diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16914 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.24999189937074312,
+  "eval_steps": 2411,
+  "global_step": 2411,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00010368805448807264,
+      "grad_norm": 3.078125,
+      "learning_rate": 2e-05,
+      "loss": 11.9266,
+      "step": 1
+    },
+    {
+      "epoch": 0.00010368805448807264,
+      "eval_loss": 11.923038482666016,
+      "eval_runtime": 0.4444,
+      "eval_samples_per_second": 335.305,
+      "eval_steps_per_second": 15.753,
+      "step": 1
+    },
+    {
+      "epoch": 0.00020737610897614527,
+      "grad_norm": 3.046875,
+      "learning_rate": 4e-05,
+      "loss": 11.9246,
+      "step": 2
+    },
+    {
+      "epoch": 0.0003110641634642179,
+      "grad_norm": 3.171875,
+      "learning_rate": 6e-05,
+      "loss": 11.8897,
+      "step": 3
+    },
+    {
+      "epoch": 0.00041475221795229055,
+      "grad_norm": 2.890625,
+      "learning_rate": 8e-05,
+      "loss": 11.7637,
+      "step": 4
+    },
+    {
+      "epoch": 0.0005184402724403631,
+      "grad_norm": 2.6875,
+      "learning_rate": 0.0001,
+      "loss": 11.6167,
+      "step": 5
+    },
+    {
+      "epoch": 0.0006221283269284358,
+      "grad_norm": 2.5625,
+      "learning_rate": 0.00012,
+      "loss": 11.3739,
+      "step": 6
+    },
+    {
+      "epoch": 0.0007258163814165084,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.00014,
+      "loss": 11.1311,
+      "step": 7
+    },
+    {
+      "epoch": 0.0008295044359045811,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.00016,
+      "loss": 10.8909,
+      "step": 8
+    },
+    {
+      "epoch": 0.0009331924903926537,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00018,
+      "loss": 10.6657,
+      "step": 9
+    },
+    {
+      "epoch": 0.0010368805448807263,
+      "grad_norm": 2.125,
+      "learning_rate": 0.0002,
+      "loss": 10.4763,
+      "step": 10
+    },
+    {
+      "epoch": 0.001140568599368799,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.00019999999941005286,
+      "loss": 10.2975,
+      "step": 11
+    },
+    {
+      "epoch": 0.0012442566538568716,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.00019999999764021143,
+      "loss": 10.1245,
+      "step": 12
+    },
+    {
+      "epoch": 0.0013479447083449443,
+      "grad_norm": 2.0,
+      "learning_rate": 0.00019999999469047573,
+      "loss": 9.9439,
+      "step": 13
+    },
+    {
+      "epoch": 0.0014516327628330168,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.0001999999905608458,
+      "loss": 9.7362,
+      "step": 14
+    },
+    {
+      "epoch": 0.0015553208173210895,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.00019999998525132166,
+      "loss": 9.5586,
+      "step": 15
+    },
+    {
+      "epoch": 0.0016590088718091622,
+      "grad_norm": 1.875,
+      "learning_rate": 0.00019999997876190344,
+      "loss": 9.4056,
+      "step": 16
+    },
+    {
+      "epoch": 0.0017626969262972347,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.00019999997109259115,
+      "loss": 9.2022,
+      "step": 17
+    },
+    {
+      "epoch": 0.0018663849807853074,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00019999996224338487,
+      "loss": 9.0599,
+      "step": 18
+    },
+    {
+      "epoch": 0.00197007303527338,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.0001999999522142848,
+      "loss": 8.9008,
+      "step": 19
+    },
+    {
+      "epoch": 0.0020737610897614525,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.000199999941005291,
+      "loss": 8.8079,
+      "step": 20
+    },
+    {
+      "epoch": 0.0021774491442495252,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00019999992861640355,
+      "loss": 8.654,
+      "step": 21
+    },
+    {
+      "epoch": 0.002281137198737598,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.0001999999150476227,
+      "loss": 8.5387,
+      "step": 22
+    },
+    {
+      "epoch": 0.0023848252532256706,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0001999999002989485,
+      "loss": 8.4015,
+      "step": 23
+    },
+    {
+      "epoch": 0.0024885133077137433,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019999988437038123,
+      "loss": 8.3069,
+      "step": 24
+    },
+    {
+      "epoch": 0.002592201362201816,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019999986726192102,
+      "loss": 8.1671,
+      "step": 25
+    },
+    {
+      "epoch": 0.0026958894166898887,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00019999984897356806,
+      "loss": 8.1281,
+      "step": 26
+    },
+    {
+      "epoch": 0.002799577471177961,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0001999998295053226,
+      "loss": 8.0346,
+      "step": 27
+    },
+    {
+      "epoch": 0.0029032655256660336,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00019999980885718487,
+      "loss": 7.9803,
+      "step": 28
+    },
+    {
+      "epoch": 0.0030069535801541063,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019999978702915508,
+      "loss": 7.9035,
+      "step": 29
+    },
+    {
+      "epoch": 0.003110641634642179,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001999997640212335,
+      "loss": 7.8359,
+      "step": 30
+    },
+    {
+      "epoch": 0.0032143296891302517,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00019999973983342043,
+      "loss": 7.8463,
+      "step": 31
+    },
+    {
+      "epoch": 0.0033180177436183244,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0001999997144657161,
+      "loss": 7.7655,
+      "step": 32
+    },
+    {
+      "epoch": 0.003421705798106397,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001999996879181209,
+      "loss": 7.7699,
+      "step": 33
+    },
+    {
+      "epoch": 0.0035253938525944693,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00019999966019063506,
+      "loss": 7.7232,
+      "step": 34
+    },
+    {
+      "epoch": 0.003629081907082542,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00019999963128325892,
+      "loss": 7.658,
+      "step": 35
+    },
+    {
+      "epoch": 0.0037327699615706147,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00019999960119599283,
+      "loss": 7.6972,
+      "step": 36
+    },
+    {
+      "epoch": 0.0038364580160586874,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.00019999956992883716,
+      "loss": 7.6441,
+      "step": 37
+    },
+    {
+      "epoch": 0.00394014607054676,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00019999953748179228,
+      "loss": 7.5872,
+      "step": 38
+    },
+    {
+      "epoch": 0.004043834125034833,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019999950385485855,
+      "loss": 7.5796,
+      "step": 39
+    },
+    {
+      "epoch": 0.004147522179522905,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00019999946904803638,
+      "loss": 7.5814,
+      "step": 40
+    },
+    {
+      "epoch": 0.004251210234010978,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00019999943306132621,
+      "loss": 7.5596,
+      "step": 41
+    },
+    {
+      "epoch": 0.0043548982884990504,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00019999939589472837,
+      "loss": 7.5181,
+      "step": 42
+    },
+    {
+      "epoch": 0.004458586342987124,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00019999935754824342,
+      "loss": 7.4972,
+      "step": 43
+    },
+    {
+      "epoch": 0.004562274397475196,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00019999931802187172,
+      "loss": 7.4948,
+      "step": 44
+    },
+    {
+      "epoch": 0.004665962451963268,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0001999992773156138,
+      "loss": 7.4437,
+      "step": 45
+    },
+    {
+      "epoch": 0.004769650506451341,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0001999992354294701,
+      "loss": 7.4184,
+      "step": 46
+    },
+    {
+      "epoch": 0.0048733385609394135,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00019999919236344114,
+      "loss": 7.3949,
+      "step": 47
+    },
+    {
+      "epoch": 0.004977026615427487,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00019999914811752738,
+      "loss": 7.3632,
+      "step": 48
+    },
+    {
+      "epoch": 0.005080714669915559,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00019999910269172938,
+      "loss": 7.3576,
+      "step": 49
+    },
+    {
+      "epoch": 0.005184402724403632,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0001999990560860477,
+      "loss": 7.3231,
+      "step": 50
+    },
+    {
+      "epoch": 0.005288090778891704,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00019999900830048283,
+      "loss": 7.3035,
+      "step": 51
+    },
+    {
+      "epoch": 0.005391778833379777,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0001999989593350354,
+      "loss": 7.2858,
+      "step": 52
+    },
+    {
+      "epoch": 0.00549546688786785,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00019999890918970592,
+      "loss": 7.2276,
+      "step": 53
+    },
+    {
+      "epoch": 0.005599154942355922,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00019999885786449505,
+      "loss": 7.2519,
+      "step": 54
+    },
+    {
+      "epoch": 0.005702842996843995,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00019999880535940333,
+      "loss": 7.2018,
+      "step": 55
+    },
+    {
+      "epoch": 0.005806531051332067,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00019999875167443142,
+      "loss": 7.1873,
+      "step": 56
+    },
+    {
+      "epoch": 0.00591021910582014,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00019999869680957993,
+      "loss": 7.1665,
+      "step": 57
+    },
+    {
+      "epoch": 0.006013907160308213,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00019999864076484955,
+      "loss": 7.1518,
+      "step": 58
+    },
+    {
+      "epoch": 0.006117595214796286,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001999985835402409,
+      "loss": 7.1188,
+      "step": 59
+    },
+    {
+      "epoch": 0.006221283269284358,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00019999852513575466,
+      "loss": 7.0633,
+      "step": 60
+    },
+    {
+      "epoch": 0.00632497132377243,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019999846555139152,
+      "loss": 7.0708,
+      "step": 61
+    },
+    {
+      "epoch": 0.006428659378260503,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0001999984047871522,
+      "loss": 7.0265,
+      "step": 62
+    },
+    {
+      "epoch": 0.006532347432748576,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.0001999983428430374,
+      "loss": 6.976,
+      "step": 63
+    },
+    {
+      "epoch": 0.006636035487236649,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019999827971904787,
+      "loss": 6.9527,
+      "step": 64
+    },
+    {
+      "epoch": 0.006739723541724721,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019999821541518437,
+      "loss": 6.9225,
+      "step": 65
+    },
+    {
+      "epoch": 0.006843411596212794,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00019999814993144755,
+      "loss": 6.9846,
+      "step": 66
+    },
+    {
+      "epoch": 0.0069470996507008664,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00019999808326783835,
+      "loss": 6.9026,
+      "step": 67
+    },
+    {
+      "epoch": 0.007050787705188939,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00019999801542435743,
+      "loss": 6.8856,
+      "step": 68
+    },
+    {
+      "epoch": 0.007154475759677012,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00019999794640100562,
+      "loss": 6.8605,
+      "step": 69
+    },
+    {
+      "epoch": 0.007258163814165084,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00019999787619778375,
+      "loss": 6.8081,
+      "step": 70
+    },
+    {
+      "epoch": 0.007361851868653157,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00019999780481469266,
+      "loss": 6.8199,
+      "step": 71
+    },
+    {
+      "epoch": 0.0074655399231412295,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00019999773225173314,
+      "loss": 6.8462,
+      "step": 72
+    },
+    {
+      "epoch": 0.007569227977629303,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019999765850890614,
+      "loss": 6.7978,
+      "step": 73
+    },
+    {
+      "epoch": 0.007672916032117375,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001999975835862124,
+      "loss": 6.7541,
+      "step": 74
+    },
+    {
+      "epoch": 0.007776604086605447,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00019999750748365294,
+      "loss": 6.7211,
+      "step": 75
+    },
+    {
+      "epoch": 0.00788029214109352,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00019999743020122855,
+      "loss": 6.7321,
+      "step": 76
+    },
+    {
+      "epoch": 0.007983980195581592,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.0001999973517389402,
+      "loss": 6.7423,
+      "step": 77
+    },
+    {
+      "epoch": 0.008087668250069666,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019999727209678883,
+      "loss": 6.6788,
+      "step": 78
+    },
+    {
+      "epoch": 0.008191356304557739,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001999971912747753,
+      "loss": 6.689,
+      "step": 79
+    },
+    {
+      "epoch": 0.00829504435904581,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00019999710927290064,
+      "loss": 6.6457,
+      "step": 80
+    },
+    {
+      "epoch": 0.008398732413533883,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00019999702609116578,
+      "loss": 6.6439,
+      "step": 81
+    },
+    {
+      "epoch": 0.008502420468021956,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019999694172957174,
+      "loss": 6.6209,
+      "step": 82
+    },
+    {
+      "epoch": 0.008606108522510028,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019999685618811948,
+      "loss": 6.5961,
+      "step": 83
+    },
+    {
+      "epoch": 0.008709796576998101,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019999676946681,
+      "loss": 6.6656,
+      "step": 84
+    },
+    {
+      "epoch": 0.008813484631486174,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00019999668156564436,
+      "loss": 6.5868,
+      "step": 85
+    },
+    {
+      "epoch": 0.008917172685974247,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00019999659248462357,
+      "loss": 6.6077,
+      "step": 86
+    },
+    {
+      "epoch": 0.009020860740462319,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0001999965022237487,
+      "loss": 6.5545,
+      "step": 87
+    },
+    {
+      "epoch": 0.009124548794950392,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00019999641078302077,
+      "loss": 6.5389,
+      "step": 88
+    },
+    {
+      "epoch": 0.009228236849438465,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019999631816244095,
+      "loss": 6.5265,
+      "step": 89
+    },
+    {
+      "epoch": 0.009331924903926536,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019999622436201025,
+      "loss": 6.5339,
+      "step": 90
+    },
+    {
+      "epoch": 0.00943561295841461,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001999961293817298,
+      "loss": 6.5092,
+      "step": 91
+    },
+    {
+      "epoch": 0.009539301012902682,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001999960332216007,
+      "loss": 6.4901,
+      "step": 92
+    },
+    {
+      "epoch": 0.009642989067390756,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019999593588162414,
+      "loss": 6.5021,
+      "step": 93
+    },
+    {
+      "epoch": 0.009746677121878827,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019999583736180122,
+      "loss": 6.4337,
+      "step": 94
+    },
+    {
+      "epoch": 0.0098503651763669,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00019999573766213313,
+      "loss": 6.4328,
+      "step": 95
+    },
+    {
+      "epoch": 0.009954053230854973,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019999563678262106,
+      "loss": 6.4444,
+      "step": 96
+    },
+    {
+      "epoch": 0.010057741285343045,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019999553472326614,
+      "loss": 6.4165,
+      "step": 97
+    },
+    {
+      "epoch": 0.010161429339831118,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0001999954314840696,
+      "loss": 6.4282,
+      "step": 98
+    },
+    {
+      "epoch": 0.01026511739431919,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001999953270650327,
+      "loss": 6.4431,
+      "step": 99
+    },
+    {
+      "epoch": 0.010368805448807264,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00019999522146615662,
+      "loss": 6.3568,
+      "step": 100
+    },
+    {
+      "epoch": 0.010472493503295335,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019999511468744263,
+      "loss": 6.3853,
+      "step": 101
+    },
+    {
+      "epoch": 0.010576181557783408,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.000199995006728892,
+      "loss": 6.4344,
+      "step": 102
+    },
+    {
+      "epoch": 0.010679869612271482,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.000199994897590506,
+      "loss": 6.3957,
+      "step": 103
+    },
+    {
+      "epoch": 0.010783557666759555,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019999478727228588,
+      "loss": 6.3347,
+      "step": 104
+    },
+    {
+      "epoch": 0.010887245721247626,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019999467577423296,
+      "loss": 6.3382,
+      "step": 105
+    },
+    {
+      "epoch": 0.0109909337757357,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001999945630963486,
+      "loss": 6.3785,
+      "step": 106
+    },
+    {
+      "epoch": 0.011094621830223772,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019999444923863405,
+      "loss": 6.3325,
+      "step": 107
+    },
+    {
+      "epoch": 0.011198309884711844,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019999433420109073,
+      "loss": 6.3663,
+      "step": 108
+    },
+    {
+      "epoch": 0.011301997939199917,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00019999421798371997,
+      "loss": 6.2954,
+      "step": 109
+    },
+    {
+      "epoch": 0.01140568599368799,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00019999410058652313,
+      "loss": 6.2503,
+      "step": 110
+    },
+    {
+      "epoch": 0.011509374048176063,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00019999398200950158,
+      "loss": 6.2885,
+      "step": 111
+    },
+    {
+      "epoch": 0.011613062102664135,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00019999386225265676,
+      "loss": 6.3214,
+      "step": 112
+    },
+    {
+      "epoch": 0.011716750157152208,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00019999374131599007,
+      "loss": 6.2841,
+      "step": 113
+    },
+    {
+      "epoch": 0.01182043821164028,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00019999361919950293,
+      "loss": 6.29,
+      "step": 114
+    },
+    {
+      "epoch": 0.011924126266128352,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019999349590319677,
+      "loss": 6.2106,
+      "step": 115
+    },
+    {
+      "epoch": 0.012027814320616425,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00019999337142707305,
+      "loss": 6.219,
+      "step": 116
+    },
+    {
+      "epoch": 0.012131502375104498,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019999324577113324,
+      "loss": 6.2419,
+      "step": 117
+    },
+    {
+      "epoch": 0.012235190429592572,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019999311893537883,
+      "loss": 6.2168,
+      "step": 118
+    },
+    {
+      "epoch": 0.012338878484080643,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00019999299091981134,
+      "loss": 6.2602,
+      "step": 119
+    },
+    {
+      "epoch": 0.012442566538568716,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00019999286172443223,
+      "loss": 6.2084,
+      "step": 120
+    },
+    {
+      "epoch": 0.01254625459305679,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.00019999273134924307,
+      "loss": 6.2672,
+      "step": 121
+    },
+    {
+      "epoch": 0.01264994264754486,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00019999259979424535,
+      "loss": 6.2597,
+      "step": 122
+    },
+    {
+      "epoch": 0.012753630702032934,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019999246705944068,
+      "loss": 6.1437,
+      "step": 123
+    },
+    {
+      "epoch": 0.012857318756521007,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00019999233314483056,
+      "loss": 6.216,
+      "step": 124
+    },
+    {
+      "epoch": 0.01296100681100908,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00019999219805041663,
+      "loss": 6.1778,
+      "step": 125
+    },
+    {
+      "epoch": 0.013064694865497151,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.00019999206177620047,
+      "loss": 6.1466,
+      "step": 126
+    },
+    {
+      "epoch": 0.013168382919985224,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019999192432218363,
+      "loss": 6.1517,
+      "step": 127
+    },
+    {
+      "epoch": 0.013272070974473298,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019999178568836783,
+      "loss": 6.1833,
+      "step": 128
+    },
+    {
+      "epoch": 0.013375759028961369,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019999164587475464,
+      "loss": 6.177,
+      "step": 129
+    },
+    {
+      "epoch": 0.013479447083449442,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001999915048813457,
+      "loss": 6.19,
+      "step": 130
+    },
+    {
+      "epoch": 0.013583135137937515,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001999913627081427,
+      "loss": 6.1394,
+      "step": 131
+    },
+    {
+      "epoch": 0.013686823192425588,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019999121935514736,
+      "loss": 6.1704,
+      "step": 132
+    },
+    {
+      "epoch": 0.01379051124691366,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019999107482236128,
+      "loss": 6.1321,
+      "step": 133
+    },
+    {
+      "epoch": 0.013894199301401733,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019999092910978625,
+      "loss": 6.1726,
+      "step": 134
+    },
+    {
+      "epoch": 0.013997887355889806,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019999078221742393,
+      "loss": 6.1438,
+      "step": 135
+    },
+    {
+      "epoch": 0.014101575410377877,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019999063414527607,
+      "loss": 6.1395,
+      "step": 136
+    },
+    {
+      "epoch": 0.01420526346486595,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019999048489334443,
+      "loss": 6.0961,
+      "step": 137
+    },
+    {
+      "epoch": 0.014308951519354024,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019999033446163077,
+      "loss": 6.1165,
+      "step": 138
+    },
+    {
+      "epoch": 0.014412639573842097,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00019999018285013685,
+      "loss": 6.1087,
+      "step": 139
+    },
+    {
+      "epoch": 0.014516327628330168,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019999003005886446,
+      "loss": 6.0861,
+      "step": 140
+    },
+    {
+      "epoch": 0.014620015682818241,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019998987608781544,
+      "loss": 6.1238,
+      "step": 141
+    },
+    {
+      "epoch": 0.014723703737306314,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.00019998972093699153,
+      "loss": 6.0859,
+      "step": 142
+    },
+    {
+      "epoch": 0.014827391791794386,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00019998956460639465,
+      "loss": 6.1074,
+      "step": 143
+    },
+    {
+      "epoch": 0.014931079846282459,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00019998940709602657,
+      "loss": 6.1206,
+      "step": 144
+    },
+    {
+      "epoch": 0.015034767900770532,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00019998924840588917,
+      "loss": 6.0488,
+      "step": 145
+    },
+    {
+      "epoch": 0.015138455955258605,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.00019998908853598434,
+      "loss": 6.058,
+      "step": 146
+    },
+    {
+      "epoch": 0.015242144009746677,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.000199988927486314,
+      "loss": 6.0474,
+      "step": 147
+    },
+    {
+      "epoch": 0.01534583206423475,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019998876525687998,
+      "loss": 6.0019,
+      "step": 148
+    },
+    {
+      "epoch": 0.015449520118722823,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001999886018476842,
+      "loss": 6.047,
+      "step": 149
+    },
+    {
+      "epoch": 0.015553208173210894,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0001999884372587286,
+      "loss": 6.0433,
+      "step": 150
+    },
+    {
+      "epoch": 0.015656896227698967,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0001999882714900152,
+      "loss": 6.0162,
+      "step": 151
+    },
+    {
+      "epoch": 0.01576058428218704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019998810454154584,
+      "loss": 6.0574,
+      "step": 152
+    },
+    {
+      "epoch": 0.015864272336675114,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019998793641332256,
+      "loss": 6.0328,
+      "step": 153
+    },
+    {
+      "epoch": 0.015967960391163185,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001999877671053473,
+      "loss": 5.9934,
+      "step": 154
+    },
+    {
+      "epoch": 0.01607164844565126,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001999875966176221,
+      "loss": 6.0257,
+      "step": 155
+    },
+    {
+      "epoch": 0.01617533650013933,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019998742495014896,
+      "loss": 5.9716,
+      "step": 156
+    },
+    {
+      "epoch": 0.016279024554627403,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001999872521029299,
+      "loss": 6.011,
+      "step": 157
+    },
+    {
+      "epoch": 0.016382712609115477,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001999870780759669,
+      "loss": 5.986,
+      "step": 158
+    },
+    {
+      "epoch": 0.01648640066360355,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0001999869028692621,
+      "loss": 5.9172,
+      "step": 159
+    },
+    {
+      "epoch": 0.01659008871809162,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019998672648281757,
+      "loss": 5.9745,
+      "step": 160
+    },
+    {
+      "epoch": 0.016693776772579695,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0001999865489166353,
+      "loss": 5.9591,
+      "step": 161
+    },
+    {
+      "epoch": 0.016797464827067767,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019998637017071752,
+      "loss": 5.982,
+      "step": 162
+    },
+    {
+      "epoch": 0.016901152881555838,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.0001999861902450662,
+      "loss": 5.9767,
+      "step": 163
+    },
+    {
+      "epoch": 0.017004840936043913,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.00019998600913968356,
+      "loss": 6.0064,
+      "step": 164
+    },
+    {
+      "epoch": 0.017108528990531984,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00019998582685457165,
+      "loss": 5.9651,
+      "step": 165
+    },
+    {
+      "epoch": 0.017212217045020056,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00019998564338973273,
+      "loss": 5.9246,
+      "step": 166
+    },
+    {
+      "epoch": 0.01731590509950813,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00019998545874516888,
+      "loss": 5.907,
+      "step": 167
+    },
+    {
+      "epoch": 0.017419593153996202,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00019998527292088228,
+      "loss": 5.9681,
+      "step": 168
+    },
+    {
+      "epoch": 0.017523281208484277,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00019998508591687522,
+      "loss": 5.962,
+      "step": 169
+    },
+    {
+      "epoch": 0.017626969262972348,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00019998489773314976,
+      "loss": 5.8911,
+      "step": 170
+    },
+    {
+      "epoch": 0.01773065731746042,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.00019998470836970827,
+      "loss": 5.924,
+      "step": 171
+    },
+    {
+      "epoch": 0.017834345371948494,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.00019998451782655282,
+      "loss": 5.9342,
+      "step": 172
+    },
+    {
+      "epoch": 0.017938033426436566,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019998432610368583,
+      "loss": 5.9439,
+      "step": 173
+    },
+    {
+      "epoch": 0.018041721480924637,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00019998413320110943,
+      "loss": 5.9079,
+      "step": 174
+    },
+    {
+      "epoch": 0.018145409535412712,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019998393911882598,
+      "loss": 5.9049,
+      "step": 175
+    },
+    {
+      "epoch": 0.018249097589900783,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001999837438568377,
+      "loss": 5.8602,
+      "step": 176
+    },
+    {
+      "epoch": 0.018352785644388855,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019998354741514694,
+      "loss": 5.9309,
+      "step": 177
+    },
+    {
+      "epoch": 0.01845647369887693,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019998334979375604,
+      "loss": 5.9288,
+      "step": 178
+    },
+    {
+      "epoch": 0.018560161753365,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019998315099266728,
+      "loss": 5.8735,
+      "step": 179
+    },
+    {
+      "epoch": 0.018663849807853072,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.000199982951011883,
+      "loss": 5.8988,
+      "step": 180
+    },
+    {
+      "epoch": 0.018767537862341147,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001999827498514056,
+      "loss": 5.8772,
+      "step": 181
+    },
+    {
+      "epoch": 0.01887122591682922,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019998254751123746,
+      "loss": 5.8569,
+      "step": 182
+    },
+    {
+      "epoch": 0.018974913971317293,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019998234399138092,
+      "loss": 5.8623,
+      "step": 183
+    },
+    {
+      "epoch": 0.019078602025805365,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019998213929183842,
+      "loss": 5.832,
+      "step": 184
+    },
+    {
+      "epoch": 0.019182290080293436,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00019998193341261238,
+      "loss": 5.8703,
+      "step": 185
+    },
+    {
+      "epoch": 0.01928597813478151,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019998172635370516,
+      "loss": 5.8666,
+      "step": 186
+    },
+    {
+      "epoch": 0.019389666189269582,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019998151811511928,
+      "loss": 5.8473,
+      "step": 187
+    },
+    {
+      "epoch": 0.019493354243757654,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00019998130869685717,
+      "loss": 5.8832,
+      "step": 188
+    },
+    {
+      "epoch": 0.01959704229824573,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019998109809892133,
+      "loss": 5.8562,
+      "step": 189
+    },
+    {
+      "epoch": 0.0197007303527338,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00019998088632131419,
+      "loss": 5.8587,
+      "step": 190
+    },
+    {
+      "epoch": 0.01980441840722187,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019998067336403827,
+      "loss": 5.8556,
+      "step": 191
+    },
+    {
+      "epoch": 0.019908106461709946,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0001999804592270961,
+      "loss": 5.8503,
+      "step": 192
+    },
+    {
+      "epoch": 0.020011794516198018,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0001999802439104902,
+      "loss": 5.8315,
+      "step": 193
+    },
+    {
+      "epoch": 0.02011548257068609,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.0001999800274142231,
+      "loss": 5.7988,
+      "step": 194
+    },
+    {
+      "epoch": 0.020219170625174164,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00019997980973829736,
+      "loss": 5.7978,
+      "step": 195
+    },
+    {
+      "epoch": 0.020322858679662235,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00019997959088271554,
+      "loss": 5.8234,
+      "step": 196
+    },
+    {
+      "epoch": 0.02042654673415031,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00019997937084748025,
+      "loss": 5.8187,
+      "step": 197
+    },
+    {
+      "epoch": 0.02053023478863838,
+      "grad_norm": 3.75,
+      "learning_rate": 0.00019997914963259405,
+      "loss": 5.8793,
+      "step": 198
+    },
+    {
+      "epoch": 0.020633922843126453,
+      "grad_norm": 3.390625,
+      "learning_rate": 0.00019997892723805957,
+      "loss": 5.8885,
+      "step": 199
+    },
+    {
+      "epoch": 0.020737610897614528,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00019997870366387943,
+      "loss": 5.8362,
+      "step": 200
+    },
+    {
+      "epoch": 0.0208412989521026,
+      "grad_norm": 3.171875,
+      "learning_rate": 0.00019997847891005627,
+      "loss": 5.8813,
+      "step": 201
+    },
+    {
+      "epoch": 0.02094498700659067,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00019997825297659273,
+      "loss": 5.9057,
+      "step": 202
+    },
+    {
+      "epoch": 0.021048675061078746,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.0001999780258634915,
+      "loss": 5.8437,
+      "step": 203
+    },
+    {
+      "epoch": 0.021152363115566817,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019997779757075526,
+      "loss": 5.8491,
+      "step": 204
+    },
+    {
+      "epoch": 0.02125605117005489,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.0001999775680983867,
+      "loss": 5.8136,
+      "step": 205
+    },
+    {
+      "epoch": 0.021359739224542963,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019997733744638846,
+      "loss": 5.8221,
+      "step": 206
+    },
+    {
+      "epoch": 0.021463427279031035,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00019997710561476335,
+      "loss": 5.8324,
+      "step": 207
+    },
+    {
+      "epoch": 0.02156711533351911,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001999768726035141,
+      "loss": 5.8158,
+      "step": 208
+    },
+    {
+      "epoch": 0.02167080338800718,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019997663841264337,
+      "loss": 5.8085,
+      "step": 209
+    },
+    {
+      "epoch": 0.021774491442495252,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00019997640304215402,
+      "loss": 5.807,
+      "step": 210
+    },
+    {
+      "epoch": 0.021878179496983327,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001999761664920488,
+      "loss": 5.8282,
+      "step": 211
+    },
+    {
+      "epoch": 0.0219818675514714,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001999759287623305,
+      "loss": 5.7731,
+      "step": 212
+    },
+    {
+      "epoch": 0.02208555560595947,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0001999756898530019,
+      "loss": 5.7734,
+      "step": 213
+    },
+    {
+      "epoch": 0.022189243660447545,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019997544976406588,
+      "loss": 5.7714,
+      "step": 214
+    },
+    {
+      "epoch": 0.022292931714935616,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019997520849552517,
+      "loss": 5.7947,
+      "step": 215
+    },
+    {
+      "epoch": 0.022396619769423688,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019997496604738272,
+      "loss": 5.8244,
+      "step": 216
+    },
+    {
+      "epoch": 0.022500307823911762,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019997472241964134,
+      "loss": 5.7605,
+      "step": 217
+    },
+    {
+      "epoch": 0.022603995878399834,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019997447761230393,
+      "loss": 5.7564,
+      "step": 218
+    },
+    {
+      "epoch": 0.022707683932887905,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019997423162537335,
+      "loss": 5.7313,
+      "step": 219
+    },
+    {
+      "epoch": 0.02281137198737598,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019997398445885248,
+      "loss": 5.7635,
+      "step": 220
+    },
+    {
+      "epoch": 0.02291506004186405,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019997373611274432,
+      "loss": 5.7636,
+      "step": 221
+    },
+    {
+      "epoch": 0.023018748096352126,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019997348658705173,
+      "loss": 5.7049,
+      "step": 222
+    },
+    {
+      "epoch": 0.023122436150840198,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019997323588177767,
+      "loss": 5.7628,
+      "step": 223
+    },
+    {
+      "epoch": 0.02322612420532827,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0001999729839969251,
+      "loss": 5.7761,
+      "step": 224
+    },
+    {
+      "epoch": 0.023329812259816344,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.000199972730932497,
+      "loss": 5.772,
+      "step": 225
+    },
+    {
+      "epoch": 0.023433500314304415,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00019997247668849638,
+      "loss": 5.7357,
+      "step": 226
+    },
+    {
+      "epoch": 0.023537188368792487,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019997222126492617,
+      "loss": 5.75,
+      "step": 227
+    },
+    {
+      "epoch": 0.02364087642328056,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00019997196466178943,
+      "loss": 5.7343,
+      "step": 228
+    },
+    {
+      "epoch": 0.023744564477768633,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00019997170687908919,
+      "loss": 5.7152,
+      "step": 229
+    },
+    {
+      "epoch": 0.023848252532256704,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00019997144791682848,
+      "loss": 5.734,
+      "step": 230
+    },
+    {
+      "epoch": 0.02395194058674478,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00019997118777501037,
+      "loss": 5.7307,
+      "step": 231
+    },
+    {
+      "epoch": 0.02405562864123285,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001999709264536379,
+      "loss": 5.7429,
+      "step": 232
+    },
+    {
+      "epoch": 0.024159316695720922,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00019997066395271418,
+      "loss": 5.7384,
+      "step": 233
+    },
+    {
+      "epoch": 0.024263004750208997,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019997040027224232,
+      "loss": 5.7171,
+      "step": 234
+    },
+    {
+      "epoch": 0.024366692804697068,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00019997013541222538,
+      "loss": 5.7003,
+      "step": 235
+    },
+    {
+      "epoch": 0.024470380859185143,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019996986937266653,
+      "loss": 5.6908,
+      "step": 236
+    },
+    {
+      "epoch": 0.024574068913673214,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001999696021535689,
+      "loss": 5.6583,
+      "step": 237
+    },
+    {
+      "epoch": 0.024677756968161286,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019996933375493562,
+      "loss": 5.724,
+      "step": 238
+    },
+    {
+      "epoch": 0.02478144502264936,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001999690641767699,
+      "loss": 5.6948,
+      "step": 239
+    },
+    {
+      "epoch": 0.024885133077137432,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019996879341907487,
+      "loss": 5.7105,
+      "step": 240
+    },
+    {
+      "epoch": 0.024988821131625504,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00019996852148185373,
+      "loss": 5.6769,
+      "step": 241
+    },
+    {
+      "epoch": 0.02509250918611358,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00019996824836510975,
+      "loss": 5.671,
+      "step": 242
+    },
+    {
+      "epoch": 0.02519619724060165,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001999679740688461,
+      "loss": 5.6549,
+      "step": 243
+    },
+    {
+      "epoch": 0.02529988529508972,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.000199967698593066,
+      "loss": 5.6671,
+      "step": 244
+    },
+    {
+      "epoch": 0.025403573349577796,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00019996742193777273,
+      "loss": 5.6164,
+      "step": 245
+    },
+    {
+      "epoch": 0.025507261404065867,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019996714410296958,
+      "loss": 5.647,
+      "step": 246
+    },
+    {
+      "epoch": 0.02561094945855394,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0001999668650886598,
+      "loss": 5.64,
+      "step": 247
+    },
+    {
+      "epoch": 0.025714637513042014,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019996658489484666,
+      "loss": 5.6899,
+      "step": 248
+    },
+    {
+      "epoch": 0.025818325567530085,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019996630352153353,
+      "loss": 5.6688,
+      "step": 249
+    },
+    {
+      "epoch": 0.02592201362201816,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.0001999660209687236,
+      "loss": 5.6829,
+      "step": 250
+    },
+    {
+      "epoch": 0.02602570167650623,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019996573723642035,
+      "loss": 5.6252,
+      "step": 251
+    },
+    {
+      "epoch": 0.026129389730994303,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019996545232462708,
+      "loss": 5.6089,
+      "step": 252
+    },
+    {
+      "epoch": 0.026233077785482378,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00019996516623334713,
+      "loss": 5.672,
+      "step": 253
+    },
+    {
+      "epoch": 0.02633676583997045,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00019996487896258388,
+      "loss": 5.6516,
+      "step": 254
+    },
+    {
+      "epoch": 0.02644045389445852,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001999645905123407,
+      "loss": 5.608,
+      "step": 255
+    },
+    {
+      "epoch": 0.026544141948946595,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.00019996430088262108,
+      "loss": 5.6102,
+      "step": 256
+    },
+    {
+      "epoch": 0.026647830003434667,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019996401007342832,
+      "loss": 5.6053,
+      "step": 257
+    },
+    {
+      "epoch": 0.026751518057922738,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019996371808476596,
+      "loss": 5.6318,
+      "step": 258
+    },
+    {
+      "epoch": 0.026855206112410813,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019996342491663733,
+      "loss": 5.612,
+      "step": 259
+    },
+    {
+      "epoch": 0.026958894166898884,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.000199963130569046,
+      "loss": 5.6529,
+      "step": 260
+    },
+    {
+      "epoch": 0.02706258222138696,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019996283504199538,
+      "loss": 5.5798,
+      "step": 261
+    },
+    {
+      "epoch": 0.02716627027587503,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019996253833548896,
+      "loss": 5.6042,
+      "step": 262
+    },
+    {
+      "epoch": 0.027269958330363102,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019996224044953028,
+      "loss": 5.6064,
+      "step": 263
+    },
+    {
+      "epoch": 0.027373646384851177,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001999619413841228,
+      "loss": 5.6279,
+      "step": 264
+    },
+    {
+      "epoch": 0.027477334439339248,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019996164113927008,
+      "loss": 5.595,
+      "step": 265
+    },
+    {
+      "epoch": 0.02758102249382732,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00019996133971497568,
+      "loss": 5.6144,
+      "step": 266
+    },
+    {
+      "epoch": 0.027684710548315394,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00019996103711124313,
+      "loss": 5.6075,
+      "step": 267
+    },
+    {
+      "epoch": 0.027788398602803466,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.000199960733328076,
+      "loss": 5.5957,
+      "step": 268
+    },
+    {
+      "epoch": 0.027892086657291537,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00019996042836547786,
+      "loss": 5.5676,
+      "step": 269
+    },
+    {
+      "epoch": 0.027995774711779612,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00019996012222345236,
+      "loss": 5.606,
+      "step": 270
+    },
+    {
+      "epoch": 0.028099462766267683,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019995981490200304,
+      "loss": 5.5437,
+      "step": 271
+    },
+    {
+      "epoch": 0.028203150820755755,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0001999595064011336,
+      "loss": 5.5388,
+      "step": 272
+    },
+    {
+      "epoch": 0.02830683887524383,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019995919672084763,
+      "loss": 5.5548,
+      "step": 273
+    },
+    {
+      "epoch": 0.0284105269297319,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0001999588858611488,
+      "loss": 5.5774,
+      "step": 274
+    },
+    {
+      "epoch": 0.028514214984219976,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019995857382204083,
+      "loss": 5.5795,
+      "step": 275
+    },
+    {
+      "epoch": 0.028617903038708047,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019995826060352728,
+      "loss": 5.5806,
+      "step": 276
+    },
+    {
+      "epoch": 0.02872159109319612,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019995794620561195,
+      "loss": 5.5304,
+      "step": 277
+    },
+    {
+      "epoch": 0.028825279147684194,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0001999576306282985,
+      "loss": 5.594,
+      "step": 278
+    },
+    {
+      "epoch": 0.028928967202172265,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019995731387159067,
+      "loss": 5.5603,
+      "step": 279
+    },
+    {
+      "epoch": 0.029032655256660336,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0001999569959354922,
+      "loss": 5.5512,
+      "step": 280
+    },
+    {
+      "epoch": 0.02913634331114841,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00019995667682000683,
+      "loss": 5.5618,
+      "step": 281
+    },
+    {
+      "epoch": 0.029240031365636483,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019995635652513835,
+      "loss": 5.5426,
+      "step": 282
+    },
+    {
+      "epoch": 0.029343719420124554,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001999560350508905,
+      "loss": 5.5263,
+      "step": 283
+    },
+    {
+      "epoch": 0.02944740747461263,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001999557123972671,
+      "loss": 5.5721,
+      "step": 284
+    },
+    {
+      "epoch": 0.0295510955291007,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00019995538856427196,
+      "loss": 5.5413,
+      "step": 285
