "epoch": 2.804642166344294,
      "grad_norm": 1.4273048639297485,
      "learning_rate": 9.761780874601272e-06,
      "loss": 0.0983,
      "step": 27550
    },
    {
      "epoch": 2.8071872136821745,
      "grad_norm": 5.444447994232178,
      "learning_rate": 9.750469425152141e-06,
      "loss": 0.1104,
      "step": 27575
    },
    {
      "epoch": 2.809732261020055,
      "grad_norm": 2.195810556411743,
      "learning_rate": 9.739157975703007e-06,
      "loss": 0.0842,
      "step": 27600
    },
    {
      "epoch": 2.8122773083579355,
      "grad_norm": 0.5844478011131287,
      "learning_rate": 9.727846526253875e-06,
      "loss": 0.0925,
      "step": 27625
    },
    {
      "epoch": 2.814822355695816,
      "grad_norm": 5.637092113494873,
      "learning_rate": 9.716535076804743e-06,
      "loss": 0.096,
      "step": 27650
    },
    {
      "epoch": 2.817367403033696,
      "grad_norm": 4.654635906219482,
      "learning_rate": 9.70522362735561e-06,
      "loss": 0.1455,
      "step": 27675
    },
    {
      "epoch": 2.819912450371577,
      "grad_norm": 4.294269561767578,
      "learning_rate": 9.693912177906479e-06,
      "loss": 0.1088,
      "step": 27700
    },
    {
      "epoch": 2.8224574977094576,
      "grad_norm": 2.964021921157837,
      "learning_rate": 9.682600728457345e-06,
      "loss": 0.0763,
      "step": 27725
    },
    {
      "epoch": 2.825002545047338,
      "grad_norm": 3.658801317214966,
      "learning_rate": 9.671289279008213e-06,
      "loss": 0.0907,
      "step": 27750
    },
    {
      "epoch": 2.8275475923852182,
      "grad_norm": 0.45649853348731995,
      "learning_rate": 9.65997782955908e-06,
      "loss": 0.1006,
      "step": 27775
    },
    {
      "epoch": 2.830092639723099,
      "grad_norm": 1.596645474433899,
      "learning_rate": 9.648666380109948e-06,
      "loss": 0.0753,
      "step": 27800
    },
    {
      "epoch": 2.8326376870609793,
      "grad_norm": 0.9637457132339478,
      "learning_rate": 9.637354930660816e-06,
      "loss": 0.0974,
      "step": 27825
    },
    {
      "epoch": 2.8351827343988596,
      "grad_norm": 4.628258228302002,
      "learning_rate": 9.626043481211684e-06,
      "loss": 0.1121,
      "step": 27850
    },
    {
      "epoch": 2.8377277817367403,
      "grad_norm": 2.2486555576324463,
      "learning_rate": 9.61473203176255e-06,
      "loss": 0.0653,
      "step": 27875
    },
    {
      "epoch": 2.8402728290746206,
      "grad_norm": 3.0560920238494873,
      "learning_rate": 9.603420582313418e-06,
      "loss": 0.1054,
      "step": 27900
    },
    {
      "epoch": 2.8428178764125014,
      "grad_norm": 5.340658187866211,
      "learning_rate": 9.592109132864286e-06,
      "loss": 0.094,
      "step": 27925
    },
    {
      "epoch": 2.8453629237503817,
      "grad_norm": 5.649479389190674,
      "learning_rate": 9.580797683415154e-06,
      "loss": 0.1088,
      "step": 27950
    },
    {
      "epoch": 2.8479079710882624,
      "grad_norm": 2.4257004261016846,
      "learning_rate": 9.569486233966022e-06,
      "loss": 0.0947,
      "step": 27975
    },
    {
      "epoch": 2.8504530184261427,
      "grad_norm": 6.8244242668151855,
      "learning_rate": 9.55817478451689e-06,
      "loss": 0.1219,
      "step": 28000
    },
    {
      "epoch": 2.852998065764023,
      "grad_norm": 4.071915149688721,
      "learning_rate": 9.546863335067757e-06,
      "loss": 0.087,
      "step": 28025
    },
    {
      "epoch": 2.855543113101904,
      "grad_norm": 2.7061493396759033,
      "learning_rate": 9.535551885618623e-06,
      "loss": 0.1076,
      "step": 28050
    },
    {
      "epoch": 2.858088160439784,
      "grad_norm": 0.8928428888320923,
      "learning_rate": 9.524240436169491e-06,
      "loss": 0.1001,
      "step": 28075
    },
    {
      "epoch": 2.8606332077776644,
      "grad_norm": 3.2439606189727783,
      "learning_rate": 9.512928986720359e-06,
      "loss": 0.0919,
      "step": 28100
    },
    {
      "epoch": 2.863178255115545,
      "grad_norm": 2.2897791862487793,
      "learning_rate": 9.501617537271227e-06,
      "loss": 0.1167,
      "step": 28125
    },
    {
      "epoch": 2.865723302453426,
      "grad_norm": 4.564056873321533,
      "learning_rate": 9.490306087822095e-06,
      "loss": 0.0769,
      "step": 28150
    },
    {
      "epoch": 2.868268349791306,
      "grad_norm": 4.80610466003418,
      "learning_rate": 9.478994638372961e-06,
      "loss": 0.1048,
      "step": 28175
    },
    {
      "epoch": 2.8708133971291865,
      "grad_norm": 3.0985665321350098,
      "learning_rate": 9.46768318892383e-06,
      "loss": 0.0918,
      "step": 28200
    },
    {
      "epoch": 2.8733584444670672,
      "grad_norm": 5.210390567779541,
      "learning_rate": 9.456371739474697e-06,
      "loss": 0.1116,
      "step": 28225
    },
    {
      "epoch": 2.8759034918049475,
      "grad_norm": 3.7207789421081543,
      "learning_rate": 9.445060290025564e-06,
      "loss": 0.106,
      "step": 28250
    },
    {
      "epoch": 2.878448539142828,
      "grad_norm": 3.766141653060913,
      "learning_rate": 9.433748840576432e-06,
      "loss": 0.1243,
      "step": 28275
    },
    {
      "epoch": 2.8809935864807086,
      "grad_norm": 5.841989994049072,
      "learning_rate": 9.422437391127298e-06,
      "loss": 0.1119,
      "step": 28300
    },
    {
      "epoch": 2.883538633818589,
      "grad_norm": 2.617666244506836,
      "learning_rate": 9.411125941678168e-06,
      "loss": 0.0793,
      "step": 28325
    },
    {
      "epoch": 2.8860836811564696,
      "grad_norm": 11.287698745727539,
      "learning_rate": 9.399814492229034e-06,
      "loss": 0.0999,
      "step": 28350
    },
    {
      "epoch": 2.88862872849435,
      "grad_norm": 4.3712358474731445,
      "learning_rate": 9.388503042779904e-06,
      "loss": 0.0853,
      "step": 28375
    },
    {
      "epoch": 2.8911737758322307,
      "grad_norm": 3.98750376701355,
      "learning_rate": 9.37719159333077e-06,
      "loss": 0.1512,
      "step": 28400
    },
    {
      "epoch": 2.893718823170111,
      "grad_norm": 0.24144740402698517,
      "learning_rate": 9.365880143881638e-06,
      "loss": 0.1036,
      "step": 28425
    },
    {
      "epoch": 2.8962638705079913,
      "grad_norm": 7.1824727058410645,
      "learning_rate": 9.354568694432506e-06,
      "loss": 0.1061,
      "step": 28450
    },
    {
      "epoch": 2.898808917845872,
      "grad_norm": 4.1219162940979,
      "learning_rate": 9.343257244983372e-06,
      "loss": 0.1099,
      "step": 28475
    },
    {
      "epoch": 2.9013539651837523,
      "grad_norm": 2.997187614440918,
      "learning_rate": 9.331945795534241e-06,
      "loss": 0.1071,
      "step": 28500
    },
    {
      "epoch": 2.9038990125216326,
      "grad_norm": 3.1705572605133057,
      "learning_rate": 9.320634346085107e-06,
      "loss": 0.0869,
      "step": 28525
    },
    {
      "epoch": 2.9064440598595134,
      "grad_norm": 1.6288036108016968,
      "learning_rate": 9.309322896635977e-06,
      "loss": 0.0949,
      "step": 28550
    },
    {
      "epoch": 2.908989107197394,
      "grad_norm": 8.080781936645508,
      "learning_rate": 9.298011447186843e-06,
      "loss": 0.1113,
      "step": 28575
    },
    {
      "epoch": 2.9115341545352744,
      "grad_norm": 5.966607093811035,
      "learning_rate": 9.286699997737711e-06,
      "loss": 0.0832,
      "step": 28600
    },
    {
      "epoch": 2.9140792018731547,
      "grad_norm": 5.3897247314453125,
      "learning_rate": 9.275388548288579e-06,
      "loss": 0.0958,
      "step": 28625
    },
    {
      "epoch": 2.9166242492110355,
      "grad_norm": 1.9493873119354248,
      "learning_rate": 9.264077098839445e-06,
      "loss": 0.0913,
      "step": 28650
    },
    {
      "epoch": 2.9191692965489158,
      "grad_norm": 1.491591215133667,
      "learning_rate": 9.252765649390314e-06,
      "loss": 0.1015,
      "step": 28675
    },
    {
      "epoch": 2.921714343886796,
      "grad_norm": 4.423450946807861,
      "learning_rate": 9.24145419994118e-06,
      "loss": 0.0838,
      "step": 28700
    },
    {
      "epoch": 2.924259391224677,
      "grad_norm": 1.92838716506958,
      "learning_rate": 9.23014275049205e-06,
      "loss": 0.0823,
      "step": 28725
    },
    {
      "epoch": 2.926804438562557,
      "grad_norm": 0.5521523952484131,
      "learning_rate": 9.218831301042916e-06,
      "loss": 0.0754,
      "step": 28750
    },
    {
      "epoch": 2.929349485900438,
      "grad_norm": 5.309305667877197,
      "learning_rate": 9.207519851593784e-06,
      "loss": 0.0894,
      "step": 28775
    },
    {
      "epoch": 2.931894533238318,
      "grad_norm": 0.7912878394126892,
      "learning_rate": 9.196208402144652e-06,
      "loss": 0.0937,
      "step": 28800
    },
    {
      "epoch": 2.934439580576199,
      "grad_norm": 0.36283740401268005,
      "learning_rate": 9.184896952695518e-06,
      "loss": 0.0749,
      "step": 28825
    },
    {
      "epoch": 2.936984627914079,
      "grad_norm": 4.372325420379639,
      "learning_rate": 9.173585503246388e-06,
      "loss": 0.0965,
      "step": 28850
    },
    {
      "epoch": 2.9395296752519595,
      "grad_norm": 1.6252275705337524,
      "learning_rate": 9.162274053797254e-06,
      "loss": 0.1027,
      "step": 28875
    },
    {
      "epoch": 2.9420747225898403,
      "grad_norm": 3.252793073654175,
      "learning_rate": 9.150962604348122e-06,
      "loss": 0.0908,
      "step": 28900
    },
    {
      "epoch": 2.9446197699277206,
      "grad_norm": 7.335224628448486,
      "learning_rate": 9.140103612876954e-06,
      "loss": 0.0792,
      "step": 28925
    },
    {
      "epoch": 2.947164817265601,
      "grad_norm": 6.060163974761963,
      "learning_rate": 9.129244621405787e-06,
      "loss": 0.0602,
      "step": 28950
    },
    {
      "epoch": 2.9497098646034816,
      "grad_norm": 4.286234378814697,
      "learning_rate": 9.117933171956655e-06,
      "loss": 0.1196,
      "step": 28975
    },
    {
      "epoch": 2.9522549119413624,
      "grad_norm": 2.4993934631347656,
      "learning_rate": 9.106621722507523e-06,
      "loss": 0.0894,
      "step": 29000
    },
    {
      "epoch": 2.9547999592792427,
      "grad_norm": 5.36308479309082,
      "learning_rate": 9.09531027305839e-06,
      "loss": 0.0854,
      "step": 29025
    },
    {
      "epoch": 2.957345006617123,
      "grad_norm": 1.1274926662445068,
      "learning_rate": 9.083998823609258e-06,
      "loss": 0.1062,
      "step": 29050
    },
    {
      "epoch": 2.9598900539550037,
      "grad_norm": 0.503273606300354,
      "learning_rate": 9.072687374160125e-06,
      "loss": 0.0947,
      "step": 29075
    },
    {
      "epoch": 2.962435101292884,
      "grad_norm": 6.292426586151123,
      "learning_rate": 9.061375924710994e-06,
      "loss": 0.1098,
      "step": 29100
    },
    {
      "epoch": 2.9649801486307643,
      "grad_norm": 3.6071207523345947,
      "learning_rate": 9.05006447526186e-06,
      "loss": 0.1041,
      "step": 29125
    },
    {
      "epoch": 2.967525195968645,
      "grad_norm": 3.5902364253997803,
      "learning_rate": 9.038753025812728e-06,
      "loss": 0.1027,
      "step": 29150
    },
    {
      "epoch": 2.9700702433065254,
      "grad_norm": 1.8288074731826782,
      "learning_rate": 9.027441576363596e-06,
      "loss": 0.0874,
      "step": 29175
    },
    {
      "epoch": 2.972615290644406,
      "grad_norm": 5.246890068054199,
      "learning_rate": 9.016130126914464e-06,
      "loss": 0.1144,
      "step": 29200
    },
    {
      "epoch": 2.9751603379822864,
      "grad_norm": 2.4337267875671387,
      "learning_rate": 9.004818677465332e-06,
      "loss": 0.0763,
      "step": 29225
    },
    {
      "epoch": 2.977705385320167,
      "grad_norm": 10.622210502624512,
      "learning_rate": 8.993507228016198e-06,
      "loss": 0.0686,
      "step": 29250
    },
    {
      "epoch": 2.9802504326580475,
      "grad_norm": 6.085145950317383,
      "learning_rate": 8.982195778567067e-06,
      "loss": 0.095,
      "step": 29275
    },
    {
      "epoch": 2.9827954799959278,
      "grad_norm": 5.365617752075195,
      "learning_rate": 8.970884329117933e-06,
      "loss": 0.0624,
      "step": 29300
    },
    {
      "epoch": 2.9853405273338085,
      "grad_norm": 6.674490451812744,
      "learning_rate": 8.959572879668801e-06,
      "loss": 0.0654,
      "step": 29325
    },
    {
      "epoch": 2.987885574671689,
      "grad_norm": 2.957017421722412,
      "learning_rate": 8.948261430219669e-06,
      "loss": 0.0963,
      "step": 29350
    },
    {
      "epoch": 2.990430622009569,
      "grad_norm": 6.887320518493652,
      "learning_rate": 8.936949980770537e-06,
      "loss": 0.0882,
      "step": 29375
    },
    {
      "epoch": 2.99297566934745,
      "grad_norm": 3.883092164993286,
      "learning_rate": 8.925638531321405e-06,
      "loss": 0.0863,
      "step": 29400
    },
    {
      "epoch": 2.9955207166853306,
      "grad_norm": 1.341077208518982,
      "learning_rate": 8.914327081872271e-06,
      "loss": 0.0833,
      "step": 29425
    },
    {
      "epoch": 2.998065764023211,
      "grad_norm": 6.598110198974609,
      "learning_rate": 8.90301563242314e-06,
      "loss": 0.1083,
      "step": 29450
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.08900922536849976,
      "eval_runtime": 7.4605,
      "eval_samples_per_second": 912.542,
      "eval_steps_per_second": 14.342,
      "step": 29469
    },
    {
      "epoch": 3.000610811361091,
      "grad_norm": 5.234329700469971,
      "learning_rate": 8.891704182974007e-06,
      "loss": 0.0903,
      "step": 29475
    },
    {
      "epoch": 3.003155858698972,
      "grad_norm": 2.3512911796569824,
      "learning_rate": 8.880392733524874e-06,
      "loss": 0.0831,
      "step": 29500
    },
    {
      "epoch": 3.0057009060368522,
      "grad_norm": 5.458704471588135,
      "learning_rate": 8.869081284075742e-06,
      "loss": 0.0656,
      "step": 29525
    },
    {
      "epoch": 3.0082459533747326,
      "grad_norm": 2.52107834815979,
      "learning_rate": 8.85776983462661e-06,
      "loss": 0.1035,
      "step": 29550
    },
    {
      "epoch": 3.0107910007126133,
      "grad_norm": 1.5311448574066162,
      "learning_rate": 8.846458385177478e-06,
      "loss": 0.0727,
      "step": 29575
    },
    {
      "epoch": 3.0133360480504936,
      "grad_norm": 5.179780960083008,
      "learning_rate": 8.835146935728344e-06,
      "loss": 0.0761,
      "step": 29600
    },
    {
      "epoch": 3.0158810953883743,
      "grad_norm": 3.7667253017425537,
      "learning_rate": 8.823835486279214e-06,
      "loss": 0.0873,
      "step": 29625
    },
    {
      "epoch": 3.0184261427262546,
      "grad_norm": 4.558624744415283,
      "learning_rate": 8.81252403683008e-06,
      "loss": 0.1085,
      "step": 29650
    },
    {
      "epoch": 3.0209711900641354,
      "grad_norm": 3.4559326171875,
      "learning_rate": 8.801212587380948e-06,
      "loss": 0.0939,
      "step": 29675
    },
    {
      "epoch": 3.0235162374020157,
      "grad_norm": 6.828591346740723,
      "learning_rate": 8.789901137931816e-06,
      "loss": 0.0735,
      "step": 29700
    },
    {
      "epoch": 3.026061284739896,
      "grad_norm": 2.5354812145233154,
      "learning_rate": 8.778589688482683e-06,
      "loss": 0.0719,
      "step": 29725
    },
    {
      "epoch": 3.0286063320777767,
      "grad_norm": 1.035030484199524,
      "learning_rate": 8.767278239033551e-06,
      "loss": 0.0957,
      "step": 29750
    },
    {
      "epoch": 3.031151379415657,
      "grad_norm": 2.077613592147827,
      "learning_rate": 8.755966789584417e-06,
      "loss": 0.0942,
      "step": 29775
    },
    {
      "epoch": 3.033696426753538,
      "grad_norm": 2.0107665061950684,
      "learning_rate": 8.744655340135285e-06,
      "loss": 0.0697,
      "step": 29800
    },
    {
      "epoch": 3.036241474091418,
      "grad_norm": 3.822760581970215,
      "learning_rate": 8.733343890686153e-06,
      "loss": 0.0969,
      "step": 29825
    },
    {
      "epoch": 3.0387865214292984,
      "grad_norm": 1.4622304439544678,
      "learning_rate": 8.722032441237021e-06,
      "loss": 0.0733,
      "step": 29850
    },
    {
      "epoch": 3.041331568767179,
      "grad_norm": 4.076181411743164,
      "learning_rate": 8.710720991787889e-06,
      "loss": 0.0528,
      "step": 29875
    },
    {
      "epoch": 3.0438766161050594,
      "grad_norm": 1.466534972190857,
      "learning_rate": 8.699409542338757e-06,
      "loss": 0.0909,
      "step": 29900
    },
    {
      "epoch": 3.04642166344294,
      "grad_norm": 1.0620015859603882,
      "learning_rate": 8.688098092889624e-06,
      "loss": 0.0576,
      "step": 29925
    },
    {
      "epoch": 3.0489667107808205,
      "grad_norm": 2.384843587875366,
      "learning_rate": 8.67678664344049e-06,
      "loss": 0.1137,
      "step": 29950
    },
    {
      "epoch": 3.051511758118701,
      "grad_norm": 4.1187334060668945,
      "learning_rate": 8.665475193991358e-06,
      "loss": 0.0854,
      "step": 29975
    },
    {
      "epoch": 3.0540568054565815,
      "grad_norm": 3.1828019618988037,
      "learning_rate": 8.654163744542226e-06,
      "loss": 0.0794,
      "step": 30000
    },
    {
      "epoch": 3.056601852794462,
      "grad_norm": 1.8246716260910034,
      "learning_rate": 8.642852295093094e-06,
      "loss": 0.0612,
      "step": 30025
    },
    {
      "epoch": 3.0591469001323426,
      "grad_norm": 6.157846450805664,
      "learning_rate": 8.631540845643962e-06,
      "loss": 0.0834,
      "step": 30050
    },
    {
      "epoch": 3.061691947470223,
      "grad_norm": 5.81326961517334,
      "learning_rate": 8.62022939619483e-06,
      "loss": 0.0833,
      "step": 30075
    },
    {
      "epoch": 3.0642369948081036,
      "grad_norm": 8.527436256408691,
      "learning_rate": 8.608917946745696e-06,
      "loss": 0.1205,
      "step": 30100
    },
    {
      "epoch": 3.066782042145984,
      "grad_norm": 2.045522928237915,
      "learning_rate": 8.597606497296564e-06,
      "loss": 0.079,
      "step": 30125
    },
    {
      "epoch": 3.0693270894838642,
      "grad_norm": 4.061382293701172,
      "learning_rate": 8.586295047847432e-06,
      "loss": 0.0598,
      "step": 30150
    },
    {
      "epoch": 3.071872136821745,
      "grad_norm": 1.9634780883789062,
      "learning_rate": 8.5749835983983e-06,
      "loss": 0.0794,
      "step": 30175
    },
    {
      "epoch": 3.0744171841596253,
      "grad_norm": 2.8861138820648193,
      "learning_rate": 8.563672148949167e-06,
      "loss": 0.079,
      "step": 30200
    },
    {
      "epoch": 3.076962231497506,
      "grad_norm": 4.507253170013428,