+    },
+    {
+      "epoch": 0.02965478358358877,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019995506355190889,
+      "loss": 5.5277,
+      "step": 286
+    },
+    {
+      "epoch": 0.029758471638076846,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00019995473736018172,
+      "loss": 5.5505,
+      "step": 287
+    },
+    {
+      "epoch": 0.029862159692564918,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00019995440998909431,
+      "loss": 5.5775,
+      "step": 288
+    },
+    {
+      "epoch": 0.029965847747052993,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00019995408143865052,
+      "loss": 5.5016,
+      "step": 289
+    },
+    {
+      "epoch": 0.030069535801541064,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00019995375170885424,
+      "loss": 5.5683,
+      "step": 290
+    },
+    {
+      "epoch": 0.030173223856029135,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019995342079970932,
+      "loss": 5.5217,
+      "step": 291
+    },
+    {
+      "epoch": 0.03027691191051721,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019995308871121971,
+      "loss": 5.562,
+      "step": 292
+    },
+    {
+      "epoch": 0.03038059996500528,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019995275544338928,
+      "loss": 5.5251,
+      "step": 293
+    },
+    {
+      "epoch": 0.030484288019493353,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00019995242099622203,
+      "loss": 5.5147,
+      "step": 294
+    },
+    {
+      "epoch": 0.030587976073981428,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00019995208536972183,
+      "loss": 5.4956,
+      "step": 295
+    },
+    {
+      "epoch": 0.0306916641284695,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0001999517485638927,
+      "loss": 5.4974,
+      "step": 296
+    },
+    {
+      "epoch": 0.03079535218295757,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019995141057873857,
+      "loss": 5.4782,
+      "step": 297
+    },
+    {
+      "epoch": 0.030899040237445646,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00019995107141426347,
+      "loss": 5.5044,
+      "step": 298
+    },
+    {
+      "epoch": 0.031002728291933717,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00019995073107047134,
+      "loss": 5.5123,
+      "step": 299
+    },
+    {
+      "epoch": 0.03110641634642179,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0001999503895473663,
+      "loss": 5.5219,
+      "step": 300
+    },
+    {
+      "epoch": 0.031210104400909863,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00019995004684495227,
+      "loss": 5.5151,
+      "step": 301
+    },
+    {
+      "epoch": 0.031313792455397935,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00019994970296323335,
+      "loss": 5.5151,
+      "step": 302
+    },
+    {
+      "epoch": 0.031417480509886006,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.00019994935790221358,
+      "loss": 5.5093,
+      "step": 303
+    },
+    {
+      "epoch": 0.03152116856437408,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00019994901166189708,
+      "loss": 5.5052,
+      "step": 304
+    },
+    {
+      "epoch": 0.031624856618862156,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00019994866424228783,
+      "loss": 5.4662,
+      "step": 305
+    },
+    {
+      "epoch": 0.03172854467335023,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.00019994831564339004,
+      "loss": 5.4841,
+      "step": 306
+    },
+    {
+      "epoch": 0.0318322327278383,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019994796586520773,
+      "loss": 5.4731,
+      "step": 307
+    },
+    {
+      "epoch": 0.03193592078232637,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.00019994761490774513,
+      "loss": 5.4515,
+      "step": 308
+    },
+    {
+      "epoch": 0.03203960883681444,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019994726277100628,
+      "loss": 5.519,
+      "step": 309
+    },
+    {
+      "epoch": 0.03214329689130252,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0001999469094549954,
+      "loss": 5.5203,
+      "step": 310
+    },
+    {
+      "epoch": 0.03224698494579059,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001999465549597166,
+      "loss": 5.5374,
+      "step": 311
+    },
+    {
+      "epoch": 0.03235067300027866,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019994619928517416,
+      "loss": 5.508,
+      "step": 312
+    },
+    {
+      "epoch": 0.032454361054766734,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019994584243137218,
+      "loss": 5.4988,
+      "step": 313
+    },
+    {
+      "epoch": 0.032558049109254805,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019994548439831487,
+      "loss": 5.4792,
+      "step": 314
+    },
+    {
+      "epoch": 0.03266173716374288,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00019994512518600654,
+      "loss": 5.4493,
+      "step": 315
+    },
+    {
+      "epoch": 0.032765425218230955,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001999447647944514,
+      "loss": 5.4875,
+      "step": 316
+    },
+    {
+      "epoch": 0.032869113272719026,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019994440322365363,
+      "loss": 5.5292,
+      "step": 317
+    },
+    {
+      "epoch": 0.0329728013272071,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019994404047361756,
+      "loss": 5.4906,
+      "step": 318
+    },
+    {
+      "epoch": 0.03307648938169517,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00019994367654434746,
+      "loss": 5.461,
+      "step": 319
+    },
+    {
+      "epoch": 0.03318017743618324,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00019994331143584763,
+      "loss": 5.439,
+      "step": 320
+    },
+    {
+      "epoch": 0.03328386549067131,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019994294514812238,
+      "loss": 5.451,
+      "step": 321
+    },
+    {
+      "epoch": 0.03338755354515939,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019994257768117602,
+      "loss": 5.4436,
+      "step": 322
+    },
+    {
+      "epoch": 0.03349124159964746,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0001999422090350129,
+      "loss": 5.4283,
+      "step": 323
+    },
+    {
+      "epoch": 0.03359492965413553,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001999418392096373,
+      "loss": 5.4542,
+      "step": 324
+    },
+    {
+      "epoch": 0.033698617708623604,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001999414682050537,
+      "loss": 5.4371,
+      "step": 325
+    },
+    {
+      "epoch": 0.033802305763111676,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00019994109602126638,
+      "loss": 5.4293,
+      "step": 326
+    },
+    {
+      "epoch": 0.033905993817599754,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019994072265827977,
+      "loss": 5.4883,
+      "step": 327
+    },
+    {
+      "epoch": 0.034009681872087826,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0001999403481160983,
+      "loss": 5.4215,
+      "step": 328
+    },
+    {
+      "epoch": 0.0341133699265759,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019993997239472634,
+      "loss": 5.4048,
+      "step": 329
+    },
+    {
+      "epoch": 0.03421705798106397,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019993959549416835,
+      "loss": 5.4454,
+      "step": 330
+    },
+    {
+      "epoch": 0.03432074603555204,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00019993921741442877,
+      "loss": 5.4441,
+      "step": 331
+    },
+    {
+      "epoch": 0.03442443409004011,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001999388381555121,
+      "loss": 5.4147,
+      "step": 332
+    },
+    {
+      "epoch": 0.03452812214452819,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00019993845771742276,
+      "loss": 5.4387,
+      "step": 333
+    },
+    {
+      "epoch": 0.03463181019901626,
+      "grad_norm": 1.921875,
+      "learning_rate": 0.00019993807610016524,
+      "loss": 5.3842,
+      "step": 334
+    },
+    {
+      "epoch": 0.03473549825350433,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019993769330374408,
+      "loss": 5.4189,
+      "step": 335
+    },
+    {
+      "epoch": 0.034839186307992404,
+      "grad_norm": 5.53125,
+      "learning_rate": 0.00019993730932816377,
+      "loss": 5.5256,
+      "step": 336
+    },
+    {
+      "epoch": 0.034942874362480475,
+      "grad_norm": 5.0,
+      "learning_rate": 0.00019993692417342884,
+      "loss": 5.5175,
+      "step": 337
+    },
+    {
+      "epoch": 0.03504656241696855,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00019993653783954388,
+      "loss": 5.4119,
+      "step": 338
+    },
+    {
+      "epoch": 0.035150250471456625,
+      "grad_norm": 3.4375,
+      "learning_rate": 0.00019993615032651337,
+      "loss": 5.5242,
+      "step": 339
+    },
+    {
+      "epoch": 0.035253938525944696,
+      "grad_norm": 3.15625,
+      "learning_rate": 0.00019993576163434193,
+      "loss": 5.53,
+      "step": 340
+    },
+    {
+      "epoch": 0.03535762658043277,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00019993537176303416,
+      "loss": 5.4046,
+      "step": 341
+    },
+    {
+      "epoch": 0.03546131463492084,
+      "grad_norm": 2.59375,
+      "learning_rate": 0.00019993498071259463,
+      "loss": 5.4244,
+      "step": 342
+    },
+    {
+      "epoch": 0.03556500268940891,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00019993458848302796,
+      "loss": 5.4798,
+      "step": 343
+    },
+    {
+      "epoch": 0.03566869074389699,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00019993419507433876,
+      "loss": 5.4492,
+      "step": 344
+    },
+    {
+      "epoch": 0.03577237879838506,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019993380048653175,
+      "loss": 5.4628,
+      "step": 345
+    },
+    {
+      "epoch": 0.03587606685287313,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001999334047196115,
+      "loss": 5.4155,
+      "step": 346
+    },
+    {
+      "epoch": 0.0359797549073612,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019993300777358268,
+      "loss": 5.4204,
+      "step": 347
+    },
+    {
+      "epoch": 0.036083442961849274,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00019993260964845,
+      "loss": 5.448,
+      "step": 348
+    },
+    {
+      "epoch": 0.03618713101633735,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001999322103442182,
+      "loss": 5.4505,
+      "step": 349
+    },
+    {
+      "epoch": 0.036290819070825424,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00019993180986089192,
+      "loss": 5.4001,
+      "step": 350
+    },
+    {
+      "epoch": 0.036394507125313495,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00019993140819847595,
+      "loss": 5.3998,
+      "step": 351
+    },
+    {
+      "epoch": 0.03649819517980157,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00019993100535697496,
+      "loss": 5.4476,
+      "step": 352
+    },
+    {
+      "epoch": 0.03660188323428964,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00019993060133639376,
+      "loss": 5.4657,
+      "step": 353
+    },
+    {
+      "epoch": 0.03670557128877771,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00019993019613673708,
+      "loss": 5.4182,
+      "step": 354
+    },
+    {
+      "epoch": 0.03680925934326579,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019992978975800972,
+      "loss": 5.4388,
+      "step": 355
+    },
+    {
+      "epoch": 0.03691294739775386,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0001999293822002165,
+      "loss": 5.4333,
+      "step": 356
+    },
+    {
+      "epoch": 0.03701663545224193,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019992897346336218,
+      "loss": 5.4078,
+      "step": 357
+    },
+    {
+      "epoch": 0.03712032350673,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019992856354745158,
+      "loss": 5.3843,
+      "step": 358
+    },
+    {
+      "epoch": 0.03722401156121807,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019992815245248958,
+      "loss": 5.4116,
+      "step": 359
+    },
+    {
+      "epoch": 0.037327699615706145,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.000199927740178481,
+      "loss": 5.3862,
+      "step": 360
+    },
+    {
+      "epoch": 0.03743138767019422,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019992732672543073,
+      "loss": 5.3852,
+      "step": 361
+    },
+    {
+      "epoch": 0.037535075724682294,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019992691209334362,
+      "loss": 5.3804,
+      "step": 362
+    },
+    {
+      "epoch": 0.037638763779170366,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.0001999264962822246,
+      "loss": 5.3745,
+      "step": 363
+    },
+    {
+      "epoch": 0.03774245183365844,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019992607929207853,
+      "loss": 5.3948,
+      "step": 364
+    },
+    {
+      "epoch": 0.03784613988814651,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00019992566112291034,
+      "loss": 5.384,
+      "step": 365
+    },
+    {
+      "epoch": 0.03794982794263459,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.000199925241774725,
+      "loss": 5.3606,
+      "step": 366
+    },
+    {
+      "epoch": 0.03805351599712266,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001999248212475274,
+      "loss": 5.4083,
+      "step": 367
+    },
+    {
+      "epoch": 0.03815720405161073,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019992439954132256,
+      "loss": 5.3636,
+      "step": 368
+    },
+    {
+      "epoch": 0.0382608921060988,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019992397665611543,
+      "loss": 5.3986,
+      "step": 369
+    },
+    {
+      "epoch": 0.03836458016058687,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00019992355259191097,
+      "loss": 5.4029,
+      "step": 370
+    },
+    {
+      "epoch": 0.038468268215074944,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019992312734871425,
+      "loss": 5.405,
+      "step": 371
+    },
+    {
+      "epoch": 0.03857195626956302,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00019992270092653022,
+      "loss": 5.3392,
+      "step": 372
+    },
+    {
+      "epoch": 0.038675644324051094,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00019992227332536397,
+      "loss": 5.3901,
+      "step": 373
+    },
+    {
+      "epoch": 0.038779332378539165,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019992184454522053,
+      "loss": 5.3694,
+      "step": 374
+    },
+    {
+      "epoch": 0.038883020433027236,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001999214145861049,
+      "loss": 5.359,
+      "step": 375
+    },
+    {
+      "epoch": 0.03898670848751531,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00019992098344802223,
+      "loss": 5.3545,
+      "step": 376
+    },
+    {
+      "epoch": 0.039090396542003386,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00019992055113097755,
+      "loss": 5.3735,
+      "step": 377
+    },
+    {
+      "epoch": 0.03919408459649146,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.000199920117634976,
+      "loss": 5.3751,
+      "step": 378
+    },
+    {
+      "epoch": 0.03929777265097953,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001999196829600227,
+      "loss": 5.2899,
+      "step": 379
+    },
+    {
+      "epoch": 0.0394014607054676,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019991924710612276,
+      "loss": 5.3484,
+      "step": 380
+    },
+    {
+      "epoch": 0.03950514875995567,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00019991881007328131,
+      "loss": 5.3171,
+      "step": 381
+    },
+    {
+      "epoch": 0.03960883681444374,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.0001999183718615035,
+      "loss": 5.3399,
+      "step": 382
+    },
+    {
+      "epoch": 0.03971252486893182,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019991793247079457,
+      "loss": 5.3487,
+      "step": 383
+    },
+    {
+      "epoch": 0.03981621292341989,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00019991749190115962,
+      "loss": 5.335,
+      "step": 384
+    },
+    {
+      "epoch": 0.039919900977907964,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.0001999170501526039,
+      "loss": 5.333,
+      "step": 385
+    },
+    {
+      "epoch": 0.040023589032396036,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019991660722513258,
+      "loss": 5.3132,
+      "step": 386
+    },
+    {
+      "epoch": 0.04012727708688411,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019991616311875092,
+      "loss": 5.3834,
+      "step": 387
+    },
+    {
+      "epoch": 0.04023096514137218,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00019991571783346416,
+      "loss": 5.3623,
+      "step": 388
+    },
+    {
+      "epoch": 0.04033465319586026,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019991527136927753,
+      "loss": 5.3597,
+      "step": 389
+    },
+    {
+      "epoch": 0.04043834125034833,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019991482372619634,
+      "loss": 5.3423,
+      "step": 390
+    },
+    {
+      "epoch": 0.0405420293048364,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0001999143749042258,
+      "loss": 5.3152,
+      "step": 391
+    },
+    {
+      "epoch": 0.04064571735932447,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019991392490337128,
+      "loss": 5.3344,
+      "step": 392
+    },
+    {
+      "epoch": 0.04074940541381254,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019991347372363806,
+      "loss": 5.3016,
+      "step": 393
+    },
+    {
+      "epoch": 0.04085309346830062,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00019991302136503148,
+      "loss": 5.3684,
+      "step": 394
+    },
+    {
+      "epoch": 0.04095678152278869,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00019991256782755684,
+      "loss": 5.3299,
+      "step": 395
+    },
+    {
+      "epoch": 0.04106046957727676,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001999121131112195,
+      "loss": 5.2799,
+      "step": 396
+    },
+    {
+      "epoch": 0.041164157631764835,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00019991165721602484,
+      "loss": 5.3184,
+      "step": 397
+    },
+    {
+      "epoch": 0.041267845686252906,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019991120014197828,
+      "loss": 5.3399,
+      "step": 398
+    },
+    {
+      "epoch": 0.04137153374074098,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00019991074188908513,
+      "loss": 5.2717,
+      "step": 399
+    },
+    {
+      "epoch": 0.041475221795229056,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019991028245735083,
+      "loss": 5.344,
+      "step": 400
+    },
+    {
+      "epoch": 0.04157890984971713,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019990982184678086,
+      "loss": 5.3392,
+      "step": 401
+    },
+    {
+      "epoch": 0.0416825979042052,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00019990936005738052,
+      "loss": 5.262,
+      "step": 402
+    },
+    {
+      "epoch": 0.04178628595869327,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0001999088970891554,
+      "loss": 5.3249,
+      "step": 403
+    },
+    {
+      "epoch": 0.04188997401318134,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019990843294211087,
+      "loss": 5.3172,
+      "step": 404
+    },
+    {
+      "epoch": 0.04199366206766942,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019990796761625246,
+      "loss": 5.3205,
+      "step": 405
+    },
+    {
+      "epoch": 0.04209735012215749,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019990750111158564,
+      "loss": 5.3259,
+      "step": 406
+    },
+    {
+      "epoch": 0.04220103817664556,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001999070334281159,
+      "loss": 5.3271,
+      "step": 407
+    },
+    {
+      "epoch": 0.042304726231133634,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019990656456584876,
+      "loss": 5.3035,
+      "step": 408
+    },
+    {
+      "epoch": 0.042408414285621705,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001999060945247898,
+      "loss": 5.2843,
+      "step": 409
+    },
+    {
+      "epoch": 0.04251210234010978,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001999056233049445,
+      "loss": 5.2869,
+      "step": 410
+    },
+    {
+      "epoch": 0.042615790394597855,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019990515090631848,
+      "loss": 5.2991,
+      "step": 411
+    },
+    {
+      "epoch": 0.042719478449085926,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019990467732891725,
+      "loss": 5.2575,
+      "step": 412
+    },
+    {
+      "epoch": 0.042823166503574,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019990420257274643,
+      "loss": 5.3164,
+      "step": 413
+    },
+    {
+      "epoch": 0.04292685455806207,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019990372663781166,
+      "loss": 5.2825,
+      "step": 414
+    },
+    {
+      "epoch": 0.04303054261255014,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019990324952411846,
+      "loss": 5.2654,
+      "step": 415
+    },
+    {
+      "epoch": 0.04313423066703822,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019990277123167258,
+      "loss": 5.2965,
+      "step": 416
+    },
+    {
+      "epoch": 0.04323791872152629,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019990229176047958,
+      "loss": 5.2539,
+      "step": 417
+    },
+    {
+      "epoch": 0.04334160677601436,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0001999018111105451,
+      "loss": 5.2908,
+      "step": 418
+    },
+    {
+      "epoch": 0.04344529483050243,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0001999013292818749,
+      "loss": 5.2979,
+      "step": 419
+    },
+    {
+      "epoch": 0.043548982884990504,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001999008462744746,
+      "loss": 5.2747,
+      "step": 420
+    },
+    {
+      "epoch": 0.043652670939478576,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00019990036208834992,
+      "loss": 5.3147,
+      "step": 421
+    },
+    {
+      "epoch": 0.043756358993966654,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00019989987672350656,
+      "loss": 5.2847,
+      "step": 422
+    },
+    {
+      "epoch": 0.043860047048454726,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019989939017995024,
+      "loss": 5.2892,
+      "step": 423
+    },
+    {
+      "epoch": 0.0439637351029428,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019989890245768673,
+      "loss": 5.2348,
+      "step": 424
+    },
+    {
+      "epoch": 0.04406742315743087,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019989841355672178,
+      "loss": 5.2523,
+      "step": 425
+    },
+    {
+      "epoch": 0.04417111121191894,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019989792347706114,
+      "loss": 5.2648,
+      "step": 426
+    },
+    {
+      "epoch": 0.04427479926640701,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019989743221871057,
+      "loss": 5.2998,
+      "step": 427
+    },
+    {
+      "epoch": 0.04437848732089509,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019989693978167595,
+      "loss": 5.1773,
+      "step": 428
+    },
+    {
+      "epoch": 0.04448217537538316,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019989644616596298,
+      "loss": 5.2602,
+      "step": 429
+    },
+    {
+      "epoch": 0.04458586342987123,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019989595137157758,
+      "loss": 5.2278,
+      "step": 430
+    },
+    {
+      "epoch": 0.044689551484359304,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0001998954553985255,
+      "loss": 5.2847,
+      "step": 431
+    },
+    {
+      "epoch": 0.044793239538847375,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0001998949582468127,
+      "loss": 5.2977,
+      "step": 432
+    },
+    {
+      "epoch": 0.04489692759333545,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00019989445991644496,
+      "loss": 5.2542,
+      "step": 433
+    },
+    {
+      "epoch": 0.045000615647823525,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019989396040742818,
+      "loss": 5.2308,
+      "step": 434
+    },
+    {
+      "epoch": 0.045104303702311596,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019989345971976828,
+      "loss": 5.2806,
+      "step": 435
+    },
+    {
+      "epoch": 0.04520799175679967,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00019989295785347112,
+      "loss": 5.2863,
+      "step": 436
+    },
+    {
+      "epoch": 0.04531167981128774,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019989245480854265,
+      "loss": 5.1997,
+      "step": 437
+    },
+    {
+      "epoch": 0.04541536786577581,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.00019989195058498882,
+      "loss": 5.264,
+      "step": 438
+    },
+    {
+      "epoch": 0.04551905592026389,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00019989144518281558,
+      "loss": 5.2664,
+      "step": 439
+    },
+    {
+      "epoch": 0.04562274397475196,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00019989093860202885,
+      "loss": 5.2437,
+      "step": 440
+    },
+    {
+      "epoch": 0.04572643202924003,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019989043084263464,
+      "loss": 5.2767,
+      "step": 441
+    },
+    {
+      "epoch": 0.0458301200837281,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019988992190463894,
+      "loss": 5.2763,
+      "step": 442
+    },
+    {
+      "epoch": 0.045933808138216174,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00019988941178804775,
+      "loss": 5.24,
+      "step": 443
+    },
+    {
+      "epoch": 0.04603749619270425,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00019988890049286705,
+      "loss": 5.2116,
+      "step": 444
+    },
+    {
+      "epoch": 0.046141184247192324,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00019988838801910297,
+      "loss": 5.269,
+      "step": 445
+    },
+    {
+      "epoch": 0.046244872301680395,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019988787436676147,
+      "loss": 5.2098,
+      "step": 446
+    },
+    {
+      "epoch": 0.04634856035616847,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019988735953584862,
+      "loss": 5.2502,
+      "step": 447
+    },
+    {
+      "epoch": 0.04645224841065654,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019988684352637056,
+      "loss": 5.2842,
+      "step": 448
+    },
+    {
+      "epoch": 0.04655593646514461,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001998863263383333,
+      "loss": 5.2292,
+      "step": 449
+    },
+    {
+      "epoch": 0.04665962451963269,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00019988580797174297,
+      "loss": 5.2584,
+      "step": 450
+    },
+    {
+      "epoch": 0.04676331257412076,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0001998852884266057,
+      "loss": 5.1916,
+      "step": 451
+    },
+    {
+      "epoch": 0.04686700062860883,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019988476770292762,
+      "loss": 5.2374,
+      "step": 452
+    },
+    {
+      "epoch": 0.0469706886830969,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019988424580071485,
+      "loss": 5.2478,
+      "step": 453
+    },
+    {
+      "epoch": 0.04707437673758497,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019988372271997356,
+      "loss": 5.2032,
+      "step": 454
+    },
+    {
+      "epoch": 0.04717806479207305,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0001998831984607099,
+      "loss": 5.2208,
+      "step": 455
+    },
+    {
+      "epoch": 0.04728175284656112,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019988267302293013,
+      "loss": 5.2389,
+      "step": 456
+    },
+    {
+      "epoch": 0.047385440901049194,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00019988214640664036,
+      "loss": 5.2707,
+      "step": 457
+    },
+    {
+      "epoch": 0.047489128955537266,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019988161861184687,
+      "loss": 5.221,
+      "step": 458
+    },
+    {
+      "epoch": 0.04759281701002534,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019988108963855586,
+      "loss": 5.1917,
+      "step": 459
+    },
+    {
+      "epoch": 0.04769650506451341,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019988055948677355,
+      "loss": 5.2225,
+      "step": 460
+    },
+    {
+      "epoch": 0.04780019311900149,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019988002815650622,
+      "loss": 5.1813,
+      "step": 461
+    },
+    {
+      "epoch": 0.04790388117348956,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019987949564776014,
+      "loss": 5.2486,
+      "step": 462
+    },
+    {
+      "epoch": 0.04800756922797763,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0001998789619605416,
+      "loss": 5.2211,
+      "step": 463
+    },
+    {
+      "epoch": 0.0481112572824657,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019987842709485686,
+      "loss": 5.2175,
+      "step": 464
+    },
+    {
+      "epoch": 0.04821494533695377,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0001998778910507123,
+      "loss": 5.2004,
+      "step": 465
+    },
+    {
+      "epoch": 0.048318633391441844,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019987735382811416,
+      "loss": 5.2169,
+      "step": 466
+    },
+    {
+      "epoch": 0.04842232144592992,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001998768154270688,
+      "loss": 5.165,
+      "step": 467
+    },
+    {
+      "epoch": 0.048526009500417994,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019987627584758263,
+      "loss": 5.2305,
+      "step": 468
+    },
+    {
+      "epoch": 0.048629697554906065,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019987573508966199,
+      "loss": 5.2377,
+      "step": 469
+    },
+    {
+      "epoch": 0.048733385609394136,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019987519315331324,
+      "loss": 5.2023,
+      "step": 470
+    },
+    {
+      "epoch": 0.04883707366388221,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019987465003854275,
+      "loss": 5.2201,
+      "step": 471
+    },
+    {
+      "epoch": 0.048940761718370286,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.000199874105745357,
+      "loss": 5.1884,
+      "step": 472
+    },
+    {
+      "epoch": 0.04904444977285836,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00019987356027376238,
+      "loss": 5.1698,
+      "step": 473
+    },
+    {
+      "epoch": 0.04914813782734643,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001998730136237653,
+      "loss": 5.2097,
+      "step": 474
+    },
+    {
+      "epoch": 0.0492518258818345,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019987246579537222,
+      "loss": 5.2154,
+      "step": 475
+    },
+    {
+      "epoch": 0.04935551393632257,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019987191678858964,
+      "loss": 5.1997,
+      "step": 476
+    },
+    {
+      "epoch": 0.04945920199081064,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019987136660342398,
+      "loss": 5.2161,
+      "step": 477
+    },
+    {
+      "epoch": 0.04956289004529872,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019987081523988178,
+      "loss": 5.2113,
+      "step": 478
+    },
+    {
+      "epoch": 0.04966657809978679,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019987026269796952,
+      "loss": 5.1944,
+      "step": 479
+    },
+    {
+      "epoch": 0.049770266154274864,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019986970897769375,
+      "loss": 5.1795,
+      "step": 480
+    },
+    {
+      "epoch": 0.049873954208762936,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019986915407906096,
+      "loss": 5.2348,
+      "step": 481
+    },
+    {
+      "epoch": 0.04997764226325101,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00019986859800207772,
+      "loss": 5.1926,
+      "step": 482
+    },
+    {
+      "epoch": 0.050081330317739085,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.00019986804074675058,
+      "loss": 5.2065,
+      "step": 483
+    },
+    {
+      "epoch": 0.05018501837222716,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019986748231308615,
+      "loss": 5.1956,
+      "step": 484
+    },
+    {
+      "epoch": 0.05028870642671523,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019986692270109098,
+      "loss": 5.1497,
+      "step": 485
+    },
+    {
+      "epoch": 0.0503923944812033,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019986636191077168,
+      "loss": 5.1756,
+      "step": 486
+    },
+    {
+      "epoch": 0.05049608253569137,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019986579994213486,
+      "loss": 5.2124,
+      "step": 487
+    },
+    {
+      "epoch": 0.05059977059017944,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00019986523679518722,
+      "loss": 5.1792,
+      "step": 488
+    },
+    {
+      "epoch": 0.05070345864466752,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00019986467246993527,
+      "loss": 5.1996,
+      "step": 489
+    },
+    {
+      "epoch": 0.05080714669915559,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.0001998641069663858,
+      "loss": 5.2116,
+      "step": 490
+    },
+    {
+      "epoch": 0.05091083475364366,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019986354028454542,
+      "loss": 5.1607,
+      "step": 491
+    },
+    {
+      "epoch": 0.051014522808131735,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0001998629724244208,
+      "loss": 5.1851,
+      "step": 492
+    },
+    {
+      "epoch": 0.051118210862619806,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.00019986240338601869,
+      "loss": 5.196,
+      "step": 493
+    },
+    {
+      "epoch": 0.05122189891710788,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00019986183316934576,
+      "loss": 5.1985,
+      "step": 494
+    },
+    {
+      "epoch": 0.051325586971595956,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0001998612617744088,
+      "loss": 5.1647,
+      "step": 495
+    },
+    {
+      "epoch": 0.05142927502608403,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0001998606892012145,
+      "loss": 5.205,
+      "step": 496
+    },
+    {
+      "epoch": 0.0515329630805721,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019986011544976956,
+      "loss": 5.2331,
+      "step": 497
+    },
+    {
+      "epoch": 0.05163665113506017,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.00019985954052008085,
+      "loss": 5.1625,
+      "step": 498
+    },
+    {
+      "epoch": 0.05174033918954824,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019985896441215514,
+      "loss": 5.1805,
+      "step": 499
+    },
+    {
+      "epoch": 0.05184402724403632,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0001998583871259992,
+      "loss": 5.1823,
+      "step": 500
+    },
+    {
+      "epoch": 0.05194771529852439,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019985780866161985,
+      "loss": 5.1178,
+      "step": 501
+    },
+    {
+      "epoch": 0.05205140335301246,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019985722901902389,
+      "loss": 5.1614,
+      "step": 502
+    },
+    {
+      "epoch": 0.052155091407500534,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001998566481982182,
+      "loss": 5.178,
+      "step": 503
+    },
+    {
+      "epoch": 0.052258779461988605,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0001998560661992096,
+      "loss": 5.1482,
+      "step": 504
+    },
+    {
+      "epoch": 0.05236246751647668,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019985548302200497,
+      "loss": 5.1702,
+      "step": 505
+    },
+    {
+      "epoch": 0.052466155570964755,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001998548986666112,
+      "loss": 5.165,
+      "step": 506
+    },
+    {
+      "epoch": 0.052569843625452826,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019985431313303517,
+      "loss": 5.1466,
+      "step": 507
+    },
+    {
+      "epoch": 0.0526735316799409,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019985372642128383,
+      "loss": 5.183,
+      "step": 508
+    },
+    {
+      "epoch": 0.05277721973442897,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019985313853136403,
+      "loss": 5.1417,
+      "step": 509
+    },
+    {
+      "epoch": 0.05288090778891704,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019985254946328274,
+      "loss": 5.1574,
+      "step": 510
+    },
+    {
+      "epoch": 0.05298459584340512,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019985195921704696,
+      "loss": 5.1722,
+      "step": 511
+    },
+    {
+      "epoch": 0.05308828389789319,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001998513677926636,
+      "loss": 5.163,
+      "step": 512
+    },
+    {
+      "epoch": 0.05319197195238126,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0001998507751901396,
+      "loss": 5.1497,
+      "step": 513
+    },
+    {
+      "epoch": 0.05329566000686933,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000199850181409482,
+      "loss": 5.1216,
+      "step": 514
+    },
+    {
+      "epoch": 0.053399348061357405,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.00019984958645069786,
+      "loss": 5.1793,
+      "step": 515
+    },
+    {
+      "epoch": 0.053503036115845476,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.0001998489903137941,
+      "loss": 5.1748,
+      "step": 516
+    },
+    {
+      "epoch": 0.053606724170333554,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001998483929987778,
+      "loss": 5.1428,
+      "step": 517
+    },
+    {
+      "epoch": 0.053710412224821626,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00019984779450565605,
+      "loss": 5.1818,
+      "step": 518
+    },
+    {
+      "epoch": 0.0538141002793097,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019984719483443587,
+      "loss": 5.1619,
+      "step": 519
+    },
+    {
+      "epoch": 0.05391778833379777,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0001998465939851243,
+      "loss": 5.1613,
+      "step": 520
+    },
+    {
+      "epoch": 0.05402147638828584,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00019984599195772845,
+      "loss": 5.1229,
+      "step": 521
+    },
+    {
+      "epoch": 0.05412516444277392,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00019984538875225547,
+      "loss": 5.1587,
+      "step": 522
+    },
+    {
+      "epoch": 0.05422885249726199,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019984478436871244,
+      "loss": 5.163,
+      "step": 523
+    },
+    {
+      "epoch": 0.05433254055175006,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00019984417880710646,
+      "loss": 5.1997,
+      "step": 524
+    },
+    {
+      "epoch": 0.05443622860623813,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.00019984357206744474,
+      "loss": 5.1593,
+      "step": 525
+    },
+    {
+      "epoch": 0.054539916660726204,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0001998429641497344,
+      "loss": 5.1307,
+      "step": 526
+    },
+    {
+      "epoch": 0.054643604715214275,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019984235505398262,
+      "loss": 5.0949,
+      "step": 527
+    },
+    {
+      "epoch": 0.05474729276970235,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0001998417447801966,
+      "loss": 5.1336,
+      "step": 528
+    },
+    {
+      "epoch": 0.054850980824190425,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019984113332838352,
+      "loss": 5.1222,
+      "step": 529
+    },
+    {
+      "epoch": 0.054954668878678496,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001998405206985506,
+      "loss": 5.1132,
+      "step": 530
+    },
+    {
+      "epoch": 0.05505835693316657,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019983990689070508,
+      "loss": 5.1254,
+      "step": 531
+    },
+    {
+      "epoch": 0.05516204498765464,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019983929190485423,
+      "loss": 5.1434,
+      "step": 532
+    },
+    {
+      "epoch": 0.05526573304214271,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0001998386757410052,
+      "loss": 5.0891,
+      "step": 533
+    },
+    {
+      "epoch": 0.05536942109663079,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001998380583991654,
+      "loss": 5.144,
+      "step": 534
+    },
+    {
+      "epoch": 0.05547310915111886,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019983743987934198,
+      "loss": 5.1256,
+      "step": 535
+    },
+    {
+      "epoch": 0.05557679720560693,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019983682018154234,
+      "loss": 5.1397,
+      "step": 536
+    },
+    {
+      "epoch": 0.055680485260095,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019983619930577374,
+      "loss": 5.1786,
+      "step": 537
+    },
+    {
+      "epoch": 0.055784173314583074,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019983557725204352,
+      "loss": 5.1161,
+      "step": 538
+    },
+    {
+      "epoch": 0.05588786136907115,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019983495402035902,
+      "loss": 5.1185,
+      "step": 539
+    },
+    {
+      "epoch": 0.055991549423559224,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0001998343296107276,
+      "loss": 5.1027,
+      "step": 540
+    },
+    {
+      "epoch": 0.056095237478047295,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001998337040231566,
+      "loss": 5.1329,
+      "step": 541
+    },
+    {
+      "epoch": 0.05619892553253537,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019983307725765346,
+      "loss": 5.15,
+      "step": 542
+    },
+    {
+      "epoch": 0.05630261358702344,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0001998324493142255,
+      "loss": 5.1271,
+      "step": 543
+    },
+    {
+      "epoch": 0.05640630164151151,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00019983182019288017,
+      "loss": 5.1476,
+      "step": 544
+    },
+    {
+      "epoch": 0.05650998969599959,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0001998311898936249,
+      "loss": 5.0857,
+      "step": 545
+    },
+    {
+      "epoch": 0.05661367775048766,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001998305584164671,
+      "loss": 5.1089,
+      "step": 546
+    },
+    {
+      "epoch": 0.05671736580497573,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019982992576141425,
+      "loss": 5.1258,
+      "step": 547
+    },
+    {
+      "epoch": 0.0568210538594638,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001998292919284738,
+      "loss": 5.1405,
+      "step": 548
+    },
+    {
+      "epoch": 0.05692474191395187,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019982865691765323,
+      "loss": 5.1548,
+      "step": 549
+    },
+    {
+      "epoch": 0.05702842996843995,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019982802072896004,
+      "loss": 5.0671,
+      "step": 550
+    },
+    {
+      "epoch": 0.05713211802292802,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019982738336240172,
+      "loss": 5.1215,
+      "step": 551
+    },
+    {
+      "epoch": 0.057235806077416095,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001998267448179858,
+      "loss": 5.0941,
+      "step": 552
+    },
+    {
+      "epoch": 0.057339494131904166,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019982610509571979,
+      "loss": 5.1049,
+      "step": 553
+    },
+    {
+      "epoch": 0.05744318218639224,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001998254641956113,
+      "loss": 5.1052,
+      "step": 554
+    },
+    {
+      "epoch": 0.05754687024088031,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0001998248221176678,
+      "loss": 5.0879,
+      "step": 555
+    },
+    {
+      "epoch": 0.05765055829536839,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019982417886189698,
+      "loss": 5.1172,
+      "step": 556
+    },
+    {
+      "epoch": 0.05775424634985646,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019982353442830634,
+      "loss": 5.0876,
+      "step": 557
+    },
+    {
+      "epoch": 0.05785793440434453,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019982288881690349,
+      "loss": 5.1051,
+      "step": 558
+    },
+    {
+      "epoch": 0.0579616224588326,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019982224202769611,
+      "loss": 5.0873,
+      "step": 559
+    },
+    {
+      "epoch": 0.05806531051332067,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019982159406069176,
+      "loss": 5.0969,
+      "step": 560
+    },
+    {
+      "epoch": 0.058168998567808744,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019982094491589813,
+      "loss": 5.1099,
+      "step": 561
+    },
+    {
+      "epoch": 0.05827268662229682,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019982029459332287,
+      "loss": 5.0659,
+      "step": 562
+    },
+    {
+      "epoch": 0.058376374676784894,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019981964309297363,
+      "loss": 5.0939,
+      "step": 563
+    },
+    {
+      "epoch": 0.058480062731272965,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019981899041485813,
+      "loss": 5.1099,
+      "step": 564
+    },
+    {
+      "epoch": 0.058583750785761037,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019981833655898404,
+      "loss": 5.1289,
+      "step": 565
+    },
+    {
+      "epoch": 0.05868743884024911,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019981768152535913,
+      "loss": 5.0745,
+      "step": 566
+    },
+    {
+      "epoch": 0.058791126894737186,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019981702531399106,
+      "loss": 5.1127,
+      "step": 567
+    },
+    {
+      "epoch": 0.05889481494922526,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0001998163679248876,
+      "loss": 5.0662,
+      "step": 568
+    },
+    {
+      "epoch": 0.05899850300371333,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0001998157093580565,
+      "loss": 5.109,
+      "step": 569
+    },
+    {
+      "epoch": 0.0591021910582014,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00019981504961350558,
+      "loss": 5.0697,
+      "step": 570
+    },
+    {
+      "epoch": 0.05920587911268947,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00019981438869124256,
+      "loss": 5.0217,
+      "step": 571
+    },
+    {
+      "epoch": 0.05930956716717754,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019981372659127523,
+      "loss": 5.0847,
+      "step": 572
+    },
+    {
+      "epoch": 0.05941325522166562,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019981306331361148,