      "learning_rate": 8.552360699500034e-06,
      "loss": 0.0752,
      "step": 30225
    },
    {
      "epoch": 3.0795072788353863,
      "grad_norm": 12.6360445022583,
      "learning_rate": 8.541049250050903e-06,
      "loss": 0.1178,
      "step": 30250
    },
    {
      "epoch": 3.0820523261732666,
      "grad_norm": 5.043088436126709,
      "learning_rate": 8.52973780060177e-06,
      "loss": 0.1003,
      "step": 30275
    },
    {
      "epoch": 3.0845973735111474,
      "grad_norm": 1.5982271432876587,
      "learning_rate": 8.518426351152637e-06,
      "loss": 0.0765,
      "step": 30300
    },
    {
      "epoch": 3.0871424208490277,
      "grad_norm": 4.377776622772217,
      "learning_rate": 8.507114901703505e-06,
      "loss": 0.0984,
      "step": 30325
    },
    {
      "epoch": 3.0896874681869084,
      "grad_norm": 2.552151679992676,
      "learning_rate": 8.495803452254373e-06,
      "loss": 0.0629,
      "step": 30350
    },
    {
      "epoch": 3.0922325155247887,
      "grad_norm": 8.379557609558105,
      "learning_rate": 8.48449200280524e-06,
      "loss": 0.059,
      "step": 30375
    },
    {
      "epoch": 3.094777562862669,
      "grad_norm": 0.7588167786598206,
      "learning_rate": 8.473180553356107e-06,
      "loss": 0.0763,
      "step": 30400
    },
    {
      "epoch": 3.0973226102005498,
      "grad_norm": 5.55032205581665,
      "learning_rate": 8.461869103906976e-06,
      "loss": 0.0731,
      "step": 30425
    },
    {
      "epoch": 3.09986765753843,
      "grad_norm": 1.346492052078247,
      "learning_rate": 8.450557654457842e-06,
      "loss": 0.0777,
      "step": 30450
    },
    {
      "epoch": 3.102412704876311,
      "grad_norm": 6.685083389282227,
      "learning_rate": 8.43924620500871e-06,
      "loss": 0.08,
      "step": 30475
    },
    {
      "epoch": 3.104957752214191,
      "grad_norm": 2.1243855953216553,
      "learning_rate": 8.427934755559578e-06,
      "loss": 0.0775,
      "step": 30500
    },
    {
      "epoch": 3.107502799552072,
      "grad_norm": 5.800358772277832,
      "learning_rate": 8.416623306110446e-06,
      "loss": 0.0833,
      "step": 30525
    },
    {
      "epoch": 3.110047846889952,
      "grad_norm": 3.737163782119751,
      "learning_rate": 8.405311856661314e-06,
      "loss": 0.0905,
      "step": 30550
    },
    {
      "epoch": 3.1125928942278325,
      "grad_norm": 0.30908119678497314,
      "learning_rate": 8.39400040721218e-06,
      "loss": 0.0782,
      "step": 30575
    },
    {
      "epoch": 3.115137941565713,
      "grad_norm": 4.908120155334473,
      "learning_rate": 8.38268895776305e-06,
      "loss": 0.0613,
      "step": 30600
    },
    {
      "epoch": 3.1176829889035935,
      "grad_norm": 2.8786909580230713,
      "learning_rate": 8.371377508313916e-06,
      "loss": 0.0618,
      "step": 30625
    },
    {
      "epoch": 3.1202280362414743,
      "grad_norm": 4.245732307434082,
      "learning_rate": 8.360066058864783e-06,
      "loss": 0.0579,
      "step": 30650
    },
    {
      "epoch": 3.1227730835793546,
      "grad_norm": 1.4124987125396729,
      "learning_rate": 8.348754609415651e-06,
      "loss": 0.0649,
      "step": 30675
    },
    {
      "epoch": 3.125318130917235,
      "grad_norm": 5.46633243560791,
      "learning_rate": 8.33744315996652e-06,
      "loss": 0.0777,
      "step": 30700
    },
    {
      "epoch": 3.1278631782551156,
      "grad_norm": 2.1597933769226074,
      "learning_rate": 8.326131710517387e-06,
      "loss": 0.0882,
      "step": 30725
    },
    {
      "epoch": 3.130408225592996,
      "grad_norm": 1.188536524772644,
      "learning_rate": 8.314820261068253e-06,
      "loss": 0.0792,
      "step": 30750
    },
    {
      "epoch": 3.1329532729308767,
      "grad_norm": 3.5105278491973877,
      "learning_rate": 8.303508811619123e-06,
      "loss": 0.0673,
      "step": 30775
    },
    {
      "epoch": 3.135498320268757,
      "grad_norm": 1.3239123821258545,
      "learning_rate": 8.292197362169989e-06,
      "loss": 0.0644,
      "step": 30800
    },
    {
      "epoch": 3.1380433676066373,
      "grad_norm": 0.6118612885475159,
      "learning_rate": 8.280885912720857e-06,
      "loss": 0.0681,
      "step": 30825
    },
    {
      "epoch": 3.140588414944518,
      "grad_norm": 3.204726457595825,
      "learning_rate": 8.269574463271725e-06,
      "loss": 0.077,
      "step": 30850
    },
    {
      "epoch": 3.1431334622823983,
      "grad_norm": 7.448500156402588,
      "learning_rate": 8.258263013822592e-06,
      "loss": 0.0897,
      "step": 30875
    },
    {
      "epoch": 3.145678509620279,
      "grad_norm": 6.173210144042969,
      "learning_rate": 8.24695156437346e-06,
      "loss": 0.0906,
      "step": 30900
    },
    {
      "epoch": 3.1482235569581594,
      "grad_norm": 1.42446768283844,
      "learning_rate": 8.235640114924326e-06,
      "loss": 0.0698,
      "step": 30925
    },
    {
      "epoch": 3.15076860429604,
      "grad_norm": 2.045053482055664,
      "learning_rate": 8.224328665475194e-06,
      "loss": 0.0682,
      "step": 30950
    },
    {
      "epoch": 3.1533136516339204,
      "grad_norm": 5.3061442375183105,
      "learning_rate": 8.213017216026062e-06,
      "loss": 0.1129,
      "step": 30975
    },
    {
      "epoch": 3.1558586989718007,
      "grad_norm": 0.9427996873855591,
      "learning_rate": 8.20170576657693e-06,
      "loss": 0.0822,
      "step": 31000
    },
    {
      "epoch": 3.1584037463096815,
      "grad_norm": 3.678173542022705,
      "learning_rate": 8.190394317127798e-06,
      "loss": 0.077,
      "step": 31025
    },
    {
      "epoch": 3.1609487936475618,
      "grad_norm": 1.559303641319275,
      "learning_rate": 8.179082867678666e-06,
      "loss": 0.0681,
      "step": 31050
    },
    {
      "epoch": 3.1634938409854425,
      "grad_norm": 9.293242454528809,
      "learning_rate": 8.167771418229533e-06,
      "loss": 0.0791,
      "step": 31075
    },
    {
      "epoch": 3.166038888323323,
      "grad_norm": 2.898684024810791,
      "learning_rate": 8.1564599687804e-06,
      "loss": 0.0937,
      "step": 31100
    },
    {
      "epoch": 3.168583935661203,
      "grad_norm": 0.6863572001457214,
      "learning_rate": 8.145148519331267e-06,
      "loss": 0.0911,
      "step": 31125
    },
    {
      "epoch": 3.171128982999084,
      "grad_norm": 1.8808056116104126,
      "learning_rate": 8.133837069882135e-06,
      "loss": 0.0686,
      "step": 31150
    },
    {
      "epoch": 3.173674030336964,
      "grad_norm": 0.8487688899040222,
      "learning_rate": 8.122525620433003e-06,
      "loss": 0.0677,
      "step": 31175
    },
    {
      "epoch": 3.176219077674845,
      "grad_norm": 2.2487783432006836,
      "learning_rate": 8.111214170983871e-06,
      "loss": 0.0722,
      "step": 31200
    },
    {
      "epoch": 3.178764125012725,
      "grad_norm": 2.0429940223693848,
      "learning_rate": 8.099902721534739e-06,
      "loss": 0.1115,
      "step": 31225
    },
    {
      "epoch": 3.1813091723506055,
      "grad_norm": 3.094607353210449,
      "learning_rate": 8.088591272085605e-06,
      "loss": 0.1071,
      "step": 31250
    },
    {
      "epoch": 3.1838542196884863,
      "grad_norm": 0.322568416595459,
      "learning_rate": 8.077279822636473e-06,
      "loss": 0.0717,
      "step": 31275
    },
    {
      "epoch": 3.1863992670263666,
      "grad_norm": 2.278010845184326,
      "learning_rate": 8.06596837318734e-06,
      "loss": 0.0865,
      "step": 31300
    },
    {
      "epoch": 3.1889443143642473,
      "grad_norm": 1.3545632362365723,
      "learning_rate": 8.054656923738209e-06,
      "loss": 0.0756,
      "step": 31325
    },
    {
      "epoch": 3.1914893617021276,
      "grad_norm": 4.890456199645996,
      "learning_rate": 8.043345474289076e-06,
      "loss": 0.0894,
      "step": 31350
    },
    {
      "epoch": 3.1940344090400083,
      "grad_norm": 3.303328514099121,
      "learning_rate": 8.032034024839943e-06,
      "loss": 0.0853,
      "step": 31375
    },
    {
      "epoch": 3.1965794563778887,
      "grad_norm": 0.5776151418685913,
      "learning_rate": 8.020722575390812e-06,
      "loss": 0.081,
      "step": 31400
    },
    {
      "epoch": 3.199124503715769,
      "grad_norm": 5.187852382659912,
      "learning_rate": 8.009411125941678e-06,
      "loss": 0.0774,
      "step": 31425
    },
    {
      "epoch": 3.2016695510536497,
      "grad_norm": 4.518702507019043,
      "learning_rate": 7.998099676492546e-06,
      "loss": 0.0559,
      "step": 31450
    },
    {
      "epoch": 3.20421459839153,
      "grad_norm": 4.617893218994141,
      "learning_rate": 7.986788227043414e-06,
      "loss": 0.0731,
      "step": 31475
    },
    {
      "epoch": 3.2067596457294107,
      "grad_norm": 0.9509522914886475,
      "learning_rate": 7.975476777594282e-06,
      "loss": 0.0882,
      "step": 31500
    },
    {
      "epoch": 3.209304693067291,
      "grad_norm": 5.752696514129639,
      "learning_rate": 7.96416532814515e-06,
      "loss": 0.1022,
      "step": 31525
    },
    {
      "epoch": 3.2118497404051713,
      "grad_norm": 2.312471389770508,
      "learning_rate": 7.952853878696016e-06,
      "loss": 0.1257,
      "step": 31550
    },
    {
      "epoch": 3.214394787743052,
      "grad_norm": 0.9991899132728577,
      "learning_rate": 7.941542429246885e-06,
      "loss": 0.0745,
      "step": 31575
    },
    {
      "epoch": 3.2169398350809324,
      "grad_norm": 1.8518400192260742,
      "learning_rate": 7.930230979797751e-06,
      "loss": 0.0905,
      "step": 31600
    },
    {
      "epoch": 3.219484882418813,
      "grad_norm": 1.001679539680481,
      "learning_rate": 7.91891953034862e-06,
      "loss": 0.0764,
      "step": 31625
    },
    {
      "epoch": 3.2220299297566934,
      "grad_norm": 1.2771847248077393,
      "learning_rate": 7.907608080899487e-06,
      "loss": 0.0649,
      "step": 31650
    },
    {
      "epoch": 3.2245749770945737,
      "grad_norm": 1.6247678995132446,
      "learning_rate": 7.896296631450355e-06,
      "loss": 0.0843,
      "step": 31675
    },
    {
      "epoch": 3.2271200244324545,
      "grad_norm": 2.9473443031311035,
      "learning_rate": 7.884985182001223e-06,
      "loss": 0.1045,
      "step": 31700
    },
    {
      "epoch": 3.229665071770335,
      "grad_norm": 0.8892804980278015,
      "learning_rate": 7.873673732552089e-06,
      "loss": 0.0714,
      "step": 31725
    },
    {
      "epoch": 3.2322101191082155,
      "grad_norm": 1.752219796180725,
      "learning_rate": 7.862362283102959e-06,
      "loss": 0.0999,
      "step": 31750
    },
    {
      "epoch": 3.234755166446096,
      "grad_norm": 2.873009443283081,
      "learning_rate": 7.851050833653825e-06,
      "loss": 0.0779,
      "step": 31775
    },
    {
      "epoch": 3.2373002137839766,
      "grad_norm": 2.4850993156433105,
      "learning_rate": 7.839739384204693e-06,
      "loss": 0.0622,
      "step": 31800
    },
    {
      "epoch": 3.239845261121857,
      "grad_norm": 0.40809884667396545,
      "learning_rate": 7.82842793475556e-06,
      "loss": 0.0775,
      "step": 31825
    },
    {
      "epoch": 3.242390308459737,
      "grad_norm": 2.4178569316864014,
      "learning_rate": 7.817116485306428e-06,
      "loss": 0.0738,
      "step": 31850
    },
    {
      "epoch": 3.244935355797618,
      "grad_norm": 3.458000421524048,
      "learning_rate": 7.805805035857296e-06,
      "loss": 0.092,
      "step": 31875
    },
    {
      "epoch": 3.2474804031354982,
      "grad_norm": 2.608797788619995,
      "learning_rate": 7.794493586408162e-06,
      "loss": 0.0772,
      "step": 31900
    },
    {
      "epoch": 3.250025450473379,
      "grad_norm": 1.444105863571167,
      "learning_rate": 7.783182136959032e-06,
      "loss": 0.063,
      "step": 31925
    },
    {
      "epoch": 3.2525704978112593,
      "grad_norm": 2.722048759460449,
      "learning_rate": 7.771870687509898e-06,
      "loss": 0.0831,
      "step": 31950
    },
    {
      "epoch": 3.2551155451491396,
      "grad_norm": 2.0308265686035156,
      "learning_rate": 7.760559238060766e-06,
      "loss": 0.0805,
      "step": 31975
    },
    {
      "epoch": 3.2576605924870203,
      "grad_norm": 1.9619109630584717,
      "learning_rate": 7.749247788611634e-06,
      "loss": 0.0866,
      "step": 32000
    },
    {
      "epoch": 3.2602056398249006,
      "grad_norm": 1.6465561389923096,
      "learning_rate": 7.737936339162501e-06,
      "loss": 0.0968,
      "step": 32025
    },
    {
      "epoch": 3.2627506871627814,
      "grad_norm": 1.229393720626831,
      "learning_rate": 7.72662488971337e-06,
      "loss": 0.0701,
      "step": 32050
    },
    {
      "epoch": 3.2652957345006617,
      "grad_norm": 3.8762965202331543,
      "learning_rate": 7.715313440264235e-06,
      "loss": 0.0815,
      "step": 32075
    },
    {
      "epoch": 3.267840781838542,
      "grad_norm": 1.9568135738372803,
      "learning_rate": 7.704001990815103e-06,
      "loss": 0.0718,
      "step": 32100
    },
    {
      "epoch": 3.2703858291764227,
      "grad_norm": 3.7460007667541504,
      "learning_rate": 7.692690541365971e-06,
      "loss": 0.09,
      "step": 32125
    },
    {
      "epoch": 3.272930876514303,
      "grad_norm": 0.3636890649795532,
      "learning_rate": 7.681379091916839e-06,
      "loss": 0.0874,
      "step": 32150
    },
    {
      "epoch": 3.275475923852184,
      "grad_norm": 4.578624725341797,
      "learning_rate": 7.670067642467707e-06,
      "loss": 0.0717,
      "step": 32175
    },
    {
      "epoch": 3.278020971190064,
      "grad_norm": 7.152780532836914,
      "learning_rate": 7.658756193018575e-06,
      "loss": 0.065,
      "step": 32200
    },
    {
      "epoch": 3.280566018527945,
      "grad_norm": 2.064504384994507,
      "learning_rate": 7.647444743569442e-06,
      "loss": 0.08,
      "step": 32225
    },
    {
      "epoch": 3.283111065865825,
      "grad_norm": 0.4257572293281555,
      "learning_rate": 7.636133294120309e-06,
      "loss": 0.0773,
      "step": 32250
    },
    {
      "epoch": 3.2856561132037054,
      "grad_norm": 0.999352216720581,
      "learning_rate": 7.6248218446711765e-06,
      "loss": 0.0728,
      "step": 32275
    },
    {
      "epoch": 3.288201160541586,
      "grad_norm": 2.5123298168182373,
      "learning_rate": 7.613510395222044e-06,
      "loss": 0.0756,
      "step": 32300
    },
    {
      "epoch": 3.2907462078794665,
      "grad_norm": 6.116274833679199,
      "learning_rate": 7.602198945772911e-06,
      "loss": 0.0733,
      "step": 32325
    },
    {
      "epoch": 3.2932912552173472,
      "grad_norm": 7.821257591247559,
      "learning_rate": 7.59088749632378e-06,
      "loss": 0.0616,
      "step": 32350
    },
    {
      "epoch": 3.2958363025552275,
      "grad_norm": 4.588683128356934,
      "learning_rate": 7.579576046874647e-06,
      "loss": 0.0935,
      "step": 32375
    },
    {
      "epoch": 3.298381349893108,
      "grad_norm": 7.06011962890625,
      "learning_rate": 7.568264597425514e-06,
      "loss": 0.0604,
      "step": 32400
    },
    {
      "epoch": 3.3009263972309886,
      "grad_norm": 5.7961201667785645,
      "learning_rate": 7.556953147976383e-06,
      "loss": 0.0774,
      "step": 32425
    },
    {
      "epoch": 3.303471444568869,
      "grad_norm": 1.789345383644104,
      "learning_rate": 7.54564169852725e-06,
      "loss": 0.0879,
      "step": 32450
    },
    {
      "epoch": 3.3060164919067496,
      "grad_norm": 5.761059284210205,
      "learning_rate": 7.5343302490781176e-06,
      "loss": 0.1144,
      "step": 32475
    },
    {
      "epoch": 3.30856153924463,
      "grad_norm": 1.1593629121780396,
      "learning_rate": 7.5230187996289846e-06,
      "loss": 0.0879,
      "step": 32500
    },
    {
      "epoch": 3.3111065865825102,
      "grad_norm": 1.7188637256622314,
      "learning_rate": 7.511707350179853e-06,
      "loss": 0.0765,
      "step": 32525
    },
    {
      "epoch": 3.313651633920391,
      "grad_norm": 0.9226068258285522,
      "learning_rate": 7.50039590073072e-06,
      "loss": 0.063,
      "step": 32550
    },
    {
      "epoch": 3.3161966812582713,
      "grad_norm": 5.403501987457275,
      "learning_rate": 7.489084451281587e-06,
      "loss": 0.0773,
      "step": 32575
    },
    {
      "epoch": 3.318741728596152,
      "grad_norm": 3.45514178276062,
      "learning_rate": 7.477773001832456e-06,
      "loss": 0.0614,
      "step": 32600
    },
    {
      "epoch": 3.3212867759340323,
      "grad_norm": 3.7157094478607178,
      "learning_rate": 7.466461552383323e-06,
      "loss": 0.0876,
      "step": 32625
    },
    {
      "epoch": 3.323831823271913,
      "grad_norm": 5.224157333374023,
      "learning_rate": 7.455150102934191e-06,
      "loss": 0.0801,
      "step": 32650
    },
    {
      "epoch": 3.3263768706097934,
      "grad_norm": 5.631719589233398,
      "learning_rate": 7.443838653485058e-06,
      "loss": 0.0864,
      "step": 32675
    },
    {
      "epoch": 3.3289219179476737,
      "grad_norm": 1.3295774459838867,
      "learning_rate": 7.432527204035926e-06,
      "loss": 0.0991,
      "step": 32700
    },
    {
      "epoch": 3.3314669652855544,
      "grad_norm": 1.4614148139953613,
      "learning_rate": 7.4212157545867935e-06,
      "loss": 0.1021,
      "step": 32725
    },
    {
      "epoch": 3.3340120126234347,
      "grad_norm": 1.3813050985336304,
      "learning_rate": 7.4099043051376605e-06,
      "loss": 0.0542,
      "step": 32750
    },
    {
      "epoch": 3.3365570599613155,
      "grad_norm": 2.403815984725952,
      "learning_rate": 7.398592855688529e-06,
      "loss": 0.0896,
      "step": 32775
    },
    {
      "epoch": 3.3391021072991958,
      "grad_norm": 4.827250957489014,
      "learning_rate": 7.387281406239396e-06,
      "loss": 0.0828,
      "step": 32800
    },
    {
      "epoch": 3.341647154637076,
      "grad_norm": 5.572385311126709,
      "learning_rate": 7.375969956790263e-06,
      "loss": 0.0814,
      "step": 32825
    },
    {
      "epoch": 3.344192201974957,
      "grad_norm": 3.9175562858581543,
      "learning_rate": 7.364658507341131e-06,
      "loss": 0.0883,
      "step": 32850
    },
    {
      "epoch": 3.346737249312837,
      "grad_norm": 0.47550782561302185,
      "learning_rate": 7.353347057891999e-06,
      "loss": 0.0829,
      "step": 32875
    },
    {
      "epoch": 3.349282296650718,
      "grad_norm": 1.9550349712371826,
      "learning_rate": 7.342035608442867e-06,
      "loss": 0.0717,
      "step": 32900
    },
    {
      "epoch": 3.351827343988598,
      "grad_norm": 4.848787307739258,
      "learning_rate": 7.330724158993734e-06,
      "loss": 0.0727,
      "step": 32925
    },
    {
      "epoch": 3.3543723913264785,
      "grad_norm": 1.4837538003921509,
      "learning_rate": 7.319412709544602e-06,
      "loss": 0.0573,
      "step": 32950
    },
    {
      "epoch": 3.356917438664359,
      "grad_norm": 6.307371139526367,
      "learning_rate": 7.308101260095469e-06,
      "loss": 0.0789,
      "step": 32975
    },
    {
      "epoch": 3.3594624860022395,
      "grad_norm": 4.478782653808594,
      "learning_rate": 7.296789810646336e-06,
      "loss": 0.0763,
      "step": 33000
    },
    {
      "epoch": 3.3620075333401203,
      "grad_norm": 3.771304130554199,
      "learning_rate": 7.285478361197204e-06,
      "loss": 0.071,
      "step": 33025
    },
    {
      "epoch": 3.3645525806780006,
      "grad_norm": 4.51546049118042,
      "learning_rate": 7.274166911748072e-06,
      "loss": 0.0943,
      "step": 33050
    },
    {
      "epoch": 3.3670976280158813,
      "grad_norm": 1.119053840637207,
      "learning_rate": 7.26285546229894e-06,
      "loss": 0.0653,
      "step": 33075
    },
    {
      "epoch": 3.3696426753537616,
      "grad_norm": 4.0295305252075195,
      "learning_rate": 7.251544012849807e-06,
      "loss": 0.098,
      "step": 33100
    },
    {
      "epoch": 3.372187722691642,
      "grad_norm": 3.882275104522705,
      "learning_rate": 7.240232563400674e-06,
      "loss": 0.0764,
      "step": 33125
    },
    {
      "epoch": 3.3747327700295227,
      "grad_norm": 4.427306175231934,
      "learning_rate": 7.228921113951543e-06,
      "loss": 0.0687,
      "step": 33150
    },
    {
      "epoch": 3.377277817367403,
      "grad_norm": 2.3375508785247803,
      "learning_rate": 7.21760966450241e-06,
      "loss": 0.083,
      "step": 33175
    },
    {
      "epoch": 3.3798228647052837,
      "grad_norm": 1.8188954591751099,
      "learning_rate": 7.2062982150532774e-06,
      "loss": 0.0731,
      "step": 33200
    },
    {
      "epoch": 3.382367912043164,
      "grad_norm": 4.421236515045166,
      "learning_rate": 7.194986765604145e-06,
      "loss": 0.0699,
      "step": 33225
    },
    {
      "epoch": 3.3849129593810443,
      "grad_norm": 3.788590669631958,
      "learning_rate": 7.183675316155013e-06,
      "loss": 0.0962,
      "step": 33250
    },
    {
      "epoch": 3.387458006718925,
      "grad_norm": 2.9976346492767334,
      "learning_rate": 7.17236386670588e-06,
      "loss": 0.0867,
      "step": 33275
    },
    {
      "epoch": 3.3900030540568054,
      "grad_norm": 4.236015319824219,
      "learning_rate": 7.161052417256747e-06,
      "loss": 0.0845,
      "step": 33300
    },
    {
      "epoch": 3.392548101394686,
      "grad_norm": 4.394545555114746,
      "learning_rate": 7.149740967807616e-06,
      "loss": 0.0667,
      "step": 33325
    },
    {
      "epoch": 3.3950931487325664,
      "grad_norm": 1.5164837837219238,
      "learning_rate": 7.138429518358483e-06,
      "loss": 0.0715,
      "step": 33350
    },
    {
      "epoch": 3.3976381960704467,
      "grad_norm": 3.266295909881592,
      "learning_rate": 7.1271180689093515e-06,
      "loss": 0.113,
      "step": 33375
    },
    {
      "epoch": 3.4001832434083274,
      "grad_norm": 1.4110231399536133,
      "learning_rate": 7.1158066194602185e-06,
      "loss": 0.076,
      "step": 33400
    },
    {
      "epoch": 3.4027282907462078,
      "grad_norm": 5.217437267303467,
      "learning_rate": 7.1044951700110855e-06,
      "loss": 0.0642,
      "step": 33425
    },
    {
      "epoch": 3.4052733380840885,
      "grad_norm": 1.9967962503433228,
      "learning_rate": 7.093183720561953e-06,
      "loss": 0.0691,
      "step": 33450
    },
    {
      "epoch": 3.407818385421969,
      "grad_norm": 1.0126267671585083,
      "learning_rate": 7.08187227111282e-06,
      "loss": 0.0813,
      "step": 33475
    },
    {
      "epoch": 3.4103634327598495,
      "grad_norm": 3.0378081798553467,
      "learning_rate": 7.070560821663689e-06,
      "loss": 0.0828,
      "step": 33500
    },
    {
      "epoch": 3.41290848009773,
      "grad_norm": 3.2735512256622314,
      "learning_rate": 7.059249372214556e-06,
      "loss": 0.0724,
      "step": 33525
    },
    {
      "epoch": 3.41545352743561,
      "grad_norm": 1.570923089981079,
      "learning_rate": 7.047937922765423e-06,
      "loss": 0.0923,
      "step": 33550
    },
    {
      "epoch": 3.417998574773491,
      "grad_norm": 1.2357620000839233,
      "learning_rate": 7.036626473316292e-06,
      "loss": 0.0926,
      "step": 33575
    },
    {
      "epoch": 3.420543622111371,
      "grad_norm": 2.3194172382354736,
      "learning_rate": 7.025315023867159e-06,
      "loss": 0.0837,
      "step": 33600
    },
    {
      "epoch": 3.423088669449252,
      "grad_norm": 4.630713939666748,
      "learning_rate": 7.014003574418027e-06,
      "loss": 0.0948,
      "step": 33625
    },
    {
      "epoch": 3.4256337167871322,
      "grad_norm": 2.8066182136535645,
      "learning_rate": 7.0026921249688936e-06,
      "loss": 0.1054,
      "step": 33650
    },
    {
      "epoch": 3.4281787641250125,
      "grad_norm": 2.1490931510925293,
      "learning_rate": 6.991380675519762e-06,
      "loss": 0.0792,
      "step": 33675
    },
    {
      "epoch": 3.4307238114628933,
      "grad_norm": 2.8342697620391846,
      "learning_rate": 6.980069226070629e-06,
      "loss": 0.0857,
      "step": 33700
    },
    {
      "epoch": 3.4332688588007736,
      "grad_norm": 3.6628048419952393,
      "learning_rate": 6.968757776621496e-06,
      "loss": 0.0751,
      "step": 33725
    },
    {
      "epoch": 3.4358139061386543,
      "grad_norm": 0.3415294587612152,
      "learning_rate": 6.957446327172365e-06,
      "loss": 0.0847,
      "step": 33750
    },
    {
      "epoch": 3.4383589534765346,
      "grad_norm": 1.9966833591461182,
      "learning_rate": 6.946134877723232e-06,
      "loss": 0.0598,
      "step": 33775
    },
    {
      "epoch": 3.440904000814415,
      "grad_norm": 6.195132255554199,
      "learning_rate": 6.9348234282741e-06,
      "loss": 0.0697,
      "step": 33800
    },
    {
      "epoch": 3.4434490481522957,
      "grad_norm": 3.475865364074707,
      "learning_rate": 6.923511978824967e-06,
      "loss": 0.0943,
      "step": 33825
    },
    {
      "epoch": 3.445994095490176,
      "grad_norm": 2.3039791584014893,
      "learning_rate": 6.912200529375835e-06,
      "loss": 0.089,
      "step": 33850
    },
    {
      "epoch": 3.4485391428280567,
      "grad_norm": 5.273120880126953,
      "learning_rate": 6.901341537904668e-06,
      "loss": 0.0887,
      "step": 33875
    },
    {
      "epoch": 3.451084190165937,
      "grad_norm": 5.2622880935668945,
      "learning_rate": 6.890030088455535e-06,
      "loss": 0.0665,
      "step": 33900
    },
    {
      "epoch": 3.453629237503818,
      "grad_norm": 4.5153422355651855,
      "learning_rate": 6.878718639006403e-06,
      "loss": 0.0661,
      "step": 33925
    },
    {
      "epoch": 3.456174284841698,
      "grad_norm": 3.4902474880218506,
      "learning_rate": 6.86740718955727e-06,
      "loss": 0.0894,
      "step": 33950
    },
    {
      "epoch": 3.4587193321795784,
      "grad_norm": 1.9006632566452026,
      "learning_rate": 6.856095740108139e-06,
      "loss": 0.0913,
      "step": 33975
    },
    {
      "epoch": 3.461264379517459,
      "grad_norm": 1.2017245292663574,
      "learning_rate": 6.844784290659006e-06,
      "loss": 0.0917,
      "step": 34000
    },
    {
      "epoch": 3.4638094268553394,
      "grad_norm": 4.50941276550293,
      "learning_rate": 6.833472841209873e-06,
      "loss": 0.1008,
      "step": 34025
    },
    {
      "epoch": 3.46635447419322,
      "grad_norm": 8.09762191772461,
      "learning_rate": 6.822161391760741e-06,
      "loss": 0.0715,
      "step": 34050
    },
    {
      "epoch": 3.4688995215311005,
      "grad_norm": 6.90067720413208,
      "learning_rate": 6.810849942311608e-06,
      "loss": 0.0698,
      "step": 34075
    },
    {
      "epoch": 3.471444568868981,
      "grad_norm": 5.395306587219238,
      "learning_rate": 6.799538492862476e-06,
      "loss": 0.0722,
      "step": 34100
    },
    {
      "epoch": 3.4739896162068615,
      "grad_norm": 2.4700376987457275,
      "learning_rate": 6.788227043413343e-06,
      "loss": 0.0854,
      "step": 34125
    },
    {
      "epoch": 3.476534663544742,
      "grad_norm": 6.859822750091553,
      "learning_rate": 6.77691559396421e-06,
      "loss": 0.0733,
      "step": 34150
    },
    {
      "epoch": 3.4790797108826226,
      "grad_norm": 0.3821323812007904,
      "learning_rate": 6.765604144515079e-06,
      "loss": 0.089,
      "step": 34175
    },
    {
      "epoch": 3.481624758220503,
      "grad_norm": 4.010385990142822,
      "learning_rate": 6.754292695065946e-06,
      "loss": 0.0828,
      "step": 34200
    },
    {
      "epoch": 3.484169805558383,
      "grad_norm": 0.7583944797515869,
      "learning_rate": 6.742981245616815e-06,
      "loss": 0.1021,
      "step": 34225
    },
    {
      "epoch": 3.486714852896264,
      "grad_norm": 1.1786764860153198,
      "learning_rate": 6.731669796167682e-06,
      "loss": 0.0603,
      "step": 34250
    },
    {
      "epoch": 3.4892599002341442,
      "grad_norm": 2.502434492111206,
      "learning_rate": 6.7203583467185494e-06,
      "loss": 0.127,
      "step": 34275
    },
    {
      "epoch": 3.491804947572025,
      "grad_norm": 4.207679271697998,
      "learning_rate": 6.7090468972694164e-06,
      "loss": 0.0795,
      "step": 34300
    },
    {
      "epoch": 3.4943499949099053,
      "grad_norm": 3.7935028076171875,
      "learning_rate": 6.6977354478202834e-06,
      "loss": 0.0925,
      "step": 34325
    },
    {
      "epoch": 3.496895042247786,
      "grad_norm": 3.029387950897217,
      "learning_rate": 6.686423998371152e-06,
      "loss": 0.0838,
      "step": 34350
    },
    {
      "epoch": 3.4994400895856663,
      "grad_norm": 3.835543394088745,
      "learning_rate": 6.675112548922019e-06,
      "loss": 0.0792,
      "step": 34375
    },
    {
      "epoch": 3.5019851369235466,
      "grad_norm": 7.944116115570068,
      "learning_rate": 6.663801099472888e-06,
      "loss": 0.0866,
      "step": 34400
    },
    {
      "epoch": 3.5045301842614274,
      "grad_norm": 3.8808681964874268,
      "learning_rate": 6.652489650023755e-06,
      "loss": 0.0775,
      "step": 34425
    },
    {
      "epoch": 3.5070752315993077,
      "grad_norm": 0.7064144611358643,
      "learning_rate": 6.641178200574622e-06,
      "loss": 0.0887,
      "step": 34450
    },
    {
      "epoch": 3.509620278937188,
      "grad_norm": 6.215606689453125,
      "learning_rate": 6.62986675112549e-06,
      "loss": 0.1262,
      "step": 34475
    },
    {
      "epoch": 3.5121653262750687,
      "grad_norm": 4.817259788513184,
      "learning_rate": 6.618555301676357e-06,
      "loss": 0.0753,
      "step": 34500
    },
    {
      "epoch": 3.5147103736129495,
      "grad_norm": 0.5267298817634583,
      "learning_rate": 6.607243852227225e-06,
      "loss": 0.1185,
      "step": 34525
    },
    {
      "epoch": 3.5172554209508298,
      "grad_norm": 8.5421781539917,
      "learning_rate": 6.595932402778092e-06,
      "loss": 0.1017,
      "step": 34550
    },
    {
      "epoch": 3.51980046828871,
      "grad_norm": 0.9383201003074646,
      "learning_rate": 6.584620953328959e-06,
      "loss": 0.0829,
      "step": 34575
    },
    {
      "epoch": 3.522345515626591,
      "grad_norm": 7.349745273590088,
      "learning_rate": 6.573309503879828e-06,
      "loss": 0.0883,
      "step": 34600
    },
    {
      "epoch": 3.524890562964471,
      "grad_norm": 1.4736689329147339,
      "learning_rate": 6.561998054430695e-06,
      "loss": 0.0912,
      "step": 34625
    },
    {
      "epoch": 3.5274356103023514,
      "grad_norm": 1.4113051891326904,
      "learning_rate": 6.550686604981563e-06,
      "loss": 0.0796,
      "step": 34650
    },
    {
      "epoch": 3.529980657640232,
      "grad_norm": 7.705810546875,
      "learning_rate": 6.539375155532431e-06,
      "loss": 0.0784,
      "step": 34675
    },
    {
      "epoch": 3.5325257049781125,
      "grad_norm": 2.766878843307495,
      "learning_rate": 6.5280637060832986e-06,
      "loss": 0.0777,
      "step": 34700
    },
    {
      "epoch": 3.535070752315993,
      "grad_norm": 1.7747935056686401,
      "learning_rate": 6.5167522566341656e-06,
      "loss": 0.0695,
      "step": 34725
    },
    {
      "epoch": 3.5376157996538735,
      "grad_norm": 2.4587645530700684,
      "learning_rate": 6.5054408071850326e-06,
      "loss": 0.0839,
      "step": 34750
    },
    {
      "epoch": 3.5401608469917543,
      "grad_norm": 4.459035873413086,
      "learning_rate": 6.494129357735901e-06,
      "loss": 0.0931,
      "step": 34775
    },
    {
      "epoch": 3.5427058943296346,
      "grad_norm": 1.7593748569488525,
      "learning_rate": 6.482817908286768e-06,
      "loss": 0.07,
      "step": 34800
    },
    {
      "epoch": 3.545250941667515,
      "grad_norm": 3.184353828430176,
      "learning_rate": 6.471506458837636e-06,
      "loss": 0.0715,
      "step": 34825
    },
    {
      "epoch": 3.5477959890053956,
      "grad_norm": 3.696401357650757,
      "learning_rate": 6.460195009388504e-06,
      "loss": 0.0697,
      "step": 34850
    },
    {
      "epoch": 3.550341036343276,
      "grad_norm": 2.8429620265960693,
      "learning_rate": 6.448883559939371e-06,
      "loss": 0.0925,
      "step": 34875
    },
    {
      "epoch": 3.552886083681156,
      "grad_norm": 1.776930809020996,
      "learning_rate": 6.437572110490239e-06,
      "loss": 0.0657,
      "step": 34900
    },
    {
      "epoch": 3.555431131019037,
      "grad_norm": 1.681552529335022,
      "learning_rate": 6.426260661041106e-06,
      "loss": 0.0755,
      "step": 34925
    },
    {
      "epoch": 3.5579761783569177,
      "grad_norm": 3.1120924949645996,
      "learning_rate": 6.4149492115919745e-06,
      "loss": 0.0978,
      "step": 34950
    },
    {
      "epoch": 3.560521225694798,
      "grad_norm": 2.5734004974365234,
      "learning_rate": 6.4036377621428415e-06,
      "loss": 0.0792,
      "step": 34975
    },
    {
      "epoch": 3.5630662730326783,
      "grad_norm": 1.1274526119232178,
      "learning_rate": 6.392326312693709e-06,
      "loss": 0.0853,
      "step": 35000
    },
    {
      "epoch": 3.565611320370559,
      "grad_norm": 2.7420918941497803,
      "learning_rate": 6.381014863244577e-06,
      "loss": 0.0852,
      "step": 35025
    },
    {
      "epoch": 3.5681563677084394,
      "grad_norm": 0.742755115032196,
      "learning_rate": 6.369703413795444e-06,
      "loss": 0.1186,
      "step": 35050
    },
    {
      "epoch": 3.5707014150463197,
      "grad_norm": 6.001101493835449,
      "learning_rate": 6.358391964346312e-06,
      "loss": 0.1153,
      "step": 35075
    },
    {
      "epoch": 3.5732464623842004,
      "grad_norm": 3.2310545444488525,
      "learning_rate": 6.347080514897179e-06,
      "loss": 0.0848,
      "step": 35100
    },
    {
      "epoch": 3.5757915097220807,
      "grad_norm": 5.802320957183838,
      "learning_rate": 6.335769065448048e-06,
      "loss": 0.0753,
      "step": 35125
    },
    {
      "epoch": 3.5783365570599615,
      "grad_norm": 4.039153099060059,
      "learning_rate": 6.324457615998915e-06,
      "loss": 0.076,
      "step": 35150
    },
    {
      "epoch": 3.5808816043978418,
      "grad_norm": 3.7043445110321045,
      "learning_rate": 6.313146166549782e-06,
      "loss": 0.0626,
      "step": 35175
    },
    {
      "epoch": 3.5834266517357225,
      "grad_norm": 2.622565984725952,
      "learning_rate": 6.30183471710065e-06,
      "loss": 0.0944,
      "step": 35200
    },
    {
      "epoch": 3.585971699073603,
      "grad_norm": 2.2240231037139893,
      "learning_rate": 6.290523267651517e-06,
      "loss": 0.0923,
      "step": 35225
    },
    {
      "epoch": 3.588516746411483,
      "grad_norm": 3.193891763687134,
      "learning_rate": 6.279211818202385e-06,
      "loss": 0.0982,
      "step": 35250
    },
    {
      "epoch": 3.591061793749364,
      "grad_norm": 3.5597784519195557,
      "learning_rate": 6.267900368753252e-06,
      "loss": 0.0916,
      "step": 35275
    },
    {
      "epoch": 3.593606841087244,
      "grad_norm": 4.625251770019531,
      "learning_rate": 6.256588919304119e-06,
      "loss": 0.1102,
      "step": 35300
    },
    {
      "epoch": 3.5961518884251245,
      "grad_norm": 4.520129203796387,
      "learning_rate": 6.245277469854988e-06,
      "loss": 0.0769,
      "step": 35325
    },
    {
      "epoch": 3.598696935763005,
      "grad_norm": 3.313098192214966,
      "learning_rate": 6.233966020405855e-06,
      "loss": 0.0769,
      "step": 35350
    },
    {
      "epoch": 3.601241983100886,
      "grad_norm": 3.836454153060913,
      "learning_rate": 6.222654570956724e-06,
      "loss": 0.0771,
      "step": 35375
    },
    {
      "epoch": 3.6037870304387662,
      "grad_norm": 3.4968671798706055,
      "learning_rate": 6.211343121507591e-06,
      "loss": 0.0928,
      "step": 35400
    },
    {
      "epoch": 3.6063320777766465,
      "grad_norm": 3.556164026260376,
      "learning_rate": 6.2000316720584585e-06,
      "loss": 0.074,
      "step": 35425
    },
    {
      "epoch": 3.6088771251145273,
      "grad_norm": 2.0920588970184326,
      "learning_rate": 6.1887202226093254e-06,
      "loss": 0.1027,
      "step": 35450
    },
    {
      "epoch": 3.6114221724524076,
      "grad_norm": 5.674222469329834,
      "learning_rate": 6.1774087731601924e-06,
      "loss": 0.0874,
      "step": 35475
    },
    {
      "epoch": 3.613967219790288,
      "grad_norm": 2.4876623153686523,
      "learning_rate": 6.166097323711061e-06,
      "loss": 0.0627,
      "step": 35500
    },
    {
      "epoch": 3.6165122671281686,
      "grad_norm": 4.723164081573486,
      "learning_rate": 6.154785874261928e-06,
      "loss": 0.108,
      "step": 35525
    },
    {
      "epoch": 3.619057314466049,
      "grad_norm": 0.5420049428939819,
      "learning_rate": 6.143474424812797e-06,
      "loss": 0.101,
      "step": 35550
    },
    {
      "epoch": 3.6216023618039297,
      "grad_norm": 0.39233437180519104,
      "learning_rate": 6.132162975363664e-06,
      "loss": 0.0738,
      "step": 35575
    },
    {
      "epoch": 3.62414740914181,
      "grad_norm": 5.043944835662842,
      "learning_rate": 6.120851525914531e-06,
      "loss": 0.0988,
      "step": 35600
    },
    {
      "epoch": 3.6266924564796907,
      "grad_norm": 8.465002059936523,
      "learning_rate": 6.109540076465399e-06,
      "loss": 0.0928,
      "step": 35625
    },
    {
      "epoch": 3.629237503817571,