+      "loss": 5.0614,
+      "step": 573
+    },
+    {
+      "epoch": 0.05951694327615369,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019981239885825906,
+      "loss": 5.0795,
+      "step": 574
+    },
+    {
+      "epoch": 0.059620631330641764,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00019981173322522586,
+      "loss": 5.0716,
+      "step": 575
+    },
+    {
+      "epoch": 0.059724319385129836,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019981106641451973,
+      "loss": 5.0803,
+      "step": 576
+    },
+    {
+      "epoch": 0.05982800743961791,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001998103984261485,
+      "loss": 5.0795,
+      "step": 577
+    },
+    {
+      "epoch": 0.059931695494105985,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019980972926012005,
+      "loss": 5.0748,
+      "step": 578
+    },
+    {
+      "epoch": 0.06003538354859406,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001998090589164423,
+      "loss": 5.0885,
+      "step": 579
+    },
+    {
+      "epoch": 0.06013907160308213,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0001998083873951232,
+      "loss": 5.0466,
+      "step": 580
+    },
+    {
+      "epoch": 0.0602427596575702,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019980771469617058,
+      "loss": 5.0745,
+      "step": 581
+    },
+    {
+      "epoch": 0.06034644771205827,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019980704081959248,
+      "loss": 5.0909,
+      "step": 582
+    },
+    {
+      "epoch": 0.06045013576654634,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019980636576539678,
+      "loss": 5.017,
+      "step": 583
+    },
+    {
+      "epoch": 0.06055382382103442,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019980568953359144,
+      "loss": 5.1101,
+      "step": 584
+    },
+    {
+      "epoch": 0.06065751187552249,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.00019980501212418447,
+      "loss": 5.0365,
+      "step": 585
+    },
+    {
+      "epoch": 0.06076119993001056,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.0001998043335371839,
+      "loss": 5.07,
+      "step": 586
+    },
+    {
+      "epoch": 0.060864887984498635,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00019980365377259763,
+      "loss": 5.0933,
+      "step": 587
+    },
+    {
+      "epoch": 0.060968576038986706,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019980297283043379,
+      "loss": 5.1316,
+      "step": 588
+    },
+    {
+      "epoch": 0.061072264093474785,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019980229071070037,
+      "loss": 5.0325,
+      "step": 589
+    },
+    {
+      "epoch": 0.061175952147962856,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00019980160741340537,
+      "loss": 5.1035,
+      "step": 590
+    },
+    {
+      "epoch": 0.06127964020245093,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.0001998009229385569,
+      "loss": 5.0554,
+      "step": 591
+    },
+    {
+      "epoch": 0.061383328256939,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019980023728616305,
+      "loss": 5.0741,
+      "step": 592
+    },
+    {
+      "epoch": 0.06148701631142707,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.0001997995504562319,
+      "loss": 5.083,
+      "step": 593
+    },
+    {
+      "epoch": 0.06159070436591514,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019979886244877158,
+      "loss": 5.0724,
+      "step": 594
+    },
+    {
+      "epoch": 0.06169439242040322,
+      "grad_norm": 3.359375,
+      "learning_rate": 0.00019979817326379012,
+      "loss": 5.1534,
+      "step": 595
+    },
+    {
+      "epoch": 0.06179808047489129,
+      "grad_norm": 3.109375,
+      "learning_rate": 0.00019979748290129573,
+      "loss": 5.1388,
+      "step": 596
+    },
+    {
+      "epoch": 0.06190176852937936,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00019979679136129653,
+      "loss": 5.0704,
+      "step": 597
+    },
+    {
+      "epoch": 0.062005456583867434,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00019979609864380067,
+      "loss": 5.1138,
+      "step": 598
+    },
+    {
+      "epoch": 0.062109144638355505,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00019979540474881634,
+      "loss": 5.1004,
+      "step": 599
+    },
+    {
+      "epoch": 0.06221283269284358,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019979470967635172,
+      "loss": 5.0521,
+      "step": 600
+    },
+    {
+      "epoch": 0.062316520747331655,
+      "grad_norm": 1.796875,
+      "learning_rate": 0.00019979401342641503,
+      "loss": 5.0985,
+      "step": 601
+    },
+    {
+      "epoch": 0.062420208801819727,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00019979331599901445,
+      "loss": 5.0998,
+      "step": 602
+    },
+    {
+      "epoch": 0.0625238968563078,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00019979261739415825,
+      "loss": 5.1008,
+      "step": 603
+    },
+    {
+      "epoch": 0.06262758491079587,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00019979191761185466,
+      "loss": 5.088,
+      "step": 604
+    },
+    {
+      "epoch": 0.06273127296528394,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00019979121665211186,
+      "loss": 5.1075,
+      "step": 605
+    },
+    {
+      "epoch": 0.06283496101977201,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00019979051451493826,
+      "loss": 5.1131,
+      "step": 606
+    },
+    {
+      "epoch": 0.06293864907426008,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00019978981120034203,
+      "loss": 5.0863,
+      "step": 607
+    },
+    {
+      "epoch": 0.06304233712874815,
+      "grad_norm": 2.875,
+      "learning_rate": 0.0001997891067083315,
+      "loss": 5.126,
+      "step": 608
+    },
+    {
+      "epoch": 0.06314602518323624,
+      "grad_norm": 2.453125,
+      "learning_rate": 0.00019978840103891505,
+      "loss": 5.1418,
+      "step": 609
+    },
+    {
+      "epoch": 0.06324971323772431,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.0001997876941921009,
+      "loss": 5.0454,
+      "step": 610
+    },
+    {
+      "epoch": 0.06335340129221238,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019978698616789745,
+      "loss": 5.0991,
+      "step": 611
+    },
+    {
+      "epoch": 0.06345708934670045,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00019978627696631306,
+      "loss": 5.0266,
+      "step": 612
+    },
+    {
+      "epoch": 0.06356077740118853,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00019978556658735606,
+      "loss": 5.096,
+      "step": 613
+    },
+    {
+      "epoch": 0.0636644654556766,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00019978485503103485,
+      "loss": 5.0584,
+      "step": 614
+    },
+    {
+      "epoch": 0.06376815351016467,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.0001997841422973578,
+      "loss": 5.092,
+      "step": 615
+    },
+    {
+      "epoch": 0.06387184156465274,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00019978342838633344,
+      "loss": 5.1038,
+      "step": 616
+    },
+    {
+      "epoch": 0.06397552961914081,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00019978271329797003,
+      "loss": 5.0346,
+      "step": 617
+    },
+    {
+      "epoch": 0.06407921767362888,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00019978199703227608,
+      "loss": 5.0628,
+      "step": 618
+    },
+    {
+      "epoch": 0.06418290572811695,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00019978127958926006,
+      "loss": 5.1023,
+      "step": 619
+    },
+    {
+      "epoch": 0.06428659378260504,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00019978056096893042,
+      "loss": 5.0591,
+      "step": 620
+    },
+    {
+      "epoch": 0.06439028183709311,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.0001997798411712956,
+      "loss": 5.0737,
+      "step": 621
+    },
+    {
+      "epoch": 0.06449396989158118,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00019977912019636415,
+      "loss": 5.0839,
+      "step": 622
+    },
+    {
+      "epoch": 0.06459765794606925,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019977839804414456,
+      "loss": 5.0516,
+      "step": 623
+    },
+    {
+      "epoch": 0.06470134600055732,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019977767471464531,
+      "loss": 5.0687,
+      "step": 624
+    },
+    {
+      "epoch": 0.0648050340550454,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019977695020787498,
+      "loss": 5.0761,
+      "step": 625
+    },
+    {
+      "epoch": 0.06490872210953347,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019977622452384212,
+      "loss": 5.0706,
+      "step": 626
+    },
+    {
+      "epoch": 0.06501241016402154,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019977549766255528,
+      "loss": 5.0099,
+      "step": 627
+    },
+    {
+      "epoch": 0.06511609821850961,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019977476962402304,
+      "loss": 5.0794,
+      "step": 628
+    },
+    {
+      "epoch": 0.06521978627299768,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019977404040825395,
+      "loss": 5.0676,
+      "step": 629
+    },
+    {
+      "epoch": 0.06532347432748575,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0001997733100152567,
+      "loss": 5.0754,
+      "step": 630
+    },
+    {
+      "epoch": 0.06542716238197384,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0001997725784450398,
+      "loss": 5.0106,
+      "step": 631
+    },
+    {
+      "epoch": 0.06553085043646191,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.000199771845697612,
+      "loss": 5.0163,
+      "step": 632
+    },
+    {
+      "epoch": 0.06563453849094998,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019977111177298183,
+      "loss": 5.0573,
+      "step": 633
+    },
+    {
+      "epoch": 0.06573822654543805,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019977037667115802,
+      "loss": 5.0315,
+      "step": 634
+    },
+    {
+      "epoch": 0.06584191459992612,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019976964039214923,
+      "loss": 5.0168,
+      "step": 635
+    },
+    {
+      "epoch": 0.0659456026544142,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019976890293596416,
+      "loss": 5.0454,
+      "step": 636
+    },
+    {
+      "epoch": 0.06604929070890227,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019976816430261146,
+      "loss": 5.0578,
+      "step": 637
+    },
+    {
+      "epoch": 0.06615297876339034,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00019976742449209992,
+      "loss": 5.0527,
+      "step": 638
+    },
+    {
+      "epoch": 0.06625666681787841,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0001997666835044382,
+      "loss": 5.083,
+      "step": 639
+    },
+    {
+      "epoch": 0.06636035487236648,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019976594133963512,
+      "loss": 5.0622,
+      "step": 640
+    },
+    {
+      "epoch": 0.06646404292685455,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00019976519799769931,
+      "loss": 5.0684,
+      "step": 641
+    },
+    {
+      "epoch": 0.06656773098134262,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019976445347863968,
+      "loss": 5.0087,
+      "step": 642
+    },
+    {
+      "epoch": 0.06667141903583071,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019976370778246495,
+      "loss": 5.0554,
+      "step": 643
+    },
+    {
+      "epoch": 0.06677510709031878,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001997629609091839,
+      "loss": 5.0249,
+      "step": 644
+    },
+    {
+      "epoch": 0.06687879514480685,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001997622128588054,
+      "loss": 5.0804,
+      "step": 645
+    },
+    {
+      "epoch": 0.06698248319929492,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0001997614636313382,
+      "loss": 5.0207,
+      "step": 646
+    },
+    {
+      "epoch": 0.067086171253783,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0001997607132267912,
+      "loss": 5.0279,
+      "step": 647
+    },
+    {
+      "epoch": 0.06718985930827107,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00019975996164517325,
+      "loss": 5.062,
+      "step": 648
+    },
+    {
+      "epoch": 0.06729354736275914,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019975920888649318,
+      "loss": 4.9891,
+      "step": 649
+    },
+    {
+      "epoch": 0.06739723541724721,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00019975845495075992,
+      "loss": 5.0609,
+      "step": 650
+    },
+    {
+      "epoch": 0.06750092347173528,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001997576998379823,
+      "loss": 5.0621,
+      "step": 651
+    },
+    {
+      "epoch": 0.06760461152622335,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0001997569435481693,
+      "loss": 4.9655,
+      "step": 652
+    },
+    {
+      "epoch": 0.06770829958071142,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00019975618608132983,
+      "loss": 5.0336,
+      "step": 653
+    },
+    {
+      "epoch": 0.06781198763519951,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0001997554274374728,
+      "loss": 5.0162,
+      "step": 654
+    },
+    {
+      "epoch": 0.06791567568968758,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00019975466761660714,
+      "loss": 4.9897,
+      "step": 655
+    },
+    {
+      "epoch": 0.06801936374417565,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019975390661874188,
+      "loss": 4.9694,
+      "step": 656
+    },
+    {
+      "epoch": 0.06812305179866372,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019975314444388597,
+      "loss": 5.0425,
+      "step": 657
+    },
+    {
+      "epoch": 0.0682267398531518,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019975238109204836,
+      "loss": 5.0513,
+      "step": 658
+    },
+    {
+      "epoch": 0.06833042790763987,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019975161656323812,
+      "loss": 5.0141,
+      "step": 659
+    },
+    {
+      "epoch": 0.06843411596212794,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019975085085746427,
+      "loss": 5.0405,
+      "step": 660
+    },
+    {
+      "epoch": 0.06853780401661601,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019975008397473578,
+      "loss": 5.0309,
+      "step": 661
+    },
+    {
+      "epoch": 0.06864149207110408,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00019974931591506176,
+      "loss": 5.0045,
+      "step": 662
+    },
+    {
+      "epoch": 0.06874518012559215,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019974854667845126,
+      "loss": 5.0045,
+      "step": 663
+    },
+    {
+      "epoch": 0.06884886818008022,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019974777626491334,
+      "loss": 5.0137,
+      "step": 664
+    },
+    {
+      "epoch": 0.06895255623456831,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.0001997470046744571,
+      "loss": 5.0363,
+      "step": 665
+    },
+    {
+      "epoch": 0.06905624428905638,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019974623190709164,
+      "loss": 5.052,
+      "step": 666
+    },
+    {
+      "epoch": 0.06915993234354445,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019974545796282606,
+      "loss": 5.0489,
+      "step": 667
+    },
+    {
+      "epoch": 0.06926362039803252,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019974468284166954,
+      "loss": 5.0173,
+      "step": 668
+    },
+    {
+      "epoch": 0.06936730845252059,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001997439065436312,
+      "loss": 4.9908,
+      "step": 669
+    },
+    {
+      "epoch": 0.06947099650700866,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019974312906872018,
+      "loss": 5.024,
+      "step": 670
+    },
+    {
+      "epoch": 0.06957468456149674,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019974235041694566,
+      "loss": 5.0214,
+      "step": 671
+    },
+    {
+      "epoch": 0.06967837261598481,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019974157058831685,
+      "loss": 5.0328,
+      "step": 672
+    },
+    {
+      "epoch": 0.06978206067047288,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019974078958284294,
+      "loss": 4.9868,
+      "step": 673
+    },
+    {
+      "epoch": 0.06988574872496095,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019974000740053316,
+      "loss": 4.9927,
+      "step": 674
+    },
+    {
+      "epoch": 0.06998943677944902,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001997392240413967,
+      "loss": 5.0442,
+      "step": 675
+    },
+    {
+      "epoch": 0.0700931248339371,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0001997384395054428,
+      "loss": 5.0179,
+      "step": 676
+    },
+    {
+      "epoch": 0.07019681288842518,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019973765379268082,
+      "loss": 5.0022,
+      "step": 677
+    },
+    {
+      "epoch": 0.07030050094291325,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019973686690311987,
+      "loss": 5.0517,
+      "step": 678
+    },
+    {
+      "epoch": 0.07040418899740132,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00019973607883676936,
+      "loss": 4.9883,
+      "step": 679
+    },
+    {
+      "epoch": 0.07050787705188939,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019973528959363855,
+      "loss": 4.9936,
+      "step": 680
+    },
+    {
+      "epoch": 0.07061156510637746,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019973449917373674,
+      "loss": 5.0083,
+      "step": 681
+    },
+    {
+      "epoch": 0.07071525316086553,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019973370757707325,
+      "loss": 5.0072,
+      "step": 682
+    },
+    {
+      "epoch": 0.0708189412153536,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019973291480365743,
+      "loss": 5.0331,
+      "step": 683
+    },
+    {
+      "epoch": 0.07092262926984168,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019973212085349867,
+      "loss": 4.9926,
+      "step": 684
+    },
+    {
+      "epoch": 0.07102631732432975,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019973132572660628,
+      "loss": 4.9922,
+      "step": 685
+    },
+    {
+      "epoch": 0.07113000537881782,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019973052942298967,
+      "loss": 5.0278,
+      "step": 686
+    },
+    {
+      "epoch": 0.0712336934333059,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019972973194265823,
+      "loss": 5.0312,
+      "step": 687
+    },
+    {
+      "epoch": 0.07133738148779398,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019972893328562137,
+      "loss": 4.9927,
+      "step": 688
+    },
+    {
+      "epoch": 0.07144106954228205,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00019972813345188852,
+      "loss": 5.0116,
+      "step": 689
+    },
+    {
+      "epoch": 0.07154475759677012,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00019972733244146912,
+      "loss": 4.9578,
+      "step": 690
+    },
+    {
+      "epoch": 0.07164844565125819,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019972653025437261,
+      "loss": 4.9803,
+      "step": 691
+    },
+    {
+      "epoch": 0.07175213370574626,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019972572689060846,
+      "loss": 4.9453,
+      "step": 692
+    },
+    {
+      "epoch": 0.07185582176023433,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00019972492235018616,
+      "loss": 4.9737,
+      "step": 693
+    },
+    {
+      "epoch": 0.0719595098147224,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019972411663311517,
+      "loss": 5.0081,
+      "step": 694
+    },
+    {
+      "epoch": 0.07206319786921048,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019972330973940503,
+      "loss": 4.9832,
+      "step": 695
+    },
+    {
+      "epoch": 0.07216688592369855,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019972250166906523,
+      "loss": 5.0042,
+      "step": 696
+    },
+    {
+      "epoch": 0.07227057397818662,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0001997216924221053,
+      "loss": 5.0134,
+      "step": 697
+    },
+    {
+      "epoch": 0.0723742620326747,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019972088199853488,
+      "loss": 4.9701,
+      "step": 698
+    },
+    {
+      "epoch": 0.07247795008716278,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001997200703983634,
+      "loss": 5.0043,
+      "step": 699
+    },
+    {
+      "epoch": 0.07258163814165085,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019971925762160054,
+      "loss": 5.0103,
+      "step": 700
+    },
+    {
+      "epoch": 0.07268532619613892,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0001997184436682558,
+      "loss": 4.9916,
+      "step": 701
+    },
+    {
+      "epoch": 0.07278901425062699,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019971762853833886,
+      "loss": 4.9837,
+      "step": 702
+    },
+    {
+      "epoch": 0.07289270230511506,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0001997168122318593,
+      "loss": 4.9554,
+      "step": 703
+    },
+    {
+      "epoch": 0.07299639035960313,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0001997159947488268,
+      "loss": 5.0302,
+      "step": 704
+    },
+    {
+      "epoch": 0.0731000784140912,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019971517608925092,
+      "loss": 5.0219,
+      "step": 705
+    },
+    {
+      "epoch": 0.07320376646857928,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019971435625314139,
+      "loss": 4.9801,
+      "step": 706
+    },
+    {
+      "epoch": 0.07330745452306735,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019971353524050783,
+      "loss": 5.0094,
+      "step": 707
+    },
+    {
+      "epoch": 0.07341114257755542,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019971271305135998,
+      "loss": 4.9897,
+      "step": 708
+    },
+    {
+      "epoch": 0.07351483063204349,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.00019971188968570752,
+      "loss": 4.9852,
+      "step": 709
+    },
+    {
+      "epoch": 0.07361851868653158,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00019971106514356018,
+      "loss": 5.0171,
+      "step": 710
+    },
+    {
+      "epoch": 0.07372220674101965,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019971023942492763,
+      "loss": 5.022,
+      "step": 711
+    },
+    {
+      "epoch": 0.07382589479550772,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.00019970941252981964,
+      "loss": 4.9833,
+      "step": 712
+    },
+    {
+      "epoch": 0.07392958284999579,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019970858445824603,
+      "loss": 5.025,
+      "step": 713
+    },
+    {
+      "epoch": 0.07403327090448386,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001997077552102165,
+      "loss": 4.9984,
+      "step": 714
+    },
+    {
+      "epoch": 0.07413695895897193,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00019970692478574084,
+      "loss": 4.9837,
+      "step": 715
+    },
+    {
+      "epoch": 0.07424064701346,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00019970609318482887,
+      "loss": 5.0337,
+      "step": 716
+    },
+    {
+      "epoch": 0.07434433506794808,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001997052604074904,
+      "loss": 5.0083,
+      "step": 717
+    },
+    {
+      "epoch": 0.07444802312243615,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00019970442645373526,
+      "loss": 4.99,
+      "step": 718
+    },
+    {
+      "epoch": 0.07455171117692422,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00019970359132357327,
+      "loss": 4.9583,
+      "step": 719
+    },
+    {
+      "epoch": 0.07465539923141229,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001997027550170143,
+      "loss": 4.9729,
+      "step": 720
+    },
+    {
+      "epoch": 0.07475908728590037,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001997019175340682,
+      "loss": 5.0005,
+      "step": 721
+    },
+    {
+      "epoch": 0.07486277534038845,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019970107887474486,
+      "loss": 4.9606,
+      "step": 722
+    },
+    {
+      "epoch": 0.07496646339487652,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001997002390390542,
+      "loss": 4.9606,
+      "step": 723
+    },
+    {
+      "epoch": 0.07507015144936459,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019969939802700606,
+      "loss": 4.9956,
+      "step": 724
+    },
+    {
+      "epoch": 0.07517383950385266,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00019969855583861046,
+      "loss": 5.0123,
+      "step": 725
+    },
+    {
+      "epoch": 0.07527752755834073,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.00019969771247387724,
+      "loss": 5.0062,
+      "step": 726
+    },
+    {
+      "epoch": 0.0753812156128288,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00019969686793281643,
+      "loss": 4.9654,
+      "step": 727
+    },
+    {
+      "epoch": 0.07548490366731687,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00019969602221543798,
+      "loss": 5.0011,
+      "step": 728
+    },
+    {
+      "epoch": 0.07558859172180495,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019969517532175183,
+      "loss": 4.9748,
+      "step": 729
+    },
+    {
+      "epoch": 0.07569227977629302,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.000199694327251768,
+      "loss": 4.9883,
+      "step": 730
+    },
+    {
+      "epoch": 0.07579596783078109,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00019969347800549646,
+      "loss": 4.9907,
+      "step": 731
+    },
+    {
+      "epoch": 0.07589965588526917,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0001996926275829473,
+      "loss": 4.9906,
+      "step": 732
+    },
+    {
+      "epoch": 0.07600334393975725,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001996917759841305,
+      "loss": 4.9916,
+      "step": 733
+    },
+    {
+      "epoch": 0.07610703199424532,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0001996909232090561,
+      "loss": 4.9873,
+      "step": 734
+    },
+    {
+      "epoch": 0.07621072004873339,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0001996900692577342,
+      "loss": 5.0099,
+      "step": 735
+    },
+    {
+      "epoch": 0.07631440810322146,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00019968921413017487,
+      "loss": 5.0039,
+      "step": 736
+    },
+    {
+      "epoch": 0.07641809615770953,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0001996883578263882,
+      "loss": 4.9653,
+      "step": 737
+    },
+    {
+      "epoch": 0.0765217842121976,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00019968750034638427,
+      "loss": 4.967,
+      "step": 738
+    },
+    {
+      "epoch": 0.07662547226668567,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0001996866416901732,
+      "loss": 4.9534,
+      "step": 739
+    },
+    {
+      "epoch": 0.07672916032117374,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019968578185776515,
+      "loss": 4.9588,
+      "step": 740
+    },
+    {
+      "epoch": 0.07683284837566182,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0001996849208491702,
+      "loss": 4.9746,
+      "step": 741
+    },
+    {
+      "epoch": 0.07693653643014989,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0001996840586643986,
+      "loss": 4.9714,
+      "step": 742
+    },
+    {
+      "epoch": 0.07704022448463797,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019968319530346048,
+      "loss": 4.9626,
+      "step": 743
+    },
+    {
+      "epoch": 0.07714391253912604,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.000199682330766366,
+      "loss": 4.9415,
+      "step": 744
+    },
+    {
+      "epoch": 0.07724760059361412,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0001996814650531254,
+      "loss": 4.9414,
+      "step": 745
+    },
+    {
+      "epoch": 0.07735128864810219,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00019968059816374888,
+      "loss": 4.9692,
+      "step": 746
+    },
+    {
+      "epoch": 0.07745497670259026,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019967973009824664,
+      "loss": 4.9797,
+      "step": 747
+    },
+    {
+      "epoch": 0.07755866475707833,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.000199678860856629,
+      "loss": 4.9844,
+      "step": 748
+    },
+    {
+      "epoch": 0.0776623528115664,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00019967799043890615,
+      "loss": 4.982,
+      "step": 749
+    },
+    {
+      "epoch": 0.07776604086605447,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00019967711884508839,
+      "loss": 4.9981,
+      "step": 750
+    },
+    {
+      "epoch": 0.07786972892054254,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00019967624607518595,
+      "loss": 4.9928,
+      "step": 751
+    },
+    {
+      "epoch": 0.07797341697503062,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001996753721292092,
+      "loss": 4.9345,
+      "step": 752
+    },
+    {
+      "epoch": 0.07807710502951869,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0001996744970071684,
+      "loss": 4.9615,
+      "step": 753
+    },
+    {
+      "epoch": 0.07818079308400677,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0001996736207090739,
+      "loss": 5.0009,
+      "step": 754
+    },
+    {
+      "epoch": 0.07828448113849484,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.00019967274323493605,
+      "loss": 4.9854,
+      "step": 755
+    },
+    {
+      "epoch": 0.07838816919298291,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0001996718645847652,
+      "loss": 4.9598,
+      "step": 756
+    },
+    {
+      "epoch": 0.07849185724747099,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0001996709847585717,
+      "loss": 4.9534,
+      "step": 757
+    },
+    {
+      "epoch": 0.07859554530195906,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0001996701037563659,
+      "loss": 4.9808,
+      "step": 758
+    },
+    {
+      "epoch": 0.07869923335644713,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019966922157815825,
+      "loss": 4.9569,
+      "step": 759
+    },
+    {
+      "epoch": 0.0788029214109352,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00019966833822395916,
+      "loss": 4.9297,
+      "step": 760
+    },
+    {
+      "epoch": 0.07890660946542327,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.000199667453693779,
+      "loss": 4.9103,
+      "step": 761
+    },
+    {
+      "epoch": 0.07901029751991134,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019966656798762827,
+      "loss": 4.9669,
+      "step": 762
+    },
+    {
+      "epoch": 0.07911398557439941,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019966568110551736,
+      "loss": 4.9627,
+      "step": 763
+    },
+    {
+      "epoch": 0.07921767362888749,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0001996647930474568,
+      "loss": 4.9544,
+      "step": 764
+    },
+    {
+      "epoch": 0.07932136168337557,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.000199663903813457,
+      "loss": 4.9443,
+      "step": 765
+    },
+    {
+      "epoch": 0.07942504973786364,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019966301340352852,
+      "loss": 4.9739,
+      "step": 766
+    },
+    {
+      "epoch": 0.07952873779235171,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019966212181768178,
+      "loss": 4.9593,
+      "step": 767
+    },
+    {
+      "epoch": 0.07963242584683979,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0001996612290559274,
+      "loss": 4.9728,
+      "step": 768
+    },
+    {
+      "epoch": 0.07973611390132786,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019966033511827584,
+      "loss": 4.968,
+      "step": 769
+    },
+    {
+      "epoch": 0.07983980195581593,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019965944000473768,
+      "loss": 4.9428,
+      "step": 770
+    },
+    {
+      "epoch": 0.079943490010304,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019965854371532346,
+      "loss": 4.9784,
+      "step": 771
+    },
+    {
+      "epoch": 0.08004717806479207,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019965764625004377,
+      "loss": 4.9521,
+      "step": 772
+    },
+    {
+      "epoch": 0.08015086611928014,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001996567476089092,
+      "loss": 4.9318,
+      "step": 773
+    },
+    {
+      "epoch": 0.08025455417376821,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019965584779193035,
+      "loss": 4.9945,
+      "step": 774
+    },
+    {
+      "epoch": 0.08035824222825629,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019965494679911782,
+      "loss": 4.9496,
+      "step": 775
+    },
+    {
+      "epoch": 0.08046193028274436,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0001996540446304823,
+      "loss": 4.9248,
+      "step": 776
+    },
+    {
+      "epoch": 0.08056561833723244,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019965314128603435,
+      "loss": 4.9327,
+      "step": 777
+    },
+    {
+      "epoch": 0.08066930639172051,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019965223676578472,
+      "loss": 4.9228,
+      "step": 778
+    },
+    {
+      "epoch": 0.08077299444620858,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019965133106974396,
+      "loss": 4.9292,
+      "step": 779
+    },
+    {
+      "epoch": 0.08087668250069666,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019965042419792288,
+      "loss": 4.944,
+      "step": 780
+    },
+    {
+      "epoch": 0.08098037055518473,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019964951615033215,
+      "loss": 4.9587,
+      "step": 781
+    },
+    {
+      "epoch": 0.0810840586096728,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001996486069269824,
+      "loss": 4.9611,
+      "step": 782
+    },
+    {
+      "epoch": 0.08118774666416087,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019964769652788448,
+      "loss": 4.9236,
+      "step": 783
+    },
+    {
+      "epoch": 0.08129143471864894,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019964678495304906,
+      "loss": 4.9575,
+      "step": 784
+    },
+    {
+      "epoch": 0.08139512277313701,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019964587220248686,
+      "loss": 4.9343,
+      "step": 785
+    },
+    {
+      "epoch": 0.08149881082762508,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019964495827620875,
+      "loss": 4.9315,
+      "step": 786
+    },
+    {
+      "epoch": 0.08160249888211316,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0001996440431742254,
+      "loss": 4.9526,
+      "step": 787
+    },
+    {
+      "epoch": 0.08170618693660124,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019964312689654777,
+      "loss": 4.9687,
+      "step": 788
+    },
+    {
+      "epoch": 0.08180987499108931,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001996422094431865,
+      "loss": 4.9541,
+      "step": 789
+    },
+    {
+      "epoch": 0.08191356304557738,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001996412908141525,
+      "loss": 5.007,
+      "step": 790
+    },
+    {
+      "epoch": 0.08201725110006546,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0001996403710094566,
+      "loss": 4.8851,
+      "step": 791
+    },
+    {
+      "epoch": 0.08212093915455353,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019963945002910964,
+      "loss": 4.8957,
+      "step": 792
+    },
+    {
+      "epoch": 0.0822246272090416,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0001996385278731225,
+      "loss": 4.9079,
+      "step": 793
+    },
+    {
+      "epoch": 0.08232831526352967,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019963760454150603,
+      "loss": 4.9504,
+      "step": 794
+    },
+    {
+      "epoch": 0.08243200331801774,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0001996366800342712,
+      "loss": 4.9559,
+      "step": 795
+    },
+    {
+      "epoch": 0.08253569137250581,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001996357543514288,
+      "loss": 4.9396,
+      "step": 796
+    },
+    {
+      "epoch": 0.08263937942699388,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019963482749298984,
+      "loss": 4.9287,
+      "step": 797
+    },
+    {
+      "epoch": 0.08274306748148196,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019963389945896527,
+      "loss": 4.922,
+      "step": 798
+    },
+    {
+      "epoch": 0.08284675553597004,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019963297024936595,
+      "loss": 4.9211,
+      "step": 799
+    },
+    {
+      "epoch": 0.08295044359045811,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00019963203986420296,
+      "loss": 4.9202,
+      "step": 800
+    },
+    {
+      "epoch": 0.08305413164494618,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00019963110830348714,
+      "loss": 4.9276,
+      "step": 801
+    },
+    {
+      "epoch": 0.08315781969943425,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019963017556722963,
+      "loss": 4.8891,
+      "step": 802
+    },
+    {
+      "epoch": 0.08326150775392233,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0001996292416554413,
+      "loss": 4.9741,
+      "step": 803
+    },
+    {
+      "epoch": 0.0833651958084104,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001996283065681333,
+      "loss": 4.9151,
+      "step": 804
+    },
+    {
+      "epoch": 0.08346888386289847,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019962737030531654,
+      "loss": 4.9354,
+      "step": 805
+    },
+    {
+      "epoch": 0.08357257191738654,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019962643286700215,
+      "loss": 4.9499,
+      "step": 806
+    },
+    {
+      "epoch": 0.08367625997187461,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019962549425320112,
+      "loss": 4.9544,
+      "step": 807
+    },
+    {
+      "epoch": 0.08377994802636268,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019962455446392461,
+      "loss": 4.9092,
+      "step": 808
+    },
+    {
+      "epoch": 0.08388363608085075,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019962361349918365,
+      "loss": 4.9512,
+      "step": 809
+    },
+    {
+      "epoch": 0.08398732413533884,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019962267135898936,
+      "loss": 4.9327,
+      "step": 810
+    },
+    {
+      "epoch": 0.08409101218982691,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019962172804335285,
+      "loss": 4.944,
+      "step": 811
+    },
+    {
+      "epoch": 0.08419470024431498,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019962078355228525,
+      "loss": 4.8733,
+      "step": 812
+    },
+    {
+      "epoch": 0.08429838829880305,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0001996198378857977,
+      "loss": 4.901,
+      "step": 813
+    },
+    {
+      "epoch": 0.08440207635329113,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019961889104390138,
+      "loss": 4.9424,
+      "step": 814
+    },
+    {
+      "epoch": 0.0845057644077792,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019961794302660746,
+      "loss": 4.9486,
+      "step": 815
+    },
+    {
+      "epoch": 0.08460945246226727,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019961699383392708,
+      "loss": 4.9393,
+      "step": 816
+    },
+    {
+      "epoch": 0.08471314051675534,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001996160434658715,
+      "loss": 4.9351,
+      "step": 817
+    },
+    {
+      "epoch": 0.08481682857124341,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001996150919224519,
+      "loss": 4.9562,
+      "step": 818
+    },
+    {
+      "epoch": 0.08492051662573148,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019961413920367948,
+      "loss": 4.9268,
+      "step": 819
+    },
+    {
+      "epoch": 0.08502420468021955,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019961318530956556,
+      "loss": 4.9037,
+      "step": 820
+    },
+    {
+      "epoch": 0.08512789273470764,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019961223024012132,
+      "loss": 4.9261,
+      "step": 821
+    },
+    {
+      "epoch": 0.08523158078919571,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0001996112739953581,
+      "loss": 4.8719,
+      "step": 822
+    },
+    {
+      "epoch": 0.08533526884368378,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019961031657528708,
+      "loss": 4.8848,
+      "step": 823
+    },
+    {
+      "epoch": 0.08543895689817185,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019960935797991967,
+      "loss": 4.9255,
+      "step": 824
+    },
+    {
+      "epoch": 0.08554264495265992,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001996083982092671,
+      "loss": 4.8943,
+      "step": 825
+    },
+    {
+      "epoch": 0.085646333007148,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019960743726334072,
+      "loss": 4.8836,
+      "step": 826
+    },
+    {
+      "epoch": 0.08575002106163607,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0001996064751421519,
+      "loss": 4.9091,
+      "step": 827
+    },
+    {
+      "epoch": 0.08585370911612414,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019960551184571192,
+      "loss": 4.8901,
+      "step": 828
+    },
+    {
+      "epoch": 0.08595739717061221,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019960454737403223,
+      "loss": 4.9074,
+      "step": 829
+    },
+    {
+      "epoch": 0.08606108522510028,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019960358172712412,
+      "loss": 4.9147,
+      "step": 830
+    },
+    {
+      "epoch": 0.08616477327958835,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00019960261490499907,
+      "loss": 4.8972,
+      "step": 831
+    },
+    {
+      "epoch": 0.08626846133407644,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019960164690766843,
+      "loss": 4.9044,
+      "step": 832
+    },
+    {
+      "epoch": 0.08637214938856451,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019960067773514364,
+      "loss": 4.9212,
+      "step": 833
+    },
+    {
+      "epoch": 0.08647583744305258,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019959970738743613,
+      "loss": 4.9064,
+      "step": 834
+    },
+    {
+      "epoch": 0.08657952549754065,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019959873586455738,
+      "loss": 4.8741,
+      "step": 835
+    },
+    {
+      "epoch": 0.08668321355202872,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001995977631665188,
+      "loss": 4.9145,
+      "step": 836
+    },
+    {
+      "epoch": 0.0867869016065168,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0001995967892933319,
+      "loss": 4.9278,
+      "step": 837
+    },
+    {
+      "epoch": 0.08689058966100487,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019959581424500817,
+      "loss": 4.9188,
+      "step": 838
+    },
+    {
+      "epoch": 0.08699427771549294,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019959483802155912,
+      "loss": 4.9492,
+      "step": 839
+    },
+    {
+      "epoch": 0.08709796576998101,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019959386062299626,
+      "loss": 4.9078,
+      "step": 840
+    },
+    {
+      "epoch": 0.08720165382446908,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0001995928820493311,
+      "loss": 4.898,
+      "step": 841
+    },
+    {
+      "epoch": 0.08730534187895715,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00019959190230057518,
+      "loss": 4.9058,
+      "step": 842
+    },
+    {
+      "epoch": 0.08740902993344522,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019959092137674013,
+      "loss": 4.8931,
+      "step": 843
+    },
+    {
+      "epoch": 0.08751271798793331,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0001995899392778375,
+      "loss": 4.9151,
+      "step": 844
+    },
+    {
+      "epoch": 0.08761640604242138,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0001995889560038788,
+      "loss": 4.9288,
+      "step": 845
+    },
+    {
+      "epoch": 0.08772009409690945,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001995879715548757,
+      "loss": 4.8856,
+      "step": 846
+    },
+    {
+      "epoch": 0.08782378215139752,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00019958698593083981,
+      "loss": 4.8822,
+      "step": 847
+    },
+    {
+      "epoch": 0.0879274702058856,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019958599913178277,
+      "loss": 4.9602,
+      "step": 848
+    },
+    {
+      "epoch": 0.08803115826037367,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.00019958501115771622,
+      "loss": 4.9163,
+      "step": 849
+    },
+    {
+      "epoch": 0.08813484631486174,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00019958402200865178,
+      "loss": 4.918,
+      "step": 850
+    },
+    {
+      "epoch": 0.08823853436934981,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019958303168460115,
+      "loss": 4.8722,
+      "step": 851
+    },
+    {
+      "epoch": 0.08834222242383788,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.000199582040185576,
+      "loss": 4.8877,
+      "step": 852
+    },
+    {
+      "epoch": 0.08844591047832595,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.00019958104751158806,
+      "loss": 4.905,
+      "step": 853
+    },
+    {
+      "epoch": 0.08854959853281402,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019958005366264901,
+      "loss": 4.8835,
+      "step": 854
+    },
+    {
+      "epoch": 0.08865328658730211,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001995790586387706,
+      "loss": 4.8851,
+      "step": 855
+    },
+    {
+      "epoch": 0.08875697464179018,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019957806243996453,
+      "loss": 4.8804,
+      "step": 856
+    },
+    {
+      "epoch": 0.08886066269627825,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001995770650662426,
+      "loss": 4.9273,
+      "step": 857
+    },
+    {
+      "epoch": 0.08896435075076632,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019957606651761656,
+      "loss": 4.872,
+      "step": 858
+    },
+    {
+      "epoch": 0.0890680388052544,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.00019957506679409818,
+      "loss": 4.9288,
+      "step": 859
+    },
+    {
+      "epoch": 0.08917172685974246,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019957406589569927,
+      "loss": 4.8692,
+      "step": 860
+    },
+    {
+      "epoch": 0.08927541491423054,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019957306382243167,
+      "loss": 4.9116,