      "grad_norm": 2.7983903884887695,
      "learning_rate": 6.098228627016266e-06,
      "loss": 0.0771,
      "step": 35650
    },
    {
      "epoch": 3.6317825511554513,
      "grad_norm": 5.560369968414307,
      "learning_rate": 6.086917177567134e-06,
      "loss": 0.0965,
      "step": 35675
    },
    {
      "epoch": 3.634327598493332,
      "grad_norm": 4.1370038986206055,
      "learning_rate": 6.075605728118001e-06,
      "loss": 0.0954,
      "step": 35700
    },
    {
      "epoch": 3.6368726458312124,
      "grad_norm": 4.174961090087891,
      "learning_rate": 6.06429427866887e-06,
      "loss": 0.0736,
      "step": 35725
    },
    {
      "epoch": 3.6394176931690927,
      "grad_norm": 11.217022895812988,
      "learning_rate": 6.052982829219737e-06,
      "loss": 0.0914,
      "step": 35750
    },
    {
      "epoch": 3.6419627405069734,
      "grad_norm": 7.772279739379883,
      "learning_rate": 6.041671379770604e-06,
      "loss": 0.0849,
      "step": 35775
    },
    {
      "epoch": 3.644507787844854,
      "grad_norm": 4.524623870849609,
      "learning_rate": 6.030359930321472e-06,
      "loss": 0.0997,
      "step": 35800
    },
    {
      "epoch": 3.6470528351827345,
      "grad_norm": 3.7422924041748047,
      "learning_rate": 6.019048480872339e-06,
      "loss": 0.0755,
      "step": 35825
    },
    {
      "epoch": 3.649597882520615,
      "grad_norm": 5.526354789733887,
      "learning_rate": 6.007737031423208e-06,
      "loss": 0.0858,
      "step": 35850
    },
    {
      "epoch": 3.6521429298584955,
      "grad_norm": 8.158432006835938,
      "learning_rate": 5.996425581974075e-06,
      "loss": 0.1136,
      "step": 35875
    },
    {
      "epoch": 3.654687977196376,
      "grad_norm": 6.096880912780762,
      "learning_rate": 5.985114132524942e-06,
      "loss": 0.0758,
      "step": 35900
    },
    {
      "epoch": 3.657233024534256,
      "grad_norm": 3.285121440887451,
      "learning_rate": 5.97380268307581e-06,
      "loss": 0.0922,
      "step": 35925
    },
    {
      "epoch": 3.659778071872137,
      "grad_norm": 1.9945199489593506,
      "learning_rate": 5.962491233626677e-06,
      "loss": 0.0701,
      "step": 35950
    },
    {
      "epoch": 3.662323119210017,
      "grad_norm": 0.6748458743095398,
      "learning_rate": 5.951632242155511e-06,
      "loss": 0.0724,
      "step": 35975
    },
    {
      "epoch": 3.664868166547898,
      "grad_norm": 1.5930081605911255,
      "learning_rate": 5.940320792706378e-06,
      "loss": 0.0842,
      "step": 36000
    },
    {
      "epoch": 3.6674132138857782,
      "grad_norm": 3.754176139831543,
      "learning_rate": 5.929009343257246e-06,
      "loss": 0.0763,
      "step": 36025
    },
    {
      "epoch": 3.669958261223659,
      "grad_norm": 3.0190577507019043,
      "learning_rate": 5.9176978938081135e-06,
      "loss": 0.0806,
      "step": 36050
    },
    {
      "epoch": 3.6725033085615393,
      "grad_norm": 4.963992118835449,
      "learning_rate": 5.9063864443589805e-06,
      "loss": 0.1168,
      "step": 36075
    },
    {
      "epoch": 3.6750483558994196,
      "grad_norm": 3.961651086807251,
      "learning_rate": 5.895074994909848e-06,
      "loss": 0.1051,
      "step": 36100
    },
    {
      "epoch": 3.6775934032373003,
      "grad_norm": 3.3032639026641846,
      "learning_rate": 5.883763545460715e-06,
      "loss": 0.0837,
      "step": 36125
    },
    {
      "epoch": 3.6801384505751806,
      "grad_norm": 8.002496719360352,
      "learning_rate": 5.872452096011584e-06,
      "loss": 0.0634,
      "step": 36150
    },
    {
      "epoch": 3.682683497913061,
      "grad_norm": 3.493532419204712,
      "learning_rate": 5.861140646562451e-06,
      "loss": 0.0702,
      "step": 36175
    },
    {
      "epoch": 3.6852285452509417,
      "grad_norm": 5.555636405944824,
      "learning_rate": 5.849829197113318e-06,
      "loss": 0.0892,
      "step": 36200
    },
    {
      "epoch": 3.6877735925888224,
      "grad_norm": 0.9933393597602844,
      "learning_rate": 5.838517747664187e-06,
      "loss": 0.073,
      "step": 36225
    },
    {
      "epoch": 3.6903186399267027,
      "grad_norm": 3.9761369228363037,
      "learning_rate": 5.827206298215054e-06,
      "loss": 0.0872,
      "step": 36250
    },
    {
      "epoch": 3.692863687264583,
      "grad_norm": 0.9500647783279419,
      "learning_rate": 5.8158948487659215e-06,
      "loss": 0.102,
      "step": 36275
    },
    {
      "epoch": 3.6954087346024638,
      "grad_norm": 0.9967595934867859,
      "learning_rate": 5.8045833993167885e-06,
      "loss": 0.098,
      "step": 36300
    },
    {
      "epoch": 3.697953781940344,
      "grad_norm": 1.624714732170105,
      "learning_rate": 5.793271949867657e-06,
      "loss": 0.0982,
      "step": 36325
    },
    {
      "epoch": 3.7004988292782244,
      "grad_norm": 0.7662064433097839,
      "learning_rate": 5.781960500418524e-06,
      "loss": 0.0527,
      "step": 36350
    },
    {
      "epoch": 3.703043876616105,
      "grad_norm": 7.585689544677734,
      "learning_rate": 5.770649050969391e-06,
      "loss": 0.0883,
      "step": 36375
    },
    {
      "epoch": 3.7055889239539854,
      "grad_norm": 8.151590347290039,
      "learning_rate": 5.75933760152026e-06,
      "loss": 0.0966,
      "step": 36400
    },
    {
      "epoch": 3.708133971291866,
      "grad_norm": 4.967852592468262,
      "learning_rate": 5.748026152071127e-06,
      "loss": 0.0637,
      "step": 36425
    },
    {
      "epoch": 3.7106790186297465,
      "grad_norm": 4.405301094055176,
      "learning_rate": 5.736714702621995e-06,
      "loss": 0.0837,
      "step": 36450
    },
    {
      "epoch": 3.713224065967627,
      "grad_norm": 2.9060306549072266,
      "learning_rate": 5.725403253172862e-06,
      "loss": 0.0747,
      "step": 36475
    },
    {
      "epoch": 3.7157691133055075,
      "grad_norm": 1.2156260013580322,
      "learning_rate": 5.71409180372373e-06,
      "loss": 0.0524,
      "step": 36500
    },
    {
      "epoch": 3.718314160643388,
      "grad_norm": 5.655495643615723,
      "learning_rate": 5.7027803542745974e-06,
      "loss": 0.094,
      "step": 36525
    },
    {
      "epoch": 3.7208592079812686,
      "grad_norm": 1.6090160608291626,
      "learning_rate": 5.6914689048254644e-06,
      "loss": 0.0596,
      "step": 36550
    },
    {
      "epoch": 3.723404255319149,
      "grad_norm": 4.2439680099487305,
      "learning_rate": 5.680157455376333e-06,
      "loss": 0.0859,
      "step": 36575
    },
    {
      "epoch": 3.725949302657029,
      "grad_norm": 1.56519615650177,
      "learning_rate": 5.6688460059272e-06,
      "loss": 0.075,
      "step": 36600
    },
    {
      "epoch": 3.72849434999491,
      "grad_norm": 4.301578521728516,
      "learning_rate": 5.657534556478067e-06,
      "loss": 0.0861,
      "step": 36625
    },
    {
      "epoch": 3.7310393973327907,
      "grad_norm": 2.4486348628997803,
      "learning_rate": 5.646223107028935e-06,
      "loss": 0.0569,
      "step": 36650
    },
    {
      "epoch": 3.733584444670671,
      "grad_norm": 4.346898078918457,
      "learning_rate": 5.634911657579803e-06,
      "loss": 0.0641,
      "step": 36675
    },
    {
      "epoch": 3.7361294920085513,
      "grad_norm": 1.4785878658294678,
      "learning_rate": 5.623600208130671e-06,
      "loss": 0.0787,
      "step": 36700
    },
    {
      "epoch": 3.738674539346432,
      "grad_norm": 1.4040265083312988,
      "learning_rate": 5.612288758681538e-06,
      "loss": 0.0642,
      "step": 36725
    },
    {
      "epoch": 3.7412195866843123,
      "grad_norm": 0.16664713621139526,
      "learning_rate": 5.600977309232406e-06,
      "loss": 0.0821,
      "step": 36750
    },
    {
      "epoch": 3.7437646340221926,
      "grad_norm": 5.200014591217041,
      "learning_rate": 5.589665859783273e-06,
      "loss": 0.0636,
      "step": 36775
    },
    {
      "epoch": 3.7463096813600734,
      "grad_norm": 0.5989975929260254,
      "learning_rate": 5.57835441033414e-06,
      "loss": 0.0769,
      "step": 36800
    },
    {
      "epoch": 3.7488547286979537,
      "grad_norm": 2.7018861770629883,
      "learning_rate": 5.567042960885008e-06,
      "loss": 0.0983,
      "step": 36825
    },
    {
      "epoch": 3.7513997760358344,
      "grad_norm": 4.699031829833984,
      "learning_rate": 5.555731511435876e-06,
      "loss": 0.0821,
      "step": 36850
    },
    {
      "epoch": 3.7539448233737147,
      "grad_norm": 6.736579895019531,
      "learning_rate": 5.544420061986744e-06,
      "loss": 0.0777,
      "step": 36875
    },
    {
      "epoch": 3.7564898707115955,
      "grad_norm": 4.7765350341796875,
      "learning_rate": 5.533108612537611e-06,
      "loss": 0.0798,
      "step": 36900
    },
    {
      "epoch": 3.7590349180494758,
      "grad_norm": 3.2896828651428223,
      "learning_rate": 5.521797163088478e-06,
      "loss": 0.0885,
      "step": 36925
    },
    {
      "epoch": 3.761579965387356,
      "grad_norm": 4.15386438369751,
      "learning_rate": 5.5104857136393466e-06,
      "loss": 0.0832,
      "step": 36950
    },
    {
      "epoch": 3.764125012725237,
      "grad_norm": 1.831699252128601,
      "learning_rate": 5.4991742641902136e-06,
      "loss": 0.0762,
      "step": 36975
    },
    {
      "epoch": 3.766670060063117,
      "grad_norm": 2.754887580871582,
      "learning_rate": 5.487862814741081e-06,
      "loss": 0.1048,
      "step": 37000
    },
    {
      "epoch": 3.7692151074009974,
      "grad_norm": 6.482685565948486,
      "learning_rate": 5.476551365291949e-06,
      "loss": 0.0884,
      "step": 37025
    },
    {
      "epoch": 3.771760154738878,
      "grad_norm": 5.080991268157959,
      "learning_rate": 5.465239915842816e-06,
      "loss": 0.0967,
      "step": 37050
    },
    {
      "epoch": 3.774305202076759,
      "grad_norm": 5.668878078460693,
      "learning_rate": 5.453928466393684e-06,
      "loss": 0.0947,
      "step": 37075
    },
    {
      "epoch": 3.776850249414639,
      "grad_norm": 2.5355238914489746,
      "learning_rate": 5.442617016944551e-06,
      "loss": 0.0829,
      "step": 37100
    },
    {
      "epoch": 3.7793952967525195,
      "grad_norm": 4.266459941864014,
      "learning_rate": 5.43130556749542e-06,
      "loss": 0.1081,
      "step": 37125
    },
    {
      "epoch": 3.7819403440904003,
      "grad_norm": 5.158119201660156,
      "learning_rate": 5.419994118046287e-06,
      "loss": 0.0677,
      "step": 37150
    },
    {
      "epoch": 3.7844853914282806,
      "grad_norm": 3.6510822772979736,
      "learning_rate": 5.408682668597155e-06,
      "loss": 0.0542,
      "step": 37175
    },
    {
      "epoch": 3.787030438766161,
      "grad_norm": 4.6219892501831055,
      "learning_rate": 5.3973712191480225e-06,
      "loss": 0.0905,
      "step": 37200
    },
    {
      "epoch": 3.7895754861040416,
      "grad_norm": 1.2882821559906006,
      "learning_rate": 5.3860597696988895e-06,
      "loss": 0.0844,
      "step": 37225
    },
    {
      "epoch": 3.792120533441922,
      "grad_norm": 1.5892689228057861,
      "learning_rate": 5.374748320249757e-06,
      "loss": 0.081,
      "step": 37250
    },
    {
      "epoch": 3.7946655807798026,
      "grad_norm": 2.733649730682373,
      "learning_rate": 5.363436870800624e-06,
      "loss": 0.085,
      "step": 37275
    },
    {
      "epoch": 3.797210628117683,
      "grad_norm": 0.7701331377029419,
      "learning_rate": 5.352125421351493e-06,
      "loss": 0.104,
      "step": 37300
    },
    {
      "epoch": 3.7997556754555637,
      "grad_norm": 2.512700080871582,
      "learning_rate": 5.34081397190236e-06,
      "loss": 0.0711,
      "step": 37325
    },
    {
      "epoch": 3.802300722793444,
      "grad_norm": 4.5707879066467285,
      "learning_rate": 5.329502522453227e-06,
      "loss": 0.0774,
      "step": 37350
    },
    {
      "epoch": 3.8048457701313243,
      "grad_norm": 8.487582206726074,
      "learning_rate": 5.318191073004096e-06,
      "loss": 0.0951,
      "step": 37375
    },
    {
      "epoch": 3.807390817469205,
      "grad_norm": 2.3424673080444336,
      "learning_rate": 5.306879623554963e-06,
      "loss": 0.0613,
      "step": 37400
    },
    {
      "epoch": 3.8099358648070853,
      "grad_norm": 1.476176381111145,
      "learning_rate": 5.2955681741058305e-06,
      "loss": 0.0795,
      "step": 37425
    },
    {
      "epoch": 3.8124809121449657,
      "grad_norm": 1.8778008222579956,
      "learning_rate": 5.2842567246566975e-06,
      "loss": 0.0954,
      "step": 37450
    },
    {
      "epoch": 3.8150259594828464,
      "grad_norm": 3.7538087368011475,
      "learning_rate": 5.272945275207566e-06,
      "loss": 0.1211,
      "step": 37475
    },
    {
      "epoch": 3.817571006820727,
      "grad_norm": 2.788107395172119,
      "learning_rate": 5.261633825758433e-06,
      "loss": 0.0835,
      "step": 37500
    },
    {
      "epoch": 3.8201160541586074,
      "grad_norm": 5.375553131103516,
      "learning_rate": 5.2503223763093e-06,
      "loss": 0.0829,
      "step": 37525
    },
    {
      "epoch": 3.8226611014964877,
      "grad_norm": 1.8406885862350464,
      "learning_rate": 5.239010926860169e-06,
      "loss": 0.0862,
      "step": 37550
    },
    {
      "epoch": 3.8252061488343685,
      "grad_norm": 1.745599389076233,
      "learning_rate": 5.227699477411036e-06,
      "loss": 0.0696,
      "step": 37575
    },
    {
      "epoch": 3.827751196172249,
      "grad_norm": 4.49993371963501,
      "learning_rate": 5.216388027961904e-06,
      "loss": 0.0963,
      "step": 37600
    },
    {
      "epoch": 3.830296243510129,
      "grad_norm": 2.113429307937622,
      "learning_rate": 5.205076578512771e-06,
      "loss": 0.0757,
      "step": 37625
    },
    {
      "epoch": 3.83284129084801,
      "grad_norm": 3.142502546310425,
      "learning_rate": 5.193765129063638e-06,
      "loss": 0.0781,
      "step": 37650
    },
    {
      "epoch": 3.83538633818589,
      "grad_norm": 2.529702663421631,
      "learning_rate": 5.1824536796145065e-06,
      "loss": 0.0585,
      "step": 37675
    },
    {
      "epoch": 3.837931385523771,
      "grad_norm": 1.782511591911316,
      "learning_rate": 5.1711422301653735e-06,
      "loss": 0.0915,
      "step": 37700
    },
    {
      "epoch": 3.840476432861651,
      "grad_norm": 3.4654805660247803,
      "learning_rate": 5.159830780716242e-06,
      "loss": 0.1021,
      "step": 37725
    },
    {
      "epoch": 3.843021480199532,
      "grad_norm": 7.175243854522705,
      "learning_rate": 5.148519331267109e-06,
      "loss": 0.0943,
      "step": 37750
    },
    {
      "epoch": 3.8455665275374122,
      "grad_norm": 3.8063061237335205,
      "learning_rate": 5.137207881817976e-06,
      "loss": 0.0666,
      "step": 37775
    },
    {
      "epoch": 3.8481115748752925,
      "grad_norm": 0.8635425567626953,
      "learning_rate": 5.125896432368844e-06,
      "loss": 0.0564,
      "step": 37800
    },
    {
      "epoch": 3.8506566222131733,
      "grad_norm": 3.5335941314697266,
      "learning_rate": 5.114584982919711e-06,
      "loss": 0.0717,
      "step": 37825
    },
    {
      "epoch": 3.8532016695510536,
      "grad_norm": 3.7828986644744873,
      "learning_rate": 5.10327353347058e-06,
      "loss": 0.0861,
      "step": 37850
    },
    {
      "epoch": 3.855746716888934,
      "grad_norm": 1.944362759590149,
      "learning_rate": 5.091962084021447e-06,
      "loss": 0.0611,
      "step": 37875
    },
    {
      "epoch": 3.8582917642268146,
      "grad_norm": 2.3826661109924316,
      "learning_rate": 5.080650634572315e-06,
      "loss": 0.0868,
      "step": 37900
    },
    {
      "epoch": 3.860836811564695,
      "grad_norm": 0.43980684876441956,
      "learning_rate": 5.069339185123182e-06,
      "loss": 0.0844,
      "step": 37925
    },
    {
      "epoch": 3.8633818589025757,
      "grad_norm": 1.001634120941162,
      "learning_rate": 5.058027735674049e-06,
      "loss": 0.0842,
      "step": 37950
    },
    {
      "epoch": 3.865926906240456,
      "grad_norm": 2.313572883605957,
      "learning_rate": 5.046716286224917e-06,
      "loss": 0.0689,
      "step": 37975
    },
    {
      "epoch": 3.8684719535783367,
      "grad_norm": 2.2224433422088623,
      "learning_rate": 5.035404836775784e-06,
      "loss": 0.0747,
      "step": 38000
    },
    {
      "epoch": 3.871017000916217,
      "grad_norm": 3.743778944015503,
      "learning_rate": 5.024093387326653e-06,
      "loss": 0.0557,
      "step": 38025
    },
    {
      "epoch": 3.8735620482540973,
      "grad_norm": 5.278766632080078,
      "learning_rate": 5.01278193787752e-06,
      "loss": 0.0761,
      "step": 38050
    },
    {
      "epoch": 3.876107095591978,
      "grad_norm": 6.019226551055908,
      "learning_rate": 5.001470488428387e-06,
      "loss": 0.0843,
      "step": 38075
    },
    {
      "epoch": 3.8786521429298584,
      "grad_norm": 1.7229478359222412,
      "learning_rate": 4.990159038979256e-06,
      "loss": 0.099,
      "step": 38100
    },
    {
      "epoch": 3.881197190267739,
      "grad_norm": 4.364188194274902,
      "learning_rate": 4.9788475895301234e-06,
      "loss": 0.0812,
      "step": 38125
    },
    {
      "epoch": 3.8837422376056194,
      "grad_norm": 0.27184224128723145,
      "learning_rate": 4.9675361400809904e-06,
      "loss": 0.0668,
      "step": 38150
    },
    {
      "epoch": 3.8862872849435,
      "grad_norm": 1.0122255086898804,
      "learning_rate": 4.9562246906318574e-06,
      "loss": 0.074,
      "step": 38175
    },
    {
      "epoch": 3.8888323322813805,
      "grad_norm": 4.959586143493652,
      "learning_rate": 4.944913241182725e-06,
      "loss": 0.0933,
      "step": 38200
    },
    {
      "epoch": 3.891377379619261,
      "grad_norm": 5.364275932312012,
      "learning_rate": 4.933601791733593e-06,
      "loss": 0.1152,
      "step": 38225
    },
    {
      "epoch": 3.8939224269571415,
      "grad_norm": 3.487792491912842,
      "learning_rate": 4.922290342284461e-06,
      "loss": 0.0639,
      "step": 38250
    },
    {
      "epoch": 3.896467474295022,
      "grad_norm": 6.711452007293701,
      "learning_rate": 4.911431350813294e-06,
      "loss": 0.0997,
      "step": 38275
    },
    {
      "epoch": 3.899012521632902,
      "grad_norm": 3.9726145267486572,
      "learning_rate": 4.900119901364161e-06,
      "loss": 0.0741,
      "step": 38300
    },
    {
      "epoch": 3.901557568970783,
      "grad_norm": 6.3305277824401855,
      "learning_rate": 4.8888084519150285e-06,
      "loss": 0.0705,
      "step": 38325
    },
    {
      "epoch": 3.904102616308663,
      "grad_norm": 1.8397938013076782,
      "learning_rate": 4.877497002465896e-06,
      "loss": 0.0663,
      "step": 38350
    },
    {
      "epoch": 3.906647663646544,
      "grad_norm": 1.2621251344680786,
      "learning_rate": 4.866185553016764e-06,
      "loss": 0.0659,
      "step": 38375
    },
    {
      "epoch": 3.9091927109844242,
      "grad_norm": 1.6359314918518066,
      "learning_rate": 4.854874103567632e-06,
      "loss": 0.0894,
      "step": 38400
    },
    {
      "epoch": 3.911737758322305,
      "grad_norm": 1.9118399620056152,
      "learning_rate": 4.843562654118499e-06,
      "loss": 0.0752,
      "step": 38425
    },
    {
      "epoch": 3.9142828056601853,
      "grad_norm": 0.7467500567436218,
      "learning_rate": 4.832251204669367e-06,
      "loss": 0.0588,
      "step": 38450
    },
    {
      "epoch": 3.9168278529980656,
      "grad_norm": 1.7692584991455078,
      "learning_rate": 4.820939755220234e-06,
      "loss": 0.0914,
      "step": 38475
    },
    {
      "epoch": 3.9193729003359463,
      "grad_norm": 2.5429513454437256,
      "learning_rate": 4.809628305771102e-06,
      "loss": 0.1153,
      "step": 38500
    },
    {
      "epoch": 3.9219179476738266,
      "grad_norm": 1.0103989839553833,
      "learning_rate": 4.7983168563219695e-06,
      "loss": 0.0536,
      "step": 38525
    },
    {
      "epoch": 3.9244629950117074,
      "grad_norm": 4.339071273803711,
      "learning_rate": 4.787005406872837e-06,
      "loss": 0.0906,
      "step": 38550
    },
    {
      "epoch": 3.9270080423495877,
      "grad_norm": 0.8115999698638916,
      "learning_rate": 4.775693957423705e-06,
      "loss": 0.0871,
      "step": 38575
    },
    {
      "epoch": 3.9295530896874684,
      "grad_norm": 1.1291712522506714,
      "learning_rate": 4.764382507974572e-06,
      "loss": 0.0968,
      "step": 38600
    },
    {
      "epoch": 3.9320981370253487,
      "grad_norm": 2.7206923961639404,
      "learning_rate": 4.75307105852544e-06,
      "loss": 0.0918,
      "step": 38625
    },
    {
      "epoch": 3.934643184363229,
      "grad_norm": 0.8271558880805969,
      "learning_rate": 4.741759609076307e-06,
      "loss": 0.0939,
      "step": 38650
    },
    {
      "epoch": 3.9371882317011098,
      "grad_norm": 0.7608330845832825,
      "learning_rate": 4.730448159627175e-06,
      "loss": 0.0861,
      "step": 38675
    },
    {
      "epoch": 3.93973327903899,
      "grad_norm": 0.4039336144924164,
      "learning_rate": 4.719136710178043e-06,
      "loss": 0.0662,
      "step": 38700
    },
    {
      "epoch": 3.9422783263768704,
      "grad_norm": 5.616768836975098,
      "learning_rate": 4.708277718706875e-06,
      "loss": 0.0845,
      "step": 38725
    },
    {
      "epoch": 3.944823373714751,
      "grad_norm": 2.859184503555298,
      "learning_rate": 4.696966269257742e-06,
      "loss": 0.0843,
      "step": 38750
    },
    {
      "epoch": 3.9473684210526314,
      "grad_norm": 7.883376598358154,
      "learning_rate": 4.68565481980861e-06,
      "loss": 0.0902,
      "step": 38775
    },
    {
      "epoch": 3.949913468390512,
      "grad_norm": 5.071816444396973,
      "learning_rate": 4.674343370359478e-06,
      "loss": 0.1097,
      "step": 38800
    },
    {
      "epoch": 3.9524585157283925,
      "grad_norm": 4.740415096282959,
      "learning_rate": 4.663031920910346e-06,
      "loss": 0.0721,
      "step": 38825
    },
    {
      "epoch": 3.955003563066273,
      "grad_norm": 0.4243851602077484,
      "learning_rate": 4.651720471461214e-06,
      "loss": 0.1045,
      "step": 38850
    },
    {
      "epoch": 3.9575486104041535,
      "grad_norm": 3.7491915225982666,
      "learning_rate": 4.640409022012082e-06,
      "loss": 0.0846,
      "step": 38875
    },
    {
      "epoch": 3.960093657742034,
      "grad_norm": 2.330118417739868,
      "learning_rate": 4.629097572562949e-06,
      "loss": 0.0772,
      "step": 38900
    },
    {
      "epoch": 3.9626387050799146,
      "grad_norm": 2.0773494243621826,
      "learning_rate": 4.617786123113816e-06,
      "loss": 0.0605,
      "step": 38925
    },
    {
      "epoch": 3.965183752417795,
      "grad_norm": 1.8872305154800415,
      "learning_rate": 4.6064746736646835e-06,
      "loss": 0.076,
      "step": 38950
    },
    {
      "epoch": 3.9677287997556756,
      "grad_norm": 4.278075218200684,
      "learning_rate": 4.595163224215551e-06,
      "loss": 0.0584,
      "step": 38975
    },
    {
      "epoch": 3.970273847093556,
      "grad_norm": 1.7767386436462402,
      "learning_rate": 4.583851774766419e-06,
      "loss": 0.076,
      "step": 39000
    },
    {
      "epoch": 3.9728188944314367,
      "grad_norm": 0.9367185235023499,
      "learning_rate": 4.572540325317287e-06,
      "loss": 0.0854,
      "step": 39025
    },
    {
      "epoch": 3.975363941769317,
      "grad_norm": 4.025914669036865,
      "learning_rate": 4.561228875868154e-06,
      "loss": 0.0836,
      "step": 39050
    },
    {
      "epoch": 3.9779089891071973,
      "grad_norm": 0.49854645133018494,
      "learning_rate": 4.549917426419022e-06,
      "loss": 0.0704,
      "step": 39075
    },
    {
      "epoch": 3.980454036445078,
      "grad_norm": 2.95552921295166,
      "learning_rate": 4.538605976969889e-06,
      "loss": 0.081,
      "step": 39100
    },
    {
      "epoch": 3.9829990837829583,
      "grad_norm": 3.494931697845459,
      "learning_rate": 4.527294527520757e-06,
      "loss": 0.0846,
      "step": 39125
    },
    {
      "epoch": 3.9855441311208386,
      "grad_norm": 2.842068910598755,
      "learning_rate": 4.5159830780716245e-06,
      "loss": 0.056,
      "step": 39150
    },
    {
      "epoch": 3.9880891784587194,
      "grad_norm": 1.898389458656311,
      "learning_rate": 4.504671628622492e-06,
      "loss": 0.063,
      "step": 39175
    },
    {
      "epoch": 3.9906342257965997,
      "grad_norm": 3.592628240585327,
      "learning_rate": 4.493360179173359e-06,
      "loss": 0.0935,
      "step": 39200
    },
    {
      "epoch": 3.9931792731344804,
      "grad_norm": 1.6398416757583618,
      "learning_rate": 4.482048729724227e-06,
      "loss": 0.0895,
      "step": 39225
    },
    {
      "epoch": 3.9957243204723607,
      "grad_norm": 3.1050050258636475,
      "learning_rate": 4.470737280275095e-06,
      "loss": 0.0748,
      "step": 39250
    },
    {
      "epoch": 3.9982693678102414,
      "grad_norm": 3.505990505218506,
      "learning_rate": 4.459425830825962e-06,
      "loss": 0.0704,
      "step": 39275
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.08321376889944077,
      "eval_runtime": 7.2624,
      "eval_samples_per_second": 937.43,
      "eval_steps_per_second": 14.733,
      "step": 39292
    },
    {
      "epoch": 4.000814415148122,
      "grad_norm": 4.233890056610107,
      "learning_rate": 4.44811438137683e-06,
      "loss": 0.0657,
      "step": 39300
    },
    {
      "epoch": 4.003359462486002,
      "grad_norm": 6.2566704750061035,
      "learning_rate": 4.436802931927698e-06,
      "loss": 0.0788,
      "step": 39325
    },
    {
      "epoch": 4.005904509823883,
      "grad_norm": 6.7541704177856445,
      "learning_rate": 4.425491482478565e-06,
      "loss": 0.0847,
      "step": 39350
    },
    {
      "epoch": 4.0084495571617635,
      "grad_norm": 2.7961132526397705,
      "learning_rate": 4.414180033029433e-06,
      "loss": 0.0667,
      "step": 39375
    },
    {
      "epoch": 4.010994604499643,
      "grad_norm": 3.7333006858825684,
      "learning_rate": 4.4028685835803005e-06,
      "loss": 0.0646,
      "step": 39400
    },
    {
      "epoch": 4.013539651837524,
      "grad_norm": 5.583388805389404,
      "learning_rate": 4.391557134131168e-06,
      "loss": 0.0528,
      "step": 39425
    },
    {
      "epoch": 4.016084699175405,
      "grad_norm": 0.5492544174194336,
      "learning_rate": 4.380245684682035e-06,
      "loss": 0.0563,
      "step": 39450
    },
    {
      "epoch": 4.018629746513285,
      "grad_norm": 3.7247314453125,
      "learning_rate": 4.368934235232903e-06,
      "loss": 0.0723,
      "step": 39475
    },
    {
      "epoch": 4.0211747938511655,
      "grad_norm": 1.7265299558639526,
      "learning_rate": 4.35762278578377e-06,
      "loss": 0.0703,
      "step": 39500
    },
    {
      "epoch": 4.023719841189046,
      "grad_norm": 4.306021690368652,
      "learning_rate": 4.346311336334638e-06,
      "loss": 0.0788,
      "step": 39525
    },
    {
      "epoch": 4.026264888526927,
      "grad_norm": 1.109270453453064,
      "learning_rate": 4.334999886885506e-06,
      "loss": 0.0572,
      "step": 39550
    },
    {
      "epoch": 4.028809935864807,
      "grad_norm": 1.2954659461975098,
      "learning_rate": 4.323688437436374e-06,
      "loss": 0.0757,
      "step": 39575
    },
    {
      "epoch": 4.031354983202688,
      "grad_norm": 1.4971513748168945,
      "learning_rate": 4.3123769879872415e-06,
      "loss": 0.0593,
      "step": 39600
    },
    {
      "epoch": 4.033900030540568,
      "grad_norm": 1.8888990879058838,
      "learning_rate": 4.3010655385381085e-06,
      "loss": 0.0739,
      "step": 39625
    },
    {
      "epoch": 4.036445077878448,
      "grad_norm": 4.024899482727051,
      "learning_rate": 4.289754089088976e-06,
      "loss": 0.06,
      "step": 39650
    },
    {
      "epoch": 4.038990125216329,
      "grad_norm": 3.701951503753662,
      "learning_rate": 4.278442639639843e-06,
      "loss": 0.0736,
      "step": 39675
    },
    {
      "epoch": 4.04153517255421,
      "grad_norm": 3.4520320892333984,
      "learning_rate": 4.267131190190711e-06,
      "loss": 0.0743,
      "step": 39700
    },
    {
      "epoch": 4.04408021989209,
      "grad_norm": 2.047182083129883,
      "learning_rate": 4.255819740741579e-06,
      "loss": 0.0655,
      "step": 39725
    },
    {
      "epoch": 4.04662526722997,
      "grad_norm": 7.164806365966797,
      "learning_rate": 4.244508291292447e-06,
      "loss": 0.0398,
      "step": 39750
    },
    {
      "epoch": 4.049170314567851,
      "grad_norm": 1.4133317470550537,
      "learning_rate": 4.233196841843314e-06,
      "loss": 0.0622,
      "step": 39775
    },
    {
      "epoch": 4.051715361905732,
      "grad_norm": 2.1031320095062256,
      "learning_rate": 4.221885392394182e-06,
      "loss": 0.0774,
      "step": 39800
    },
    {
      "epoch": 4.054260409243612,
      "grad_norm": 2.162343740463257,
      "learning_rate": 4.21057394294505e-06,
      "loss": 0.074,
      "step": 39825
    },
    {
      "epoch": 4.056805456581492,
      "grad_norm": 7.961151123046875,
      "learning_rate": 4.199262493495917e-06,
      "loss": 0.0537,
      "step": 39850
    },
    {
      "epoch": 4.059350503919373,
      "grad_norm": 3.6725287437438965,
      "learning_rate": 4.1879510440467844e-06,
      "loss": 0.0635,
      "step": 39875
    },
    {
      "epoch": 4.061895551257253,
      "grad_norm": 2.4533803462982178,
      "learning_rate": 4.176639594597652e-06,
      "loss": 0.0578,
      "step": 39900
    },
    {
      "epoch": 4.064440598595134,
      "grad_norm": 7.141175270080566,
      "learning_rate": 4.165328145148519e-06,
      "loss": 0.0987,
      "step": 39925
    },
    {
      "epoch": 4.0669856459330145,
      "grad_norm": 0.6580816507339478,
      "learning_rate": 4.154016695699387e-06,
      "loss": 0.0931,
      "step": 39950
    },
    {
      "epoch": 4.069530693270895,
      "grad_norm": 2.766205310821533,
      "learning_rate": 4.142705246250255e-06,
      "loss": 0.0899,
      "step": 39975
    },
    {
      "epoch": 4.072075740608775,
      "grad_norm": 4.015225410461426,
      "learning_rate": 4.131393796801123e-06,
      "loss": 0.0797,
      "step": 40000
    },
    {
      "epoch": 4.074620787946656,
      "grad_norm": 5.227399826049805,
      "learning_rate": 4.12008234735199e-06,
      "loss": 0.0803,
      "step": 40025
    },
    {
      "epoch": 4.077165835284537,
      "grad_norm": 1.7184827327728271,
      "learning_rate": 4.108770897902858e-06,
      "loss": 0.0851,
      "step": 40050
    },
    {
      "epoch": 4.079710882622416,
      "grad_norm": 3.887936592102051,
      "learning_rate": 4.097459448453725e-06,
      "loss": 0.0826,
      "step": 40075
    },
    {
      "epoch": 4.082255929960297,
      "grad_norm": 3.3515472412109375,
      "learning_rate": 4.0861479990045925e-06,
      "loss": 0.0808,
      "step": 40100
    },
    {
      "epoch": 4.084800977298178,
      "grad_norm": 1.4721667766571045,
      "learning_rate": 4.07483654955546e-06,
      "loss": 0.0734,
      "step": 40125
    },
    {
      "epoch": 4.087346024636059,
      "grad_norm": 3.627595901489258,
      "learning_rate": 4.063525100106328e-06,
      "loss": 0.0634,
      "step": 40150
    },
    {
      "epoch": 4.0898910719739385,
      "grad_norm": 0.0573771595954895,
      "learning_rate": 4.052213650657196e-06,
      "loss": 0.0792,
      "step": 40175
    },
    {
      "epoch": 4.092436119311819,
      "grad_norm": 2.682100772857666,
      "learning_rate": 4.040902201208063e-06,
      "loss": 0.0921,
      "step": 40200
    },
    {
      "epoch": 4.0949811666497,
      "grad_norm": 0.7701199054718018,
      "learning_rate": 4.029590751758931e-06,
      "loss": 0.0754,
      "step": 40225
    },
    {
      "epoch": 4.09752621398758,
      "grad_norm": 8.216882705688477,
      "learning_rate": 4.018279302309798e-06,
      "loss": 0.0925,
      "step": 40250
    },
    {
      "epoch": 4.100071261325461,
      "grad_norm": 5.027645111083984,
      "learning_rate": 4.006967852860666e-06,
      "loss": 0.0777,
      "step": 40275
    },
    {
      "epoch": 4.102616308663341,
      "grad_norm": 0.8864279985427856,
      "learning_rate": 3.9956564034115336e-06,
      "loss": 0.0746,
      "step": 40300
    },
    {
      "epoch": 4.105161356001221,
      "grad_norm": 1.5097590684890747,
      "learning_rate": 3.984344953962401e-06,
      "loss": 0.058,
      "step": 40325
    },
    {
      "epoch": 4.107706403339102,
      "grad_norm": 4.6194963455200195,
      "learning_rate": 3.973033504513268e-06,
      "loss": 0.0549,
      "step": 40350
    },
    {
      "epoch": 4.110251450676983,
      "grad_norm": 1.5456056594848633,
      "learning_rate": 3.961722055064136e-06,
      "loss": 0.0958,
      "step": 40375
    },
    {
      "epoch": 4.1127964980148635,
      "grad_norm": 6.5456647872924805,
      "learning_rate": 3.950410605615004e-06,
      "loss": 0.1041,
      "step": 40400
    },
    {
      "epoch": 4.115341545352743,
      "grad_norm": 3.12103009223938,
      "learning_rate": 3.939099156165871e-06,
      "loss": 0.0841,
      "step": 40425
    },
    {
      "epoch": 4.117886592690624,
      "grad_norm": 1.520664095878601,
      "learning_rate": 3.927787706716739e-06,
      "loss": 0.1077,
      "step": 40450
    },
    {
      "epoch": 4.120431640028505,
      "grad_norm": 2.8787078857421875,
      "learning_rate": 3.916476257267607e-06,
      "loss": 0.0653,
      "step": 40475
    },
    {
      "epoch": 4.122976687366385,
      "grad_norm": 0.7856115102767944,
      "learning_rate": 3.905164807818474e-06,
      "loss": 0.065,
      "step": 40500
    },
    {
      "epoch": 4.125521734704265,
      "grad_norm": 1.089637041091919,
      "learning_rate": 3.893853358369342e-06,
      "loss": 0.094,
      "step": 40525
    },
    {
      "epoch": 4.128066782042146,
      "grad_norm": 4.611286163330078,
      "learning_rate": 3.8825419089202095e-06,
      "loss": 0.0612,
      "step": 40550
    },
    {
      "epoch": 4.130611829380026,
      "grad_norm": 2.995384693145752,
      "learning_rate": 3.871230459471077e-06,
      "loss": 0.0677,
      "step": 40575
    },
    {
      "epoch": 4.133156876717907,
      "grad_norm": 2.4871647357940674,
      "learning_rate": 3.859919010021944e-06,
      "loss": 0.092,
      "step": 40600
    },
    {
      "epoch": 4.1357019240557875,
      "grad_norm": 0.4613381028175354,
      "learning_rate": 3.848607560572812e-06,
      "loss": 0.0927,
      "step": 40625
    },
    {
      "epoch": 4.138246971393668,
      "grad_norm": 4.346590042114258,
      "learning_rate": 3.837296111123679e-06,
      "loss": 0.073,
      "step": 40650
    },
    {
      "epoch": 4.140792018731548,
      "grad_norm": 4.693612575531006,
      "learning_rate": 3.825984661674547e-06,
      "loss": 0.0581,
      "step": 40675
    },
    {
      "epoch": 4.143337066069429,
      "grad_norm": 1.904029130935669,
      "learning_rate": 3.814673212225415e-06,
      "loss": 0.0616,
      "step": 40700
    },
    {
      "epoch": 4.14588211340731,
      "grad_norm": 4.891720294952393,
      "learning_rate": 3.8033617627762827e-06,
      "loss": 0.0797,
      "step": 40725
    },
    {
      "epoch": 4.1484271607451895,
      "grad_norm": 1.2688778638839722,
      "learning_rate": 3.79205031332715e-06,
      "loss": 0.0869,
      "step": 40750
    },
    {
      "epoch": 4.15097220808307,
      "grad_norm": 4.826283931732178,
      "learning_rate": 3.7807388638780175e-06,
      "loss": 0.0808,
      "step": 40775
    },
    {
      "epoch": 4.153517255420951,
      "grad_norm": 8.116747856140137,
      "learning_rate": 3.769427414428885e-06,
      "loss": 0.0762,
      "step": 40800
    },
    {
      "epoch": 4.156062302758832,
      "grad_norm": 5.6922807693481445,
      "learning_rate": 3.758568422957718e-06,
      "loss": 0.0831,
      "step": 40825
    },
    {
      "epoch": 4.158607350096712,
      "grad_norm": 2.4496312141418457,
      "learning_rate": 3.7472569735085855e-06,
      "loss": 0.0775,
      "step": 40850
    },
    {
      "epoch": 4.161152397434592,
      "grad_norm": 3.060380697250366,
      "learning_rate": 3.7359455240594533e-06,
      "loss": 0.0685,
      "step": 40875
    },
    {
      "epoch": 4.163697444772473,
      "grad_norm": 4.143198490142822,
      "learning_rate": 3.724634074610321e-06,
      "loss": 0.055,
      "step": 40900
    },
    {
      "epoch": 4.166242492110353,
      "grad_norm": 5.605116367340088,
      "learning_rate": 3.7133226251611886e-06,
      "loss": 0.0893,
      "step": 40925
    },
    {
      "epoch": 4.168787539448234,
      "grad_norm": 4.88701868057251,
      "learning_rate": 3.702011175712056e-06,
      "loss": 0.0536,
      "step": 40950
    },
    {
      "epoch": 4.171332586786114,
      "grad_norm": 3.8386881351470947,
      "learning_rate": 3.6906997262629234e-06,
      "loss": 0.0943,
      "step": 40975
    },
    {
      "epoch": 4.173877634123995,
      "grad_norm": 0.8875072598457336,
      "learning_rate": 3.6793882768137913e-06,
      "loss": 0.0643,
      "step": 41000
    },
    {
      "epoch": 4.176422681461875,
      "grad_norm": 3.3152079582214355,
      "learning_rate": 3.6680768273646587e-06,
      "loss": 0.0601,
      "step": 41025