+      "step": 861
+    },
+    {
+      "epoch": 0.08937910296871861,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019957206057430712,
+      "loss": 4.8783,
+      "step": 862
+    },
+    {
+      "epoch": 0.08948279102320668,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.00019957105615133754,
+      "loss": 4.8542,
+      "step": 863
+    },
+    {
+      "epoch": 0.08958647907769475,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019957005055353474,
+      "loss": 4.8961,
+      "step": 864
+    },
+    {
+      "epoch": 0.08969016713218282,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001995690437809106,
+      "loss": 4.848,
+      "step": 865
+    },
+    {
+      "epoch": 0.0897938551866709,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019956803583347696,
+      "loss": 4.8486,
+      "step": 866
+    },
+    {
+      "epoch": 0.08989754324115898,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0001995670267112458,
+      "loss": 4.8902,
+      "step": 867
+    },
+    {
+      "epoch": 0.09000123129564705,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019956601641422892,
+      "loss": 4.8824,
+      "step": 868
+    },
+    {
+      "epoch": 0.09010491935013512,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019956500494243832,
+      "loss": 4.8438,
+      "step": 869
+    },
+    {
+      "epoch": 0.09020860740462319,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019956399229588588,
+      "loss": 4.9038,
+      "step": 870
+    },
+    {
+      "epoch": 0.09031229545911126,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001995629784745836,
+      "loss": 4.8895,
+      "step": 871
+    },
+    {
+      "epoch": 0.09041598351359934,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001995619634785434,
+      "loss": 4.9058,
+      "step": 872
+    },
+    {
+      "epoch": 0.0905196715680874,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001995609473077773,
+      "loss": 4.8509,
+      "step": 873
+    },
+    {
+      "epoch": 0.09062335962257548,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019955992996229728,
+      "loss": 4.8988,
+      "step": 874
+    },
+    {
+      "epoch": 0.09072704767706355,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019955891144211524,
+      "loss": 4.9242,
+      "step": 875
+    },
+    {
+      "epoch": 0.09083073573155162,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019955789174724338,
+      "loss": 4.8703,
+      "step": 876
+    },
+    {
+      "epoch": 0.0909344237860397,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00019955687087769357,
+      "loss": 4.8834,
+      "step": 877
+    },
+    {
+      "epoch": 0.09103811184052778,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00019955584883347792,
+      "loss": 4.9105,
+      "step": 878
+    },
+    {
+      "epoch": 0.09114179989501585,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.00019955482561460853,
+      "loss": 4.9056,
+      "step": 879
+    },
+    {
+      "epoch": 0.09124548794950392,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00019955380122109738,
+      "loss": 4.8805,
+      "step": 880
+    },
+    {
+      "epoch": 0.09134917600399199,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00019955277565295666,
+      "loss": 4.8183,
+      "step": 881
+    },
+    {
+      "epoch": 0.09145286405848006,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.00019955174891019838,
+      "loss": 4.8363,
+      "step": 882
+    },
+    {
+      "epoch": 0.09155655211296813,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00019955072099283472,
+      "loss": 4.8429,
+      "step": 883
+    },
+    {
+      "epoch": 0.0916602401674562,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019954969190087777,
+      "loss": 4.8869,
+      "step": 884
+    },
+    {
+      "epoch": 0.09176392822194428,
+      "grad_norm": 0.875,
+      "learning_rate": 0.0001995486616343397,
+      "loss": 4.9302,
+      "step": 885
+    },
+    {
+      "epoch": 0.09186761627643235,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00019954763019323265,
+      "loss": 4.8797,
+      "step": 886
+    },
+    {
+      "epoch": 0.09197130433092042,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019954659757756877,
+      "loss": 4.8558,
+      "step": 887
+    },
+    {
+      "epoch": 0.0920749923854085,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019954556378736028,
+      "loss": 4.9109,
+      "step": 888
+    },
+    {
+      "epoch": 0.09217868043989658,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00019954452882261933,
+      "loss": 4.8751,
+      "step": 889
+    },
+    {
+      "epoch": 0.09228236849438465,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.0001995434926833582,
+      "loss": 4.8356,
+      "step": 890
+    },
+    {
+      "epoch": 0.09238605654887272,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019954245536958908,
+      "loss": 4.8736,
+      "step": 891
+    },
+    {
+      "epoch": 0.09248974460336079,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00019954141688132419,
+      "loss": 4.922,
+      "step": 892
+    },
+    {
+      "epoch": 0.09259343265784886,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001995403772185758,
+      "loss": 4.8539,
+      "step": 893
+    },
+    {
+      "epoch": 0.09269712071233693,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019953933638135616,
+      "loss": 4.9001,
+      "step": 894
+    },
+    {
+      "epoch": 0.092800808766825,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00019953829436967759,
+      "loss": 4.8676,
+      "step": 895
+    },
+    {
+      "epoch": 0.09290449682131308,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00019953725118355235,
+      "loss": 4.8616,
+      "step": 896
+    },
+    {
+      "epoch": 0.09300818487580115,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00019953620682299278,
+      "loss": 4.8614,
+      "step": 897
+    },
+    {
+      "epoch": 0.09311187293028922,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.0001995351612880112,
+      "loss": 4.8647,
+      "step": 898
+    },
+    {
+      "epoch": 0.0932155609847773,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001995341145786199,
+      "loss": 4.8408,
+      "step": 899
+    },
+    {
+      "epoch": 0.09331924903926538,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00019953306669483127,
+      "loss": 4.9216,
+      "step": 900
+    },
+    {
+      "epoch": 0.09342293709375345,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.00019953201763665766,
+      "loss": 4.8706,
+      "step": 901
+    },
+    {
+      "epoch": 0.09352662514824152,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00019953096740411144,
+      "loss": 4.8723,
+      "step": 902
+    },
+    {
+      "epoch": 0.09363031320272959,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00019952991599720503,
+      "loss": 4.8612,
+      "step": 903
+    },
+    {
+      "epoch": 0.09373400125721766,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0001995288634159508,
+      "loss": 4.9012,
+      "step": 904
+    },
+    {
+      "epoch": 0.09383768931170573,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00019952780966036123,
+      "loss": 4.8422,
+      "step": 905
+    },
+    {
+      "epoch": 0.0939413773661938,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019952675473044868,
+      "loss": 4.8762,
+      "step": 906
+    },
+    {
+      "epoch": 0.09404506542068188,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00019952569862622562,
+      "loss": 4.8799,
+      "step": 907
+    },
+    {
+      "epoch": 0.09414875347516995,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.00019952464134770454,
+      "loss": 4.8731,
+      "step": 908
+    },
+    {
+      "epoch": 0.09425244152965802,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.0001995235828948979,
+      "loss": 4.8628,
+      "step": 909
+    },
+    {
+      "epoch": 0.0943561295841461,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00019952252326781815,
+      "loss": 4.8959,
+      "step": 910
+    },
+    {
+      "epoch": 0.09445981763863417,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.00019952146246647785,
+      "loss": 4.8658,
+      "step": 911
+    },
+    {
+      "epoch": 0.09456350569312225,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0001995204004908895,
+      "loss": 4.8525,
+      "step": 912
+    },
+    {
+      "epoch": 0.09466719374761032,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001995193373410656,
+      "loss": 4.8731,
+      "step": 913
+    },
+    {
+      "epoch": 0.09477088180209839,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019951827301701872,
+      "loss": 4.862,
+      "step": 914
+    },
+    {
+      "epoch": 0.09487456985658646,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019951720751876142,
+      "loss": 4.8848,
+      "step": 915
+    },
+    {
+      "epoch": 0.09497825791107453,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019951614084630625,
+      "loss": 4.8816,
+      "step": 916
+    },
+    {
+      "epoch": 0.0950819459655626,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00019951507299966585,
+      "loss": 4.8559,
+      "step": 917
+    },
+    {
+      "epoch": 0.09518563402005067,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019951400397885273,
+      "loss": 4.8733,
+      "step": 918
+    },
+    {
+      "epoch": 0.09528932207453875,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.00019951293378387962,
+      "loss": 4.7999,
+      "step": 919
+    },
+    {
+      "epoch": 0.09539301012902682,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.000199511862414759,
+      "loss": 4.829,
+      "step": 920
+    },
+    {
+      "epoch": 0.09549669818351489,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019951078987150365,
+      "loss": 4.893,
+      "step": 921
+    },
+    {
+      "epoch": 0.09560038623800297,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00019950971615412616,
+      "loss": 4.853,
+      "step": 922
+    },
+    {
+      "epoch": 0.09570407429249105,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019950864126263917,
+      "loss": 4.8412,
+      "step": 923
+    },
+    {
+      "epoch": 0.09580776234697912,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019950756519705544,
+      "loss": 4.8648,
+      "step": 924
+    },
+    {
+      "epoch": 0.09591145040146719,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001995064879573876,
+      "loss": 4.8454,
+      "step": 925
+    },
+    {
+      "epoch": 0.09601513845595526,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001995054095436484,
+      "loss": 4.8425,
+      "step": 926
+    },
+    {
+      "epoch": 0.09611882651044333,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019950432995585054,
+      "loss": 4.8581,
+      "step": 927
+    },
+    {
+      "epoch": 0.0962225145649314,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019950324919400676,
+      "loss": 4.8878,
+      "step": 928
+    },
+    {
+      "epoch": 0.09632620261941947,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019950216725812982,
+      "loss": 4.7899,
+      "step": 929
+    },
+    {
+      "epoch": 0.09642989067390755,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001995010841482325,
+      "loss": 4.8928,
+      "step": 930
+    },
+    {
+      "epoch": 0.09653357872839562,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019949999986432757,
+      "loss": 4.9347,
+      "step": 931
+    },
+    {
+      "epoch": 0.09663726678288369,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0001994989144064278,
+      "loss": 4.8777,
+      "step": 932
+    },
+    {
+      "epoch": 0.09674095483737177,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019949782777454602,
+      "loss": 4.9026,
+      "step": 933
+    },
+    {
+      "epoch": 0.09684464289185984,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019949673996869506,
+      "loss": 4.8867,
+      "step": 934
+    },
+    {
+      "epoch": 0.09694833094634792,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019949565098888771,
+      "loss": 4.847,
+      "step": 935
+    },
+    {
+      "epoch": 0.09705201900083599,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019949456083513686,
+      "loss": 4.8488,
+      "step": 936
+    },
+    {
+      "epoch": 0.09715570705532406,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019949346950745537,
+      "loss": 4.8131,
+      "step": 937
+    },
+    {
+      "epoch": 0.09725939510981213,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001994923770058561,
+      "loss": 4.8802,
+      "step": 938
+    },
+    {
+      "epoch": 0.0973630831643002,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019949128333035198,
+      "loss": 4.8309,
+      "step": 939
+    },
+    {
+      "epoch": 0.09746677121878827,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019949018848095586,
+      "loss": 4.8319,
+      "step": 940
+    },
+    {
+      "epoch": 0.09757045927327634,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019948909245768066,
+      "loss": 4.8677,
+      "step": 941
+    },
+    {
+      "epoch": 0.09767414732776442,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019948799526053938,
+      "loss": 4.8561,
+      "step": 942
+    },
+    {
+      "epoch": 0.09777783538225249,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019948689688954489,
+      "loss": 4.8883,
+      "step": 943
+    },
+    {
+      "epoch": 0.09788152343674057,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019948579734471017,
+      "loss": 4.826,
+      "step": 944
+    },
+    {
+      "epoch": 0.09798521149122864,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019948469662604823,
+      "loss": 4.7702,
+      "step": 945
+    },
+    {
+      "epoch": 0.09808889954571672,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019948359473357202,
+      "loss": 4.8859,
+      "step": 946
+    },
+    {
+      "epoch": 0.09819258760020479,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019948249166729454,
+      "loss": 4.8573,
+      "step": 947
+    },
+    {
+      "epoch": 0.09829627565469286,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001994813874272288,
+      "loss": 4.8528,
+      "step": 948
+    },
+    {
+      "epoch": 0.09839996370918093,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019948028201338788,
+      "loss": 4.9022,
+      "step": 949
+    },
+    {
+      "epoch": 0.098503651763669,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019947917542578478,
+      "loss": 4.8382,
+      "step": 950
+    },
+    {
+      "epoch": 0.09860733981815707,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019947806766443255,
+      "loss": 4.8507,
+      "step": 951
+    },
+    {
+      "epoch": 0.09871102787264514,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001994769587293443,
+      "loss": 4.7939,
+      "step": 952
+    },
+    {
+      "epoch": 0.09881471592713321,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019947584862053307,
+      "loss": 4.8567,
+      "step": 953
+    },
+    {
+      "epoch": 0.09891840398162129,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019947473733801196,
+      "loss": 4.8055,
+      "step": 954
+    },
+    {
+      "epoch": 0.09902209203610937,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019947362488179413,
+      "loss": 4.8541,
+      "step": 955
+    },
+    {
+      "epoch": 0.09912578009059744,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00019947251125189264,
+      "loss": 4.8555,
+      "step": 956
+    },
+    {
+      "epoch": 0.09922946814508551,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0001994713964483207,
+      "loss": 4.8593,
+      "step": 957
+    },
+    {
+      "epoch": 0.09933315619957359,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019947028047109143,
+      "loss": 4.8419,
+      "step": 958
+    },
+    {
+      "epoch": 0.09943684425406166,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019946916332021797,
+      "loss": 4.8051,
+      "step": 959
+    },
+    {
+      "epoch": 0.09954053230854973,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00019946804499571354,
+      "loss": 4.8403,
+      "step": 960
+    },
+    {
+      "epoch": 0.0996442203630378,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00019946692549759133,
+      "loss": 4.8334,
+      "step": 961
+    },
+    {
+      "epoch": 0.09974790841752587,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00019946580482586452,
+      "loss": 4.8314,
+      "step": 962
+    },
+    {
+      "epoch": 0.09985159647201394,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00019946468298054636,
+      "loss": 4.8393,
+      "step": 963
+    },
+    {
+      "epoch": 0.09995528452650201,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00019946355996165006,
+      "loss": 4.8274,
+      "step": 964
+    },
+    {
+      "epoch": 0.10005897258099009,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00019946243576918893,
+      "loss": 4.844,
+      "step": 965
+    },
+    {
+      "epoch": 0.10016266063547817,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00019946131040317618,
+      "loss": 4.8438,
+      "step": 966
+    },
+    {
+      "epoch": 0.10026634868996624,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019946018386362508,
+      "loss": 4.8389,
+      "step": 967
+    },
+    {
+      "epoch": 0.10037003674445431,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00019945905615054898,
+      "loss": 4.822,
+      "step": 968
+    },
+    {
+      "epoch": 0.10047372479894238,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.00019945792726396114,
+      "loss": 4.8085,
+      "step": 969
+    },
+    {
+      "epoch": 0.10057741285343046,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.00019945679720387486,
+      "loss": 4.8656,
+      "step": 970
+    },
+    {
+      "epoch": 0.10068110090791853,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00019945566597030353,
+      "loss": 4.8728,
+      "step": 971
+    },
+    {
+      "epoch": 0.1007847889624066,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00019945453356326045,
+      "loss": 4.8481,
+      "step": 972
+    },
+    {
+      "epoch": 0.10088847701689467,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019945339998275903,
+      "loss": 4.8336,
+      "step": 973
+    },
+    {
+      "epoch": 0.10099216507138274,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0001994522652288126,
+      "loss": 4.8323,
+      "step": 974
+    },
+    {
+      "epoch": 0.10109585312587081,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019945112930143456,
+      "loss": 4.8381,
+      "step": 975
+    },
+    {
+      "epoch": 0.10119954118035888,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.00019944999220063834,
+      "loss": 4.8525,
+      "step": 976
+    },
+    {
+      "epoch": 0.10130322923484697,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00019944885392643734,
+      "loss": 4.8475,
+      "step": 977
+    },
+    {
+      "epoch": 0.10140691728933504,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.00019944771447884496,
+      "loss": 4.8167,
+      "step": 978
+    },
+    {
+      "epoch": 0.10151060534382311,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00019944657385787467,
+      "loss": 4.8381,
+      "step": 979
+    },
+    {
+      "epoch": 0.10161429339831118,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00019944543206353995,
+      "loss": 4.8668,
+      "step": 980
+    },
+    {
+      "epoch": 0.10171798145279926,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00019944428909585423,
+      "loss": 4.8108,
+      "step": 981
+    },
+    {
+      "epoch": 0.10182166950728733,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00019944314495483104,
+      "loss": 4.8041,
+      "step": 982
+    },
+    {
+      "epoch": 0.1019253575617754,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001994419996404838,
+      "loss": 4.7916,
+      "step": 983
+    },
+    {
+      "epoch": 0.10202904561626347,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00019944085315282614,
+      "loss": 4.8384,
+      "step": 984
+    },
+    {
+      "epoch": 0.10213273367075154,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0001994397054918715,
+      "loss": 4.8389,
+      "step": 985
+    },
+    {
+      "epoch": 0.10223642172523961,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00019943855665763345,
+      "loss": 4.8711,
+      "step": 986
+    },
+    {
+      "epoch": 0.10234010977972768,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019943740665012553,
+      "loss": 4.8159,
+      "step": 987
+    },
+    {
+      "epoch": 0.10244379783421576,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00019943625546936134,
+      "loss": 4.7974,
+      "step": 988
+    },
+    {
+      "epoch": 0.10254748588870384,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00019943510311535445,
+      "loss": 4.8683,
+      "step": 989
+    },
+    {
+      "epoch": 0.10265117394319191,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00019943394958811842,
+      "loss": 4.8304,
+      "step": 990
+    },
+    {
+      "epoch": 0.10275486199767998,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019943279488766693,
+      "loss": 4.8283,
+      "step": 991
+    },
+    {
+      "epoch": 0.10285855005216805,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019943163901401355,
+      "loss": 4.8337,
+      "step": 992
+    },
+    {
+      "epoch": 0.10296223810665613,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0001994304819671719,
+      "loss": 4.8154,
+      "step": 993
+    },
+    {
+      "epoch": 0.1030659261611442,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001994293237471557,
+      "loss": 4.8173,
+      "step": 994
+    },
+    {
+      "epoch": 0.10316961421563227,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0001994281643539786,
+      "loss": 4.8074,
+      "step": 995
+    },
+    {
+      "epoch": 0.10327330227012034,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019942700378765423,
+      "loss": 4.8224,
+      "step": 996
+    },
+    {
+      "epoch": 0.10337699032460841,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019942584204819632,
+      "loss": 4.8471,
+      "step": 997
+    },
+    {
+      "epoch": 0.10348067837909648,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019942467913561859,
+      "loss": 4.776,
+      "step": 998
+    },
+    {
+      "epoch": 0.10358436643358455,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001994235150499347,
+      "loss": 4.8177,
+      "step": 999
+    },
+    {
+      "epoch": 0.10368805448807264,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00019942234979115848,
+      "loss": 4.8716,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10379174254256071,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0001994211833593036,
+      "loss": 4.8509,
+      "step": 1001
+    },
+    {
+      "epoch": 0.10389543059704878,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019942001575438384,
+      "loss": 4.7899,
+      "step": 1002
+    },
+    {
+      "epoch": 0.10399911865153685,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019941884697641298,
+      "loss": 4.8424,
+      "step": 1003
+    },
+    {
+      "epoch": 0.10410280670602493,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019941767702540483,
+      "loss": 4.8242,
+      "step": 1004
+    },
+    {
+      "epoch": 0.104206494760513,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001994165059013732,
+      "loss": 4.8327,
+      "step": 1005
+    },
+    {
+      "epoch": 0.10431018281500107,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019941533360433184,
+      "loss": 4.8073,
+      "step": 1006
+    },
+    {
+      "epoch": 0.10441387086948914,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019941416013429468,
+      "loss": 4.82,
+      "step": 1007
+    },
+    {
+      "epoch": 0.10451755892397721,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0001994129854912755,
+      "loss": 4.8072,
+      "step": 1008
+    },
+    {
+      "epoch": 0.10462124697846528,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019941180967528817,
+      "loss": 4.8173,
+      "step": 1009
+    },
+    {
+      "epoch": 0.10472493503295335,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019941063268634655,
+      "loss": 4.7941,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10482862308744144,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.00019940945452446456,
+      "loss": 4.8366,
+      "step": 1011
+    },
+    {
+      "epoch": 0.10493231114192951,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0001994082751896561,
+      "loss": 4.832,
+      "step": 1012
+    },
+    {
+      "epoch": 0.10503599919641758,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019940709468193509,
+      "loss": 4.8127,
+      "step": 1013
+    },
+    {
+      "epoch": 0.10513968725090565,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00019940591300131538,
+      "loss": 4.8416,
+      "step": 1014
+    },
+    {
+      "epoch": 0.10524337530539372,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019940473014781103,
+      "loss": 4.8164,
+      "step": 1015
+    },
+    {
+      "epoch": 0.1053470633598818,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0001994035461214359,
+      "loss": 4.8007,
+      "step": 1016
+    },
+    {
+      "epoch": 0.10545075141436987,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.00019940236092220404,
+      "loss": 4.7715,
+      "step": 1017
+    },
+    {
+      "epoch": 0.10555443946885794,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019940117455012935,
+      "loss": 4.8128,
+      "step": 1018
+    },
+    {
+      "epoch": 0.10565812752334601,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019939998700522587,
+      "loss": 4.8322,
+      "step": 1019
+    },
+    {
+      "epoch": 0.10576181557783408,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019939879828750768,
+      "loss": 4.7856,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10586550363232215,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019939760839698866,
+      "loss": 4.7796,
+      "step": 1021
+    },
+    {
+      "epoch": 0.10596919168681024,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019939641733368298,
+      "loss": 4.812,
+      "step": 1022
+    },
+    {
+      "epoch": 0.10607287974129831,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019939522509760462,
+      "loss": 4.8104,
+      "step": 1023
+    },
+    {
+      "epoch": 0.10617656779578638,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0001993940316887677,
+      "loss": 4.8058,
+      "step": 1024
+    },
+    {
+      "epoch": 0.10628025585027445,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.00019939283710718623,
+      "loss": 4.8323,
+      "step": 1025
+    },
+    {
+      "epoch": 0.10638394390476252,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00019939164135287433,
+      "loss": 4.8172,
+      "step": 1026
+    },
+    {
+      "epoch": 0.1064876319592506,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019939044442584617,
+      "loss": 4.8447,
+      "step": 1027
+    },
+    {
+      "epoch": 0.10659132001373867,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00019938924632611582,
+      "loss": 4.8457,
+      "step": 1028
+    },
+    {
+      "epoch": 0.10669500806822674,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019938804705369741,
+      "loss": 4.8359,
+      "step": 1029
+    },
+    {
+      "epoch": 0.10679869612271481,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019938684660860513,
+      "loss": 4.8462,
+      "step": 1030
+    },
+    {
+      "epoch": 0.10690238417720288,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019938564499085305,
+      "loss": 4.853,
+      "step": 1031
+    },
+    {
+      "epoch": 0.10700607223169095,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019938444220045547,
+      "loss": 4.7956,
+      "step": 1032
+    },
+    {
+      "epoch": 0.10710976028617904,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001993832382374265,
+      "loss": 4.8215,
+      "step": 1033
+    },
+    {
+      "epoch": 0.10721344834066711,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0001993820331017804,
+      "loss": 4.8004,
+      "step": 1034
+    },
+    {
+      "epoch": 0.10731713639515518,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019938082679353132,
+      "loss": 4.8516,
+      "step": 1035
+    },
+    {
+      "epoch": 0.10742082444964325,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019937961931269357,
+      "loss": 4.7898,
+      "step": 1036
+    },
+    {
+      "epoch": 0.10752451250413132,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019937841065928135,
+      "loss": 4.8164,
+      "step": 1037
+    },
+    {
+      "epoch": 0.1076282005586194,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019937720083330893,
+      "loss": 4.8158,
+      "step": 1038
+    },
+    {
+      "epoch": 0.10773188861310747,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019937598983479058,
+      "loss": 4.8225,
+      "step": 1039
+    },
+    {
+      "epoch": 0.10783557666759554,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0001993747776637406,
+      "loss": 4.7566,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10793926472208361,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019937356432017328,
+      "loss": 4.8108,
+      "step": 1041
+    },
+    {
+      "epoch": 0.10804295277657168,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019937234980410296,
+      "loss": 4.7864,
+      "step": 1042
+    },
+    {
+      "epoch": 0.10814664083105975,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019937113411554395,
+      "loss": 4.8389,
+      "step": 1043
+    },
+    {
+      "epoch": 0.10825032888554784,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019936991725451057,
+      "loss": 4.8236,
+      "step": 1044
+    },
+    {
+      "epoch": 0.10835401694003591,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019936869922101727,
+      "loss": 4.8259,
+      "step": 1045
+    },
+    {
+      "epoch": 0.10845770499452398,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001993674800150783,
+      "loss": 4.7816,
+      "step": 1046
+    },
+    {
+      "epoch": 0.10856139304901205,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019936625963670813,
+      "loss": 4.7689,
+      "step": 1047
+    },
+    {
+      "epoch": 0.10866508110350012,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0001993650380859211,
+      "loss": 4.8243,
+      "step": 1048
+    },
+    {
+      "epoch": 0.1087687691579882,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019936381536273166,
+      "loss": 4.828,
+      "step": 1049
+    },
+    {
+      "epoch": 0.10887245721247626,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00019936259146715425,
+      "loss": 4.8372,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10897614526696434,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0001993613663992033,
+      "loss": 4.7794,
+      "step": 1051
+    },
+    {
+      "epoch": 0.10907983332145241,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00019936014015889321,
+      "loss": 4.7735,
+      "step": 1052
+    },
+    {
+      "epoch": 0.10918352137594048,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.00019935891274623852,
+      "loss": 4.7367,
+      "step": 1053
+    },
+    {
+      "epoch": 0.10928720943042855,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019935768416125371,
+      "loss": 4.8041,
+      "step": 1054
+    },
+    {
+      "epoch": 0.10939089748491662,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00019935645440395322,
+      "loss": 4.8382,
+      "step": 1055
+    },
+    {
+      "epoch": 0.1094945855394047,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.0001993552234743516,
+      "loss": 4.7804,
+      "step": 1056
+    },
+    {
+      "epoch": 0.10959827359389278,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019935399137246338,
+      "loss": 4.8278,
+      "step": 1057
+    },
+    {
+      "epoch": 0.10970196164838085,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019935275809830307,
+      "loss": 4.7946,
+      "step": 1058
+    },
+    {
+      "epoch": 0.10980564970286892,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019935152365188526,
+      "loss": 4.8324,
+      "step": 1059
+    },
+    {
+      "epoch": 0.10990933775735699,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019935028803322444,
+      "loss": 4.7871,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11001302581184506,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019934905124233528,
+      "loss": 4.7859,
+      "step": 1061
+    },
+    {
+      "epoch": 0.11011671386633314,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019934781327923232,
+      "loss": 4.7975,
+      "step": 1062
+    },
+    {
+      "epoch": 0.1102204019208212,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019934657414393016,
+      "loss": 4.785,
+      "step": 1063
+    },
+    {
+      "epoch": 0.11032408997530928,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.00019934533383644348,
+      "loss": 4.8142,
+      "step": 1064
+    },
+    {
+      "epoch": 0.11042777802979735,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00019934409235678683,
+      "loss": 4.8027,
+      "step": 1065
+    },
+    {
+      "epoch": 0.11053146608428542,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00019934284970497492,
+      "loss": 4.8281,
+      "step": 1066
+    },
+    {
+      "epoch": 0.1106351541387735,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00019934160588102242,
+      "loss": 4.8019,
+      "step": 1067
+    },
+    {
+      "epoch": 0.11073884219326158,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00019934036088494394,
+      "loss": 4.7709,
+      "step": 1068
+    },
+    {
+      "epoch": 0.11084253024774965,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00019933911471675423,
+      "loss": 4.7505,
+      "step": 1069
+    },
+    {
+      "epoch": 0.11094621830223772,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00019933786737646797,
+      "loss": 4.8155,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11104990635672579,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019933661886409988,
+      "loss": 4.8044,
+      "step": 1071
+    },
+    {
+      "epoch": 0.11115359441121386,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00019933536917966468,
+      "loss": 4.8024,
+      "step": 1072
+    },
+    {
+      "epoch": 0.11125728246570193,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00019933411832317712,
+      "loss": 4.7923,
+      "step": 1073
+    },
+    {
+      "epoch": 0.11136097052019,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.000199332866294652,
+      "loss": 4.7929,
+      "step": 1074
+    },
+    {
+      "epoch": 0.11146465857467808,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.00019933161309410402,
+      "loss": 4.8082,
+      "step": 1075
+    },
+    {
+      "epoch": 0.11156834662916615,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019933035872154802,
+      "loss": 4.7484,
+      "step": 1076
+    },
+    {
+      "epoch": 0.11167203468365422,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00019932910317699878,
+      "loss": 4.7837,
+      "step": 1077
+    },
+    {
+      "epoch": 0.1117757227381423,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00019932784646047112,
+      "loss": 4.8484,
+      "step": 1078
+    },
+    {
+      "epoch": 0.11187941079263038,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00019932658857197986,
+      "loss": 4.8276,
+      "step": 1079
+    },
+    {
+      "epoch": 0.11198309884711845,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019932532951153986,
+      "loss": 4.7676,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11208678690160652,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.00019932406927916595,
+      "loss": 4.8098,
+      "step": 1081
+    },
+    {
+      "epoch": 0.11219047495609459,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.000199322807874873,
+      "loss": 4.8138,
+      "step": 1082
+    },
+    {
+      "epoch": 0.11229416301058266,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019932154529867595,
+      "loss": 4.8027,
+      "step": 1083
+    },
+    {
+      "epoch": 0.11239785106507073,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019932028155058963,
+      "loss": 4.8091,
+      "step": 1084
+    },
+    {
+      "epoch": 0.1125015391195588,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00019931901663062894,
+      "loss": 4.787,
+      "step": 1085
+    },
+    {
+      "epoch": 0.11260522717404688,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.00019931775053880888,
+      "loss": 4.7998,
+      "step": 1086
+    },
+    {
+      "epoch": 0.11270891522853495,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00019931648327514435,
+      "loss": 4.819,
+      "step": 1087
+    },
+    {
+      "epoch": 0.11281260328302302,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0001993152148396503,
+      "loss": 4.7823,
+      "step": 1088
+    },
+    {
+      "epoch": 0.1129162913375111,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019931394523234165,
+      "loss": 4.804,
+      "step": 1089
+    },
+    {
+      "epoch": 0.11301997939199918,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019931267445323346,
+      "loss": 4.8188,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11312366744648725,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.00019931140250234068,
+      "loss": 4.8397,
+      "step": 1091
+    },
+    {
+      "epoch": 0.11322735550097532,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019931012937967834,
+      "loss": 4.7646,
+      "step": 1092
+    },
+    {
+      "epoch": 0.11333104355546339,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019930885508526145,
+      "loss": 4.8125,
+      "step": 1093
+    },
+    {
+      "epoch": 0.11343473160995146,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.000199307579619105,
+      "loss": 4.7948,
+      "step": 1094
+    },
+    {
+      "epoch": 0.11353841966443953,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019930630298122415,
+      "loss": 4.7597,
+      "step": 1095
+    },
+    {
+      "epoch": 0.1136421077189276,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019930502517163386,
+      "loss": 4.7938,
+      "step": 1096
+    },
+    {
+      "epoch": 0.11374579577341568,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00019930374619034927,
+      "loss": 4.8352,
+      "step": 1097
+    },
+    {
+      "epoch": 0.11384948382790375,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0001993024660373854,
+      "loss": 4.8116,
+      "step": 1098
+    },
+    {
+      "epoch": 0.11395317188239182,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00019930118471275744,
+      "loss": 4.7617,
+      "step": 1099
+    },
+    {
+      "epoch": 0.1140568599368799,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019929990221648043,
+      "loss": 4.7716,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11416054799136797,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019929861854856956,
+      "loss": 4.8149,
+      "step": 1101
+    },
+    {
+      "epoch": 0.11426423604585605,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019929733370903995,
+      "loss": 4.7617,
+      "step": 1102
+    },
+    {
+      "epoch": 0.11436792410034412,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019929604769790675,
+      "loss": 4.7973,
+      "step": 1103
+    },
+    {
+      "epoch": 0.11447161215483219,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019929476051518518,
+      "loss": 4.8084,
+      "step": 1104
+    },
+    {
+      "epoch": 0.11457530020932026,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00019929347216089037,
+      "loss": 4.8131,
+      "step": 1105
+    },
+    {
+      "epoch": 0.11467898826380833,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.00019929218263503752,
+      "loss": 4.7745,
+      "step": 1106
+    },
+    {
+      "epoch": 0.1147826763182964,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00019929089193764193,
+      "loss": 4.7861,
+      "step": 1107
+    },
+    {
+      "epoch": 0.11488636437278447,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.00019928960006871874,
+      "loss": 4.7646,
+      "step": 1108
+    },
+    {
+      "epoch": 0.11499005242727255,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019928830702828322,
+      "loss": 4.8098,
+      "step": 1109
+    },
+    {
+      "epoch": 0.11509374048176062,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019928701281635066,
+      "loss": 4.7654,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1151974285362487,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00019928571743293625,
+      "loss": 4.7427,
+      "step": 1111
+    },
+    {
+      "epoch": 0.11530111659073677,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00019928442087805537,
+      "loss": 4.8149,
+      "step": 1112
+    },
+    {
+      "epoch": 0.11540480464522485,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019928312315172324,
+      "loss": 4.8035,
+      "step": 1113
+    },
+    {
+      "epoch": 0.11550849269971292,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0001992818242539552,
+      "loss": 4.7691,
+      "step": 1114
+    },
+    {
+      "epoch": 0.11561218075420099,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0001992805241847666,
+      "loss": 4.7827,
+      "step": 1115
+    },
+    {
+      "epoch": 0.11571586880868906,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00019927922294417277,
+      "loss": 4.7606,
+      "step": 1116
+    },
+    {
+      "epoch": 0.11581955686317713,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019927792053218903,
+      "loss": 4.7606,
+      "step": 1117
+    },
+    {
+      "epoch": 0.1159232449176652,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019927661694883075,
+      "loss": 4.7667,
+      "step": 1118
+    },
+    {
+      "epoch": 0.11602693297215327,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00019927531219411337,
+      "loss": 4.7663,
+      "step": 1119
+    },
+    {
+      "epoch": 0.11613062102664135,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019927400626805223,
+      "loss": 4.7812,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11623430908112942,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019927269917066273,
+      "loss": 4.7531,
+      "step": 1121
+    },
+    {
+      "epoch": 0.11633799713561749,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019927139090196035,
+      "loss": 4.7844,
+      "step": 1122
+    },
+    {
+      "epoch": 0.11644168519010557,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019927008146196048,
+      "loss": 4.7869,
+      "step": 1123
+    },
+    {
+      "epoch": 0.11654537324459364,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0001992687708506786,
+      "loss": 4.78,
+      "step": 1124
+    },
+    {
+      "epoch": 0.11664906129908172,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019926745906813012,
+      "loss": 4.7729,
+      "step": 1125
+    },
+    {
+      "epoch": 0.11675274935356979,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019926614611433057,
+      "loss": 4.8032,
+      "step": 1126
+    },
+    {
+      "epoch": 0.11685643740805786,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0001992648319892954,
+      "loss": 4.7693,
+      "step": 1127
+    },
+    {
+      "epoch": 0.11696012546254593,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019926351669304017,
+      "loss": 4.7452,
+      "step": 1128
+    },
+    {
+      "epoch": 0.117063813517034,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019926220022558036,
+      "loss": 4.7493,
+      "step": 1129
+    },
+    {
+      "epoch": 0.11716750157152207,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019926088258693153,
+      "loss": 4.7705,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11727118962601014,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0001992595637771092,
+      "loss": 4.7883,
+      "step": 1131
+    },
+    {
+      "epoch": 0.11737487768049822,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0001992582437961289,
+      "loss": 4.767,
+      "step": 1132
+    },
+    {
+      "epoch": 0.11747856573498629,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019925692264400629,
+      "loss": 4.7713,
+      "step": 1133
+    },
+    {
+      "epoch": 0.11758225378947437,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001992556003207569,
+      "loss": 4.7332,
+      "step": 1134
+    },
+    {
+      "epoch": 0.11768594184396244,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019925427682639636,
+      "loss": 4.7582,
+      "step": 1135
+    },
+    {
+      "epoch": 0.11778962989845052,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019925295216094023,
+      "loss": 4.8117,
+      "step": 1136
+    },
+    {
+      "epoch": 0.11789331795293859,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019925162632440419,
+      "loss": 4.7745,
+      "step": 1137
+    },
+    {
+      "epoch": 0.11799700600742666,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0001992502993168039,
+      "loss": 4.7928,
+      "step": 1138
+    },
+    {
+      "epoch": 0.11810069406191473,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019924897113815496,
+      "loss": 4.7954,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1182043821164028,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001992476417884731,
+      "loss": 4.7575,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11830807017089087,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019924631126777396,
+      "loss": 4.7496,
+      "step": 1141
+    },
+    {
+      "epoch": 0.11841175822537894,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019924497957607326,
+      "loss": 4.7866,
+      "step": 1142
+    },
+    {
+      "epoch": 0.11851544627986701,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00019924364671338672,
+      "loss": 4.773,
+      "step": 1143
+    },
+    {
+      "epoch": 0.11861913433435509,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019924231267973004,
+      "loss": 4.7703,
+      "step": 1144
+    },
+    {
+      "epoch": 0.11872282238884317,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019924097747511896,
+      "loss": 4.7405,
+      "step": 1145
+    },
+    {
+      "epoch": 0.11882651044333124,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019923964109956925,
+      "loss": 4.7476,
+      "step": 1146
+    },
+    {
+      "epoch": 0.11893019849781931,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019923830355309669,
+      "loss": 4.7496,
+      "step": 1147
+    },
+    {
+      "epoch": 0.11903388655230739,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00019923696483571703,
+      "loss": 4.7838,
+      "step": 1148
+    },
+    {