    },
    {
      "epoch": 4.178967728799756,
      "grad_norm": 2.363868236541748,
      "learning_rate": 3.6567653779155265e-06,
      "loss": 0.0836,
      "step": 41050
    },
    {
      "epoch": 4.1815127761376365,
      "grad_norm": 1.1816363334655762,
      "learning_rate": 3.6454539284663944e-06,
      "loss": 0.0687,
      "step": 41075
    },
    {
      "epoch": 4.184057823475516,
      "grad_norm": 4.630997180938721,
      "learning_rate": 3.6341424790172614e-06,
      "loss": 0.0577,
      "step": 41100
    },
    {
      "epoch": 4.186602870813397,
      "grad_norm": 3.6776726245880127,
      "learning_rate": 3.6228310295681292e-06,
      "loss": 0.0838,
      "step": 41125
    },
    {
      "epoch": 4.189147918151278,
      "grad_norm": 4.161885738372803,
      "learning_rate": 3.6115195801189966e-06,
      "loss": 0.0694,
      "step": 41150
    },
    {
      "epoch": 4.191692965489158,
      "grad_norm": 3.5846166610717773,
      "learning_rate": 3.6002081306698645e-06,
      "loss": 0.0835,
      "step": 41175
    },
    {
      "epoch": 4.1942380128270385,
      "grad_norm": 3.8520216941833496,
      "learning_rate": 3.588896681220732e-06,
      "loss": 0.0767,
      "step": 41200
    },
    {
      "epoch": 4.196783060164919,
      "grad_norm": 4.335388660430908,
      "learning_rate": 3.5775852317715997e-06,
      "loss": 0.0542,
      "step": 41225
    },
    {
      "epoch": 4.1993281075028,
      "grad_norm": 1.6342220306396484,
      "learning_rate": 3.5662737823224667e-06,
      "loss": 0.0838,
      "step": 41250
    },
    {
      "epoch": 4.20187315484068,
      "grad_norm": 5.2642011642456055,
      "learning_rate": 3.5549623328733346e-06,
      "loss": 0.06,
      "step": 41275
    },
    {
      "epoch": 4.2044182021785605,
      "grad_norm": 3.4589884281158447,
      "learning_rate": 3.5436508834242024e-06,
      "loss": 0.067,
      "step": 41300
    },
    {
      "epoch": 4.206963249516441,
      "grad_norm": 6.109384536743164,
      "learning_rate": 3.53233943397507e-06,
      "loss": 0.0531,
      "step": 41325
    },
    {
      "epoch": 4.209508296854321,
      "grad_norm": 4.634017467498779,
      "learning_rate": 3.5210279845259377e-06,
      "loss": 0.0502,
      "step": 41350
    },
    {
      "epoch": 4.212053344192202,
      "grad_norm": 3.371673107147217,
      "learning_rate": 3.5097165350768047e-06,
      "loss": 0.0748,
      "step": 41375
    },
    {
      "epoch": 4.214598391530083,
      "grad_norm": 1.5770014524459839,
      "learning_rate": 3.4984050856276725e-06,
      "loss": 0.0845,
      "step": 41400
    },
    {
      "epoch": 4.2171434388679625,
      "grad_norm": 5.147093772888184,
      "learning_rate": 3.48709363617854e-06,
      "loss": 0.0725,
      "step": 41425
    },
    {
      "epoch": 4.219688486205843,
      "grad_norm": 0.8180754780769348,
      "learning_rate": 3.475782186729408e-06,
      "loss": 0.07,
      "step": 41450
    },
    {
      "epoch": 4.222233533543724,
      "grad_norm": 0.36477014422416687,
      "learning_rate": 3.4644707372802757e-06,
      "loss": 0.0587,
      "step": 41475
    },
    {
      "epoch": 4.224778580881605,
      "grad_norm": 2.6513445377349854,
      "learning_rate": 3.453159287831143e-06,
      "loss": 0.0811,
      "step": 41500
    },
    {
      "epoch": 4.227323628219485,
      "grad_norm": 8.273859977722168,
      "learning_rate": 3.4418478383820105e-06,
      "loss": 0.0902,
      "step": 41525
    },
    {
      "epoch": 4.229868675557365,
      "grad_norm": 1.1668052673339844,
      "learning_rate": 3.430536388932878e-06,
      "loss": 0.0587,
      "step": 41550
    },
    {
      "epoch": 4.232413722895246,
      "grad_norm": 4.071606636047363,
      "learning_rate": 3.4192249394837458e-06,
      "loss": 0.0882,
      "step": 41575
    },
    {
      "epoch": 4.234958770233126,
      "grad_norm": 2.3633975982666016,
      "learning_rate": 3.407913490034613e-06,
      "loss": 0.0491,
      "step": 41600
    },
    {
      "epoch": 4.237503817571007,
      "grad_norm": 5.4014363288879395,
      "learning_rate": 3.396602040585481e-06,
      "loss": 0.0747,
      "step": 41625
    },
    {
      "epoch": 4.240048864908887,
      "grad_norm": 5.445051193237305,
      "learning_rate": 3.385290591136349e-06,
      "loss": 0.0634,
      "step": 41650
    },
    {
      "epoch": 4.242593912246768,
      "grad_norm": 0.34296977519989014,
      "learning_rate": 3.373979141687216e-06,
      "loss": 0.0718,
      "step": 41675
    },
    {
      "epoch": 4.245138959584648,
      "grad_norm": 5.036444664001465,
      "learning_rate": 3.3626676922380837e-06,
      "loss": 0.0572,
      "step": 41700
    },
    {
      "epoch": 4.247684006922529,
      "grad_norm": 5.49483060836792,
      "learning_rate": 3.351356242788951e-06,
      "loss": 0.0779,
      "step": 41725
    },
    {
      "epoch": 4.2502290542604095,
      "grad_norm": 1.3740614652633667,
      "learning_rate": 3.340044793339819e-06,
      "loss": 0.0938,
      "step": 41750
    },
    {
      "epoch": 4.252774101598289,
      "grad_norm": 2.491666793823242,
      "learning_rate": 3.3287333438906864e-06,
      "loss": 0.0509,
      "step": 41775
    },
    {
      "epoch": 4.25531914893617,
      "grad_norm": 6.901576995849609,
      "learning_rate": 3.3174218944415543e-06,
      "loss": 0.0669,
      "step": 41800
    },
    {
      "epoch": 4.257864196274051,
      "grad_norm": 3.491737127304077,
      "learning_rate": 3.3061104449924213e-06,
      "loss": 0.0862,
      "step": 41825
    },
    {
      "epoch": 4.260409243611932,
      "grad_norm": 1.0229004621505737,
      "learning_rate": 3.294798995543289e-06,
      "loss": 0.0594,
      "step": 41850
    },
    {
      "epoch": 4.2629542909498115,
      "grad_norm": 4.715412139892578,
      "learning_rate": 3.283487546094157e-06,
      "loss": 0.0773,
      "step": 41875
    },
    {
      "epoch": 4.265499338287692,
      "grad_norm": 5.1960530281066895,
      "learning_rate": 3.2721760966450244e-06,
      "loss": 0.0618,
      "step": 41900
    },
    {
      "epoch": 4.268044385625573,
      "grad_norm": 1.639708161354065,
      "learning_rate": 3.260864647195892e-06,
      "loss": 0.0524,
      "step": 41925
    },
    {
      "epoch": 4.270589432963453,
      "grad_norm": 1.5220186710357666,
      "learning_rate": 3.249553197746759e-06,
      "loss": 0.0597,
      "step": 41950
    },
    {
      "epoch": 4.273134480301334,
      "grad_norm": 0.25770363211631775,
      "learning_rate": 3.238241748297627e-06,
      "loss": 0.0994,
      "step": 41975
    },
    {
      "epoch": 4.275679527639214,
      "grad_norm": 2.196244478225708,
      "learning_rate": 3.2269302988484945e-06,
      "loss": 0.055,
      "step": 42000
    },
    {
      "epoch": 4.278224574977094,
      "grad_norm": 0.6241010427474976,
      "learning_rate": 3.2156188493993623e-06,
      "loss": 0.0937,
      "step": 42025
    },
    {
      "epoch": 4.280769622314975,
      "grad_norm": 4.9643096923828125,
      "learning_rate": 3.20430739995023e-06,
      "loss": 0.0678,
      "step": 42050
    },
    {
      "epoch": 4.283314669652856,
      "grad_norm": 1.383632779121399,
      "learning_rate": 3.1929959505010976e-06,
      "loss": 0.0581,
      "step": 42075
    },
    {
      "epoch": 4.285859716990736,
      "grad_norm": 2.5081474781036377,
      "learning_rate": 3.181684501051965e-06,
      "loss": 0.0821,
      "step": 42100
    },
    {
      "epoch": 4.288404764328616,
      "grad_norm": 1.9261081218719482,
      "learning_rate": 3.1703730516028324e-06,
      "loss": 0.088,
      "step": 42125
    },
    {
      "epoch": 4.290949811666497,
      "grad_norm": 2.8013453483581543,
      "learning_rate": 3.1590616021537003e-06,
      "loss": 0.0634,
      "step": 42150
    },
    {
      "epoch": 4.293494859004378,
      "grad_norm": 2.583024501800537,
      "learning_rate": 3.1477501527045677e-06,
      "loss": 0.0826,
      "step": 42175
    },
    {
      "epoch": 4.296039906342258,
      "grad_norm": 2.5302445888519287,
      "learning_rate": 3.1364387032554355e-06,
      "loss": 0.1037,
      "step": 42200
    },
    {
      "epoch": 4.298584953680138,
      "grad_norm": 1.894404649734497,
      "learning_rate": 3.1251272538063034e-06,
      "loss": 0.0693,
      "step": 42225
    },
    {
      "epoch": 4.301130001018019,
      "grad_norm": 4.712514877319336,
      "learning_rate": 3.1138158043571704e-06,
      "loss": 0.074,
      "step": 42250
    },
    {
      "epoch": 4.303675048355899,
      "grad_norm": 6.2421183586120605,
      "learning_rate": 3.1025043549080382e-06,
      "loss": 0.0831,
      "step": 42275
    },
    {
      "epoch": 4.30622009569378,
      "grad_norm": 6.7714924812316895,
      "learning_rate": 3.0911929054589057e-06,
      "loss": 0.0691,
      "step": 42300
    },
    {
      "epoch": 4.3087651430316605,
      "grad_norm": 1.5627204179763794,
      "learning_rate": 3.0798814560097735e-06,
      "loss": 0.0689,
      "step": 42325
    },
    {
      "epoch": 4.311310190369541,
      "grad_norm": 0.7110241651535034,
      "learning_rate": 3.068570006560641e-06,
      "loss": 0.0668,
      "step": 42350
    },
    {
      "epoch": 4.313855237707421,
      "grad_norm": 4.596431732177734,
      "learning_rate": 3.0572585571115088e-06,
      "loss": 0.0782,
      "step": 42375
    },
    {
      "epoch": 4.316400285045302,
      "grad_norm": 4.947603702545166,
      "learning_rate": 3.0459471076623758e-06,
      "loss": 0.0566,
      "step": 42400
    },
    {
      "epoch": 4.318945332383183,
      "grad_norm": 3.2261035442352295,
      "learning_rate": 3.0346356582132436e-06,
      "loss": 0.0602,
      "step": 42425
    },
    {
      "epoch": 4.321490379721062,
      "grad_norm": 1.8598774671554565,
      "learning_rate": 3.0233242087641115e-06,
      "loss": 0.0789,
      "step": 42450
    },
    {
      "epoch": 4.324035427058943,
      "grad_norm": 5.585264682769775,
      "learning_rate": 3.012012759314979e-06,
      "loss": 0.088,
      "step": 42475
    },
    {
      "epoch": 4.326580474396824,
      "grad_norm": 6.0666046142578125,
      "learning_rate": 3.0007013098658467e-06,
      "loss": 0.0787,
      "step": 42500
    },
    {
      "epoch": 4.329125521734705,
      "grad_norm": 3.998551607131958,
      "learning_rate": 2.989389860416714e-06,
      "loss": 0.1013,
      "step": 42525
    },
    {
      "epoch": 4.3316705690725845,
      "grad_norm": 0.7877366542816162,
      "learning_rate": 2.9780784109675816e-06,
      "loss": 0.0691,
      "step": 42550
    },
    {
      "epoch": 4.334215616410465,
      "grad_norm": 4.090976238250732,
      "learning_rate": 2.966766961518449e-06,
      "loss": 0.0796,
      "step": 42575
    },
    {
      "epoch": 4.336760663748346,
      "grad_norm": 2.965317726135254,
      "learning_rate": 2.955455512069317e-06,
      "loss": 0.0927,
      "step": 42600
    },
    {
      "epoch": 4.339305711086226,
      "grad_norm": 0.9894925355911255,
      "learning_rate": 2.9441440626201847e-06,
      "loss": 0.0613,
      "step": 42625
    },
    {
      "epoch": 4.341850758424107,
      "grad_norm": 3.499403953552246,
      "learning_rate": 2.932832613171052e-06,
      "loss": 0.0857,
      "step": 42650
    },
    {
      "epoch": 4.344395805761987,
      "grad_norm": 2.901555061340332,
      "learning_rate": 2.921521163721919e-06,
      "loss": 0.0583,
      "step": 42675
    },
    {
      "epoch": 4.346940853099868,
      "grad_norm": 1.397181749343872,
      "learning_rate": 2.910209714272787e-06,
      "loss": 0.0684,
      "step": 42700
    },
    {
      "epoch": 4.349485900437748,
      "grad_norm": 2.98871111869812,
      "learning_rate": 2.8988982648236548e-06,
      "loss": 0.0699,
      "step": 42725
    },
    {
      "epoch": 4.352030947775629,
      "grad_norm": 2.4545464515686035,
      "learning_rate": 2.887586815374522e-06,
      "loss": 0.0771,
      "step": 42750
    },
    {
      "epoch": 4.3545759951135095,
      "grad_norm": 1.2507919073104858,
      "learning_rate": 2.87627536592539e-06,
      "loss": 0.0912,
      "step": 42775
    },
    {
      "epoch": 4.357121042451389,
      "grad_norm": 3.777596950531006,
      "learning_rate": 2.864963916476258e-06,
      "loss": 0.0766,
      "step": 42800
    },
    {
      "epoch": 4.35966608978927,
      "grad_norm": 3.1019132137298584,
      "learning_rate": 2.853652467027125e-06,
      "loss": 0.0637,
      "step": 42825
    },
    {
      "epoch": 4.362211137127151,
      "grad_norm": 1.4555113315582275,
      "learning_rate": 2.8423410175779923e-06,
      "loss": 0.0932,
      "step": 42850
    },
    {
      "epoch": 4.364756184465031,
      "grad_norm": 1.262918472290039,
      "learning_rate": 2.83102956812886e-06,
      "loss": 0.0606,
      "step": 42875
    },
    {
      "epoch": 4.367301231802911,
      "grad_norm": 1.907347321510315,
      "learning_rate": 2.819718118679728e-06,
      "loss": 0.0749,
      "step": 42900
    },
    {
      "epoch": 4.369846279140792,
      "grad_norm": 0.9305869340896606,
      "learning_rate": 2.8084066692305954e-06,
      "loss": 0.0793,
      "step": 42925
    },
    {
      "epoch": 4.372391326478673,
      "grad_norm": 2.1973392963409424,
      "learning_rate": 2.7970952197814633e-06,
      "loss": 0.091,
      "step": 42950
    },
    {
      "epoch": 4.374936373816553,
      "grad_norm": 4.098190784454346,
      "learning_rate": 2.7857837703323303e-06,
      "loss": 0.0951,
      "step": 42975
    },
    {
      "epoch": 4.3774814211544335,
      "grad_norm": 3.911076307296753,
      "learning_rate": 2.774472320883198e-06,
      "loss": 0.081,
      "step": 43000
    },
    {
      "epoch": 4.380026468492314,
      "grad_norm": 4.323380470275879,
      "learning_rate": 2.7631608714340655e-06,
      "loss": 0.0565,
      "step": 43025
    },
    {
      "epoch": 4.382571515830194,
      "grad_norm": 1.0312169790267944,
      "learning_rate": 2.7518494219849334e-06,
      "loss": 0.0604,
      "step": 43050
    },
    {
      "epoch": 4.385116563168075,
      "grad_norm": 3.453737258911133,
      "learning_rate": 2.7405379725358012e-06,
      "loss": 0.0684,
      "step": 43075
    },
    {
      "epoch": 4.387661610505956,
      "grad_norm": 1.7941606044769287,
      "learning_rate": 2.7292265230866687e-06,
      "loss": 0.0412,
      "step": 43100
    },
    {
      "epoch": 4.3902066578438355,
      "grad_norm": 1.500749945640564,
      "learning_rate": 2.717915073637536e-06,
      "loss": 0.0753,
      "step": 43125
    },
    {
      "epoch": 4.392751705181716,
      "grad_norm": 6.635108470916748,
      "learning_rate": 2.7066036241884035e-06,
      "loss": 0.0708,
      "step": 43150
    },
    {
      "epoch": 4.395296752519597,
      "grad_norm": 7.838114261627197,
      "learning_rate": 2.6952921747392713e-06,
      "loss": 0.0713,
      "step": 43175
    },
    {
      "epoch": 4.397841799857478,
      "grad_norm": 5.980000972747803,
      "learning_rate": 2.683980725290139e-06,
      "loss": 0.0757,
      "step": 43200
    },
    {
      "epoch": 4.4003868471953576,
      "grad_norm": 2.0675323009490967,
      "learning_rate": 2.6726692758410066e-06,
      "loss": 0.0795,
      "step": 43225
    },
    {
      "epoch": 4.402931894533238,
      "grad_norm": 4.80326509475708,
      "learning_rate": 2.6613578263918745e-06,
      "loss": 0.055,
      "step": 43250
    },
    {
      "epoch": 4.405476941871119,
      "grad_norm": 2.1413259506225586,
      "learning_rate": 2.6500463769427414e-06,
      "loss": 0.0775,
      "step": 43275
    },
    {
      "epoch": 4.408021989208999,
      "grad_norm": 1.2402043342590332,
      "learning_rate": 2.6387349274936093e-06,
      "loss": 0.0661,
      "step": 43300
    },
    {
      "epoch": 4.41056703654688,
      "grad_norm": 2.842992067337036,
      "learning_rate": 2.6274234780444767e-06,
      "loss": 0.0664,
      "step": 43325
    },
    {
      "epoch": 4.41311208388476,
      "grad_norm": 4.652822017669678,
      "learning_rate": 2.6161120285953446e-06,
      "loss": 0.077,
      "step": 43350
    },
    {
      "epoch": 4.415657131222641,
      "grad_norm": 2.196883201599121,
      "learning_rate": 2.6048005791462124e-06,
      "loss": 0.0595,
      "step": 43375
    },
    {
      "epoch": 4.418202178560521,
      "grad_norm": 3.021293878555298,
      "learning_rate": 2.5934891296970794e-06,
      "loss": 0.0633,
      "step": 43400
    },
    {
      "epoch": 4.420747225898402,
      "grad_norm": 7.387508392333984,
      "learning_rate": 2.582177680247947e-06,
      "loss": 0.0617,
      "step": 43425
    },
    {
      "epoch": 4.4232922732362825,
      "grad_norm": 1.5187947750091553,
      "learning_rate": 2.5708662307988147e-06,
      "loss": 0.0466,
      "step": 43450
    },
    {
      "epoch": 4.425837320574162,
      "grad_norm": 5.505544662475586,
      "learning_rate": 2.5595547813496825e-06,
      "loss": 0.0636,
      "step": 43475
    },
    {
      "epoch": 4.428382367912043,
      "grad_norm": 0.9313585162162781,
      "learning_rate": 2.54824333190055e-06,
      "loss": 0.0765,
      "step": 43500
    },
    {
      "epoch": 4.430927415249924,
      "grad_norm": 0.5325255393981934,
      "learning_rate": 2.5369318824514178e-06,
      "loss": 0.0807,
      "step": 43525
    },
    {
      "epoch": 4.433472462587805,
      "grad_norm": 7.443079471588135,
      "learning_rate": 2.5256204330022848e-06,
      "loss": 0.0659,
      "step": 43550
    },
    {
      "epoch": 4.436017509925684,
      "grad_norm": 1.6353135108947754,
      "learning_rate": 2.5143089835531526e-06,
      "loss": 0.0854,
      "step": 43575
    },
    {
      "epoch": 4.438562557263565,
      "grad_norm": 5.86698055267334,
      "learning_rate": 2.50299753410402e-06,
      "loss": 0.0658,
      "step": 43600
    },
    {
      "epoch": 4.441107604601446,
      "grad_norm": 5.454168319702148,
      "learning_rate": 2.491686084654888e-06,
      "loss": 0.0696,
      "step": 43625
    },
    {
      "epoch": 4.443652651939326,
      "grad_norm": 4.088606834411621,
      "learning_rate": 2.4803746352057553e-06,
      "loss": 0.0687,
      "step": 43650
    },
    {
      "epoch": 4.4461976992772065,
      "grad_norm": 3.597466468811035,
      "learning_rate": 2.469063185756623e-06,
      "loss": 0.0821,
      "step": 43675
    },
    {
      "epoch": 4.448742746615087,
      "grad_norm": 4.886236667633057,
      "learning_rate": 2.4577517363074906e-06,
      "loss": 0.0924,
      "step": 43700
    },
    {
      "epoch": 4.451287793952967,
      "grad_norm": 1.914562702178955,
      "learning_rate": 2.446440286858358e-06,
      "loss": 0.0541,
      "step": 43725
    },
    {
      "epoch": 4.453832841290848,