+      "epoch": 0.11913757460679546,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.00019923562494744611,
+      "loss": 4.7803,
+      "step": 1149
+    },
+    {
+      "epoch": 0.11924126266128353,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001992342838882997,
+      "loss": 4.7226,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1193449507157716,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019923294165829364,
+      "loss": 4.7638,
+      "step": 1151
+    },
+    {
+      "epoch": 0.11944863877025967,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00019923159825744376,
+      "loss": 4.7377,
+      "step": 1152
+    },
+    {
+      "epoch": 0.11955232682474774,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0001992302536857659,
+      "loss": 4.7789,
+      "step": 1153
+    },
+    {
+      "epoch": 0.11965601487923581,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019922890794327595,
+      "loss": 4.7689,
+      "step": 1154
+    },
+    {
+      "epoch": 0.11975970293372389,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019922756102998975,
+      "loss": 4.7883,
+      "step": 1155
+    },
+    {
+      "epoch": 0.11986339098821197,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019922621294592326,
+      "loss": 4.7345,
+      "step": 1156
+    },
+    {
+      "epoch": 0.11996707904270004,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019922486369109232,
+      "loss": 4.7087,
+      "step": 1157
+    },
+    {
+      "epoch": 0.12007076709718811,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00019922351326551288,
+      "loss": 4.764,
+      "step": 1158
+    },
+    {
+      "epoch": 0.12017445515167618,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00019922216166920088,
+      "loss": 4.751,
+      "step": 1159
+    },
+    {
+      "epoch": 0.12027814320616426,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019922080890217222,
+      "loss": 4.7546,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12038183126065233,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019921945496444293,
+      "loss": 4.7621,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1204855193151404,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019921809985602894,
+      "loss": 4.729,
+      "step": 1162
+    },
+    {
+      "epoch": 0.12058920736962847,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019921674357694624,
+      "loss": 4.8148,
+      "step": 1163
+    },
+    {
+      "epoch": 0.12069289542411654,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019921538612721084,
+      "loss": 4.7172,
+      "step": 1164
+    },
+    {
+      "epoch": 0.12079658347860461,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0001992140275068388,
+      "loss": 4.7725,
+      "step": 1165
+    },
+    {
+      "epoch": 0.12090027153309268,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019921266771584604,
+      "loss": 4.7636,
+      "step": 1166
+    },
+    {
+      "epoch": 0.12100395958758077,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0001992113067542487,
+      "loss": 4.7482,
+      "step": 1167
+    },
+    {
+      "epoch": 0.12110764764206884,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019920994462206283,
+      "loss": 4.7445,
+      "step": 1168
+    },
+    {
+      "epoch": 0.12121133569655691,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0001992085813193045,
+      "loss": 4.8108,
+      "step": 1169
+    },
+    {
+      "epoch": 0.12131502375104498,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019920721684598975,
+      "loss": 4.7589,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12141871180553306,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001992058512021347,
+      "loss": 4.7789,
+      "step": 1171
+    },
+    {
+      "epoch": 0.12152239986002113,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001992044843877555,
+      "loss": 4.757,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1216260879145092,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019920311640286823,
+      "loss": 4.7454,
+      "step": 1173
+    },
+    {
+      "epoch": 0.12172977596899727,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019920174724748902,
+      "loss": 4.7181,
+      "step": 1174
+    },
+    {
+      "epoch": 0.12183346402348534,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019920037692163409,
+      "loss": 4.7726,
+      "step": 1175
+    },
+    {
+      "epoch": 0.12193715207797341,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019919900542531956,
+      "loss": 4.7689,
+      "step": 1176
+    },
+    {
+      "epoch": 0.12204084013246148,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019919763275856164,
+      "loss": 4.7677,
+      "step": 1177
+    },
+    {
+      "epoch": 0.12214452818694957,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0001991962589213765,
+      "loss": 4.77,
+      "step": 1178
+    },
+    {
+      "epoch": 0.12224821624143764,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019919488391378034,
+      "loss": 4.7638,
+      "step": 1179
+    },
+    {
+      "epoch": 0.12235190429592571,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001991935077357894,
+      "loss": 4.7877,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12245559235041378,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019919213038741996,
+      "loss": 4.7555,
+      "step": 1181
+    },
+    {
+      "epoch": 0.12255928040490185,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019919075186868824,
+      "loss": 4.7536,
+      "step": 1182
+    },
+    {
+      "epoch": 0.12266296845938993,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00019918937217961043,
+      "loss": 4.7577,
+      "step": 1183
+    },
+    {
+      "epoch": 0.122766656513878,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001991879913202029,
+      "loss": 4.7453,
+      "step": 1184
+    },
+    {
+      "epoch": 0.12287034456836607,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019918660929048196,
+      "loss": 4.8053,
+      "step": 1185
+    },
+    {
+      "epoch": 0.12297403262285414,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019918522609046387,
+      "loss": 4.7143,
+      "step": 1186
+    },
+    {
+      "epoch": 0.12307772067734221,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019918384172016494,
+      "loss": 4.7592,
+      "step": 1187
+    },
+    {
+      "epoch": 0.12318140873183028,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0001991824561796015,
+      "loss": 4.7616,
+      "step": 1188
+    },
+    {
+      "epoch": 0.12328509678631835,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019918106946878995,
+      "loss": 4.7261,
+      "step": 1189
+    },
+    {
+      "epoch": 0.12338878484080644,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019917968158774657,
+      "loss": 4.7459,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12349247289529451,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019917829253648784,
+      "loss": 4.7121,
+      "step": 1191
+    },
+    {
+      "epoch": 0.12359616094978258,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019917690231503006,
+      "loss": 4.7695,
+      "step": 1192
+    },
+    {
+      "epoch": 0.12369984900427065,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019917551092338967,
+      "loss": 4.768,
+      "step": 1193
+    },
+    {
+      "epoch": 0.12380353705875873,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019917411836158308,
+      "loss": 4.7858,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1239072251132468,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019917272462962674,
+      "loss": 4.776,
+      "step": 1195
+    },
+    {
+      "epoch": 0.12401091316773487,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0001991713297275371,
+      "loss": 4.77,
+      "step": 1196
+    },
+    {
+      "epoch": 0.12411460122222294,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019916993365533056,
+      "loss": 4.7629,
+      "step": 1197
+    },
+    {
+      "epoch": 0.12421828927671101,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019916853641302365,
+      "loss": 4.7412,
+      "step": 1198
+    },
+    {
+      "epoch": 0.12432197733119908,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0001991671380006328,
+      "loss": 4.7251,
+      "step": 1199
+    },
+    {
+      "epoch": 0.12442566538568715,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0001991657384181746,
+      "loss": 4.7247,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12452935344017524,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019916433766566547,
+      "loss": 4.744,
+      "step": 1201
+    },
+    {
+      "epoch": 0.12463304149466331,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000199162935743122,
+      "loss": 4.7852,
+      "step": 1202
+    },
+    {
+      "epoch": 0.12473672954915138,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0001991615326505607,
+      "loss": 4.76,
+      "step": 1203
+    },
+    {
+      "epoch": 0.12484041760363945,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00019916012838799813,
+      "loss": 4.7716,
+      "step": 1204
+    },
+    {
+      "epoch": 0.12494410565812752,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00019915872295545086,
+      "loss": 4.7749,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1250477937126156,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00019915731635293548,
+      "loss": 4.7514,
+      "step": 1206
+    },
+    {
+      "epoch": 0.12515148176710367,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00019915590858046858,
+      "loss": 4.765,
+      "step": 1207
+    },
+    {
+      "epoch": 0.12525516982159174,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00019915449963806676,
+      "loss": 4.7111,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1253588578760798,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0001991530895257467,
+      "loss": 4.806,
+      "step": 1209
+    },
+    {
+      "epoch": 0.12546254593056788,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00019915167824352493,
+      "loss": 4.7864,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12556623398505595,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001991502657914182,
+      "loss": 4.7845,
+      "step": 1211
+    },
+    {
+      "epoch": 0.12566992203954402,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00019914885216944312,
+      "loss": 4.7299,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1257736100940321,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001991474373776164,
+      "loss": 4.7712,
+      "step": 1213
+    },
+    {
+      "epoch": 0.12587729814852017,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019914602141595476,
+      "loss": 4.7366,
+      "step": 1214
+    },
+    {
+      "epoch": 0.12598098620300824,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0001991446042844748,
+      "loss": 4.7595,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1260846742574963,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019914318598319338,
+      "loss": 4.7051,
+      "step": 1216
+    },
+    {
+      "epoch": 0.12618836231198438,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0001991417665121271,
+      "loss": 4.7561,
+      "step": 1217
+    },
+    {
+      "epoch": 0.12629205036647248,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0001991403458712928,
+      "loss": 4.7613,
+      "step": 1218
+    },
+    {
+      "epoch": 0.12639573842096055,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00019913892406070723,
+      "loss": 4.7776,
+      "step": 1219
+    },
+    {
+      "epoch": 0.12649942647544862,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.00019913750108038715,
+      "loss": 4.7397,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1266031145299367,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00019913607693034934,
+      "loss": 4.7404,
+      "step": 1221
+    },
+    {
+      "epoch": 0.12670680258442477,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0001991346516106106,
+      "loss": 4.7672,
+      "step": 1222
+    },
+    {
+      "epoch": 0.12681049063891284,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019913322512118774,
+      "loss": 4.7471,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1269141786934009,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00019913179746209765,
+      "loss": 4.8062,
+      "step": 1224
+    },
+    {
+      "epoch": 0.12701786674788898,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019913036863335713,
+      "loss": 4.776,
+      "step": 1225
+    },
+    {
+      "epoch": 0.12712155480237705,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00019912893863498305,
+      "loss": 4.7452,
+      "step": 1226
+    },
+    {
+      "epoch": 0.12722524285686512,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00019912750746699226,
+      "loss": 4.7816,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1273289309113532,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001991260751294017,
+      "loss": 4.719,
+      "step": 1228
+    },
+    {
+      "epoch": 0.12743261896584127,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019912464162222818,
+      "loss": 4.7678,
+      "step": 1229
+    },
+    {
+      "epoch": 0.12753630702032934,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001991232069454887,
+      "loss": 4.7393,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1276399950748174,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00019912177109920016,
+      "loss": 4.7238,
+      "step": 1231
+    },
+    {
+      "epoch": 0.12774368312930548,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001991203340833795,
+      "loss": 4.7208,
+      "step": 1232
+    },
+    {
+      "epoch": 0.12784737118379355,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00019911889589804366,
+      "loss": 4.7451,
+      "step": 1233
+    },
+    {
+      "epoch": 0.12795105923828162,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.00019911745654320963,
+      "loss": 4.7744,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1280547472927697,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00019911601601889438,
+      "loss": 4.7203,
+      "step": 1235
+    },
+    {
+      "epoch": 0.12815843534725777,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001991145743251149,
+      "loss": 4.7109,
+      "step": 1236
+    },
+    {
+      "epoch": 0.12826212340174584,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00019911313146188823,
+      "loss": 4.7083,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1283658114562339,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019911168742923138,
+      "loss": 4.711,
+      "step": 1238
+    },
+    {
+      "epoch": 0.12846949951072198,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00019911024222716138,
+      "loss": 4.754,
+      "step": 1239
+    },
+    {
+      "epoch": 0.12857318756521008,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0001991087958556953,
+      "loss": 4.7559,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12867687561969815,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019910734831485015,
+      "loss": 4.7452,
+      "step": 1241
+    },
+    {
+      "epoch": 0.12878056367418622,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019910589960464304,
+      "loss": 4.7699,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1288842517286743,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00019910444972509112,
+      "loss": 4.7216,
+      "step": 1243
+    },
+    {
+      "epoch": 0.12898793978316236,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019910299867621146,
+      "loss": 4.7461,
+      "step": 1244
+    },
+    {
+      "epoch": 0.12909162783765044,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019910154645802112,
+      "loss": 4.7232,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1291953158921385,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00019910009307053735,
+      "loss": 4.7075,
+      "step": 1246
+    },
+    {
+      "epoch": 0.12929900394662658,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019909863851377718,
+      "loss": 4.7002,
+      "step": 1247
+    },
+    {
+      "epoch": 0.12940269200111465,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019909718278775785,
+      "loss": 4.734,
+      "step": 1248
+    },
+    {
+      "epoch": 0.12950638005560272,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0001990957258924965,
+      "loss": 4.731,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1296100681100908,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019909426782801037,
+      "loss": 4.7217,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12971375616457886,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019909280859431658,
+      "loss": 4.723,
+      "step": 1251
+    },
+    {
+      "epoch": 0.12981744421906694,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019909134819143243,
+      "loss": 4.7431,
+      "step": 1252
+    },
+    {
+      "epoch": 0.129921132273555,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019908988661937512,
+      "loss": 4.7613,
+      "step": 1253
+    },
+    {
+      "epoch": 0.13002482032804308,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019908842387816183,
+      "loss": 4.7186,
+      "step": 1254
+    },
+    {
+      "epoch": 0.13012850838253115,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019908695996780993,
+      "loss": 4.7484,
+      "step": 1255
+    },
+    {
+      "epoch": 0.13023219643701922,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019908549488833663,
+      "loss": 4.7378,
+      "step": 1256
+    },
+    {
+      "epoch": 0.1303358844915073,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019908402863975925,
+      "loss": 4.7338,
+      "step": 1257
+    },
+    {
+      "epoch": 0.13043957254599536,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019908256122209506,
+      "loss": 4.7363,
+      "step": 1258
+    },
+    {
+      "epoch": 0.13054326060048344,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019908109263536134,
+      "loss": 4.7391,
+      "step": 1259
+    },
+    {
+      "epoch": 0.1306469486549715,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019907962287957548,
+      "loss": 4.7361,
+      "step": 1260
+    },
+    {
+      "epoch": 0.13075063670945958,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019907815195475484,
+      "loss": 4.7187,
+      "step": 1261
+    },
+    {
+      "epoch": 0.13085432476394768,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001990766798609167,
+      "loss": 4.7298,
+      "step": 1262
+    },
+    {
+      "epoch": 0.13095801281843575,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0001990752065980785,
+      "loss": 4.7505,
+      "step": 1263
+    },
+    {
+      "epoch": 0.13106170087292382,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019907373216625755,
+      "loss": 4.6917,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1311653889274119,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0001990722565654713,
+      "loss": 4.7232,
+      "step": 1265
+    },
+    {
+      "epoch": 0.13126907698189996,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019907077979573713,
+      "loss": 4.722,
+      "step": 1266
+    },
+    {
+      "epoch": 0.13137276503638803,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001990693018570725,
+      "loss": 4.734,
+      "step": 1267
+    },
+    {
+      "epoch": 0.1314764530908761,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019906782274949482,
+      "loss": 4.7565,
+      "step": 1268
+    },
+    {
+      "epoch": 0.13158014114536418,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001990663424730216,
+      "loss": 4.7149,
+      "step": 1269
+    },
+    {
+      "epoch": 0.13168382919985225,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001990648610276702,
+      "loss": 4.7146,
+      "step": 1270
+    },
+    {
+      "epoch": 0.13178751725434032,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019906337841345818,
+      "loss": 4.7258,
+      "step": 1271
+    },
+    {
+      "epoch": 0.1318912053088284,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00019906189463040299,
+      "loss": 4.7316,
+      "step": 1272
+    },
+    {
+      "epoch": 0.13199489336331646,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019906040967852215,
+      "loss": 4.7284,
+      "step": 1273
+    },
+    {
+      "epoch": 0.13209858141780453,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001990589235578332,
+      "loss": 4.7609,
+      "step": 1274
+    },
+    {
+      "epoch": 0.1322022694722926,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019905743626835368,
+      "loss": 4.7654,
+      "step": 1275
+    },
+    {
+      "epoch": 0.13230595752678068,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001990559478101011,
+      "loss": 4.7341,
+      "step": 1276
+    },
+    {
+      "epoch": 0.13240964558126875,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019905445818309305,
+      "loss": 4.7512,
+      "step": 1277
+    },
+    {
+      "epoch": 0.13251333363575682,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019905296738734709,
+      "loss": 4.6994,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1326170216902449,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019905147542288086,
+      "loss": 4.7447,
+      "step": 1279
+    },
+    {
+      "epoch": 0.13272070974473296,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0001990499822897119,
+      "loss": 4.7525,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13282439779922103,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019904848798785781,
+      "loss": 4.7446,
+      "step": 1281
+    },
+    {
+      "epoch": 0.1329280858537091,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019904699251733628,
+      "loss": 4.7214,
+      "step": 1282
+    },
+    {
+      "epoch": 0.13303177390819718,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019904549587816494,
+      "loss": 4.7364,
+      "step": 1283
+    },
+    {
+      "epoch": 0.13313546196268525,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019904399807036145,
+      "loss": 4.7707,
+      "step": 1284
+    },
+    {
+      "epoch": 0.13323915001717335,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019904249909394347,
+      "loss": 4.7137,
+      "step": 1285
+    },
+    {
+      "epoch": 0.13334283807166142,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0001990409989489287,
+      "loss": 4.7243,
+      "step": 1286
+    },
+    {
+      "epoch": 0.1334465261261495,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019903949763533483,
+      "loss": 4.7646,
+      "step": 1287
+    },
+    {
+      "epoch": 0.13355021418063756,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019903799515317956,
+      "loss": 4.6683,
+      "step": 1288
+    },
+    {
+      "epoch": 0.13365390223512563,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019903649150248068,
+      "loss": 4.7398,
+      "step": 1289
+    },
+    {
+      "epoch": 0.1337575902896137,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019903498668325583,
+      "loss": 4.7181,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13386127834410178,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019903348069552285,
+      "loss": 4.7208,
+      "step": 1291
+    },
+    {
+      "epoch": 0.13396496639858985,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001990319735392995,
+      "loss": 4.7317,
+      "step": 1292
+    },
+    {
+      "epoch": 0.13406865445307792,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019903046521460352,
+      "loss": 4.7149,
+      "step": 1293
+    },
+    {
+      "epoch": 0.134172342507566,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019902895572145274,
+      "loss": 4.6968,
+      "step": 1294
+    },
+    {
+      "epoch": 0.13427603056205406,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019902744505986494,
+      "loss": 4.7335,
+      "step": 1295
+    },
+    {
+      "epoch": 0.13437971861654213,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019902593322985797,
+      "loss": 4.6848,
+      "step": 1296
+    },
+    {
+      "epoch": 0.1344834066710302,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001990244202314497,
+      "loss": 4.7019,
+      "step": 1297
+    },
+    {
+      "epoch": 0.13458709472551827,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001990229060646579,
+      "loss": 4.7149,
+      "step": 1298
+    },
+    {
+      "epoch": 0.13469078278000635,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0001990213907295005,
+      "loss": 4.7474,
+      "step": 1299
+    },
+    {
+      "epoch": 0.13479447083449442,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019901987422599535,
+      "loss": 4.694,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1348981588889825,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019901835655416038,
+      "loss": 4.677,
+      "step": 1301
+    },
+    {
+      "epoch": 0.13500184694347056,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019901683771401344,
+      "loss": 4.7183,
+      "step": 1302
+    },
+    {
+      "epoch": 0.13510553499795863,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001990153177055725,
+      "loss": 4.7433,
+      "step": 1303
+    },
+    {
+      "epoch": 0.1352092230524467,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019901379652885543,
+      "loss": 4.7238,
+      "step": 1304
+    },
+    {
+      "epoch": 0.13531291110693477,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019901227418388028,
+      "loss": 4.7238,
+      "step": 1305
+    },
+    {
+      "epoch": 0.13541659916142285,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019901075067066493,
+      "loss": 4.7488,
+      "step": 1306
+    },
+    {
+      "epoch": 0.13552028721591095,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019900922598922738,
+      "loss": 4.7578,
+      "step": 1307
+    },
+    {
+      "epoch": 0.13562397527039902,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019900770013958562,
+      "loss": 4.7248,
+      "step": 1308
+    },
+    {
+      "epoch": 0.1357276633248871,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019900617312175768,
+      "loss": 4.6949,
+      "step": 1309
+    },
+    {
+      "epoch": 0.13583135137937516,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019900464493576153,
+      "loss": 4.7147,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13593503943386323,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019900311558161522,
+      "loss": 4.7156,
+      "step": 1311
+    },
+    {
+      "epoch": 0.1360387274883513,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019900158505933678,
+      "loss": 4.7342,
+      "step": 1312
+    },
+    {
+      "epoch": 0.13614241554283937,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0001990000533689443,
+      "loss": 4.7456,
+      "step": 1313
+    },
+    {
+      "epoch": 0.13624610359732744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019899852051045587,
+      "loss": 4.7326,
+      "step": 1314
+    },
+    {
+      "epoch": 0.13634979165181552,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019899698648388948,
+      "loss": 4.7254,
+      "step": 1315
+    },
+    {
+      "epoch": 0.1364534797063036,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019899545128926333,
+      "loss": 4.7298,
+      "step": 1316
+    },
+    {
+      "epoch": 0.13655716776079166,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019899391492659551,
+      "loss": 4.7243,
+      "step": 1317
+    },
+    {
+      "epoch": 0.13666085581527973,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019899237739590413,
+      "loss": 4.723,
+      "step": 1318
+    },
+    {
+      "epoch": 0.1367645438697678,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019899083869720735,
+      "loss": 4.7191,
+      "step": 1319
+    },
+    {
+      "epoch": 0.13686823192425587,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019898929883052326,
+      "loss": 4.6991,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13697191997874394,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0001989877577958701,
+      "loss": 4.6526,
+      "step": 1321
+    },
+    {
+      "epoch": 0.13707560803323202,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019898621559326607,
+      "loss": 4.6876,
+      "step": 1322
+    },
+    {
+      "epoch": 0.1371792960877201,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019898467222272928,
+      "loss": 4.6841,
+      "step": 1323
+    },
+    {
+      "epoch": 0.13728298414220816,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00019898312768427802,
+      "loss": 4.677,
+      "step": 1324
+    },
+    {
+      "epoch": 0.13738667219669623,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00019898158197793046,
+      "loss": 4.7136,
+      "step": 1325
+    },
+    {
+      "epoch": 0.1374903602511843,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019898003510370488,
+      "loss": 4.6611,
+      "step": 1326
+    },
+    {
+      "epoch": 0.13759404830567237,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0001989784870616195,
+      "loss": 4.6735,
+      "step": 1327
+    },
+    {
+      "epoch": 0.13769773636016044,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00019897693785169261,
+      "loss": 4.7221,
+      "step": 1328
+    },
+    {
+      "epoch": 0.13780142441464854,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019897538747394247,
+      "loss": 4.6936,
+      "step": 1329
+    },
+    {
+      "epoch": 0.13790511246913661,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019897383592838738,
+      "loss": 4.7377,
+      "step": 1330
+    },
+    {
+      "epoch": 0.1380088005236247,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019897228321504563,
+      "loss": 4.7499,
+      "step": 1331
+    },
+    {
+      "epoch": 0.13811248857811276,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00019897072933393559,
+      "loss": 4.705,
+      "step": 1332
+    },
+    {
+      "epoch": 0.13821617663260083,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00019896917428507553,
+      "loss": 4.719,
+      "step": 1333
+    },
+    {
+      "epoch": 0.1383198646870889,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00019896761806848385,
+      "loss": 4.7109,
+      "step": 1334
+    },
+    {
+      "epoch": 0.13842355274157697,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00019896606068417887,
+      "loss": 4.7314,
+      "step": 1335
+    },
+    {
+      "epoch": 0.13852724079606504,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.000198964502132179,
+      "loss": 4.7162,
+      "step": 1336
+    },
+    {
+      "epoch": 0.13863092885055311,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019896294241250262,
+      "loss": 4.7474,
+      "step": 1337
+    },
+    {
+      "epoch": 0.13873461690504119,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019896138152516812,
+      "loss": 4.6801,
+      "step": 1338
+    },
+    {
+      "epoch": 0.13883830495952926,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00019895981947019392,
+      "loss": 4.6727,
+      "step": 1339
+    },
+    {
+      "epoch": 0.13894199301401733,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00019895825624759845,
+      "loss": 4.7344,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1390456810685054,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00019895669185740017,
+      "loss": 4.737,
+      "step": 1341
+    },
+    {
+      "epoch": 0.13914936912299347,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00019895512629961753,
+      "loss": 4.6886,
+      "step": 1342
+    },
+    {
+      "epoch": 0.13925305717748154,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.000198953559574269,
+      "loss": 4.658,
+      "step": 1343
+    },
+    {
+      "epoch": 0.13935674523196961,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019895199168137306,
+      "loss": 4.6986,
+      "step": 1344
+    },
+    {
+      "epoch": 0.13946043328645769,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001989504226209482,
+      "loss": 4.7303,
+      "step": 1345
+    },
+    {
+      "epoch": 0.13956412134094576,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00019894885239301298,
+      "loss": 4.6842,
+      "step": 1346
+    },
+    {
+      "epoch": 0.13966780939543383,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0001989472809975859,
+      "loss": 4.6811,
+      "step": 1347
+    },
+    {
+      "epoch": 0.1397714974499219,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019894570843468544,
+      "loss": 4.713,
+      "step": 1348
+    },
+    {
+      "epoch": 0.13987518550440997,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00019894413470433026,
+      "loss": 4.6777,
+      "step": 1349
+    },
+    {
+      "epoch": 0.13997887355889804,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00019894255980653887,
+      "loss": 4.7175,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14008256161338611,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00019894098374132987,
+      "loss": 4.7008,
+      "step": 1351
+    },
+    {
+      "epoch": 0.1401862496678742,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0001989394065087218,
+      "loss": 4.76,
+      "step": 1352
+    },
+    {
+      "epoch": 0.14028993772236228,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019893782810873338,
+      "loss": 4.7276,
+      "step": 1353
+    },
+    {
+      "epoch": 0.14039362577685036,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00019893624854138312,
+      "loss": 4.7048,
+      "step": 1354
+    },
+    {
+      "epoch": 0.14049731383133843,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00019893466780668972,
+      "loss": 4.6713,
+      "step": 1355
+    },
+    {
+      "epoch": 0.1406010018858265,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00019893308590467185,
+      "loss": 4.7421,
+      "step": 1356
+    },
+    {
+      "epoch": 0.14070468994031457,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0001989315028353481,
+      "loss": 4.7306,
+      "step": 1357
+    },
+    {
+      "epoch": 0.14080837799480264,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00019892991859873723,
+      "loss": 4.7135,
+      "step": 1358
+    },
+    {
+      "epoch": 0.1409120660492907,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00019892833319485787,
+      "loss": 4.7376,
+      "step": 1359
+    },
+    {
+      "epoch": 0.14101575410377878,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00019892674662372876,
+      "loss": 4.7339,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14111944215826686,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.0001989251588853686,
+      "loss": 4.717,
+      "step": 1361
+    },
+    {
+      "epoch": 0.14122313021275493,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00019892356997979613,
+      "loss": 4.7236,
+      "step": 1362
+    },
+    {
+      "epoch": 0.141326818267243,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.0001989219799070301,
+      "loss": 4.6997,
+      "step": 1363
+    },
+    {
+      "epoch": 0.14143050632173107,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019892038866708932,
+      "loss": 4.6979,
+      "step": 1364
+    },
+    {
+      "epoch": 0.14153419437621914,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00019891879625999245,
+      "loss": 4.6499,
+      "step": 1365
+    },
+    {
+      "epoch": 0.1416378824307072,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00019891720268575837,
+      "loss": 4.7374,
+      "step": 1366
+    },
+    {
+      "epoch": 0.14174157048519528,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00019891560794440587,
+      "loss": 4.6959,
+      "step": 1367
+    },
+    {
+      "epoch": 0.14184525853968336,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00019891401203595374,
+      "loss": 4.7039,
+      "step": 1368
+    },
+    {
+      "epoch": 0.14194894659417143,
+      "grad_norm": 2.609375,
+      "learning_rate": 0.00019891241496042082,
+      "loss": 4.7391,
+      "step": 1369
+    },
+    {
+      "epoch": 0.1420526346486595,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.000198910816717826,
+      "loss": 4.7635,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14215632270314757,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00019890921730818806,
+      "loss": 4.7121,
+      "step": 1371
+    },
+    {
+      "epoch": 0.14226001075763564,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00019890761673152591,
+      "loss": 4.7292,
+      "step": 1372
+    },
+    {
+      "epoch": 0.1423636988121237,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00019890601498785844,
+      "loss": 4.71,
+      "step": 1373
+    },
+    {
+      "epoch": 0.1424673868666118,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019890441207720454,
+      "loss": 4.7202,
+      "step": 1374
+    },
+    {
+      "epoch": 0.14257107492109988,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001989028079995831,
+      "loss": 4.7506,
+      "step": 1375
+    },
+    {
+      "epoch": 0.14267476297558795,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00019890120275501308,
+      "loss": 4.684,
+      "step": 1376
+    },
+    {
+      "epoch": 0.14277845103007603,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019889959634351344,
+      "loss": 4.736,
+      "step": 1377
+    },
+    {
+      "epoch": 0.1428821390845641,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.0001988979887651031,
+      "loss": 4.7732,
+      "step": 1378
+    },
+    {
+      "epoch": 0.14298582713905217,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00019889638001980103,
+      "loss": 4.7442,
+      "step": 1379
+    },
+    {
+      "epoch": 0.14308951519354024,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00019889477010762618,
+      "loss": 4.6948,
+      "step": 1380
+    },
+    {
+      "epoch": 0.1431932032480283,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00019889315902859762,
+      "loss": 4.7068,
+      "step": 1381
+    },
+    {
+      "epoch": 0.14329689130251638,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.0001988915467827343,
+      "loss": 4.7583,
+      "step": 1382
+    },
+    {
+      "epoch": 0.14340057935700445,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019888993337005526,
+      "loss": 4.7032,
+      "step": 1383
+    },
+    {
+      "epoch": 0.14350426741149253,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00019888831879057953,
+      "loss": 4.7053,
+      "step": 1384
+    },
+    {
+      "epoch": 0.1436079554659806,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00019888670304432619,
+      "loss": 4.6836,
+      "step": 1385
+    },
+    {
+      "epoch": 0.14371164352046867,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.00019888508613131426,
+      "loss": 4.7571,
+      "step": 1386
+    },
+    {
+      "epoch": 0.14381533157495674,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00019888346805156283,
+      "loss": 4.6962,
+      "step": 1387
+    },
+    {
+      "epoch": 0.1439190196294448,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00019888184880509103,
+      "loss": 4.7161,
+      "step": 1388
+    },
+    {
+      "epoch": 0.14402270768393288,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00019888022839191792,
+      "loss": 4.704,
+      "step": 1389
+    },
+    {
+      "epoch": 0.14412639573842095,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00019887860681206266,
+      "loss": 4.7299,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14423008379290903,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00019887698406554431,
+      "loss": 4.7452,
+      "step": 1391
+    },
+    {
+      "epoch": 0.1443337718473971,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00019887536015238212,
+      "loss": 4.7157,
+      "step": 1392
+    },
+    {
+      "epoch": 0.14443745990188517,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00019887373507259518,
+      "loss": 4.6886,
+      "step": 1393
+    },
+    {
+      "epoch": 0.14454114795637324,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019887210882620266,
+      "loss": 4.7412,
+      "step": 1394
+    },
+    {
+      "epoch": 0.1446448360108613,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00019887048141322376,
+      "loss": 4.7263,
+      "step": 1395
+    },
+    {
+      "epoch": 0.1447485240653494,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00019886885283367772,
+      "loss": 4.7014,
+      "step": 1396
+    },
+    {
+      "epoch": 0.14485221211983748,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019886722308758373,
+      "loss": 4.6781,
+      "step": 1397
+    },
+    {
+      "epoch": 0.14495590017432555,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00019886559217496098,
+      "loss": 4.6926,
+      "step": 1398
+    },
+    {
+      "epoch": 0.14505958822881362,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00019886396009582876,
+      "loss": 4.7069,
+      "step": 1399
+    },
+    {
+      "epoch": 0.1451632762833017,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00019886232685020633,
+      "loss": 4.6784,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14526696433778977,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00019886069243811293,
+      "loss": 4.6576,
+      "step": 1401
+    },
+    {
+      "epoch": 0.14537065239227784,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0001988590568595679,
+      "loss": 4.6882,
+      "step": 1402
+    },
+    {
+      "epoch": 0.1454743404467659,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00019885742011459045,
+      "loss": 4.7041,
+      "step": 1403
+    },
+    {
+      "epoch": 0.14557802850125398,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019885578220319995,
+      "loss": 4.714,
+      "step": 1404
+    },
+    {
+      "epoch": 0.14568171655574205,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019885414312541573,
+      "loss": 4.6943,
+      "step": 1405
+    },
+    {
+      "epoch": 0.14578540461023012,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019885250288125713,
+      "loss": 4.6866,
+      "step": 1406
+    },
+    {
+      "epoch": 0.1458890926647182,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019885086147074344,
+      "loss": 4.6937,
+      "step": 1407
+    },
+    {
+      "epoch": 0.14599278071920627,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0001988492188938941,
+      "loss": 4.7627,
+      "step": 1408
+    },
+    {
+      "epoch": 0.14609646877369434,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019884757515072844,
+      "loss": 4.7113,
+      "step": 1409
+    },
+    {
+      "epoch": 0.1462001568281824,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019884593024126592,
+      "loss": 4.6828,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14630384488267048,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001988442841655259,
+      "loss": 4.7033,
+      "step": 1411
+    },
+    {
+      "epoch": 0.14640753293715855,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019884263692352777,
+      "loss": 4.7208,
+      "step": 1412
+    },
+    {
+      "epoch": 0.14651122099164662,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019884098851529104,
+      "loss": 4.7189,
+      "step": 1413
+    },
+    {
+      "epoch": 0.1466149090461347,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019883933894083514,
+      "loss": 4.6782,
+      "step": 1414
+    },
+    {
+      "epoch": 0.14671859710062277,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019883768820017948,
+      "loss": 4.7375,
+      "step": 1415
+    },
+    {
+      "epoch": 0.14682228515511084,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001988360362933436,
+      "loss": 4.7187,
+      "step": 1416
+    },
+    {
+      "epoch": 0.1469259732095989,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019883438322034695,
+      "loss": 4.7011,
+      "step": 1417
+    },
+    {
+      "epoch": 0.14702966126408698,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019883272898120905,
+      "loss": 4.7445,
+      "step": 1418
+    },
+    {
+      "epoch": 0.14713334931857508,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019883107357594943,
+      "loss": 4.724,
+      "step": 1419
+    },
+    {
+      "epoch": 0.14723703737306315,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0001988294170045876,
+      "loss": 4.6845,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14734072542755122,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019882775926714313,
+      "loss": 4.7319,
+      "step": 1421
+    },
+    {
+      "epoch": 0.1474444134820393,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019882610036363557,
+      "loss": 4.71,
+      "step": 1422
+    },
+    {
+      "epoch": 0.14754810153652737,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019882444029408448,
+      "loss": 4.7078,
+      "step": 1423
+    },
+    {
+      "epoch": 0.14765178959101544,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019882277905850946,
+      "loss": 4.6876,
+      "step": 1424
+    },
+    {
+      "epoch": 0.1477554776455035,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019882111665693011,
+      "loss": 4.6975,
+      "step": 1425
+    },
+    {
+      "epoch": 0.14785916569999158,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019881945308936603,
+      "loss": 4.6452,
+      "step": 1426
+    },
+    {
+      "epoch": 0.14796285375447965,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019881778835583686,
+      "loss": 4.7131,
+      "step": 1427
+    },
+    {
+      "epoch": 0.14806654180896772,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019881612245636226,
+      "loss": 4.681,
+      "step": 1428
+    },
+    {
+      "epoch": 0.1481702298634558,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019881445539096185,
+      "loss": 4.7163,
+      "step": 1429
+    },
+    {
+      "epoch": 0.14827391791794386,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019881278715965534,
+      "loss": 4.7312,
+      "step": 1430
+    },
+    {
+      "epoch": 0.14837760597243194,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019881111776246234,
+      "loss": 4.7217,
+      "step": 1431
+    },
+    {
+      "epoch": 0.14848129402692,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019880944719940263,
+      "loss": 4.7345,
+      "step": 1432
+    },
+    {
+      "epoch": 0.14858498208140808,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0001988077754704959,
+      "loss": 4.6661,
+      "step": 1433
+    },
+    {
+      "epoch": 0.14868867013589615,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001988061025757619,
+      "loss": 4.662,
+      "step": 1434
+    },
+    {
+      "epoch": 0.14879235819038422,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00019880442851522029,
+      "loss": 4.7057,
+      "step": 1435
+    },
+    {
+      "epoch": 0.1488960462448723,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00019880275328889083,
+      "loss": 4.6554,
+      "step": 1436
+    },
+    {
+      "epoch": 0.14899973429936036,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019880107689679337,
+      "loss": 4.673,
+      "step": 1437
+    },
+    {
+      "epoch": 0.14910342235384844,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019879939933894762,
+      "loss": 4.6325,
+      "step": 1438
+    },
+    {
+      "epoch": 0.1492071104083365,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0001987977206153734,
+      "loss": 4.7093,
+      "step": 1439
+    },
+    {
+      "epoch": 0.14931079846282458,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0001987960407260905,
+      "loss": 4.6711,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14941448651731268,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019879435967111876,
+      "loss": 4.6034,
+      "step": 1441
+    },
+    {
+      "epoch": 0.14951817457180075,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.000198792677450478,
+      "loss": 4.713,
+      "step": 1442
+    },
+    {
+      "epoch": 0.14962186262628882,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00019879099406418807,
+      "loss": 4.7042,
+      "step": 1443
+    },
+    {
+      "epoch": 0.1497255506807769,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019878930951226887,
+      "loss": 4.6731,
+      "step": 1444
+    },
+    {
+      "epoch": 0.14982923873526496,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019878762379474022,
+      "loss": 4.6662,
+      "step": 1445
+    },
+    {
+      "epoch": 0.14993292678975303,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019878593691162203,
+      "loss": 4.6774,
+      "step": 1446
+    },
+    {
+      "epoch": 0.1500366148442411,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00019878424886293422,
+      "loss": 4.6983,
+      "step": 1447
+    },
+    {
+      "epoch": 0.15014030289872918,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019878255964869666,
+      "loss": 4.6673,
+      "step": 1448
+    },
+    {
+      "epoch": 0.15024399095321725,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019878086926892934,
+      "loss": 4.6991,
+      "step": 1449
+    },
+    {
+      "epoch": 0.15034767900770532,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019877917772365215,
+      "loss": 4.7173,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1504513670621934,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0001987774850128851,
+      "loss": 4.7385,
+      "step": 1451
+    },
+    {
+      "epoch": 0.15055505511668146,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019877579113664816,
+      "loss": 4.7129,
+      "step": 1452
+    },
+    {
+      "epoch": 0.15065874317116953,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019877409609496126,
+      "loss": 4.6592,
+      "step": 1453
+    },
+    {
+      "epoch": 0.1507624312256576,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019877239988784444,
+      "loss": 4.7102,
+      "step": 1454
+    },
+    {
+      "epoch": 0.15086611928014568,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019877070251531772,
+      "loss": 4.7239,
+      "step": 1455
+    },
+    {
+      "epoch": 0.15096980733463375,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0001987690039774011,
+      "loss": 4.6529,
+      "step": 1456
+    },
+    {
+      "epoch": 0.15107349538912182,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00019876730427411467,
+      "loss": 4.7155,
+      "step": 1457
+    },
+    {
+      "epoch": 0.1511771834436099,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00019876560340547844,
+      "loss": 4.7025,
+      "step": 1458
+    },
+    {
+      "epoch": 0.15128087149809796,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019876390137151247,
+      "loss": 4.6933,
+      "step": 1459
+    },
+    {
+      "epoch": 0.15138455955258603,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019876219817223687,
+      "loss": 4.6941,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1514882476070741,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00019876049380767173,
+      "loss": 4.6765,
+      "step": 1461
+    },
+    {
+      "epoch": 0.15159193566156218,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00019875878827783713,
+      "loss": 4.6645,
+      "step": 1462
+    },
+    {
+      "epoch": 0.15169562371605028,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00019875708158275326,
+      "loss": 4.6656,
+      "step": 1463
+    },
+    {
+      "epoch": 0.15179931177053835,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019875537372244023,
+      "loss": 4.6743,
+      "step": 1464
+    },
+    {
+      "epoch": 0.15190299982502642,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019875366469691814,
+      "loss": 4.7085,
+      "step": 1465
+    },
+    {
+      "epoch": 0.1520066878795145,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0001987519545062072,
+      "loss": 4.7214,
+      "step": 1466
+    },
+    {
+      "epoch": 0.15211037593400256,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0001987502431503276,
+      "loss": 4.6776,
+      "step": 1467
+    },
+    {
+      "epoch": 0.15221406398849063,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0001987485306292995,
+      "loss": 4.7105,
+      "step": 1468
+    },
+    {
+      "epoch": 0.1523177520429787,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0001987468169431431,
+      "loss": 4.685,
+      "step": 1469
+    },
+    {
+      "epoch": 0.15242144009746678,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001987451020918787,
+      "loss": 4.6538,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15252512815195485,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019874338607552642,
+      "loss": 4.6843,
+      "step": 1471
+    },
+    {
+      "epoch": 0.15262881620644292,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019874166889410658,
+      "loss": 4.6666,
+      "step": 1472
+    },
+    {
+      "epoch": 0.152732504260931,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0001987399505476394,
+      "loss": 4.6916,
+      "step": 1473
+    },
+    {
+      "epoch": 0.15283619231541906,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001987382310361452,
+      "loss": 4.6385,
+      "step": 1474
+    },
+    {
+      "epoch": 0.15293988036990713,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019873651035964425,
+      "loss": 4.6691,
+      "step": 1475
+    },
+    {
+      "epoch": 0.1530435684243952,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019873478851815683,
+      "loss": 4.6766,
+      "step": 1476
+    },
+    {
+      "epoch": 0.15314725647888328,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019873306551170328,
+      "loss": 4.6835,
+      "step": 1477
+    },
+    {
+      "epoch": 0.15325094453337135,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019873134134030393,
+      "loss": 4.6331,
+      "step": 1478
+    },
+    {
+      "epoch": 0.15335463258785942,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001987296160039791,
+      "loss": 4.6557,
+      "step": 1479
+    },
+    {
+      "epoch": 0.1534583206423475,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019872788950274918,
+      "loss": 4.6338,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15356200869683556,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019872616183663451,
+      "loss": 4.6594,
+      "step": 1481
+    },
+    {
+      "epoch": 0.15366569675132363,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019872443300565548,
+      "loss": 4.6934,
+      "step": 1482
+    },
+    {
+      "epoch": 0.1537693848058117,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001987227030098325,
+      "loss": 4.664,
+      "step": 1483
+    },
+    {
+      "epoch": 0.15387307286029978,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.000198720971849186,
+      "loss": 4.6917,
+      "step": 1484
+    },
+    {
+      "epoch": 0.15397676091478785,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019871923952373635,
+      "loss": 4.6441,
+      "step": 1485
+    },
+    {
+      "epoch": 0.15408044896927595,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019871750603350408,
+      "loss": 4.6871,
+      "step": 1486
+    },
+    {
+      "epoch": 0.15418413702376402,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019871577137850954,
+      "loss": 4.6767,
+      "step": 1487
+    },
+    {
+      "epoch": 0.1542878250782521,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019871403555877327,
+      "loss": 4.6942,
+      "step": 1488
+    },
+    {
+      "epoch": 0.15439151313274016,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019871229857431572,
+      "loss": 4.6879,
+      "step": 1489
+    },
+    {
+      "epoch": 0.15449520118722823,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001987105604251574,
+      "loss": 4.6799,
+      "step": 1490
+    },
+    {
+      "epoch": 0.1545988892417163,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001987088211113188,
+      "loss": 4.6938,
+      "step": 1491
+    },
+    {
+      "epoch": 0.15470257729620437,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019870708063282044,
+      "loss": 4.6613,
+      "step": 1492
+    },
+    {
+      "epoch": 0.15480626535069245,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019870533898968287,
+      "loss": 4.6938,
+      "step": 1493
+    },
+    {
+      "epoch": 0.15490995340518052,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019870359618192663,
+      "loss": 4.6946,
+      "step": 1494
+    },
+    {
+      "epoch": 0.1550136414596686,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001987018522095723,
+      "loss": 4.6413,
+      "step": 1495
+    },
+    {
+      "epoch": 0.15511732951415666,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019870010707264045,
+      "loss": 4.6721,
+      "step": 1496
+    },
+    {
+      "epoch": 0.15522101756864473,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00019869836077115164,
+      "loss": 4.7004,
+      "step": 1497
+    },
+    {
+      "epoch": 0.1553247056231328,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019869661330512654,
+      "loss": 4.6585,
+      "step": 1498
+    },
+    {
+      "epoch": 0.15542839367762087,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001986948646745857,
+      "loss": 4.6284,
+      "step": 1499
+    },
+    {
+      "epoch": 0.15553208173210895,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.0001986931148795498,
+      "loss": 4.6635,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15563576978659702,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019869136392003945,
+      "loss": 4.6898,
+      "step": 1501
+    },
+    {
+      "epoch": 0.1557394578410851,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019868961179607536,
+      "loss": 4.7494,
+      "step": 1502
+    },
+    {
+      "epoch": 0.15584314589557316,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019868785850767813,
+      "loss": 4.6805,
+      "step": 1503
+    },
+    {
+      "epoch": 0.15594683395006123,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0001986861040548685,
+      "loss": 4.6183,
+      "step": 1504
+    },
+    {
+      "epoch": 0.1560505220045493,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00019868434843766717,
+      "loss": 4.6631,
+      "step": 1505
+    },
+    {
+      "epoch": 0.15615421005903737,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00019868259165609482,
+      "loss": 4.6455,
+      "step": 1506
+    },
+    {
+      "epoch": 0.15625789811352545,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00019868083371017223,
+      "loss": 4.6403,
+      "step": 1507
+    },
+    {
+      "epoch": 0.15636158616801354,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001986790745999201,
+      "loss": 4.6782,
+      "step": 1508
+    },
+    {
+      "epoch": 0.15646527422250162,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019867731432535922,
+      "loss": 4.6598,
+      "step": 1509
+    },
+    {
+      "epoch": 0.1565689622769897,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001986755528865103,
+      "loss": 4.6547,
+      "step": 1510
+    },
+    {
+      "epoch": 0.15667265033147776,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019867379028339416,
+      "loss": 4.6414,
+      "step": 1511
+    },
+    {
+      "epoch": 0.15677633838596583,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001986720265160316,
+      "loss": 4.6567,
+      "step": 1512
+    },
+    {
+      "epoch": 0.1568800264404539,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019867026158444344,
+      "loss": 4.6425,
+      "step": 1513
+    },
+    {
+      "epoch": 0.15698371449494197,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0001986684954886505,
+      "loss": 4.6946,
+      "step": 1514
+    },
+    {
+      "epoch": 0.15708740254943004,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001986667282286736,
+      "loss": 4.693,
+      "step": 1515
+    },
+    {
+      "epoch": 0.15719109060391812,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0001986649598045336,
+      "loss": 4.673,
+      "step": 1516
+    },
+    {
+      "epoch": 0.1572947786584062,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019866319021625138,
+      "loss": 4.6624,
+      "step": 1517
+    },
+    {
+      "epoch": 0.15739846671289426,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0001986614194638478,
+      "loss": 4.6387,
+      "step": 1518
+    },
+    {
+      "epoch": 0.15750215476738233,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00019865964754734377,
+      "loss": 4.6426,
+      "step": 1519
+    },
+    {
+      "epoch": 0.1576058428218704,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019865787446676016,
+      "loss": 4.6452,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15770953087635847,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019865610022211795,
+      "loss": 4.6182,
+      "step": 1521
+    },
+    {
+      "epoch": 0.15781321893084654,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.000198654324813438,
+      "loss": 4.6999,
+      "step": 1522
+    },
+    {
+      "epoch": 0.15791690698533462,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.00019865254824074136,
+      "loss": 4.6654,
+      "step": 1523
+    },
+    {
+      "epoch": 0.1580205950398227,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0001986507705040489,
+      "loss": 4.7078,
+      "step": 1524
+    },
+    {
+      "epoch": 0.15812428309431076,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001986489916033816,
+      "loss": 4.7079,
+      "step": 1525
+    },
+    {
+      "epoch": 0.15822797114879883,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0001986472115387605,
+      "loss": 4.6934,
+      "step": 1526
+    },
+    {
+      "epoch": 0.1583316592032869,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00019864543031020658,
+      "loss": 4.6808,
+      "step": 1527
+    },
+    {
+      "epoch": 0.15843534725777497,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019864364791774084,
+      "loss": 4.6659,
+      "step": 1528
+    },
+    {
+      "epoch": 0.15853903531226304,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019864186436138433,
+      "loss": 4.6941,
+      "step": 1529
+    },
+    {
+      "epoch": 0.15864272336675114,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0001986400796411581,
+      "loss": 4.682,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15874641142123921,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0001986382937570832,
+      "loss": 4.6594,
+      "step": 1531
+    },
+    {
+      "epoch": 0.15885009947572729,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0001986365067091807,
+      "loss": 4.681,
+      "step": 1532
+    },
+    {
+      "epoch": 0.15895378753021536,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019863471849747167,
+      "loss": 4.7219,
+      "step": 1533
+    },
+    {
+      "epoch": 0.15905747558470343,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0001986329291219772,
+      "loss": 4.6721,
+      "step": 1534
+    },
+    {
+      "epoch": 0.1591611636391915,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019863113858271846,
+      "loss": 4.6567,
+      "step": 1535
+    },
+    {
+      "epoch": 0.15926485169367957,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019862934687971655,
+      "loss": 4.6987,
+      "step": 1536
+    },
+    {
+      "epoch": 0.15936853974816764,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.00019862755401299257,
+      "loss": 4.6722,
+      "step": 1537
+    },
+    {
+      "epoch": 0.1594722278026557,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.00019862575998256773,
+      "loss": 4.6555,
+      "step": 1538
+    },
+    {
+      "epoch": 0.15957591585714379,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00019862396478846316,
+      "loss": 4.6425,
+      "step": 1539
+    },
+    {
+      "epoch": 0.15967960391163186,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0001986221684307001,
+      "loss": 4.6574,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15978329196611993,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019862037090929966,
+      "loss": 4.6743,
+      "step": 1541
+    },
+    {
+      "epoch": 0.159886980020608,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00019861857222428308,
+      "loss": 4.6539,
+      "step": 1542
+    },
+    {
+      "epoch": 0.15999066807509607,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019861677237567162,
+      "loss": 4.6581,
+      "step": 1543
+    },
+    {
+      "epoch": 0.16009435612958414,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00019861497136348648,
+      "loss": 4.6538,
+      "step": 1544
+    },
+    {
+      "epoch": 0.1601980441840722,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0001986131691877489,
+      "loss": 4.6637,
+      "step": 1545
+    },
+    {
+      "epoch": 0.16030173223856028,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00019861136584848019,
+      "loss": 4.6621,
+      "step": 1546
+    },
+    {
+      "epoch": 0.16040542029304836,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001986095613457016,
+      "loss": 4.6788,
+      "step": 1547
+    },
+    {
+      "epoch": 0.16050910834753643,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0001986077556794344,
+      "loss": 4.6588,
+      "step": 1548
+    },
+    {
+      "epoch": 0.1606127964020245,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00019860594884969993,
+      "loss": 4.6647,
+      "step": 1549
+    },
+    {
+      "epoch": 0.16071648445651257,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001986041408565195,
+      "loss": 4.6423,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16082017251100064,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.0001986023316999144,
+      "loss": 4.6748,
+      "step": 1551
+    },
+    {
+      "epoch": 0.1609238605654887,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019860052137990605,
+      "loss": 4.6763,
+      "step": 1552
+    },
+    {
+      "epoch": 0.1610275486199768,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00019859870989651576,
+      "loss": 4.6478,
+      "step": 1553
+    },
+    {
+      "epoch": 0.16113123667446488,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00019859689724976488,
+      "loss": 4.6574,
+      "step": 1554
+    },
+    {
+      "epoch": 0.16123492472895296,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.0001985950834396749,
+      "loss": 4.6907,
+      "step": 1555
+    },
+    {
+      "epoch": 0.16133861278344103,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00019859326846626712,
+      "loss": 4.6621,
+      "step": 1556
+    },
+    {
+      "epoch": 0.1614423008379291,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.000198591452329563,
+      "loss": 4.6589,
+      "step": 1557
+    },
+    {
+      "epoch": 0.16154598889241717,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019858963502958395,
+      "loss": 4.7071,
+      "step": 1558
+    },
+    {
+      "epoch": 0.16164967694690524,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019858781656635142,
+      "loss": 4.6513,
+      "step": 1559
+    },
+    {
+      "epoch": 0.1617533650013933,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00019858599693988688,
+      "loss": 4.6913,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16185705305588138,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00019858417615021176,
+      "loss": 4.6505,
+      "step": 1561
+    },
+    {
+      "epoch": 0.16196074111036945,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019858235419734758,
+      "loss": 4.6975,
+      "step": 1562
+    },
+    {
+      "epoch": 0.16206442916485753,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00019858053108131585,
+      "loss": 4.6967,
+      "step": 1563
+    },
+    {
+      "epoch": 0.1621681172193456,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019857870680213804,
+      "loss": 4.6619,
+      "step": 1564
+    },
+    {
+      "epoch": 0.16227180527383367,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0001985768813598357,
+      "loss": 4.6586,
+      "step": 1565
+    },
+    {
+      "epoch": 0.16237549332832174,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.00019857505475443033,
+      "loss": 4.6451,
+      "step": 1566
+    },
+    {
+      "epoch": 0.1624791813828098,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00019857322698594353,
+      "loss": 4.6634,
+      "step": 1567
+    },
+    {
+      "epoch": 0.16258286943729788,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.00019857139805439688,
+      "loss": 4.6611,
+      "step": 1568
+    },
+    {
+      "epoch": 0.16268655749178595,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0001985695679598119,
+      "loss": 4.6922,
+      "step": 1569
+    },
+    {
+      "epoch": 0.16279024554627403,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0001985677367022102,
+      "loss": 4.5804,
+      "step": 1570
+    },
+    {
+      "epoch": 0.1628939336007621,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00019856590428161342,
+      "loss": 4.6266,
+      "step": 1571
+    },
+    {
+      "epoch": 0.16299762165525017,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.00019856407069804316,
+      "loss": 4.6377,
+      "step": 1572
+    },
+    {
+      "epoch": 0.16310130970973824,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00019856223595152104,
+      "loss": 4.6668,
+      "step": 1573
+    },
+    {
+      "epoch": 0.1632049977642263,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0001985604000420687,
+      "loss": 4.6548,
+      "step": 1574
+    },
+    {
+      "epoch": 0.1633086858187144,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019855856296970784,
+      "loss": 4.7052,
+      "step": 1575
+    },
+    {
+      "epoch": 0.16341237387320248,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019855672473446012,
+      "loss": 4.6898,
+      "step": 1576
+    },
+    {
+      "epoch": 0.16351606192769055,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019855488533634724,
+      "loss": 4.6651,
+      "step": 1577
+    },
+    {
+      "epoch": 0.16361974998217862,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019855304477539085,
+      "loss": 4.6913,
+      "step": 1578
+    },
+    {
+      "epoch": 0.1637234380366667,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019855120305161273,
+      "loss": 4.6939,
+      "step": 1579
+    },
+    {
+      "epoch": 0.16382712609115477,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0001985493601650346,
+      "loss": 4.677,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16393081414564284,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001985475161156782,
+      "loss": 4.6388,
+      "step": 1581
+    },
+    {
+      "epoch": 0.1640345022001309,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019854567090356526,
+      "loss": 4.6185,
+      "step": 1582
+    },
+    {
+      "epoch": 0.16413819025461898,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00019854382452871757,
+      "loss": 4.6091,
+      "step": 1583
+    },
+    {
+      "epoch": 0.16424187830910705,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00019854197699115692,
+      "loss": 4.683,
+      "step": 1584
+    },
+    {
+      "epoch": 0.16434556636359512,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001985401282909051,
+      "loss": 4.6445,
+      "step": 1585
+    },
+    {
+      "epoch": 0.1644492544180832,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00019853827842798393,
+      "loss": 4.6586,
+      "step": 1586
+    },
+    {
+      "epoch": 0.16455294247257127,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019853642740241522,
+      "loss": 4.6508,
+      "step": 1587
+    },
+    {
+      "epoch": 0.16465663052705934,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019853457521422084,
+      "loss": 4.6936,
+      "step": 1588
+    },
+    {
+      "epoch": 0.1647603185815474,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019853272186342262,
+      "loss": 4.639,
+      "step": 1589
+    },
+    {
+      "epoch": 0.16486400663603548,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019853086735004247,
+      "loss": 4.6583,
+      "step": 1590
+    },
+    {
+      "epoch": 0.16496769469052355,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0001985290116741022,
+      "loss": 4.6363,
+      "step": 1591
+    },
+    {
+      "epoch": 0.16507138274501162,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019852715483562375,
+      "loss": 4.6199,
+      "step": 1592
+    },
+    {
+      "epoch": 0.1651750707994997,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019852529683462902,
+      "loss": 4.6458,
+      "step": 1593
+    },
+    {
+      "epoch": 0.16527875885398777,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019852343767113993,
+      "loss": 4.6503,
+      "step": 1594
+    },
+    {
+      "epoch": 0.16538244690847584,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019852157734517843,
+      "loss": 4.6741,
+      "step": 1595
+    },
+    {
+      "epoch": 0.1654861349629639,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019851971585676644,
+      "loss": 4.6378,
+      "step": 1596
+    },
+    {
+      "epoch": 0.165589823017452,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019851785320592596,
+      "loss": 4.6136,
+      "step": 1597
+    },
+    {
+      "epoch": 0.16569351107194008,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019851598939267894,
+      "loss": 4.6617,
+      "step": 1598
+    },
+    {
+      "epoch": 0.16579719912642815,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001985141244170474,
+      "loss": 4.67,
+      "step": 1599
+    },
+    {
+      "epoch": 0.16590088718091622,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001985122582790533,
+      "loss": 4.6461,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1660045752354043,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019851039097871872,
+      "loss": 4.6634,
+      "step": 1601
+    },
+    {
+      "epoch": 0.16610826328989237,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019850852251606562,
+      "loss": 4.6191,
+      "step": 1602
+    },
+    {
+      "epoch": 0.16621195134438044,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0001985066528911161,
+      "loss": 4.6047,
+      "step": 1603
+    },
+    {
+      "epoch": 0.1663156393988685,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00019850478210389218,
+      "loss": 4.6388,
+      "step": 1604
+    },
+    {
+      "epoch": 0.16641932745335658,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019850291015441598,
+      "loss": 4.6735,
+      "step": 1605
+    },
+    {
+      "epoch": 0.16652301550784465,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019850103704270957,
+      "loss": 4.6201,
+      "step": 1606
+    },
+    {
+      "epoch": 0.16662670356233272,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.000198499162768795,
+      "loss": 4.6704,
+      "step": 1607
+    },
+    {
+      "epoch": 0.1667303916168208,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.00019849728733269446,
+      "loss": 4.6382,
+      "step": 1608
+    },
+    {
+      "epoch": 0.16683407967130887,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00019849541073443006,
+      "loss": 4.6594,
+      "step": 1609
+    },
+    {
+      "epoch": 0.16693776772579694,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00019849353297402388,
+      "loss": 4.6754,
+      "step": 1610
+    },
+    {
+      "epoch": 0.167041455780285,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00019849165405149818,
+      "loss": 4.6522,
+      "step": 1611
+    },
+    {
+      "epoch": 0.16714514383477308,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019848977396687504,
+      "loss": 4.636,
+      "step": 1612
+    },
+    {
+      "epoch": 0.16724883188926115,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00019848789272017668,
+      "loss": 4.6058,
+      "step": 1613
+    },
+    {
+      "epoch": 0.16735251994374922,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001984860103114253,
+      "loss": 4.6336,
+      "step": 1614
+    },
+    {
+      "epoch": 0.1674562079982373,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019848412674064306,
+      "loss": 4.6957,
+      "step": 1615
+    },
+    {
+      "epoch": 0.16755989605272537,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019848224200785228,
+      "loss": 4.6593,
+      "step": 1616
+    },
+    {
+      "epoch": 0.16766358410721344,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019848035611307513,
+      "loss": 4.6497,
+      "step": 1617
+    },
+    {
+      "epoch": 0.1677672721617015,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019847846905633385,
+      "loss": 4.598,
+      "step": 1618
+    },
+    {
+      "epoch": 0.16787096021618958,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.00019847658083765076,
+      "loss": 4.6646,
+      "step": 1619
+    },
+    {
+      "epoch": 0.16797464827067768,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001984746914570481,
+      "loss": 4.6261,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16807833632516575,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0001984728009145482,
+      "loss": 4.6637,
+      "step": 1621
+    },
+    {
+      "epoch": 0.16818202437965382,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0001984709092101733,
+      "loss": 4.6613,
+      "step": 1622
+    },
+    {
+      "epoch": 0.1682857124341419,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019846901634394576,
+      "loss": 4.6403,
+      "step": 1623
+    },
+    {
+      "epoch": 0.16838940048862996,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019846712231588796,
+      "loss": 4.647,
+      "step": 1624
+    },
+    {
+      "epoch": 0.16849308854311804,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019846522712602216,
+      "loss": 4.6536,
+      "step": 1625
+    },
+    {
+      "epoch": 0.1685967765976061,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019846333077437077,
+      "loss": 4.6393,
+      "step": 1626
+    },
+    {
+      "epoch": 0.16870046465209418,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019846143326095615,
+      "loss": 4.6717,
+      "step": 1627
+    },
+    {
+      "epoch": 0.16880415270658225,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001984595345858007,
+      "loss": 4.6327,
+      "step": 1628
+    },
+    {
+      "epoch": 0.16890784076107032,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019845763474892681,
+      "loss": 4.6776,
+      "step": 1629
+    },
+    {
+      "epoch": 0.1690115288155584,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019845573375035694,
+      "loss": 4.6468,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16911521687004646,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00019845383159011347,
+      "loss": 4.6192,
+      "step": 1631
+    },
+    {
+      "epoch": 0.16921890492453454,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019845192826821884,
+      "loss": 4.6868,
+      "step": 1632
+    },
+    {
+      "epoch": 0.1693225929790226,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019845002378469554,
+      "loss": 4.6603,
+      "step": 1633
+    },
+    {
+      "epoch": 0.16942628103351068,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.000198448118139566,
+      "loss": 4.6291,
+      "step": 1634
+    },
+    {
+      "epoch": 0.16952996908799875,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019844621133285276,
+      "loss": 4.5767,
+      "step": 1635
+    },
+    {
+      "epoch": 0.16963365714248682,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001984443033645783,
+      "loss": 4.571,
+      "step": 1636
+    },
+    {
+      "epoch": 0.1697373451969749,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019844239423476507,
+      "loss": 4.6399,
+      "step": 1637
+    },
+    {
+      "epoch": 0.16984103325146296,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00019844048394343568,
+      "loss": 4.6592,
+      "step": 1638
+    },
+    {
+      "epoch": 0.16994472130595104,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019843857249061264,
+      "loss": 4.6768,
+      "step": 1639
+    },
+    {
+      "epoch": 0.1700484093604391,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019843665987631849,
+      "loss": 4.671,
+      "step": 1640
+    },
+    {
+      "epoch": 0.17015209741492718,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019843474610057576,
+      "loss": 4.6523,
+      "step": 1641
+    },
+    {
+      "epoch": 0.17025578546941528,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019843283116340713,
+      "loss": 4.6186,
+      "step": 1642
+    },
+    {
+      "epoch": 0.17035947352390335,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019843091506483514,
+      "loss": 4.6413,
+      "step": 1643
+    },
+    {
+      "epoch": 0.17046316157839142,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019842899780488237,
+      "loss": 4.66,
+      "step": 1644
+    },
+    {
+      "epoch": 0.1705668496328795,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0001984270793835715,
+      "loss": 4.6728,
+      "step": 1645
+    },
+    {
+      "epoch": 0.17067053768736756,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019842515980092514,
+      "loss": 4.6588,
+      "step": 1646
+    },
+    {
+      "epoch": 0.17077422574185563,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019842323905696588,
+      "loss": 4.6629,
+      "step": 1647
+    },
+    {
+      "epoch": 0.1708779137963437,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0001984213171517165,
+      "loss": 4.6732,
+      "step": 1648
+    },
+    {
+      "epoch": 0.17098160185083178,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019841939408519958,
+      "loss": 4.6467,
+      "step": 1649
+    },
+    {
+      "epoch": 0.17108528990531985,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019841746985743786,
+      "loss": 4.6683,
+      "step": 1650
+    },
+    {
+      "epoch": 0.17118897795980792,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019841554446845404,
+      "loss": 4.6328,
+      "step": 1651
+    },
+    {
+      "epoch": 0.171292666014296,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0001984136179182708,
+      "loss": 4.6829,
+      "step": 1652
+    },
+    {
+      "epoch": 0.17139635406878406,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001984116902069109,
+      "loss": 4.6603,
+      "step": 1653
+    },
+    {
+      "epoch": 0.17150004212327213,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019840976133439706,
+      "loss": 4.6468,
+      "step": 1654
+    },
+    {
+      "epoch": 0.1716037301777602,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019840783130075206,
+      "loss": 4.6419,
+      "step": 1655
+    },
+    {
+      "epoch": 0.17170741823224828,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0001984059001059987,
+      "loss": 4.5958,
+      "step": 1656
+    },
+    {
+      "epoch": 0.17181110628673635,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00019840396775015976,
+      "loss": 4.6597,
+      "step": 1657
+    },
+    {
+      "epoch": 0.17191479434122442,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.000198402034233258,
+      "loss": 4.6056,
+      "step": 1658
+    },
+    {
+      "epoch": 0.1720184823957125,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00019840009955531622,
+      "loss": 4.6602,
+      "step": 1659
+    },
+    {
+      "epoch": 0.17212217045020056,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001983981637163573,
+      "loss": 4.6517,
+      "step": 1660
+    },
+    {
+      "epoch": 0.17222585850468863,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00019839622671640405,
+      "loss": 4.6581,
+      "step": 1661
+    },
+    {
+      "epoch": 0.1723295465591767,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00019839428855547935,
+      "loss": 4.6584,
+      "step": 1662
+    },
+    {
+      "epoch": 0.17243323461366478,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00019839234923360604,
+      "loss": 4.6582,
+      "step": 1663
+    },
+    {
+      "epoch": 0.17253692266815288,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019839040875080702,
+      "loss": 4.6181,
+      "step": 1664
+    },
+    {
+      "epoch": 0.17264061072264095,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0001983884671071052,
+      "loss": 4.6752,
+      "step": 1665
+    },
+    {
+      "epoch": 0.17274429877712902,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00019838652430252346,
+      "loss": 4.6136,
+      "step": 1666
+    },
+    {
+      "epoch": 0.1728479868316171,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.0001983845803370847,
+      "loss": 4.6409,
+      "step": 1667
+    },
+    {
+      "epoch": 0.17295167488610516,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.00019838263521081191,
+      "loss": 4.578,
+      "step": 1668
+    },
+    {
+      "epoch": 0.17305536294059323,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000198380688923728,
+      "loss": 4.6316,
+      "step": 1669
+    },
+    {
+      "epoch": 0.1731590509950813,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.000198378741475856,
+      "loss": 4.6198,
+      "step": 1670
+    },
+    {
+      "epoch": 0.17326273904956938,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0001983767928672188,
+      "loss": 4.6566,
+      "step": 1671
+    },
+    {
+      "epoch": 0.17336642710405745,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019837484309783945,
+      "loss": 4.6642,
+      "step": 1672
+    },
+    {
+      "epoch": 0.17347011515854552,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019837289216774093,
+      "loss": 4.6507,
+      "step": 1673
+    },
+    {
+      "epoch": 0.1735738032130336,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0001983709400769463,
+      "loss": 4.5731,
+      "step": 1674
+    },
+    {
+      "epoch": 0.17367749126752166,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019836898682547852,
+      "loss": 4.6674,
+      "step": 1675
+    },
+    {
+      "epoch": 0.17378117932200973,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019836703241336067,
+      "loss": 4.6437,
+      "step": 1676
+    },
+    {
+      "epoch": 0.1738848673764978,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00019836507684061584,
+      "loss": 4.6252,
+      "step": 1677
+    },
+    {
+      "epoch": 0.17398855543098588,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.00019836312010726708,
+      "loss": 4.6167,
+      "step": 1678
+    },
+    {
+      "epoch": 0.17409224348547395,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019836116221333747,
+      "loss": 4.6435,
+      "step": 1679
+    },
+    {
+      "epoch": 0.17419593153996202,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.0001983592031588501,
+      "loss": 4.6064,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1742996195944501,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019835724294382814,
+      "loss": 4.5754,
+      "step": 1681
+    },
+    {
+      "epoch": 0.17440330764893816,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019835528156829466,
+      "loss": 4.5984,
+      "step": 1682
+    },
+    {
+      "epoch": 0.17450699570342623,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019835331903227284,
+      "loss": 4.6164,
+      "step": 1683
+    },
+    {
+      "epoch": 0.1746106837579143,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0001983513553357858,
+      "loss": 4.6206,
+      "step": 1684
+    },
+    {
+      "epoch": 0.17471437181240237,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019834939047885675,
+      "loss": 4.5944,
+      "step": 1685
+    },
+    {
+      "epoch": 0.17481805986689045,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001983474244615088,
+      "loss": 4.6713,
+      "step": 1686
+    },
+    {
+      "epoch": 0.17492174792137855,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.00019834545728376527,
+      "loss": 4.646,
+      "step": 1687
+    },
+    {
+      "epoch": 0.17502543597586662,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019834348894564924,
+      "loss": 4.6358,
+      "step": 1688
+    },
+    {
+      "epoch": 0.1751291240303547,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019834151944718404,
+      "loss": 4.6095,
+      "step": 1689
+    },
+    {
+      "epoch": 0.17523281208484276,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00019833954878839283,
+      "loss": 4.6115,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17533650013933083,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0001983375769692989,
+      "loss": 4.6866,
+      "step": 1691
+    },
+    {
+      "epoch": 0.1754401881938189,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.00019833560398992552,
+      "loss": 4.6569,
+      "step": 1692
+    },
+    {
+      "epoch": 0.17554387624830697,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019833362985029594,
+      "loss": 4.6235,
+      "step": 1693
+    },
+    {
+      "epoch": 0.17564756430279505,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0001983316545504335,
+      "loss": 4.6602,
+      "step": 1694
+    },
+    {
+      "epoch": 0.17575125235728312,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00019832967809036144,
+      "loss": 4.6386,
+      "step": 1695
+    },
+    {
+      "epoch": 0.1758549404117712,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019832770047010316,
+      "loss": 4.6023,
+      "step": 1696
+    },
+    {
+      "epoch": 0.17595862846625926,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019832572168968193,
+      "loss": 4.6724,
+      "step": 1697
+    },
+    {
+      "epoch": 0.17606231652074733,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00019832374174912111,
+      "loss": 4.5923,
+      "step": 1698
+    },
+    {
+      "epoch": 0.1761660045752354,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019832176064844408,
+      "loss": 4.6562,
+      "step": 1699
+    },
+    {
+      "epoch": 0.17626969262972347,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019831977838767422,
+      "loss": 4.5973,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17637338068421154,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0001983177949668349,
+      "loss": 4.6508,
+      "step": 1701
+    },
+    {
+      "epoch": 0.17647706873869962,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001983158103859495,
+      "loss": 4.643,
+      "step": 1702
+    },
+    {
+      "epoch": 0.1765807567931877,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019831382464504147,
+      "loss": 4.6174,
+      "step": 1703
+    },
+    {
+      "epoch": 0.17668444484767576,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019831183774413424,
+      "loss": 4.6328,
+      "step": 1704
+    },
+    {
+      "epoch": 0.17678813290216383,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019830984968325122,
+      "loss": 4.6157,
+      "step": 1705
+    },
+    {
+      "epoch": 0.1768918209566519,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019830786046241592,
+      "loss": 4.6173,
+      "step": 1706
+    },
+    {
+      "epoch": 0.17699550901113997,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019830587008165177,
+      "loss": 4.606,
+      "step": 1707
+    },
+    {
+      "epoch": 0.17709919706562804,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001983038785409823,
+      "loss": 4.5933,
+      "step": 1708
+    },
+    {
+      "epoch": 0.17720288512011614,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019830188584043094,
+      "loss": 4.6165,
+      "step": 1709
+    },
+    {
+      "epoch": 0.17730657317460422,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019829989198002124,
+      "loss": 4.6634,
+      "step": 1710
+    },
+    {
+      "epoch": 0.1774102612290923,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019829789695977672,
+      "loss": 4.5776,
+      "step": 1711
+    },
+    {
+      "epoch": 0.17751394928358036,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019829590077972094,
+      "loss": 4.6129,
+      "step": 1712
+    },
+    {
+      "epoch": 0.17761763733806843,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.00019829390343987743,
+      "loss": 4.5699,
+      "step": 1713
+    },
+    {
+      "epoch": 0.1777213253925565,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019829190494026974,
+      "loss": 4.6348,
+      "step": 1714
+    },
+    {
+      "epoch": 0.17782501344704457,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00019828990528092147,
+      "loss": 4.635,
+      "step": 1715
+    },
+    {
+      "epoch": 0.17792870150153264,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019828790446185622,
+      "loss": 4.6516,
+      "step": 1716
+    },
+    {
+      "epoch": 0.17803238955602071,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0001982859024830976,
+      "loss": 4.6052,
+      "step": 1717
+    },
+    {
+      "epoch": 0.1781360776105088,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001982838993446692,
+      "loss": 4.6519,
+      "step": 1718
+    },
+    {
+      "epoch": 0.17823976566499686,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019828189504659472,
+      "loss": 4.6294,
+      "step": 1719
+    },
+    {
+      "epoch": 0.17834345371948493,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019827988958889776,
+      "loss": 4.6044,
+      "step": 1720
+    },
+    {
+      "epoch": 0.178447141773973,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019827788297160196,
+      "loss": 4.6074,
+      "step": 1721
+    },
+    {
+      "epoch": 0.17855082982846107,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019827587519473107,
+      "loss": 4.6163,
+      "step": 1722
+    },
+    {
+      "epoch": 0.17865451788294914,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019827386625830871,