      "grad_norm": 5.678765773773193,
      "learning_rate": 2.435128837409226e-06,
      "loss": 0.061,
      "step": 43750
    },
    {
      "epoch": 4.456377888628729,
      "grad_norm": 1.6028286218643188,
      "learning_rate": 2.4238173879600933e-06,
      "loss": 0.069,
      "step": 43775
    },
    {
      "epoch": 4.458922935966609,
      "grad_norm": 4.589537620544434,
      "learning_rate": 2.4125059385109607e-06,
      "loss": 0.0944,
      "step": 43800
    },
    {
      "epoch": 4.461467983304489,
      "grad_norm": 0.4559231400489807,
      "learning_rate": 2.4011944890618285e-06,
      "loss": 0.0735,
      "step": 43825
    },
    {
      "epoch": 4.46401303064237,
      "grad_norm": 3.7404086589813232,
      "learning_rate": 2.3898830396126964e-06,
      "loss": 0.0507,
      "step": 43850
    },
    {
      "epoch": 4.466558077980251,
      "grad_norm": 0.8973321914672852,
      "learning_rate": 2.378571590163564e-06,
      "loss": 0.085,
      "step": 43875
    },
    {
      "epoch": 4.469103125318131,
      "grad_norm": 0.6244153380393982,
      "learning_rate": 2.3672601407144312e-06,
      "loss": 0.0778,
      "step": 43900
    },
    {
      "epoch": 4.471648172656011,
      "grad_norm": 0.38073861598968506,
      "learning_rate": 2.355948691265299e-06,
      "loss": 0.0658,
      "step": 43925
    },
    {
      "epoch": 4.474193219993892,
      "grad_norm": 4.970991611480713,
      "learning_rate": 2.3446372418161665e-06,
      "loss": 0.073,
      "step": 43950
    },
    {
      "epoch": 4.476738267331772,
      "grad_norm": 5.226351261138916,
      "learning_rate": 2.333325792367034e-06,
      "loss": 0.0753,
      "step": 43975
    },
    {
      "epoch": 4.479283314669653,
      "grad_norm": 2.576582431793213,
      "learning_rate": 2.3220143429179018e-06,
      "loss": 0.0623,
      "step": 44000
    },
    {
      "epoch": 4.481828362007533,
      "grad_norm": 1.2187014818191528,
      "learning_rate": 2.310702893468769e-06,
      "loss": 0.078,
      "step": 44025
    },
    {
      "epoch": 4.484373409345414,
      "grad_norm": 4.498028755187988,
      "learning_rate": 2.299391444019637e-06,
      "loss": 0.069,
      "step": 44050
    },
    {
      "epoch": 4.486918456683294,
      "grad_norm": 1.715052604675293,
      "learning_rate": 2.2880799945705044e-06,
      "loss": 0.0685,
      "step": 44075
    },
    {
      "epoch": 4.489463504021175,
      "grad_norm": 5.307081699371338,
      "learning_rate": 2.276768545121372e-06,
      "loss": 0.0679,
      "step": 44100
    },
    {
      "epoch": 4.4920085513590555,
      "grad_norm": 4.219576835632324,
      "learning_rate": 2.2654570956722397e-06,
      "loss": 0.0607,
      "step": 44125
    },
    {
      "epoch": 4.494553598696935,
      "grad_norm": 4.1396355628967285,
      "learning_rate": 2.254145646223107e-06,
      "loss": 0.0845,
      "step": 44150
    },
    {
      "epoch": 4.497098646034816,
      "grad_norm": 2.164240837097168,
      "learning_rate": 2.2428341967739746e-06,
      "loss": 0.0707,
      "step": 44175
    },
    {
      "epoch": 4.499643693372697,
      "grad_norm": 5.686645984649658,
      "learning_rate": 2.2315227473248424e-06,
      "loss": 0.0898,
      "step": 44200
    },
    {
      "epoch": 4.502188740710578,
      "grad_norm": 6.213311672210693,
      "learning_rate": 2.2202112978757102e-06,
      "loss": 0.0622,
      "step": 44225
    },
    {
      "epoch": 4.5047337880484575,
      "grad_norm": 1.4041837453842163,
      "learning_rate": 2.2088998484265777e-06,
      "loss": 0.0477,
      "step": 44250
    },
    {
      "epoch": 4.507278835386338,
      "grad_norm": 6.660885810852051,
      "learning_rate": 2.197588398977445e-06,
      "loss": 0.0729,
      "step": 44275
    },
    {
      "epoch": 4.509823882724219,
      "grad_norm": 4.362778663635254,
      "learning_rate": 2.186276949528313e-06,
      "loss": 0.0633,
      "step": 44300
    },
    {
      "epoch": 4.512368930062099,
      "grad_norm": 3.135101556777954,
      "learning_rate": 2.1749655000791804e-06,
      "loss": 0.0804,
      "step": 44325
    },
    {
      "epoch": 4.51491397739998,
      "grad_norm": 1.2437292337417603,
      "learning_rate": 2.1636540506300478e-06,
      "loss": 0.0768,
      "step": 44350
    },
    {
      "epoch": 4.51745902473786,
      "grad_norm": 4.525523662567139,
      "learning_rate": 2.152342601180915e-06,
      "loss": 0.0566,
      "step": 44375
    },
    {
      "epoch": 4.520004072075741,
      "grad_norm": 4.344095230102539,
      "learning_rate": 2.141031151731783e-06,
      "loss": 0.0749,
      "step": 44400
    },
    {
      "epoch": 4.522549119413621,
      "grad_norm": 3.743915319442749,
      "learning_rate": 2.129719702282651e-06,
      "loss": 0.0769,
      "step": 44425
    },
    {
      "epoch": 4.525094166751502,
      "grad_norm": 3.3865807056427,
      "learning_rate": 2.1184082528335183e-06,
      "loss": 0.0692,
      "step": 44450
    },
    {
      "epoch": 4.527639214089382,
      "grad_norm": 1.5223188400268555,
      "learning_rate": 2.1070968033843857e-06,
      "loss": 0.0508,
      "step": 44475
    },
    {
      "epoch": 4.530184261427262,
      "grad_norm": 1.2191075086593628,
      "learning_rate": 2.0957853539352536e-06,
      "loss": 0.0626,
      "step": 44500
    },
    {
      "epoch": 4.532729308765143,
      "grad_norm": 1.1970452070236206,
      "learning_rate": 2.084473904486121e-06,
      "loss": 0.0684,
      "step": 44525
    },
    {
      "epoch": 4.535274356103024,
      "grad_norm": 2.423342704772949,
      "learning_rate": 2.0731624550369884e-06,
      "loss": 0.0585,
      "step": 44550
    },
    {
      "epoch": 4.537819403440904,
      "grad_norm": 4.101266384124756,
      "learning_rate": 2.0618510055878563e-06,
      "loss": 0.0692,
      "step": 44575
    },
    {
      "epoch": 4.540364450778784,
      "grad_norm": 4.081343173980713,
      "learning_rate": 2.0505395561387237e-06,
      "loss": 0.0932,
      "step": 44600
    },
    {
      "epoch": 4.542909498116665,
      "grad_norm": 2.0687055587768555,
      "learning_rate": 2.0392281066895915e-06,
      "loss": 0.0529,
      "step": 44625
    },
    {
      "epoch": 4.545454545454545,
      "grad_norm": 5.054721832275391,
      "learning_rate": 2.027916657240459e-06,
      "loss": 0.0739,
      "step": 44650
    },
    {
      "epoch": 4.547999592792426,
      "grad_norm": 0.6480259299278259,
      "learning_rate": 2.0166052077913264e-06,
      "loss": 0.0529,
      "step": 44675
    },
    {
      "epoch": 4.5505446401303065,
      "grad_norm": 0.2223999947309494,
      "learning_rate": 2.0052937583421942e-06,
      "loss": 0.0812,
      "step": 44700
    },
    {
      "epoch": 4.553089687468187,
      "grad_norm": 9.688573837280273,
      "learning_rate": 1.9939823088930616e-06,
      "loss": 0.0608,
      "step": 44725
    },
    {
      "epoch": 4.555634734806067,
      "grad_norm": 2.712524652481079,
      "learning_rate": 1.982670859443929e-06,
      "loss": 0.0618,
      "step": 44750
    },
    {
      "epoch": 4.558179782143948,
      "grad_norm": 4.108211040496826,
      "learning_rate": 1.971359409994797e-06,
      "loss": 0.0673,
      "step": 44775
    },
    {
      "epoch": 4.560724829481829,
      "grad_norm": 1.9700902700424194,
      "learning_rate": 1.9600479605456648e-06,
      "loss": 0.0721,
      "step": 44800
    },
    {
      "epoch": 4.563269876819708,
      "grad_norm": 3.264569044113159,
      "learning_rate": 1.948736511096532e-06,
      "loss": 0.0691,
      "step": 44825
    },
    {
      "epoch": 4.565814924157589,
      "grad_norm": 1.5905423164367676,
      "learning_rate": 1.9374250616473996e-06,
      "loss": 0.0738,
      "step": 44850
    },
    {
      "epoch": 4.56835997149547,
      "grad_norm": 1.5995819568634033,
      "learning_rate": 1.9261136121982674e-06,
      "loss": 0.0853,
      "step": 44875
    },
    {
      "epoch": 4.570905018833351,
      "grad_norm": 2.281751871109009,
      "learning_rate": 1.914802162749135e-06,
      "loss": 0.0587,
      "step": 44900
    },
    {
      "epoch": 4.5734500661712305,
      "grad_norm": 1.2428480386734009,
      "learning_rate": 1.9034907133000025e-06,
      "loss": 0.0597,
      "step": 44925
    },
    {
      "epoch": 4.575995113509111,
      "grad_norm": 3.320080518722534,
      "learning_rate": 1.8921792638508701e-06,
      "loss": 0.04,
      "step": 44950
    },
    {
      "epoch": 4.578540160846992,
      "grad_norm": 4.8358330726623535,
      "learning_rate": 1.8808678144017376e-06,
      "loss": 0.0716,
      "step": 44975
    },
    {
      "epoch": 4.581085208184872,
      "grad_norm": 2.170146942138672,
      "learning_rate": 1.8695563649526052e-06,
      "loss": 0.0595,
      "step": 45000
    },
    {
      "epoch": 4.583630255522753,
      "grad_norm": 6.489071846008301,
      "learning_rate": 1.8582449155034726e-06,
      "loss": 0.0798,
      "step": 45025
    },
    {
      "epoch": 4.586175302860633,
      "grad_norm": 2.01780366897583,
      "learning_rate": 1.8469334660543402e-06,
      "loss": 0.0537,
      "step": 45050
    },
    {
      "epoch": 4.588720350198514,
      "grad_norm": 6.3545637130737305,
      "learning_rate": 1.835622016605208e-06,
      "loss": 0.08,
      "step": 45075
    },
    {
      "epoch": 4.591265397536394,
      "grad_norm": 1.5675911903381348,
      "learning_rate": 1.8243105671560755e-06,
      "loss": 0.0953,
      "step": 45100
    },
    {
      "epoch": 4.593810444874275,
      "grad_norm": 1.2118556499481201,
      "learning_rate": 1.8129991177069431e-06,
      "loss": 0.0519,
      "step": 45125
    },
    {
      "epoch": 4.5963554922121554,
      "grad_norm": 1.4789040088653564,
      "learning_rate": 1.8016876682578108e-06,
      "loss": 0.1003,
      "step": 45150
    },
    {
      "epoch": 4.598900539550035,
      "grad_norm": 4.09182071685791,
      "learning_rate": 1.7903762188086782e-06,
      "loss": 0.0774,
      "step": 45175
    },
    {
      "epoch": 4.601445586887916,
      "grad_norm": 5.658119201660156,
      "learning_rate": 1.7790647693595458e-06,
      "loss": 0.0947,
      "step": 45200
    },
    {
      "epoch": 4.603990634225797,
      "grad_norm": 1.70206618309021,
      "learning_rate": 1.7677533199104135e-06,
      "loss": 0.0708,
      "step": 45225
    },
    {
      "epoch": 4.6065356815636775,
      "grad_norm": 1.5394465923309326,
      "learning_rate": 1.7564418704612809e-06,
      "loss": 0.1066,
      "step": 45250
    },
    {
      "epoch": 4.609080728901557,
      "grad_norm": 1.8833317756652832,
      "learning_rate": 1.7451304210121487e-06,
      "loss": 0.0555,
      "step": 45275
    },
    {
      "epoch": 4.611625776239438,
      "grad_norm": 1.0039482116699219,
      "learning_rate": 1.7342714295409816e-06,
      "loss": 0.0826,
      "step": 45300
    },
    {
      "epoch": 4.614170823577319,
      "grad_norm": 2.0755767822265625,
      "learning_rate": 1.7229599800918492e-06,
      "loss": 0.0528,
      "step": 45325
    },
    {
      "epoch": 4.616715870915199,
      "grad_norm": 4.706948280334473,
      "learning_rate": 1.7116485306427167e-06,
      "loss": 0.0594,
      "step": 45350
    },
    {
      "epoch": 4.6192609182530795,
      "grad_norm": 5.853977680206299,
      "learning_rate": 1.7003370811935843e-06,
      "loss": 0.0731,
      "step": 45375
    },
    {
      "epoch": 4.62180596559096,
      "grad_norm": 1.9859325885772705,
      "learning_rate": 1.689025631744452e-06,
      "loss": 0.0808,
      "step": 45400
    },
    {
      "epoch": 4.62435101292884,
      "grad_norm": 2.232928514480591,
      "learning_rate": 1.6777141822953193e-06,
      "loss": 0.0684,
      "step": 45425
    },
    {
      "epoch": 4.626896060266721,
      "grad_norm": 1.054357886314392,
      "learning_rate": 1.6664027328461872e-06,
      "loss": 0.079,
      "step": 45450
    },
    {
      "epoch": 4.629441107604602,
      "grad_norm": 3.0795228481292725,
      "learning_rate": 1.6550912833970548e-06,
      "loss": 0.0828,
      "step": 45475
    },
    {
      "epoch": 4.6319861549424814,
      "grad_norm": 2.889328718185425,
      "learning_rate": 1.6437798339479222e-06,
      "loss": 0.0683,
      "step": 45500
    },
    {
      "epoch": 4.634531202280362,
      "grad_norm": 2.0245718955993652,
      "learning_rate": 1.6324683844987899e-06,
      "loss": 0.0757,
      "step": 45525
    },
    {
      "epoch": 4.637076249618243,
      "grad_norm": 6.983794212341309,
      "learning_rate": 1.6211569350496573e-06,
      "loss": 0.0748,
      "step": 45550
    },
    {
      "epoch": 4.639621296956124,
      "grad_norm": 2.034707546234131,
      "learning_rate": 1.609845485600525e-06,
      "loss": 0.0844,
      "step": 45575
    },
    {
      "epoch": 4.6421663442940035,
      "grad_norm": 2.9529507160186768,
      "learning_rate": 1.5985340361513926e-06,
      "loss": 0.078,
      "step": 45600
    },
    {
      "epoch": 4.644711391631884,
      "grad_norm": 3.2529144287109375,
      "learning_rate": 1.58722258670226e-06,
      "loss": 0.0762,
      "step": 45625
    },
    {
      "epoch": 4.647256438969765,
      "grad_norm": 3.109240770339966,
      "learning_rate": 1.5759111372531276e-06,
      "loss": 0.0811,
      "step": 45650
    },
    {
      "epoch": 4.649801486307645,
      "grad_norm": 3.7589831352233887,
      "learning_rate": 1.5645996878039955e-06,
      "loss": 0.0643,
      "step": 45675
    },
    {
      "epoch": 4.652346533645526,
      "grad_norm": 8.635754585266113,
      "learning_rate": 1.5532882383548629e-06,
      "loss": 0.0701,
      "step": 45700
    },
    {
      "epoch": 4.654891580983406,
      "grad_norm": 1.9032418727874756,
      "learning_rate": 1.5419767889057305e-06,
      "loss": 0.0684,
      "step": 45725
    },
    {
      "epoch": 4.657436628321287,
      "grad_norm": 2.2510008811950684,
      "learning_rate": 1.5306653394565982e-06,
      "loss": 0.0579,
      "step": 45750
    },
    {
      "epoch": 4.659981675659167,
      "grad_norm": 4.032014846801758,
      "learning_rate": 1.5193538900074656e-06,
      "loss": 0.0824,
      "step": 45775
    },
    {
      "epoch": 4.662526722997048,
      "grad_norm": 4.007706642150879,
      "learning_rate": 1.5080424405583332e-06,
      "loss": 0.0851,
      "step": 45800
    },
    {
      "epoch": 4.6650717703349285,
      "grad_norm": 0.07311418652534485,
      "learning_rate": 1.496730991109201e-06,
      "loss": 0.1039,
      "step": 45825
    },
    {
      "epoch": 4.667616817672808,
      "grad_norm": 1.875131607055664,
      "learning_rate": 1.4854195416600683e-06,
      "loss": 0.0788,
      "step": 45850
    },
    {
      "epoch": 4.670161865010689,
      "grad_norm": 2.887582778930664,
      "learning_rate": 1.4741080922109361e-06,
      "loss": 0.0558,
      "step": 45875
    },
    {
      "epoch": 4.67270691234857,
      "grad_norm": 1.3971534967422485,
      "learning_rate": 1.4627966427618037e-06,
      "loss": 0.073,
      "step": 45900
    },
    {
      "epoch": 4.675251959686451,
      "grad_norm": 1.5338753461837769,
      "learning_rate": 1.4514851933126712e-06,
      "loss": 0.0696,
      "step": 45925
    },
    {
      "epoch": 4.67779700702433,
      "grad_norm": 6.803866863250732,
      "learning_rate": 1.4401737438635388e-06,
      "loss": 0.0676,
      "step": 45950
    },
    {
      "epoch": 4.680342054362211,
      "grad_norm": 0.43175041675567627,
      "learning_rate": 1.4288622944144064e-06,
      "loss": 0.0737,
      "step": 45975
    },
    {
      "epoch": 4.682887101700092,
      "grad_norm": 2.761035919189453,
      "learning_rate": 1.4175508449652739e-06,
      "loss": 0.0587,
      "step": 46000
    },
    {
      "epoch": 4.685432149037972,
      "grad_norm": 4.648847579956055,
      "learning_rate": 1.4062393955161415e-06,
      "loss": 0.0832,
      "step": 46025
    },
    {
      "epoch": 4.6879771963758525,
      "grad_norm": 3.9929113388061523,
      "learning_rate": 1.3949279460670093e-06,
      "loss": 0.0854,
      "step": 46050
    },
    {
      "epoch": 4.690522243713733,
      "grad_norm": 2.7865846157073975,
      "learning_rate": 1.3836164966178768e-06,
      "loss": 0.0774,
      "step": 46075
    },
    {
      "epoch": 4.693067291051614,
      "grad_norm": 2.160618543624878,
      "learning_rate": 1.3723050471687444e-06,
      "loss": 0.0674,
      "step": 46100
    },
    {
      "epoch": 4.695612338389494,
      "grad_norm": 1.6433202028274536,
      "learning_rate": 1.360993597719612e-06,
      "loss": 0.0912,
      "step": 46125
    },
    {
      "epoch": 4.698157385727375,
      "grad_norm": 1.3290517330169678,
      "learning_rate": 1.3496821482704794e-06,
      "loss": 0.095,
      "step": 46150
    },
    {
      "epoch": 4.700702433065255,
      "grad_norm": 0.6854248046875,
      "learning_rate": 1.338370698821347e-06,
      "loss": 0.0713,
      "step": 46175
    },
    {
      "epoch": 4.703247480403135,
      "grad_norm": 1.6427950859069824,
      "learning_rate": 1.3270592493722145e-06,
      "loss": 0.054,
      "step": 46200
    },
    {
      "epoch": 4.705792527741016,
      "grad_norm": 2.2108538150787354,
      "learning_rate": 1.3157477999230821e-06,
      "loss": 0.0655,
      "step": 46225
    },
    {
      "epoch": 4.708337575078897,
      "grad_norm": 3.9831509590148926,
      "learning_rate": 1.30443635047395e-06,
      "loss": 0.0738,
      "step": 46250
    },
    {
      "epoch": 4.710882622416777,
      "grad_norm": 4.917258262634277,
      "learning_rate": 1.2931249010248174e-06,
      "loss": 0.0916,
      "step": 46275
    },
    {
      "epoch": 4.713427669754657,
      "grad_norm": 0.8042176365852356,
      "learning_rate": 1.281813451575685e-06,
      "loss": 0.068,
      "step": 46300
    },
    {
      "epoch": 4.715972717092538,
      "grad_norm": 0.9012075662612915,
      "learning_rate": 1.2705020021265527e-06,
      "loss": 0.045,
      "step": 46325
    },
    {
      "epoch": 4.718517764430418,
      "grad_norm": 0.9209958910942078,
      "learning_rate": 1.25919055267742e-06,
      "loss": 0.0564,
      "step": 46350
    },
    {
      "epoch": 4.721062811768299,
      "grad_norm": 1.8551143407821655,
      "learning_rate": 1.2478791032282877e-06,
      "loss": 0.0805,
      "step": 46375
    },
    {
      "epoch": 4.723607859106179,
      "grad_norm": 0.4579830467700958,
      "learning_rate": 1.2365676537791554e-06,
      "loss": 0.0493,
      "step": 46400
    },
    {
      "epoch": 4.72615290644406,
      "grad_norm": 0.38200685381889343,
      "learning_rate": 1.225256204330023e-06,
      "loss": 0.0859,
      "step": 46425
    },
    {
      "epoch": 4.72869795378194,
      "grad_norm": 3.643800735473633,