+      "loss": 4.6524,
+      "step": 1723
+    },
+    {
+      "epoch": 0.17875820593743721,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0001982718561623586,
+      "loss": 4.5932,
+      "step": 1724
+    },
+    {
+      "epoch": 0.17886189399192529,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019826984490690447,
+      "loss": 4.5631,
+      "step": 1725
+    },
+    {
+      "epoch": 0.17896558204641336,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019826783249197004,
+      "loss": 4.6035,
+      "step": 1726
+    },
+    {
+      "epoch": 0.17906927010090143,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019826581891757908,
+      "loss": 4.6006,
+      "step": 1727
+    },
+    {
+      "epoch": 0.1791729581553895,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019826380418375532,
+      "loss": 4.6126,
+      "step": 1728
+    },
+    {
+      "epoch": 0.17927664620987757,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019826178829052254,
+      "loss": 4.6404,
+      "step": 1729
+    },
+    {
+      "epoch": 0.17938033426436564,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001982597712379045,
+      "loss": 4.6289,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17948402231885374,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019825775302592503,
+      "loss": 4.6162,
+      "step": 1731
+    },
+    {
+      "epoch": 0.1795877103733418,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019825573365460798,
+      "loss": 4.615,
+      "step": 1732
+    },
+    {
+      "epoch": 0.17969139842782988,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0001982537131239771,
+      "loss": 4.6037,
+      "step": 1733
+    },
+    {
+      "epoch": 0.17979508648231796,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019825169143405623,
+      "loss": 4.6038,
+      "step": 1734
+    },
+    {
+      "epoch": 0.17989877453680603,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019824966858486933,
+      "loss": 4.6219,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1800024625912941,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00019824764457644016,
+      "loss": 4.6097,
+      "step": 1736
+    },
+    {
+      "epoch": 0.18010615064578217,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019824561940879262,
+      "loss": 4.5909,
+      "step": 1737
+    },
+    {
+      "epoch": 0.18020983870027024,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019824359308195068,
+      "loss": 4.5772,
+      "step": 1738
+    },
+    {
+      "epoch": 0.1803135267547583,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019824156559593813,
+      "loss": 4.6719,
+      "step": 1739
+    },
+    {
+      "epoch": 0.18041721480924638,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019823953695077896,
+      "loss": 4.5728,
+      "step": 1740
+    },
+    {
+      "epoch": 0.18052090286373446,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0001982375071464971,
+      "loss": 4.6433,
+      "step": 1741
+    },
+    {
+      "epoch": 0.18062459091822253,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019823547618311654,
+      "loss": 4.6155,
+      "step": 1742
+    },
+    {
+      "epoch": 0.1807282789727106,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019823344406066115,
+      "loss": 4.6704,
+      "step": 1743
+    },
+    {
+      "epoch": 0.18083196702719867,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.00019823141077915496,
+      "loss": 4.6568,
+      "step": 1744
+    },
+    {
+      "epoch": 0.18093565508168674,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00019822937633862198,
+      "loss": 4.6189,
+      "step": 1745
+    },
+    {
+      "epoch": 0.1810393431361748,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00019822734073908618,
+      "loss": 4.631,
+      "step": 1746
+    },
+    {
+      "epoch": 0.18114303119066288,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0001982253039805716,
+      "loss": 4.5911,
+      "step": 1747
+    },
+    {
+      "epoch": 0.18124671924515096,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00019822326606310227,
+      "loss": 4.6091,
+      "step": 1748
+    },
+    {
+      "epoch": 0.18135040729963903,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001982212269867022,
+      "loss": 4.5828,
+      "step": 1749
+    },
+    {
+      "epoch": 0.1814540953541271,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.00019821918675139548,
+      "loss": 4.6346,
+      "step": 1750
+    },
+    {
+      "epoch": 0.18155778340861517,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0001982171453572062,
+      "loss": 4.6333,
+      "step": 1751
+    },
+    {
+      "epoch": 0.18166147146310324,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00019821510280415837,
+      "loss": 4.658,
+      "step": 1752
+    },
+    {
+      "epoch": 0.1817651595175913,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019821305909227622,
+      "loss": 4.6648,
+      "step": 1753
+    },
+    {
+      "epoch": 0.1818688475720794,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00019821101422158374,
+      "loss": 4.6353,
+      "step": 1754
+    },
+    {
+      "epoch": 0.18197253562656748,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00019820896819210514,
+      "loss": 4.5773,
+      "step": 1755
+    },
+    {
+      "epoch": 0.18207622368105555,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001982069210038645,
+      "loss": 4.5819,
+      "step": 1756
+    },
+    {
+      "epoch": 0.18217991173554363,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00019820487265688602,
+      "loss": 4.6235,
+      "step": 1757
+    },
+    {
+      "epoch": 0.1822835997900317,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019820282315119382,
+      "loss": 4.6436,
+      "step": 1758
+    },
+    {
+      "epoch": 0.18238728784451977,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0001982007724868121,
+      "loss": 4.6404,
+      "step": 1759
+    },
+    {
+      "epoch": 0.18249097589900784,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00019819872066376512,
+      "loss": 4.6505,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1825946639534959,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.000198196667682077,
+      "loss": 4.6425,
+      "step": 1761
+    },
+    {
+      "epoch": 0.18269835200798398,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00019819461354177205,
+      "loss": 4.666,
+      "step": 1762
+    },
+    {
+      "epoch": 0.18280204006247205,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.0001981925582428744,
+      "loss": 4.5952,
+      "step": 1763
+    },
+    {
+      "epoch": 0.18290572811696013,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0001981905017854084,
+      "loss": 4.6084,
+      "step": 1764
+    },
+    {
+      "epoch": 0.1830094161714482,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019818844416939822,
+      "loss": 4.6354,
+      "step": 1765
+    },
+    {
+      "epoch": 0.18311310422593627,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00019818638539486822,
+      "loss": 4.6241,
+      "step": 1766
+    },
+    {
+      "epoch": 0.18321679228042434,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019818432546184266,
+      "loss": 4.6408,
+      "step": 1767
+    },
+    {
+      "epoch": 0.1833204803349124,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019818226437034583,
+      "loss": 4.5669,
+      "step": 1768
+    },
+    {
+      "epoch": 0.18342416838940048,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.0001981802021204021,
+      "loss": 4.6347,
+      "step": 1769
+    },
+    {
+      "epoch": 0.18352785644388855,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019817813871203573,
+      "loss": 4.6193,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18363154449837663,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019817607414527108,
+      "loss": 4.6221,
+      "step": 1771
+    },
+    {
+      "epoch": 0.1837352325528647,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00019817400842013258,
+      "loss": 4.6161,
+      "step": 1772
+    },
+    {
+      "epoch": 0.18383892060735277,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.00019817194153664455,
+      "loss": 4.5937,
+      "step": 1773
+    },
+    {
+      "epoch": 0.18394260866184084,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019816987349483136,
+      "loss": 4.6,
+      "step": 1774
+    },
+    {
+      "epoch": 0.1840462967163289,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00019816780429471743,
+      "loss": 4.6343,
+      "step": 1775
+    },
+    {
+      "epoch": 0.184149984770817,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0001981657339363272,
+      "loss": 4.592,
+      "step": 1776
+    },
+    {
+      "epoch": 0.18425367282530508,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00019816366241968506,
+      "loss": 4.6029,
+      "step": 1777
+    },
+    {
+      "epoch": 0.18435736087979315,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00019816158974481548,
+      "loss": 4.6375,
+      "step": 1778
+    },
+    {
+      "epoch": 0.18446104893428122,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0001981595159117429,
+      "loss": 4.6048,
+      "step": 1779
+    },
+    {
+      "epoch": 0.1845647369887693,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0001981574409204918,
+      "loss": 4.5995,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18466842504325737,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019815536477108662,
+      "loss": 4.6003,
+      "step": 1781
+    },
+    {
+      "epoch": 0.18477211309774544,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019815328746355192,
+      "loss": 4.6188,
+      "step": 1782
+    },
+    {
+      "epoch": 0.1848758011522335,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019815120899791216,
+      "loss": 4.6491,
+      "step": 1783
+    },
+    {
+      "epoch": 0.18497948920672158,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0001981491293741919,
+      "loss": 4.6405,
+      "step": 1784
+    },
+    {
+      "epoch": 0.18508317726120965,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019814704859241565,
+      "loss": 4.6342,
+      "step": 1785
+    },
+    {
+      "epoch": 0.18518686531569772,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019814496665260798,
+      "loss": 4.6543,
+      "step": 1786
+    },
+    {
+      "epoch": 0.1852905533701858,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019814288355479346,
+      "loss": 4.6173,
+      "step": 1787
+    },
+    {
+      "epoch": 0.18539424142467387,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00019814079929899662,
+      "loss": 4.6002,
+      "step": 1788
+    },
+    {
+      "epoch": 0.18549792947916194,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019813871388524215,
+      "loss": 4.5986,
+      "step": 1789
+    },
+    {
+      "epoch": 0.18560161753365,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019813662731355452,
+      "loss": 4.6107,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18570530558813808,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019813453958395847,
+      "loss": 4.6364,
+      "step": 1791
+    },
+    {
+      "epoch": 0.18580899364262615,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019813245069647857,
+      "loss": 4.5867,
+      "step": 1792
+    },
+    {
+      "epoch": 0.18591268169711422,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001981303606511395,
+      "loss": 4.5961,
+      "step": 1793
+    },
+    {
+      "epoch": 0.1860163697516023,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019812826944796586,
+      "loss": 4.6213,
+      "step": 1794
+    },
+    {
+      "epoch": 0.18612005780609037,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00019812617708698241,
+      "loss": 4.5809,
+      "step": 1795
+    },
+    {
+      "epoch": 0.18622374586057844,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019812408356821378,
+      "loss": 4.6406,
+      "step": 1796
+    },
+    {
+      "epoch": 0.1863274339150665,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019812198889168468,
+      "loss": 4.6107,
+      "step": 1797
+    },
+    {
+      "epoch": 0.1864311219695546,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019811989305741984,
+      "loss": 4.5858,
+      "step": 1798
+    },
+    {
+      "epoch": 0.18653481002404268,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00019811779606544397,
+      "loss": 4.6217,
+      "step": 1799
+    },
+    {
+      "epoch": 0.18663849807853075,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019811569791578182,
+      "loss": 4.5735,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18674218613301882,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019811359860845814,
+      "loss": 4.6024,
+      "step": 1801
+    },
+    {
+      "epoch": 0.1868458741875069,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019811149814349773,
+      "loss": 4.5761,
+      "step": 1802
+    },
+    {
+      "epoch": 0.18694956224199497,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019810939652092533,
+      "loss": 4.5857,
+      "step": 1803
+    },
+    {
+      "epoch": 0.18705325029648304,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019810729374076575,
+      "loss": 4.633,
+      "step": 1804
+    },
+    {
+      "epoch": 0.1871569383509711,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019810518980304385,
+      "loss": 4.628,
+      "step": 1805
+    },
+    {
+      "epoch": 0.18726062640545918,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019810308470778436,
+      "loss": 4.5949,
+      "step": 1806
+    },
+    {
+      "epoch": 0.18736431445994725,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0001981009784550122,
+      "loss": 4.5659,
+      "step": 1807
+    },
+    {
+      "epoch": 0.18746800251443532,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001980988710447522,
+      "loss": 4.5788,
+      "step": 1808
+    },
+    {
+      "epoch": 0.1875716905689234,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0001980967624770292,
+      "loss": 4.5945,
+      "step": 1809
+    },
+    {
+      "epoch": 0.18767537862341147,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0001980946527518681,
+      "loss": 4.6004,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18777906667789954,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0001980925418692938,
+      "loss": 4.6061,
+      "step": 1811
+    },
+    {
+      "epoch": 0.1878827547323876,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019809042982933117,
+      "loss": 4.6141,
+      "step": 1812
+    },
+    {
+      "epoch": 0.18798644278687568,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019808831663200517,
+      "loss": 4.641,
+      "step": 1813
+    },
+    {
+      "epoch": 0.18809013084136375,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001980862022773407,
+      "loss": 4.542,
+      "step": 1814
+    },
+    {
+      "epoch": 0.18819381889585182,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019808408676536275,
+      "loss": 4.587,
+      "step": 1815
+    },
+    {
+      "epoch": 0.1882975069503399,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019808197009609624,
+      "loss": 4.6407,
+      "step": 1816
+    },
+    {
+      "epoch": 0.18840119500482796,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019807985226956616,
+      "loss": 4.6117,
+      "step": 1817
+    },
+    {
+      "epoch": 0.18850488305931604,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0001980777332857975,
+      "loss": 4.6044,
+      "step": 1818
+    },
+    {
+      "epoch": 0.1886085711138041,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019807561314481525,
+      "loss": 4.5919,
+      "step": 1819
+    },
+    {
+      "epoch": 0.1887122591682922,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019807349184664447,
+      "loss": 4.5995,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18881594722278028,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0001980713693913101,
+      "loss": 4.6098,
+      "step": 1821
+    },
+    {
+      "epoch": 0.18891963527726835,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0001980692457788373,
+      "loss": 4.6336,
+      "step": 1822
+    },
+    {
+      "epoch": 0.18902332333175642,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019806712100925103,
+      "loss": 4.5602,
+      "step": 1823
+    },
+    {
+      "epoch": 0.1891270113862445,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019806499508257636,
+      "loss": 4.585,
+      "step": 1824
+    },
+    {
+      "epoch": 0.18923069944073256,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019806286799883846,
+      "loss": 4.6508,
+      "step": 1825
+    },
+    {
+      "epoch": 0.18933438749522064,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019806073975806235,
+      "loss": 4.5661,
+      "step": 1826
+    },
+    {
+      "epoch": 0.1894380755497087,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019805861036027318,
+      "loss": 4.6343,
+      "step": 1827
+    },
+    {
+      "epoch": 0.18954176360419678,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019805647980549606,
+      "loss": 4.6045,
+      "step": 1828
+    },
+    {
+      "epoch": 0.18964545165868485,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001980543480937561,
+      "loss": 4.6495,
+      "step": 1829
+    },
+    {
+      "epoch": 0.18974913971317292,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001980522152250785,
+      "loss": 4.5932,
+      "step": 1830
+    },
+    {
+      "epoch": 0.189852827767661,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019805008119948842,
+      "loss": 4.6119,
+      "step": 1831
+    },
+    {
+      "epoch": 0.18995651582214906,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.000198047946017011,
+      "loss": 4.6007,
+      "step": 1832
+    },
+    {
+      "epoch": 0.19006020387663713,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00019804580967767147,
+      "loss": 4.5951,
+      "step": 1833
+    },
+    {
+      "epoch": 0.1901638919311252,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00019804367218149505,
+      "loss": 4.5742,
+      "step": 1834
+    },
+    {
+      "epoch": 0.19026757998561328,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0001980415335285069,
+      "loss": 4.5971,
+      "step": 1835
+    },
+    {
+      "epoch": 0.19037126804010135,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0001980393937187323,
+      "loss": 4.6173,
+      "step": 1836
+    },
+    {
+      "epoch": 0.19047495609458942,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019803725275219648,
+      "loss": 4.6155,
+      "step": 1837
+    },
+    {
+      "epoch": 0.1905786441490775,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0001980351106289247,
+      "loss": 4.5651,
+      "step": 1838
+    },
+    {
+      "epoch": 0.19068233220356556,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019803296734894227,
+      "loss": 4.6264,
+      "step": 1839
+    },
+    {
+      "epoch": 0.19078602025805363,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019803082291227443,
+      "loss": 4.599,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1908897083125417,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0001980286773189465,
+      "loss": 4.5985,
+      "step": 1841
+    },
+    {
+      "epoch": 0.19099339636702978,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0001980265305689838,
+      "loss": 4.5839,
+      "step": 1842
+    },
+    {
+      "epoch": 0.19109708442151788,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0001980243826624117,
+      "loss": 4.5921,
+      "step": 1843
+    },
+    {
+      "epoch": 0.19120077247600595,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00019802223359925545,
+      "loss": 4.6503,
+      "step": 1844
+    },
+    {
+      "epoch": 0.19130446053049402,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00019802008337954047,
+      "loss": 4.5806,
+      "step": 1845
+    },
+    {
+      "epoch": 0.1914081485849821,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00019801793200329213,
+      "loss": 4.5883,
+      "step": 1846
+    },
+    {
+      "epoch": 0.19151183663947016,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019801577947053579,
+      "loss": 4.6491,
+      "step": 1847
+    },
+    {
+      "epoch": 0.19161552469395823,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019801362578129685,
+      "loss": 4.6137,
+      "step": 1848
+    },
+    {
+      "epoch": 0.1917192127484463,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.00019801147093560076,
+      "loss": 4.6222,
+      "step": 1849
+    },
+    {
+      "epoch": 0.19182290080293438,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019800931493347288,
+      "loss": 4.6001,
+      "step": 1850
+    },
+    {
+      "epoch": 0.19192658885742245,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019800715777493872,
+      "loss": 4.6157,
+      "step": 1851
+    },
+    {
+      "epoch": 0.19203027691191052,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019800499946002366,
+      "loss": 4.581,
+      "step": 1852
+    },
+    {
+      "epoch": 0.1921339649663986,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.00019800283998875324,
+      "loss": 4.6079,
+      "step": 1853
+    },
+    {
+      "epoch": 0.19223765302088666,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019800067936115288,
+      "loss": 4.577,
+      "step": 1854
+    },
+    {
+      "epoch": 0.19234134107537473,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019799851757724808,
+      "loss": 4.575,
+      "step": 1855
+    },
+    {
+      "epoch": 0.1924450291298628,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019799635463706438,
+      "loss": 4.6081,
+      "step": 1856
+    },
+    {
+      "epoch": 0.19254871718435088,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019799419054062728,
+      "loss": 4.6058,
+      "step": 1857
+    },
+    {
+      "epoch": 0.19265240523883895,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019799202528796231,
+      "loss": 4.618,
+      "step": 1858
+    },
+    {
+      "epoch": 0.19275609329332702,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019798985887909502,
+      "loss": 4.6002,
+      "step": 1859
+    },
+    {
+      "epoch": 0.1928597813478151,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019798769131405098,
+      "loss": 4.6159,
+      "step": 1860
+    },
+    {
+      "epoch": 0.19296346940230316,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019798552259285579,
+      "loss": 4.5643,
+      "step": 1861
+    },
+    {
+      "epoch": 0.19306715745679123,
+      "grad_norm": 0.625,
+      "learning_rate": 0.000197983352715535,
+      "loss": 4.6061,
+      "step": 1862
+    },
+    {
+      "epoch": 0.1931708455112793,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0001979811816821142,
+      "loss": 4.5693,
+      "step": 1863
+    },
+    {
+      "epoch": 0.19327453356576738,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.000197979009492619,
+      "loss": 4.6038,
+      "step": 1864
+    },
+    {
+      "epoch": 0.19337822162025547,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.00019797683614707512,
+      "loss": 4.5976,
+      "step": 1865
+    },
+    {
+      "epoch": 0.19348190967474355,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0001979746616455081,
+      "loss": 4.5936,
+      "step": 1866
+    },
+    {
+      "epoch": 0.19358559772923162,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.00019797248598794364,
+      "loss": 4.5944,
+      "step": 1867
+    },
+    {
+      "epoch": 0.1936892857837197,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0001979703091744074,
+      "loss": 4.6682,
+      "step": 1868
+    },
+    {
+      "epoch": 0.19379297383820776,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00019796813120492507,
+      "loss": 4.562,
+      "step": 1869
+    },
+    {
+      "epoch": 0.19389666189269583,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.00019796595207952238,
+      "loss": 4.5784,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1940003499471839,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.000197963771798225,
+      "loss": 4.5825,
+      "step": 1871
+    },
+    {
+      "epoch": 0.19410403800167197,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019796159036105865,
+      "loss": 4.6164,
+      "step": 1872
+    },
+    {
+      "epoch": 0.19420772605616005,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019795940776804906,
+      "loss": 4.5685,
+      "step": 1873
+    },
+    {
+      "epoch": 0.19431141411064812,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019795722401922205,
+      "loss": 4.6074,
+      "step": 1874
+    },
+    {
+      "epoch": 0.1944151021651362,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0001979550391146033,
+      "loss": 4.5823,
+      "step": 1875
+    },
+    {
+      "epoch": 0.19451879021962426,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0001979528530542187,
+      "loss": 4.6129,
+      "step": 1876
+    },
+    {
+      "epoch": 0.19462247827411233,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019795066583809393,
+      "loss": 4.5628,
+      "step": 1877
+    },
+    {
+      "epoch": 0.1947261663286004,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019794847746625483,
+      "loss": 4.5917,
+      "step": 1878
+    },
+    {
+      "epoch": 0.19482985438308847,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019794628793872723,
+      "loss": 4.6206,
+      "step": 1879
+    },
+    {
+      "epoch": 0.19493354243757655,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019794409725553699,
+      "loss": 4.5926,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19503723049206462,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019794190541670993,
+      "loss": 4.6335,
+      "step": 1881
+    },
+    {
+      "epoch": 0.1951409185465527,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0001979397124222719,
+      "loss": 4.6138,
+      "step": 1882
+    },
+    {
+      "epoch": 0.19524460660104076,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0001979375182722488,
+      "loss": 4.622,
+      "step": 1883
+    },
+    {
+      "epoch": 0.19534829465552883,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0001979353229666665,
+      "loss": 4.5461,
+      "step": 1884
+    },
+    {
+      "epoch": 0.1954519827100169,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00019793312650555093,
+      "loss": 4.5757,
+      "step": 1885
+    },
+    {
+      "epoch": 0.19555567076450497,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.00019793092888892799,
+      "loss": 4.6087,
+      "step": 1886
+    },
+    {
+      "epoch": 0.19565935881899307,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00019792873011682357,
+      "loss": 4.5903,
+      "step": 1887
+    },
+    {
+      "epoch": 0.19576304687348114,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0001979265301892637,
+      "loss": 4.5752,
+      "step": 1888
+    },
+    {
+      "epoch": 0.19586673492796922,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00019792432910627425,
+      "loss": 4.5782,
+      "step": 1889
+    },
+    {
+      "epoch": 0.1959704229824573,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019792212686788122,
+      "loss": 4.5527,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19607411103694536,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00019791992347411064,
+      "loss": 4.6206,
+      "step": 1891
+    },
+    {
+      "epoch": 0.19617779909143343,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00019791771892498843,
+      "loss": 4.6059,
+      "step": 1892
+    },
+    {
+      "epoch": 0.1962814871459215,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.00019791551322054067,
+      "loss": 4.5715,
+      "step": 1893
+    },
+    {
+      "epoch": 0.19638517520040957,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00019791330636079332,
+      "loss": 4.5888,
+      "step": 1894
+    },
+    {
+      "epoch": 0.19648886325489764,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0001979110983457725,
+      "loss": 4.5763,
+      "step": 1895
+    },
+    {
+      "epoch": 0.19659255130938572,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0001979088891755042,
+      "loss": 4.5754,
+      "step": 1896
+    },
+    {
+      "epoch": 0.1966962393638738,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.00019790667885001448,
+      "loss": 4.6025,
+      "step": 1897
+    },
+    {
+      "epoch": 0.19679992741836186,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.00019790446736932946,
+      "loss": 4.6139,
+      "step": 1898
+    },
+    {
+      "epoch": 0.19690361547284993,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0001979022547334752,
+      "loss": 4.5304,
+      "step": 1899
+    },
+    {
+      "epoch": 0.197007303527338,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00019790004094247783,
+      "loss": 4.6188,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19711099158182607,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019789782599636348,
+      "loss": 4.6019,
+      "step": 1901
+    },
+    {
+      "epoch": 0.19721467963631414,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.00019789560989515824,
+      "loss": 4.5996,
+      "step": 1902
+    },
+    {
+      "epoch": 0.19731836769080222,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001978933926388883,
+      "loss": 4.585,
+      "step": 1903
+    },
+    {
+      "epoch": 0.1974220557452903,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0001978911742275798,
+      "loss": 4.6249,
+      "step": 1904
+    },
+    {
+      "epoch": 0.19752574379977836,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0001978889546612589,
+      "loss": 4.5999,
+      "step": 1905
+    },
+    {
+      "epoch": 0.19762943185426643,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00019788673393995182,
+      "loss": 4.6135,
+      "step": 1906
+    },
+    {
+      "epoch": 0.1977331199087545,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019788451206368475,
+      "loss": 4.6007,
+      "step": 1907
+    },
+    {
+      "epoch": 0.19783680796324257,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00019788228903248393,
+      "loss": 4.6105,
+      "step": 1908
+    },
+    {
+      "epoch": 0.19794049601773064,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019788006484637553,
+      "loss": 4.6072,
+      "step": 1909
+    },
+    {
+      "epoch": 0.19804418407221874,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019787783950538587,
+      "loss": 4.59,
+      "step": 1910
+    },
+    {
+      "epoch": 0.19814787212670681,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001978756130095411,
+      "loss": 4.5976,
+      "step": 1911
+    },
+    {
+      "epoch": 0.19825156018119489,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001978733853588676,
+      "loss": 4.5981,
+      "step": 1912
+    },
+    {
+      "epoch": 0.19835524823568296,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019787115655339163,
+      "loss": 4.6151,
+      "step": 1913
+    },
+    {
+      "epoch": 0.19845893629017103,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00019786892659313945,
+      "loss": 4.6266,
+      "step": 1914
+    },
+    {
+      "epoch": 0.1985626243446591,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019786669547813737,
+      "loss": 4.6112,
+      "step": 1915
+    },
+    {
+      "epoch": 0.19866631239914717,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019786446320841172,
+      "loss": 4.5899,
+      "step": 1916
+    },
+    {
+      "epoch": 0.19877000045363524,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00019786222978398889,
+      "loss": 4.546,
+      "step": 1917
+    },
+    {
+      "epoch": 0.19887368850812331,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00019785999520489518,
+      "loss": 4.537,
+      "step": 1918
+    },
+    {
+      "epoch": 0.19897737656261139,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019785775947115696,
+      "loss": 4.5717,
+      "step": 1919
+    },
+    {
+      "epoch": 0.19908106461709946,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.00019785552258280064,
+      "loss": 4.5215,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19918475267158753,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019785328453985257,
+      "loss": 4.6153,
+      "step": 1921
+    },
+    {
+      "epoch": 0.1992884407260756,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019785104534233918,
+      "loss": 4.5823,
+      "step": 1922
+    },
+    {
+      "epoch": 0.19939212878056367,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019784880499028692,
+      "loss": 4.5774,
+      "step": 1923
+    },
+    {
+      "epoch": 0.19949581683505174,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00019784656348372214,
+      "loss": 4.6047,
+      "step": 1924
+    },
+    {
+      "epoch": 0.1995995048895398,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.00019784432082267138,
+      "loss": 4.612,
+      "step": 1925
+    },
+    {
+      "epoch": 0.19970319294402789,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00019784207700716103,
+      "loss": 4.6276,
+      "step": 1926
+    },
+    {
+      "epoch": 0.19980688099851596,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019783983203721758,
+      "loss": 4.5811,
+      "step": 1927
+    },
+    {
+      "epoch": 0.19991056905300403,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.0001978375859128676,
+      "loss": 4.5809,
+      "step": 1928
+    },
+    {
+      "epoch": 0.2000142571074921,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001978353386341375,
+      "loss": 4.6231,
+      "step": 1929
+    },
+    {
+      "epoch": 0.20011794516198017,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019783309020105375,
+      "loss": 4.5642,
+      "step": 1930
+    },
+    {
+      "epoch": 0.20022163321646824,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00019783084061364303,
+      "loss": 4.581,
+      "step": 1931
+    },
+    {
+      "epoch": 0.20032532127095634,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019782858987193178,
+      "loss": 4.5868,
+      "step": 1932
+    },
+    {
+      "epoch": 0.2004290093254444,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019782633797594659,
+      "loss": 4.617,
+      "step": 1933
+    },
+    {
+      "epoch": 0.20053269737993248,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00019782408492571399,
+      "loss": 4.6109,
+      "step": 1934
+    },
+    {
+      "epoch": 0.20063638543442056,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00019782183072126062,
+      "loss": 4.6252,
+      "step": 1935
+    },
+    {
+      "epoch": 0.20074007348890863,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00019781957536261303,
+      "loss": 4.6283,
+      "step": 1936
+    },
+    {
+      "epoch": 0.2008437615433967,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00019781731884979786,
+      "loss": 4.606,
+      "step": 1937
+    },
+    {
+      "epoch": 0.20094744959788477,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00019781506118284173,
+      "loss": 4.6014,
+      "step": 1938
+    },
+    {
+      "epoch": 0.20105113765237284,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019781280236177127,
+      "loss": 4.6413,
+      "step": 1939
+    },
+    {
+      "epoch": 0.2011548257068609,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0001978105423866131,
+      "loss": 4.5813,
+      "step": 1940
+    },
+    {
+      "epoch": 0.20125851376134898,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019780828125739398,
+      "loss": 4.5541,
+      "step": 1941
+    },
+    {
+      "epoch": 0.20136220181583706,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0001978060189741405,
+      "loss": 4.6159,
+      "step": 1942
+    },
+    {
+      "epoch": 0.20146588987032513,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00019780375553687937,
+      "loss": 4.6638,
+      "step": 1943
+    },
+    {
+      "epoch": 0.2015695779248132,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00019780149094563733,
+      "loss": 4.6151,
+      "step": 1944
+    },
+    {
+      "epoch": 0.20167326597930127,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00019779922520044108,
+      "loss": 4.5835,
+      "step": 1945
+    },
+    {
+      "epoch": 0.20177695403378934,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00019779695830131732,
+      "loss": 4.5769,
+      "step": 1946
+    },
+    {
+      "epoch": 0.2018806420882774,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00019779469024829285,
+      "loss": 4.6143,
+      "step": 1947
+    },
+    {
+      "epoch": 0.20198433014276548,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001977924210413944,
+      "loss": 4.604,
+      "step": 1948
+    },
+    {
+      "epoch": 0.20208801819725355,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019779015068064877,
+      "loss": 4.5636,
+      "step": 1949
+    },
+    {
+      "epoch": 0.20219170625174163,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00019778787916608273,
+      "loss": 4.5581,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2022953943062297,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019778560649772305,
+      "loss": 4.5624,
+      "step": 1951
+    },
+    {
+      "epoch": 0.20239908236071777,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019778333267559658,
+      "loss": 4.6108,
+      "step": 1952
+    },
+    {
+      "epoch": 0.20250277041520584,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00019778105769973018,
+      "loss": 4.5904,
+      "step": 1953
+    },
+    {
+      "epoch": 0.20260645846969394,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00019777878157015063,
+      "loss": 4.5902,
+      "step": 1954
+    },
+    {
+      "epoch": 0.202710146524182,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019777650428688483,
+      "loss": 4.6012,
+      "step": 1955
+    },
+    {
+      "epoch": 0.20281383457867008,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00019777422584995965,
+      "loss": 4.544,
+      "step": 1956
+    },
+    {
+      "epoch": 0.20291752263315815,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019777194625940193,
+      "loss": 4.6165,
+      "step": 1957
+    },
+    {
+      "epoch": 0.20302121068764623,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00019776966551523858,
+      "loss": 4.5806,
+      "step": 1958
+    },
+    {
+      "epoch": 0.2031248987421343,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00019776738361749655,
+      "loss": 4.5912,
+      "step": 1959
+    },
+    {
+      "epoch": 0.20322858679662237,
+      "grad_norm": 1.9375,
+      "learning_rate": 0.00019776510056620272,
+      "loss": 4.6098,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20333227485111044,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00019776281636138407,
+      "loss": 4.5712,
+      "step": 1961
+    },
+    {
+      "epoch": 0.2034359629055985,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.0001977605310030675,
+      "loss": 4.6034,
+      "step": 1962
+    },
+    {
+      "epoch": 0.20353965096008658,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00019775824449128003,
+      "loss": 4.6355,
+      "step": 1963
+    },
+    {
+      "epoch": 0.20364333901457465,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0001977559568260486,
+      "loss": 4.5732,
+      "step": 1964
+    },
+    {
+      "epoch": 0.20374702706906272,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.0001977536680074002,
+      "loss": 4.5894,
+      "step": 1965
+    },
+    {
+      "epoch": 0.2038507151235508,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00019775137803536186,
+      "loss": 4.6234,
+      "step": 1966
+    },
+    {
+      "epoch": 0.20395440317803887,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.00019774908690996056,
+      "loss": 4.646,
+      "step": 1967
+    },
+    {
+      "epoch": 0.20405809123252694,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.0001977467946312234,
+      "loss": 4.518,
+      "step": 1968
+    },
+    {
+      "epoch": 0.204161779287015,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00019774450119917737,
+      "loss": 4.6341,
+      "step": 1969
+    },
+    {
+      "epoch": 0.20426546734150308,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00019774220661384956,
+      "loss": 4.5852,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20436915539599115,
+      "grad_norm": 1.5,
+      "learning_rate": 0.000197739910875267,
+      "loss": 4.5918,
+      "step": 1971
+    },
+    {
+      "epoch": 0.20447284345047922,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00019773761398345682,
+      "loss": 4.5398,
+      "step": 1972
+    },
+    {
+      "epoch": 0.2045765315049673,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00019773531593844613,
+      "loss": 4.5249,
+      "step": 1973
+    },
+    {
+      "epoch": 0.20468021955945537,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019773301674026197,
+      "loss": 4.5895,
+      "step": 1974
+    },
+    {
+      "epoch": 0.20478390761394344,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019773071638893157,
+      "loss": 4.6123,
+      "step": 1975
+    },
+    {
+      "epoch": 0.2048875956684315,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00019772841488448198,
+      "loss": 4.5979,
+      "step": 1976
+    },
+    {
+      "epoch": 0.2049912837229196,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.00019772611222694045,
+      "loss": 4.5542,
+      "step": 1977
+    },
+    {
+      "epoch": 0.20509497177740768,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00019772380841633406,
+      "loss": 4.5651,
+      "step": 1978
+    },
+    {
+      "epoch": 0.20519865983189575,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00019772150345269003,
+      "loss": 4.5454,
+      "step": 1979
+    },
+    {
+      "epoch": 0.20530234788638382,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00019771919733603557,
+      "loss": 4.6146,
+      "step": 1980
+    },
+    {
+      "epoch": 0.2054060359408719,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019771689006639785,
+      "loss": 4.5667,
+      "step": 1981
+    },
+    {
+      "epoch": 0.20550972399535997,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00019771458164380415,
+      "loss": 4.5801,
+      "step": 1982
+    },
+    {
+      "epoch": 0.20561341204984804,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019771227206828167,
+      "loss": 4.6104,
+      "step": 1983
+    },
+    {
+      "epoch": 0.2057171001043361,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00019770996133985767,
+      "loss": 4.6062,
+      "step": 1984
+    },
+    {
+      "epoch": 0.20582078815882418,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00019770764945855937,
+      "loss": 4.5738,
+      "step": 1985
+    },
+    {
+      "epoch": 0.20592447621331225,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00019770533642441413,
+      "loss": 4.5966,
+      "step": 1986
+    },
+    {
+      "epoch": 0.20602816426780032,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.0001977030222374492,
+      "loss": 4.6273,
+      "step": 1987
+    },
+    {
+      "epoch": 0.2061318523222884,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00019770070689769184,
+      "loss": 4.5446,
+      "step": 1988
+    },
+    {
+      "epoch": 0.20623554037677647,
+      "grad_norm": 1.921875,
+      "learning_rate": 0.00019769839040516946,
+      "loss": 4.5785,
+      "step": 1989
+    },
+    {
+      "epoch": 0.20633922843126454,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00019769607275990934,
+      "loss": 4.5787,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2064429164857526,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00019769375396193881,
+      "loss": 4.5644,
+      "step": 1991
+    },
+    {
+      "epoch": 0.20654660454024068,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019769143401128525,
+      "loss": 4.5818,
+      "step": 1992
+    },
+    {
+      "epoch": 0.20665029259472875,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00019768911290797604,
+      "loss": 4.5749,
+      "step": 1993
+    },
+    {
+      "epoch": 0.20675398064921682,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019768679065203855,
+      "loss": 4.5796,
+      "step": 1994
+    },
+    {
+      "epoch": 0.2068576687037049,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00019768446724350024,
+      "loss": 4.5841,
+      "step": 1995
+    },
+    {
+      "epoch": 0.20696135675819297,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00019768214268238842,
+      "loss": 4.569,
+      "step": 1996
+    },
+    {
+      "epoch": 0.20706504481268104,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00019767981696873057,
+      "loss": 4.6043,
+      "step": 1997
+    },
+    {
+      "epoch": 0.2071687328671691,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00019767749010255416,
+      "loss": 4.6222,
+      "step": 1998
+    },
+    {
+      "epoch": 0.2072724209216572,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0001976751620838866,
+      "loss": 4.5898,
+      "step": 1999
+    },
+    {
+      "epoch": 0.20737610897614528,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00019767283291275537,
+      "loss": 4.5915,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20747979703063335,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.00019767050258918798,
+      "loss": 4.5657,
+      "step": 2001
+    },
+    {
+      "epoch": 0.20758348508512142,
+      "grad_norm": 2.0,
+      "learning_rate": 0.00019766817111321186,
+      "loss": 4.6221,
+      "step": 2002
+    },
+    {
+      "epoch": 0.2076871731396095,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.0001976658384848546,
+      "loss": 4.6072,
+      "step": 2003
+    },
+    {
+      "epoch": 0.20779086119409756,
+      "grad_norm": 1.796875,
+      "learning_rate": 0.00019766350470414365,
+      "loss": 4.576,
+      "step": 2004
+    },
+    {
+      "epoch": 0.20789454924858564,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00019766116977110661,
+      "loss": 4.5551,
+      "step": 2005
+    },
+    {
+      "epoch": 0.2079982373030737,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.000197658833685771,
+      "loss": 4.5634,
+      "step": 2006
+    },
+    {
+      "epoch": 0.20810192535756178,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00019765649644816436,
+      "loss": 4.5981,
+      "step": 2007
+    },
+    {
+      "epoch": 0.20820561341204985,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.0001976541580583143,
+      "loss": 4.5633,
+      "step": 2008
+    },
+    {
+      "epoch": 0.20830930146653792,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001976518185162484,
+      "loss": 4.6089,
+      "step": 2009
+    },