      "learning_rate": 1.2139447548808906e-06,
      "loss": 0.0613,
      "step": 46450
    },
    {
      "epoch": 4.731243001119821,
      "grad_norm": 3.732656240463257,
      "learning_rate": 1.202633305431758e-06,
      "loss": 0.0685,
      "step": 46475
    },
    {
      "epoch": 4.7337880484577015,
      "grad_norm": 5.9201579093933105,
      "learning_rate": 1.1913218559826257e-06,
      "loss": 0.0697,
      "step": 46500
    },
    {
      "epoch": 4.736333095795581,
      "grad_norm": 6.302225112915039,
      "learning_rate": 1.1800104065334933e-06,
      "loss": 0.0808,
      "step": 46525
    },
    {
      "epoch": 4.738878143133462,
      "grad_norm": 0.6190359592437744,
      "learning_rate": 1.168698957084361e-06,
      "loss": 0.081,
      "step": 46550
    },
    {
      "epoch": 4.741423190471343,
      "grad_norm": 9.110676765441895,
      "learning_rate": 1.1578399656131938e-06,
      "loss": 0.0815,
      "step": 46575
    },
    {
      "epoch": 4.743968237809224,
      "grad_norm": 5.75714111328125,
      "learning_rate": 1.1465285161640612e-06,
      "loss": 0.0875,
      "step": 46600
    },
    {
      "epoch": 4.7465132851471035,
      "grad_norm": 1.9609627723693848,
      "learning_rate": 1.135217066714929e-06,
      "loss": 0.0711,
      "step": 46625
    },
    {
      "epoch": 4.749058332484984,
      "grad_norm": 2.6252379417419434,
      "learning_rate": 1.1239056172657965e-06,
      "loss": 0.0544,
      "step": 46650
    },
    {
      "epoch": 4.751603379822865,
      "grad_norm": 1.7580702304840088,
      "learning_rate": 1.1125941678166641e-06,
      "loss": 0.0975,
      "step": 46675
    },
    {
      "epoch": 4.754148427160745,
      "grad_norm": 1.8180975914001465,
      "learning_rate": 1.1012827183675318e-06,
      "loss": 0.0843,
      "step": 46700
    },
    {
      "epoch": 4.756693474498626,
      "grad_norm": 4.706656455993652,
      "learning_rate": 1.0899712689183994e-06,
      "loss": 0.1002,
      "step": 46725
    },
    {
      "epoch": 4.759238521836506,
      "grad_norm": 2.2304418087005615,
      "learning_rate": 1.0786598194692668e-06,
      "loss": 0.0497,
      "step": 46750
    },
    {
      "epoch": 4.761783569174387,
      "grad_norm": 3.291037082672119,
      "learning_rate": 1.0673483700201345e-06,
      "loss": 0.0993,
      "step": 46775
    },
    {
      "epoch": 4.764328616512267,
      "grad_norm": 1.777855634689331,
      "learning_rate": 1.056036920571002e-06,
      "loss": 0.0683,
      "step": 46800
    },
    {
      "epoch": 4.766873663850148,
      "grad_norm": 2.230701208114624,
      "learning_rate": 1.0447254711218697e-06,
      "loss": 0.0892,
      "step": 46825
    },
    {
      "epoch": 4.769418711188028,
      "grad_norm": 5.757084846496582,
      "learning_rate": 1.0334140216727371e-06,
      "loss": 0.0594,
      "step": 46850
    },
    {
      "epoch": 4.771963758525908,
      "grad_norm": 2.717808961868286,
      "learning_rate": 1.0221025722236048e-06,
      "loss": 0.0766,
      "step": 46875
    },
    {
      "epoch": 4.774508805863789,
      "grad_norm": 4.532754898071289,
      "learning_rate": 1.0107911227744724e-06,
      "loss": 0.0603,
      "step": 46900
    },
    {
      "epoch": 4.77705385320167,
      "grad_norm": 2.240419626235962,
      "learning_rate": 9.994796733253398e-07,
      "loss": 0.0797,
      "step": 46925
    },
    {
      "epoch": 4.7795989005395505,
      "grad_norm": 1.5470459461212158,
      "learning_rate": 9.881682238762077e-07,
      "loss": 0.0602,
      "step": 46950
    },
    {
      "epoch": 4.78214394787743,
      "grad_norm": 3.4554860591888428,
      "learning_rate": 9.76856774427075e-07,
      "loss": 0.0658,
      "step": 46975
    },
    {
      "epoch": 4.784688995215311,
      "grad_norm": 5.003359794616699,
      "learning_rate": 9.655453249779427e-07,
      "loss": 0.0669,
      "step": 47000
    },
    {
      "epoch": 4.787234042553192,
      "grad_norm": 2.6245150566101074,
      "learning_rate": 9.542338755288104e-07,
      "loss": 0.0566,
      "step": 47025
    },
    {
      "epoch": 4.789779089891072,
      "grad_norm": 3.7440950870513916,
      "learning_rate": 9.429224260796779e-07,
      "loss": 0.0832,
      "step": 47050
    },
    {
      "epoch": 4.7923241372289525,
      "grad_norm": 3.2227530479431152,
      "learning_rate": 9.316109766305455e-07,
      "loss": 0.0925,
      "step": 47075
    },
    {
      "epoch": 4.794869184566833,
      "grad_norm": 0.83717280626297,
      "learning_rate": 9.202995271814132e-07,
      "loss": 0.0611,
      "step": 47100
    },
    {
      "epoch": 4.797414231904713,
      "grad_norm": 1.0189921855926514,
      "learning_rate": 9.089880777322807e-07,
      "loss": 0.0941,
      "step": 47125
    },
    {
      "epoch": 4.799959279242594,
      "grad_norm": 7.039854526519775,
      "learning_rate": 8.976766282831482e-07,
      "loss": 0.0621,
      "step": 47150
    },
    {
      "epoch": 4.8025043265804745,
      "grad_norm": 2.496320962905884,
      "learning_rate": 8.86365178834016e-07,
      "loss": 0.0617,
      "step": 47175
    },
    {
      "epoch": 4.805049373918354,
      "grad_norm": 2.810091972351074,
      "learning_rate": 8.750537293848835e-07,
      "loss": 0.0708,
      "step": 47200
    },
    {
      "epoch": 4.807594421256235,
      "grad_norm": 4.1311469078063965,
      "learning_rate": 8.63742279935751e-07,
      "loss": 0.0732,
      "step": 47225
    },
    {
      "epoch": 4.810139468594116,
      "grad_norm": 3.5702366828918457,
      "learning_rate": 8.524308304866185e-07,
      "loss": 0.0798,
      "step": 47250
    },
    {
      "epoch": 4.812684515931997,
      "grad_norm": 2.6123664379119873,
      "learning_rate": 8.411193810374863e-07,
      "loss": 0.0673,
      "step": 47275
    },
    {
      "epoch": 4.8152295632698765,
      "grad_norm": 1.623597264289856,
      "learning_rate": 8.298079315883538e-07,
      "loss": 0.0998,
      "step": 47300
    },
    {
      "epoch": 4.817774610607757,
      "grad_norm": 4.469209671020508,
      "learning_rate": 8.184964821392213e-07,
      "loss": 0.081,
      "step": 47325
    },
    {
      "epoch": 4.820319657945638,
      "grad_norm": 3.1409494876861572,
      "learning_rate": 8.071850326900891e-07,
      "loss": 0.0713,
      "step": 47350
    },
    {
      "epoch": 4.822864705283518,
      "grad_norm": 3.3526580333709717,
      "learning_rate": 7.958735832409566e-07,
      "loss": 0.0848,
      "step": 47375
    },
    {
      "epoch": 4.825409752621399,
      "grad_norm": 1.9072054624557495,
      "learning_rate": 7.845621337918241e-07,
      "loss": 0.0883,
      "step": 47400
    },
    {
      "epoch": 4.827954799959279,
      "grad_norm": 1.9279520511627197,
      "learning_rate": 7.732506843426918e-07,
      "loss": 0.0579,
      "step": 47425
    },
    {
      "epoch": 4.83049984729716,
      "grad_norm": 4.597317695617676,
      "learning_rate": 7.619392348935594e-07,
      "loss": 0.0817,
      "step": 47450
    },
    {
      "epoch": 4.83304489463504,
      "grad_norm": 0.9331306219100952,
      "learning_rate": 7.506277854444269e-07,
      "loss": 0.083,
      "step": 47475
    },
    {
      "epoch": 4.835589941972921,
      "grad_norm": 2.566190719604492,
      "learning_rate": 7.393163359952945e-07,
      "loss": 0.0661,
      "step": 47500
    },
    {
      "epoch": 4.838134989310801,
      "grad_norm": 1.445909023284912,
      "learning_rate": 7.280048865461621e-07,
      "loss": 0.0838,
      "step": 47525
    },
    {
      "epoch": 4.840680036648681,
      "grad_norm": 1.3846561908721924,
      "learning_rate": 7.166934370970296e-07,
      "loss": 0.0773,
      "step": 47550
    },
    {
      "epoch": 4.843225083986562,
      "grad_norm": 4.2333269119262695,
      "learning_rate": 7.053819876478972e-07,
      "loss": 0.0868,
      "step": 47575
    },
    {
      "epoch": 4.845770131324443,
      "grad_norm": 4.511714935302734,
      "learning_rate": 6.940705381987649e-07,
      "loss": 0.0762,
      "step": 47600
    },
    {
      "epoch": 4.8483151786623235,
      "grad_norm": 3.7210798263549805,
      "learning_rate": 6.827590887496324e-07,
      "loss": 0.081,
      "step": 47625
    },
    {
      "epoch": 4.850860226000203,
      "grad_norm": 1.1287949085235596,
      "learning_rate": 6.714476393004999e-07,
      "loss": 0.0911,
      "step": 47650
    },
    {
      "epoch": 4.853405273338084,
      "grad_norm": 1.98590087890625,
      "learning_rate": 6.601361898513677e-07,
      "loss": 0.0695,
      "step": 47675
    },
    {
      "epoch": 4.855950320675965,
      "grad_norm": 1.839471459388733,
      "learning_rate": 6.488247404022352e-07,
      "loss": 0.077,
      "step": 47700
    },
    {
      "epoch": 4.858495368013845,
      "grad_norm": 2.250749111175537,
      "learning_rate": 6.375132909531027e-07,
      "loss": 0.0641,
      "step": 47725
    },
    {
      "epoch": 4.8610404153517255,
      "grad_norm": 0.87042635679245,
      "learning_rate": 6.262018415039705e-07,
      "loss": 0.0646,
      "step": 47750
    },
    {
      "epoch": 4.863585462689606,
      "grad_norm": 1.301599144935608,
      "learning_rate": 6.14890392054838e-07,
      "loss": 0.0801,
      "step": 47775
    },
    {
      "epoch": 4.866130510027487,
      "grad_norm": 2.1233930587768555,
      "learning_rate": 6.035789426057055e-07,
      "loss": 0.0714,
      "step": 47800
    },
    {
      "epoch": 4.868675557365367,
      "grad_norm": 1.3377180099487305,
      "learning_rate": 5.922674931565731e-07,
      "loss": 0.0921,
      "step": 47825
    },
    {
      "epoch": 4.871220604703248,
      "grad_norm": 2.148926258087158,
      "learning_rate": 5.809560437074408e-07,
      "loss": 0.0669,
      "step": 47850
    },
    {
      "epoch": 4.873765652041128,
      "grad_norm": 1.1764349937438965,
      "learning_rate": 5.696445942583083e-07,
      "loss": 0.0996,
      "step": 47875
    },
    {
      "epoch": 4.876310699379008,
      "grad_norm": 5.230283737182617,
      "learning_rate": 5.583331448091759e-07,
      "loss": 0.091,
      "step": 47900
    },
    {
      "epoch": 4.878855746716889,
      "grad_norm": 3.0122835636138916,
      "learning_rate": 5.470216953600435e-07,
      "loss": 0.084,
      "step": 47925
    },
    {
      "epoch": 4.88140079405477,
      "grad_norm": 0.7090607285499573,
      "learning_rate": 5.357102459109111e-07,
      "loss": 0.0652,
      "step": 47950
    },
    {
      "epoch": 4.8839458413926495,
      "grad_norm": 4.014002799987793,
      "learning_rate": 5.243987964617786e-07,
      "loss": 0.0904,
      "step": 47975
    },
    {
      "epoch": 4.88649088873053,
      "grad_norm": 1.4843350648880005,
      "learning_rate": 5.130873470126463e-07,
      "loss": 0.06,
      "step": 48000
    },
    {
      "epoch": 4.889035936068411,
      "grad_norm": 0.7930936813354492,
      "learning_rate": 5.017758975635138e-07,
      "loss": 0.0668,
      "step": 48025
    },
    {
      "epoch": 4.891580983406291,
      "grad_norm": 2.7590086460113525,
      "learning_rate": 4.904644481143814e-07,
      "loss": 0.1019,
      "step": 48050
    },
    {
      "epoch": 4.894126030744172,
      "grad_norm": 2.253613233566284,
      "learning_rate": 4.79152998665249e-07,
      "loss": 0.0593,
      "step": 48075
    },
    {
      "epoch": 4.896671078082052,
      "grad_norm": 3.688272714614868,
      "learning_rate": 4.678415492161166e-07,
      "loss": 0.0817,
      "step": 48100
    },
    {
      "epoch": 4.899216125419933,
      "grad_norm": 3.342242956161499,
      "learning_rate": 4.5653009976698417e-07,
      "loss": 0.0633,
      "step": 48125
    },
    {
      "epoch": 4.901761172757813,
      "grad_norm": 2.113290786743164,
      "learning_rate": 4.4521865031785175e-07,
      "loss": 0.0762,
      "step": 48150
    },
    {
      "epoch": 4.904306220095694,
      "grad_norm": 1.4051839113235474,
      "learning_rate": 4.339072008687194e-07,
      "loss": 0.0879,
      "step": 48175
    },
    {
      "epoch": 4.9068512674335745,
      "grad_norm": 2.42224383354187,
      "learning_rate": 4.225957514195869e-07,
      "loss": 0.063,
      "step": 48200
    },
    {
      "epoch": 4.909396314771454,
      "grad_norm": 0.3117251694202423,
      "learning_rate": 4.1128430197045454e-07,
      "loss": 0.0613,
      "step": 48225
    },
    {
      "epoch": 4.911941362109335,
      "grad_norm": 4.431111812591553,
      "learning_rate": 3.999728525213221e-07,
      "loss": 0.0905,
      "step": 48250
    },
    {
      "epoch": 4.914486409447216,
      "grad_norm": 3.4864866733551025,
      "learning_rate": 3.886614030721897e-07,
      "loss": 0.097,
      "step": 48275
    },
    {
      "epoch": 4.917031456785097,
      "grad_norm": 4.725757598876953,
      "learning_rate": 3.773499536230573e-07,
      "loss": 0.0863,
      "step": 48300
    },
    {
      "epoch": 4.919576504122976,
      "grad_norm": 7.7111496925354,
      "learning_rate": 3.6603850417392486e-07,
      "loss": 0.0558,
      "step": 48325
    },
    {
      "epoch": 4.922121551460857,
      "grad_norm": 3.0076076984405518,
      "learning_rate": 3.5472705472479244e-07,
      "loss": 0.0819,
      "step": 48350
    },
    {
      "epoch": 4.924666598798738,
      "grad_norm": 4.618623733520508,
      "learning_rate": 3.434156052756601e-07,
      "loss": 0.0883,
      "step": 48375
    },
    {
      "epoch": 4.927211646136618,
      "grad_norm": 2.2570934295654297,
      "learning_rate": 3.321041558265276e-07,
      "loss": 0.0725,
      "step": 48400
    },
    {
      "epoch": 4.9297566934744985,
      "grad_norm": 1.5796055793762207,
      "learning_rate": 3.2079270637739524e-07,
      "loss": 0.1031,
      "step": 48425
    },
    {
      "epoch": 4.932301740812379,
      "grad_norm": 7.036586284637451,
      "learning_rate": 3.094812569282628e-07,
      "loss": 0.0724,
      "step": 48450
    },
    {
      "epoch": 4.93484678815026,
      "grad_norm": 4.79809045791626,
      "learning_rate": 2.981698074791304e-07,
      "loss": 0.0745,
      "step": 48475
    },
    {
      "epoch": 4.93739183548814,
      "grad_norm": 0.8551825284957886,
      "learning_rate": 2.86858358029998e-07,
      "loss": 0.0655,
      "step": 48500
    },
    {
      "epoch": 4.939936882826021,
      "grad_norm": 2.2786452770233154,
      "learning_rate": 2.755469085808656e-07,
      "loss": 0.0623,
      "step": 48525
    },
    {
      "epoch": 4.942481930163901,
      "grad_norm": 2.4696829319000244,
      "learning_rate": 2.642354591317332e-07,
      "loss": 0.0768,
      "step": 48550
    },
    {
      "epoch": 4.945026977501781,
      "grad_norm": 2.277616024017334,
      "learning_rate": 2.529240096826007e-07,
      "loss": 0.072,
      "step": 48575
    },
    {
      "epoch": 4.947572024839662,
      "grad_norm": 4.403280735015869,
      "learning_rate": 2.416125602334683e-07,
      "loss": 0.0721,
      "step": 48600
    },
    {
      "epoch": 4.950117072177543,
      "grad_norm": 1.1744745969772339,
      "learning_rate": 2.3030111078433593e-07,
      "loss": 0.049,
      "step": 48625
    },
    {
      "epoch": 4.9526621195154235,
      "grad_norm": 5.233054161071777,
      "learning_rate": 2.1898966133520351e-07,
      "loss": 0.0608,
      "step": 48650
    },
    {
      "epoch": 4.955207166853303,
      "grad_norm": 2.6371543407440186,
      "learning_rate": 2.076782118860711e-07,
      "loss": 0.0593,
      "step": 48675
    },
    {
      "epoch": 4.957752214191184,
      "grad_norm": 3.5058162212371826,
      "learning_rate": 1.9636676243693867e-07,
      "loss": 0.0855,
      "step": 48700
    },
    {
      "epoch": 4.960297261529065,
      "grad_norm": 2.090693235397339,
      "learning_rate": 1.8505531298780628e-07,
      "loss": 0.0626,
      "step": 48725
    },
    {
      "epoch": 4.962842308866945,
      "grad_norm": 0.4760298430919647,
      "learning_rate": 1.7374386353867386e-07,
      "loss": 0.0859,
      "step": 48750
    },
    {
      "epoch": 4.965387356204825,
      "grad_norm": 7.703958988189697,
      "learning_rate": 1.6243241408954144e-07,
      "loss": 0.0896,
      "step": 48775
    },
    {
      "epoch": 4.967932403542706,
      "grad_norm": 0.33076685667037964,
      "learning_rate": 1.5112096464040902e-07,
      "loss": 0.0748,
      "step": 48800
    },
    {
      "epoch": 4.970477450880586,
      "grad_norm": 2.938126564025879,
      "learning_rate": 1.3980951519127663e-07,
      "loss": 0.0663,
      "step": 48825
    },
    {
      "epoch": 4.973022498218467,
      "grad_norm": 2.024932861328125,
      "learning_rate": 1.284980657421442e-07,
      "loss": 0.0539,
      "step": 48850
    },
    {
      "epoch": 4.9755675455563475,
      "grad_norm": 5.843230247497559,
      "learning_rate": 1.171866162930118e-07,
      "loss": 0.0824,
      "step": 48875
    },
    {
      "epoch": 4.978112592894227,
      "grad_norm": 3.76930570602417,
      "learning_rate": 1.0587516684387938e-07,
      "loss": 0.0556,
      "step": 48900
    },
    {
      "epoch": 4.980657640232108,
      "grad_norm": 3.4306960105895996,
      "learning_rate": 9.456371739474698e-08,
      "loss": 0.0587,
      "step": 48925
    },
    {
      "epoch": 4.983202687569989,
      "grad_norm": 3.4686169624328613,
      "learning_rate": 8.325226794561456e-08,
      "loss": 0.0681,
      "step": 48950
    },
    {
      "epoch": 4.98574773490787,
      "grad_norm": 3.3935706615448,
      "learning_rate": 7.194081849648214e-08,
      "loss": 0.0651,
      "step": 48975
    },
    {
      "epoch": 4.9882927822457495,
      "grad_norm": 2.937288999557495,
      "learning_rate": 6.062936904734973e-08,
      "loss": 0.0616,
      "step": 49000
    },
    {
      "epoch": 4.99083782958363,
      "grad_norm": 0.9331411719322205,
      "learning_rate": 4.931791959821732e-08,
      "loss": 0.071,
      "step": 49025
    },
    {
      "epoch": 4.993382876921511,
      "grad_norm": 4.932324409484863,
      "learning_rate": 3.8006470149084906e-08,
      "loss": 0.0742,
      "step": 49050
    },
    {
      "epoch": 4.995927924259391,
      "grad_norm": 1.9142768383026123,
      "learning_rate": 2.6695020699952493e-08,
      "loss": 0.0889,
      "step": 49075
    },
    {
      "epoch": 4.9984729715972716,
      "grad_norm": 3.7317426204681396,
      "learning_rate": 1.538357125082008e-08,
      "loss": 0.0777,
      "step": 49100
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.08259893208742142,
      "eval_runtime": 7.0041,
      "eval_samples_per_second": 971.995,
      "eval_steps_per_second": 15.277,
      "step": 49115
    }
  ],
  "logging_steps": 25,
  "max_steps": 49115,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.01
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}