+    {
+      "epoch": 0.208412989521026,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00019764947782199426,
+      "loss": 4.5937,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20851667757551406,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00019764713597557952,
+      "loss": 4.5695,
+      "step": 2011
+    },
+    {
+      "epoch": 0.20862036563000214,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.00019764479297703177,
+      "loss": 4.6169,
+      "step": 2012
+    },
+    {
+      "epoch": 0.2087240536844902,
+      "grad_norm": 1.8984375,
+      "learning_rate": 0.00019764244882637867,
+      "loss": 4.6056,
+      "step": 2013
+    },
+    {
+      "epoch": 0.20882774173897828,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00019764010352364792,
+      "loss": 4.5951,
+      "step": 2014
+    },
+    {
+      "epoch": 0.20893142979346635,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00019763775706886714,
+      "loss": 4.5894,
+      "step": 2015
+    },
+    {
+      "epoch": 0.20903511784795442,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00019763540946206404,
+      "loss": 4.6161,
+      "step": 2016
+    },
+    {
+      "epoch": 0.2091388059024425,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00019763306070326632,
+      "loss": 4.5434,
+      "step": 2017
+    },
+    {
+      "epoch": 0.20924249395693056,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00019763071079250164,
+      "loss": 4.5784,
+      "step": 2018
+    },
+    {
+      "epoch": 0.20934618201141864,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00019762835972979783,
+      "loss": 4.5505,
+      "step": 2019
+    },
+    {
+      "epoch": 0.2094498700659067,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00019762600751518255,
+      "loss": 4.5706,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2095535581203948,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00019762365414868356,
+      "loss": 4.5687,
+      "step": 2021
+    },
+    {
+      "epoch": 0.20965724617488288,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00019762129963032867,
+      "loss": 4.5933,
+      "step": 2022
+    },
+    {
+      "epoch": 0.20976093422937095,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0001976189439601456,
+      "loss": 4.6114,
+      "step": 2023
+    },
+    {
+      "epoch": 0.20986462228385902,
+      "grad_norm": 1.8203125,
+      "learning_rate": 0.0001976165871381622,
+      "loss": 4.5804,
+      "step": 2024
+    },
+    {
+      "epoch": 0.2099683103383471,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0001976142291644063,
+      "loss": 4.6148,
+      "step": 2025
+    },
+    {
+      "epoch": 0.21007199839283516,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.00019761187003890563,
+      "loss": 4.5602,
+      "step": 2026
+    },
+    {
+      "epoch": 0.21017568644732323,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.0001976095097616881,
+      "loss": 4.6301,
+      "step": 2027
+    },
+    {
+      "epoch": 0.2102793745018113,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00019760714833278148,
+      "loss": 4.5808,
+      "step": 2028
+    },
+    {
+      "epoch": 0.21038306255629938,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00019760478575221372,
+      "loss": 4.5701,
+      "step": 2029
+    },
+    {
+      "epoch": 0.21048675061078745,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00019760242202001267,
+      "loss": 4.5741,
+      "step": 2030
+    },
+    {
+      "epoch": 0.21059043866527552,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019760005713620623,
+      "loss": 4.5976,
+      "step": 2031
+    },
+    {
+      "epoch": 0.2106941267197636,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019759769110082223,
+      "loss": 4.6137,
+      "step": 2032
+    },
+    {
+      "epoch": 0.21079781477425166,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00019759532391388867,
+      "loss": 4.5868,
+      "step": 2033
+    },
+    {
+      "epoch": 0.21090150282873973,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00019759295557543344,
+      "loss": 4.5967,
+      "step": 2034
+    },
+    {
+      "epoch": 0.2110051908832278,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0001975905860854845,
+      "loss": 4.5903,
+      "step": 2035
+    },
+    {
+      "epoch": 0.21110887893771588,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.0001975882154440698,
+      "loss": 4.6173,
+      "step": 2036
+    },
+    {
+      "epoch": 0.21121256699220395,
+      "grad_norm": 1.8984375,
+      "learning_rate": 0.00019758584365121734,
+      "loss": 4.6118,
+      "step": 2037
+    },
+    {
+      "epoch": 0.21131625504669202,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00019758347070695503,
+      "loss": 4.5663,
+      "step": 2038
+    },
+    {
+      "epoch": 0.2114199431011801,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00019758109661131092,
+      "loss": 4.5646,
+      "step": 2039
+    },
+    {
+      "epoch": 0.21152363115566816,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019757872136431305,
+      "loss": 4.6037,
+      "step": 2040
+    },
+    {
+      "epoch": 0.21162731921015623,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001975763449659894,
+      "loss": 4.5981,
+      "step": 2041
+    },
+    {
+      "epoch": 0.2117310072646443,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00019757396741636803,
+      "loss": 4.6128,
+      "step": 2042
+    },
+    {
+      "epoch": 0.21183469531913238,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019757158871547699,
+      "loss": 4.523,
+      "step": 2043
+    },
+    {
+      "epoch": 0.21193838337362048,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00019756920886334432,
+      "loss": 4.5688,
+      "step": 2044
+    },
+    {
+      "epoch": 0.21204207142810855,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00019756682785999812,
+      "loss": 4.5693,
+      "step": 2045
+    },
+    {
+      "epoch": 0.21214575948259662,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0001975644457054665,
+      "loss": 4.5807,
+      "step": 2046
+    },
+    {
+      "epoch": 0.2122494475370847,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00019756206239977751,
+      "loss": 4.5522,
+      "step": 2047
+    },
+    {
+      "epoch": 0.21235313559157276,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00019755967794295938,
+      "loss": 4.5546,
+      "step": 2048
+    },
+    {
+      "epoch": 0.21245682364606083,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.00019755729233504013,
+      "loss": 4.573,
+      "step": 2049
+    },
+    {
+      "epoch": 0.2125605117005489,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00019755490557604795,
+      "loss": 4.5695,
+      "step": 2050
+    },
+    {
+      "epoch": 0.21266419975503698,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.000197552517666011,
+      "loss": 4.5793,
+      "step": 2051
+    },
+    {
+      "epoch": 0.21276788780952505,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00019755012860495747,
+      "loss": 4.601,
+      "step": 2052
+    },
+    {
+      "epoch": 0.21287157586401312,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00019754773839291556,
+      "loss": 4.6343,
+      "step": 2053
+    },
+    {
+      "epoch": 0.2129752639185012,
+      "grad_norm": 1.96875,
+      "learning_rate": 0.00019754534702991344,
+      "loss": 4.6069,
+      "step": 2054
+    },
+    {
+      "epoch": 0.21307895197298926,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019754295451597932,
+      "loss": 4.5174,
+      "step": 2055
+    },
+    {
+      "epoch": 0.21318264002747733,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019754056085114144,
+      "loss": 4.5313,
+      "step": 2056
+    },
+    {
+      "epoch": 0.2132863280819654,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001975381660354281,
+      "loss": 4.5743,
+      "step": 2057
+    },
+    {
+      "epoch": 0.21339001613645348,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00019753577006886744,
+      "loss": 4.5637,
+      "step": 2058
+    },
+    {
+      "epoch": 0.21349370419094155,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001975333729514878,
+      "loss": 4.5662,
+      "step": 2059
+    },
+    {
+      "epoch": 0.21359739224542962,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00019753097468331747,
+      "loss": 4.5688,
+      "step": 2060
+    },
+    {
+      "epoch": 0.2137010802999177,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00019752857526438472,
+      "loss": 4.5697,
+      "step": 2061
+    },
+    {
+      "epoch": 0.21380476835440576,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0001975261746947179,
+      "loss": 4.6017,
+      "step": 2062
+    },
+    {
+      "epoch": 0.21390845640889383,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001975237729743453,
+      "loss": 4.5543,
+      "step": 2063
+    },
+    {
+      "epoch": 0.2140121444633819,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.00019752137010329527,
+      "loss": 4.5657,
+      "step": 2064
+    },
+    {
+      "epoch": 0.21411583251786998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00019751896608159614,
+      "loss": 4.5717,
+      "step": 2065
+    },
+    {
+      "epoch": 0.21421952057235807,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001975165609092763,
+      "loss": 4.5827,
+      "step": 2066
+    },
+    {
+      "epoch": 0.21432320862684615,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.00019751415458636414,
+      "loss": 4.5571,
+      "step": 2067
+    },
+    {
+      "epoch": 0.21442689668133422,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.000197511747112888,
+      "loss": 4.6079,
+      "step": 2068
+    },
+    {
+      "epoch": 0.2145305847358223,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00019750933848887634,
+      "loss": 4.5673,
+      "step": 2069
+    },
+    {
+      "epoch": 0.21463427279031036,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00019750692871435755,
+      "loss": 4.5605,
+      "step": 2070
+    },
+    {
+      "epoch": 0.21473796084479843,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00019750451778936007,
+      "loss": 4.5711,
+      "step": 2071
+    },
+    {
+      "epoch": 0.2148416488992865,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00019750210571391232,
+      "loss": 4.563,
+      "step": 2072
+    },
+    {
+      "epoch": 0.21494533695377457,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.00019749969248804283,
+      "loss": 4.5555,
+      "step": 2073
+    },
+    {
+      "epoch": 0.21504902500826265,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00019749727811178,
+      "loss": 4.589,
+      "step": 2074
+    },
+    {
+      "epoch": 0.21515271306275072,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019749486258515238,
+      "loss": 4.5742,
+      "step": 2075
+    },
+    {
+      "epoch": 0.2152564011172388,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00019749244590818842,
+      "loss": 4.5774,
+      "step": 2076
+    },
+    {
+      "epoch": 0.21536008917172686,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.00019749002808091667,
+      "loss": 4.5611,
+      "step": 2077
+    },
+    {
+      "epoch": 0.21546377722621493,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00019748760910336558,
+      "loss": 4.551,
+      "step": 2078
+    },
+    {
+      "epoch": 0.215567465280703,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00019748518897556383,
+      "loss": 4.6089,
+      "step": 2079
+    },
+    {
+      "epoch": 0.21567115333519107,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00019748276769753983,
+      "loss": 4.5747,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21577484138967915,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00019748034526932226,
+      "loss": 4.5606,
+      "step": 2081
+    },
+    {
+      "epoch": 0.21587852944416722,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00019747792169093963,
+      "loss": 4.5935,
+      "step": 2082
+    },
+    {
+      "epoch": 0.2159822174986553,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00019747549696242059,
+      "loss": 4.6101,
+      "step": 2083
+    },
+    {
+      "epoch": 0.21608590555314336,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.0001974730710837937,
+      "loss": 4.5796,
+      "step": 2084
+    },
+    {
+      "epoch": 0.21618959360763143,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00019747064405508763,
+      "loss": 4.5739,
+      "step": 2085
+    },
+    {
+      "epoch": 0.2162932816621195,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00019746821587633099,
+      "loss": 4.5887,
+      "step": 2086
+    },
+    {
+      "epoch": 0.21639696971660757,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.0001974657865475524,
+      "loss": 4.544,
+      "step": 2087
+    },
+    {
+      "epoch": 0.21650065777109567,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00019746335606878054,
+      "loss": 4.5873,
+      "step": 2088
+    },
+    {
+      "epoch": 0.21660434582558374,
+      "grad_norm": 1.625,
+      "learning_rate": 0.00019746092444004412,
+      "loss": 4.6125,
+      "step": 2089
+    },
+    {
+      "epoch": 0.21670803388007182,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00019745849166137185,
+      "loss": 4.5585,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2168117219345599,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00019745605773279236,
+      "loss": 4.5969,
+      "step": 2091
+    },
+    {
+      "epoch": 0.21691540998904796,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019745362265433442,
+      "loss": 4.5379,
+      "step": 2092
+    },
+    {
+      "epoch": 0.21701909804353603,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00019745118642602673,
+      "loss": 4.5126,
+      "step": 2093
+    },
+    {
+      "epoch": 0.2171227860980241,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00019744874904789806,
+      "loss": 4.5135,
+      "step": 2094
+    },
+    {
+      "epoch": 0.21722647415251217,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.00019744631051997718,
+      "loss": 4.5868,
+      "step": 2095
+    },
+    {
+      "epoch": 0.21733016220700024,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0001974438708422928,
+      "loss": 4.5675,
+      "step": 2096
+    },
+    {
+      "epoch": 0.21743385026148832,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00019744143001487378,
+      "loss": 4.5875,
+      "step": 2097
+    },
+    {
+      "epoch": 0.2175375383159764,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00019743898803774884,
+      "loss": 4.6141,
+      "step": 2098
+    },
+    {
+      "epoch": 0.21764122637046446,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0001974365449109469,
+      "loss": 4.5761,
+      "step": 2099
+    },
+    {
+      "epoch": 0.21774491442495253,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001974341006344967,
+      "loss": 4.5581,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2178486024794406,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001974316552084271,
+      "loss": 4.5847,
+      "step": 2101
+    },
+    {
+      "epoch": 0.21795229053392867,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00019742920863276693,
+      "loss": 4.5691,
+      "step": 2102
+    },
+    {
+      "epoch": 0.21805597858841674,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019742676090754512,
+      "loss": 4.5477,
+      "step": 2103
+    },
+    {
+      "epoch": 0.21815966664290481,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.0001974243120327905,
+      "loss": 4.5926,
+      "step": 2104
+    },
+    {
+      "epoch": 0.2182633546973929,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.000197421862008532,
+      "loss": 4.588,
+      "step": 2105
+    },
+    {
+      "epoch": 0.21836704275188096,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.0001974194108347985,
+      "loss": 4.5484,
+      "step": 2106
+    },
+    {
+      "epoch": 0.21847073080636903,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00019741695851161893,
+      "loss": 4.6028,
+      "step": 2107
+    },
+    {
+      "epoch": 0.2185744188608571,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00019741450503902222,
+      "loss": 4.6113,
+      "step": 2108
+    },
+    {
+      "epoch": 0.21867810691534517,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019741205041703733,
+      "loss": 4.5785,
+      "step": 2109
+    },
+    {
+      "epoch": 0.21878179496983324,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.0001974095946456932,
+      "loss": 4.5964,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21888548302432134,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.0001974071377250188,
+      "loss": 4.5467,
+      "step": 2111
+    },
+    {
+      "epoch": 0.2189891710788094,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001974046796550432,
+      "loss": 4.5953,
+      "step": 2112
+    },
+    {
+      "epoch": 0.21909285913329749,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019740222043579527,
+      "loss": 4.6259,
+      "step": 2113
+    },
+    {
+      "epoch": 0.21919654718778556,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019739976006730414,
+      "loss": 4.5235,
+      "step": 2114
+    },
+    {
+      "epoch": 0.21930023524227363,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001973972985495988,
+      "loss": 4.5858,
+      "step": 2115
+    },
+    {
+      "epoch": 0.2194039232967617,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.00019739483588270828,
+      "loss": 4.5866,
+      "step": 2116
+    },
+    {
+      "epoch": 0.21950761135124977,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00019739237206666164,
+      "loss": 4.6022,
+      "step": 2117
+    },
+    {
+      "epoch": 0.21961129940573784,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.00019738990710148796,
+      "loss": 4.6095,
+      "step": 2118
+    },
+    {
+      "epoch": 0.2197149874602259,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.00019738744098721632,
+      "loss": 4.5528,
+      "step": 2119
+    },
+    {
+      "epoch": 0.21981867551471398,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00019738497372387586,
+      "loss": 4.5543,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21992236356920206,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0001973825053114956,
+      "loss": 4.5641,
+      "step": 2121
+    },
+    {
+      "epoch": 0.22002605162369013,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00019738003575010474,
+      "loss": 4.5842,
+      "step": 2122
+    },
+    {
+      "epoch": 0.2201297396781782,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.0001973775650397324,
+      "loss": 4.5878,
+      "step": 2123
+    },
+    {
+      "epoch": 0.22023342773266627,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.0001973750931804077,
+      "loss": 4.5831,
+      "step": 2124
+    },
+    {
+      "epoch": 0.22033711578715434,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.00019737262017215982,
+      "loss": 4.5747,
+      "step": 2125
+    },
+    {
+      "epoch": 0.2204408038416424,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.000197370146015018,
+      "loss": 4.5351,
+      "step": 2126
+    },
+    {
+      "epoch": 0.22054449189613048,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00019736767070901133,
+      "loss": 4.5474,
+      "step": 2127
+    },
+    {
+      "epoch": 0.22064817995061856,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00019736519425416908,
+      "loss": 4.6022,
+      "step": 2128
+    },
+    {
+      "epoch": 0.22075186800510663,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019736271665052047,
+      "loss": 4.5988,
+      "step": 2129
+    },
+    {
+      "epoch": 0.2208555560595947,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00019736023789809472,
+      "loss": 4.5532,
+      "step": 2130
+    },
+    {
+      "epoch": 0.22095924411408277,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001973577579969211,
+      "loss": 4.5863,
+      "step": 2131
+    },
+    {
+      "epoch": 0.22106293216857084,
+      "grad_norm": 2.0,
+      "learning_rate": 0.0001973552769470288,
+      "loss": 4.5876,
+      "step": 2132
+    },
+    {
+      "epoch": 0.22116662022305894,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00019735279474844718,
+      "loss": 4.6137,
+      "step": 2133
+    },
+    {
+      "epoch": 0.221270308277547,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00019735031140120547,
+      "loss": 4.5873,
+      "step": 2134
+    },
+    {
+      "epoch": 0.22137399633203508,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00019734782690533298,
+      "loss": 4.605,
+      "step": 2135
+    },
+    {
+      "epoch": 0.22147768438652315,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00019734534126085904,
+      "loss": 4.603,
+      "step": 2136
+    },
+    {
+      "epoch": 0.22158137244101123,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00019734285446781297,
+      "loss": 4.5563,
+      "step": 2137
+    },
+    {
+      "epoch": 0.2216850604954993,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.00019734036652622412,
+      "loss": 4.5683,
+      "step": 2138
+    },
+    {
+      "epoch": 0.22178874854998737,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019733787743612185,
+      "loss": 4.5973,
+      "step": 2139
+    },
+    {
+      "epoch": 0.22189243660447544,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00019733538719753552,
+      "loss": 4.6036,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2219961246589635,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019733289581049448,
+      "loss": 4.5687,
+      "step": 2141
+    },
+    {
+      "epoch": 0.22209981271345158,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019733040327502815,
+      "loss": 4.6214,
+      "step": 2142
+    },
+    {
+      "epoch": 0.22220350076793965,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.000197327909591166,
+      "loss": 4.5759,
+      "step": 2143
+    },
+    {
+      "epoch": 0.22230718882242773,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.00019732541475893733,
+      "loss": 4.5674,
+      "step": 2144
+    },
+    {
+      "epoch": 0.2224108768769158,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00019732291877837165,
+      "loss": 4.5504,
+      "step": 2145
+    },
+    {
+      "epoch": 0.22251456493140387,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00019732042164949845,
+      "loss": 4.5623,
+      "step": 2146
+    },
+    {
+      "epoch": 0.22261825298589194,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001973179233723471,
+      "loss": 4.5639,
+      "step": 2147
+    },
+    {
+      "epoch": 0.22272194104038,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00019731542394694715,
+      "loss": 4.5888,
+      "step": 2148
+    },
+    {
+      "epoch": 0.22282562909486808,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00019731292337332807,
+      "loss": 4.5787,
+      "step": 2149
+    },
+    {
+      "epoch": 0.22292931714935615,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00019731042165151936,
+      "loss": 4.5718,
+      "step": 2150
+    },
+    {
+      "epoch": 0.22303300520384423,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019730791878155052,
+      "loss": 4.5511,
+      "step": 2151
+    },
+    {
+      "epoch": 0.2231366932583323,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.0001973054147634511,
+      "loss": 4.588,
+      "step": 2152
+    },
+    {
+      "epoch": 0.22324038131282037,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00019730290959725063,
+      "loss": 4.5352,
+      "step": 2153
+    },
+    {
+      "epoch": 0.22334406936730844,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.0001973004032829787,
+      "loss": 4.5747,
+      "step": 2154
+    },
+    {
+      "epoch": 0.22344775742179654,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00019729789582066486,
+      "loss": 4.5886,
+      "step": 2155
+    },
+    {
+      "epoch": 0.2235514454762846,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.0001972953872103387,
+      "loss": 4.5761,
+      "step": 2156
+    },
+    {
+      "epoch": 0.22365513353077268,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001972928774520298,
+      "loss": 4.5831,
+      "step": 2157
+    },
+    {
+      "epoch": 0.22375882158526075,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.0001972903665457678,
+      "loss": 4.5881,
+      "step": 2158
+    },
+    {
+      "epoch": 0.22386250963974882,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019728785449158232,
+      "loss": 4.5763,
+      "step": 2159
+    },
+    {
+      "epoch": 0.2239661976942369,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00019728534128950299,
+      "loss": 4.5638,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22406988574872497,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019728282693955946,
+      "loss": 4.5464,
+      "step": 2161
+    },
+    {
+      "epoch": 0.22417357380321304,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00019728031144178142,
+      "loss": 4.589,
+      "step": 2162
+    },
+    {
+      "epoch": 0.2242772618577011,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00019727779479619852,
+      "loss": 4.5857,
+      "step": 2163
+    },
+    {
+      "epoch": 0.22438094991218918,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.00019727527700284046,
+      "loss": 4.5437,
+      "step": 2164
+    },
+    {
+      "epoch": 0.22448463796667725,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00019727275806173696,
+      "loss": 4.573,
+      "step": 2165
+    },
+    {
+      "epoch": 0.22458832602116532,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019727023797291778,
+      "loss": 4.5802,
+      "step": 2166
+    },
+    {
+      "epoch": 0.2246920140756534,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00019726771673641256,
+      "loss": 4.5805,
+      "step": 2167
+    },
+    {
+      "epoch": 0.22479570213014147,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.00019726519435225113,
+      "loss": 4.593,
+      "step": 2168
+    },
+    {
+      "epoch": 0.22489939018462954,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.0001972626708204632,
+      "loss": 4.5521,
+      "step": 2169
+    },
+    {
+      "epoch": 0.2250030782391176,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00019726014614107856,
+      "loss": 4.5509,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22510676629360568,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00019725762031412702,
+      "loss": 4.5501,
+      "step": 2171
+    },
+    {
+      "epoch": 0.22521045434809375,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.0001972550933396384,
+      "loss": 4.6034,
+      "step": 2172
+    },
+    {
+      "epoch": 0.22531414240258182,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001972525652176424,
+      "loss": 4.5989,
+      "step": 2173
+    },
+    {
+      "epoch": 0.2254178304570699,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.000197250035948169,
+      "loss": 4.6182,
+      "step": 2174
+    },
+    {
+      "epoch": 0.22552151851155797,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019724750553124794,
+      "loss": 4.54,
+      "step": 2175
+    },
+    {
+      "epoch": 0.22562520656604604,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00019724497396690911,
+      "loss": 4.6102,
+      "step": 2176
+    },
+    {
+      "epoch": 0.2257288946205341,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0001972424412551824,
+      "loss": 4.6087,
+      "step": 2177
+    },
+    {
+      "epoch": 0.2258325826750222,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00019723990739609765,
+      "loss": 4.5766,
+      "step": 2178
+    },
+    {
+      "epoch": 0.22593627072951028,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001972373723896848,
+      "loss": 4.5922,
+      "step": 2179
+    },
+    {
+      "epoch": 0.22603995878399835,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0001972348362359737,
+      "loss": 4.6109,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22614364683848642,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019723229893499436,
+      "loss": 4.5587,
+      "step": 2181
+    },
+    {
+      "epoch": 0.2262473348929745,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00019722976048677668,
+      "loss": 4.6016,
+      "step": 2182
+    },
+    {
+      "epoch": 0.22635102294746257,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00019722722089135058,
+      "loss": 4.6004,
+      "step": 2183
+    },
+    {
+      "epoch": 0.22645471100195064,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00019722468014874602,
+      "loss": 4.5439,
+      "step": 2184
+    },
+    {
+      "epoch": 0.2265583990564387,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00019722213825899306,
+      "loss": 4.538,
+      "step": 2185
+    },
+    {
+      "epoch": 0.22666208711092678,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0001972195952221216,
+      "loss": 4.5763,
+      "step": 2186
+    },
+    {
+      "epoch": 0.22676577516541485,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019721705103816167,
+      "loss": 4.5971,
+      "step": 2187
+    },
+    {
+      "epoch": 0.22686946321990292,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00019721450570714332,
+      "loss": 4.5941,
+      "step": 2188
+    },
+    {
+      "epoch": 0.226973151274391,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019721195922909658,
+      "loss": 4.5869,
+      "step": 2189
+    },
+    {
+      "epoch": 0.22707683932887907,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00019720941160405146,
+      "loss": 4.5875,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22718052738336714,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019720686283203803,
+      "loss": 4.5523,
+      "step": 2191
+    },
+    {
+      "epoch": 0.2272842154378552,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.0001972043129130864,
+      "loss": 4.5619,
+      "step": 2192
+    },
+    {
+      "epoch": 0.22738790349234328,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019720176184722662,
+      "loss": 4.55,
+      "step": 2193
+    },
+    {
+      "epoch": 0.22749159154683135,
+      "grad_norm": 1.9375,
+      "learning_rate": 0.00019719920963448876,
+      "loss": 4.5597,
+      "step": 2194
+    },
+    {
+      "epoch": 0.22759527960131942,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.000197196656274903,
+      "loss": 4.5254,
+      "step": 2195
+    },
+    {
+      "epoch": 0.2276989676558075,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00019719410176849943,
+      "loss": 4.5352,
+      "step": 2196
+    },
+    {
+      "epoch": 0.22780265571029557,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.0001971915461153082,
+      "loss": 4.5494,
+      "step": 2197
+    },
+    {
+      "epoch": 0.22790634376478364,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019718898931535948,
+      "loss": 4.5682,
+      "step": 2198
+    },
+    {
+      "epoch": 0.2280100318192717,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001971864313686834,
+      "loss": 4.562,
+      "step": 2199
+    },
+    {
+      "epoch": 0.2281137198737598,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00019718387227531014,
+      "loss": 4.5403,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22821740792824788,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019718131203526996,
+      "loss": 4.5693,
+      "step": 2201
+    },
+    {
+      "epoch": 0.22832109598273595,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00019717875064859298,
+      "loss": 4.5906,
+      "step": 2202
+    },
+    {
+      "epoch": 0.22842478403722402,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001971761881153095,
+      "loss": 4.535,
+      "step": 2203
+    },
+    {
+      "epoch": 0.2285284720917121,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.0001971736244354497,
+      "loss": 4.5743,
+      "step": 2204
+    },
+    {
+      "epoch": 0.22863216014620016,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019717105960904386,
+      "loss": 4.5418,
+      "step": 2205
+    },
+    {
+      "epoch": 0.22873584820068824,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.00019716849363612222,
+      "loss": 4.581,
+      "step": 2206
+    },
+    {
+      "epoch": 0.2288395362551763,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00019716592651671506,
+      "loss": 4.5945,
+      "step": 2207
+    },
+    {
+      "epoch": 0.22894322430966438,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.00019716335825085269,
+      "loss": 4.5853,
+      "step": 2208
+    },
+    {
+      "epoch": 0.22904691236415245,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.0001971607888385654,
+      "loss": 4.574,
+      "step": 2209
+    },
+    {
+      "epoch": 0.22915060041864052,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0001971582182798835,
+      "loss": 4.5358,
+      "step": 2210
+    },
+    {
+      "epoch": 0.2292542884731286,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001971556465748373,
+      "loss": 4.5675,
+      "step": 2211
+    },
+    {
+      "epoch": 0.22935797652761666,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.0001971530737234572,
+      "loss": 4.5541,
+      "step": 2212
+    },
+    {
+      "epoch": 0.22946166458210474,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00019715049972577353,
+      "loss": 4.5301,
+      "step": 2213
+    },
+    {
+      "epoch": 0.2295653526365928,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00019714792458181663,
+      "loss": 4.5643,
+      "step": 2214
+    },
+    {
+      "epoch": 0.22966904069108088,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00019714534829161693,
+      "loss": 4.5358,
+      "step": 2215
+    },
+    {
+      "epoch": 0.22977272874556895,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0001971427708552048,
+      "loss": 4.5537,
+      "step": 2216
+    },
+    {
+      "epoch": 0.22987641680005702,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00019714019227261067,
+      "loss": 4.593,
+      "step": 2217
+    },
+    {
+      "epoch": 0.2299801048545451,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00019713761254386495,
+      "loss": 4.5416,
+      "step": 2218
+    },
+    {
+      "epoch": 0.23008379290903316,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00019713503166899807,
+      "loss": 4.5677,
+      "step": 2219
+    },
+    {
+      "epoch": 0.23018748096352123,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.0001971324496480405,
+      "loss": 4.5802,
+      "step": 2220
+    },
+    {
+      "epoch": 0.2302911690180093,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0001971298664810227,
+      "loss": 4.516,
+      "step": 2221
+    },
+    {
+      "epoch": 0.2303948570724974,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00019712728216797514,
+      "loss": 4.5601,
+      "step": 2222
+    },
+    {
+      "epoch": 0.23049854512698548,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001971246967089283,
+      "loss": 4.5543,
+      "step": 2223
+    },
+    {
+      "epoch": 0.23060223318147355,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00019712211010391274,
+      "loss": 4.5348,
+      "step": 2224
+    },
+    {
+      "epoch": 0.23070592123596162,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001971195223529589,
+      "loss": 4.5637,
+      "step": 2225
+    },
+    {
+      "epoch": 0.2308096092904497,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.00019711693345609739,
+      "loss": 4.608,
+      "step": 2226
+    },
+    {
+      "epoch": 0.23091329734493776,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.0001971143434133587,
+      "loss": 4.5851,
+      "step": 2227
+    },
+    {
+      "epoch": 0.23101698539942583,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00019711175222477344,
+      "loss": 4.5839,
+      "step": 2228
+    },
+    {
+      "epoch": 0.2311206734539139,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00019710915989037213,
+      "loss": 4.5727,
+      "step": 2229
+    },
+    {
+      "epoch": 0.23122436150840198,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0001971065664101854,
+      "loss": 4.5646,
+      "step": 2230
+    },
+    {
+      "epoch": 0.23132804956289005,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00019710397178424383,
+      "loss": 4.5607,
+      "step": 2231
+    },
+    {
+      "epoch": 0.23143173761737812,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019710137601257804,
+      "loss": 4.5645,
+      "step": 2232
+    },
+    {
+      "epoch": 0.2315354256718662,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00019709877909521864,
+      "loss": 4.5555,
+      "step": 2233
+    },
+    {
+      "epoch": 0.23163911372635426,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001970961810321963,
+      "loss": 4.5427,
+      "step": 2234
+    },
+    {
+      "epoch": 0.23174280178084233,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00019709358182354162,
+      "loss": 4.5674,
+      "step": 2235
+    },
+    {
+      "epoch": 0.2318464898353304,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00019709098146928535,
+      "loss": 4.53,
+      "step": 2236
+    },
+    {
+      "epoch": 0.23195017788981848,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0001970883799694581,
+      "loss": 4.5445,
+      "step": 2237
+    },
+    {
+      "epoch": 0.23205386594430655,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00019708577732409062,
+      "loss": 4.5639,
+      "step": 2238
+    },
+    {
+      "epoch": 0.23215755399879462,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019708317353321357,
+      "loss": 4.5738,
+      "step": 2239
+    },
+    {
+      "epoch": 0.2322612420532827,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0001970805685968577,
+      "loss": 4.5423,
+      "step": 2240
+    },
+    {
+      "epoch": 0.23236493010777076,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00019707796251505375,
+      "loss": 4.5786,
+      "step": 2241
+    },
+    {
+      "epoch": 0.23246861816225883,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00019707535528783244,
+      "loss": 4.562,
+      "step": 2242
+    },
+    {
+      "epoch": 0.2325723062167469,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00019707274691522456,
+      "loss": 4.547,
+      "step": 2243
+    },
+    {
+      "epoch": 0.23267599427123498,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.0001970701373972609,
+      "loss": 4.5492,
+      "step": 2244
+    },
+    {
+      "epoch": 0.23277968232572308,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.00019706752673397218,
+      "loss": 4.5297,
+      "step": 2245
+    },
+    {
+      "epoch": 0.23288337038021115,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00019706491492538927,
+      "loss": 4.5933,
+      "step": 2246
+    },
+    {
+      "epoch": 0.23298705843469922,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.00019706230197154298,
+      "loss": 4.5682,
+      "step": 2247
+    },
+    {
+      "epoch": 0.2330907464891873,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00019705968787246412,
+      "loss": 4.5609,
+      "step": 2248
+    },
+    {
+      "epoch": 0.23319443454367536,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00019705707262818354,
+      "loss": 4.5367,
+      "step": 2249
+    },
+    {
+      "epoch": 0.23329812259816343,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0001970544562387321,
+      "loss": 4.5668,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2334018106526515,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00019705183870414062,
+      "loss": 4.5505,
+      "step": 2251
+    },
+    {
+      "epoch": 0.23350549870713957,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019704922002444008,
+      "loss": 4.5626,
+      "step": 2252
+    },
+    {
+      "epoch": 0.23360918676162765,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019704660019966133,
+      "loss": 4.5656,
+      "step": 2253
+    },
+    {
+      "epoch": 0.23371287481611572,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019704397922983526,
+      "loss": 4.5666,
+      "step": 2254
+    },
+    {
+      "epoch": 0.2338165628706038,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019704135711499286,
+      "loss": 4.5267,
+      "step": 2255
+    },
+    {
+      "epoch": 0.23392025092509186,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019703873385516497,
+      "loss": 4.5764,
+      "step": 2256
+    },
+    {
+      "epoch": 0.23402393897957993,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001970361094503826,
+      "loss": 4.5261,
+      "step": 2257
+    },
+    {
+      "epoch": 0.234127627034068,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019703348390067674,
+      "loss": 4.5832,
+      "step": 2258
+    },
+    {
+      "epoch": 0.23423131508855607,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001970308572060783,
+      "loss": 4.5381,
+      "step": 2259
+    },
+    {
+      "epoch": 0.23433500314304415,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019702822936661836,
+      "loss": 4.5449,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23443869119753222,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019702560038232782,
+      "loss": 4.5032,
+      "step": 2261
+    },
+    {
+      "epoch": 0.2345423792520203,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0001970229702532378,
+      "loss": 4.5321,
+      "step": 2262
+    },
+    {
+      "epoch": 0.23464606730650836,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019702033897937927,
+      "loss": 4.5649,
+      "step": 2263
+    },
+    {
+      "epoch": 0.23474975536099643,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001970177065607833,
+      "loss": 4.6028,
+      "step": 2264
+    },
+    {
+      "epoch": 0.2348534434154845,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019701507299748095,
+      "loss": 4.539,
+      "step": 2265
+    },
+    {
+      "epoch": 0.23495713146997257,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019701243828950329,
+      "loss": 4.5593,
+      "step": 2266
+    },
+    {
+      "epoch": 0.23506081952446067,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001970098024368814,
+      "loss": 4.5882,
+      "step": 2267
+    },
+    {
+      "epoch": 0.23516450757894874,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019700716543964638,
+      "loss": 4.5798,
+      "step": 2268
+    },
+    {
+      "epoch": 0.23526819563343682,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019700452729782934,
+      "loss": 4.5523,
+      "step": 2269
+    },
+    {
+      "epoch": 0.2353718836879249,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001970018880114614,
+      "loss": 4.5734,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23547557174241296,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019699924758057377,
+      "loss": 4.5664,
+      "step": 2271
+    },
+    {
+      "epoch": 0.23557925979690103,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019699660600519753,
+      "loss": 4.557,
+      "step": 2272
+    },
+    {
+      "epoch": 0.2356829478513891,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019699396328536384,
+      "loss": 4.5388,
+      "step": 2273
+    },
+    {
+      "epoch": 0.23578663590587717,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019699131942110397,
+      "loss": 4.5299,
+      "step": 2274
+    },
+    {
+      "epoch": 0.23589032396036524,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.000196988674412449,
+      "loss": 4.5497,
+      "step": 2275
+    },
+    {
+      "epoch": 0.23599401201485332,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001969860282594302,
+      "loss": 4.5566,
+      "step": 2276
+    },
+    {
+      "epoch": 0.2360977000693414,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019698338096207883,
+      "loss": 4.5407,
+      "step": 2277
+    },
+    {
+      "epoch": 0.23620138812382946,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019698073252042605,
+      "loss": 4.5606,
+      "step": 2278
+    },
+    {
+      "epoch": 0.23630507617831753,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019697808293450312,
+      "loss": 4.5585,
+      "step": 2279
+    },
+    {
+      "epoch": 0.2364087642328056,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019697543220434133,
+      "loss": 4.5512,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23651245228729367,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019697278032997198,
+      "loss": 4.5865,
+      "step": 2281
+    },
+    {
+      "epoch": 0.23661614034178174,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001969701273114263,
+      "loss": 4.5634,
+      "step": 2282
+    },
+    {
+      "epoch": 0.23671982839626982,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001969674731487356,
+      "loss": 4.5198,
+      "step": 2283
+    },
+    {
+      "epoch": 0.2368235164507579,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019696481784193127,
+      "loss": 4.4959,
+      "step": 2284
+    },
+    {
+      "epoch": 0.23692720450524596,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019696216139104453,
+      "loss": 4.5441,
+      "step": 2285
+    },
+    {
+      "epoch": 0.23703089255973403,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019695950379610682,
+      "loss": 4.5769,
+      "step": 2286
+    },
+    {
+      "epoch": 0.2371345806142221,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019695684505714942,
+      "loss": 4.5464,
+      "step": 2287
+    },
+    {
+      "epoch": 0.23723826866871017,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019695418517420377,
+      "loss": 4.545,
+      "step": 2288
+    },
+    {
+      "epoch": 0.23734195672319827,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001969515241473012,
+      "loss": 4.5357,
+      "step": 2289
+    },
+    {
+      "epoch": 0.23744564477768634,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019694886197647312,
+      "loss": 4.5784,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23754933283217441,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019694619866175098,
+      "loss": 4.5894,
+      "step": 2291
+    },
+    {
+      "epoch": 0.23765302088666249,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019694353420316615,
+      "loss": 4.5586,
+      "step": 2292
+    },
+    {
+      "epoch": 0.23775670894115056,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0001969408686007501,
+      "loss": 4.5119,
+      "step": 2293
+    },
+    {
+      "epoch": 0.23786039699563863,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019693820185453427,
+      "loss": 4.5694,
+      "step": 2294
+    },
+    {
+      "epoch": 0.2379640850501267,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019693553396455012,
+      "loss": 4.5731,
+      "step": 2295
+    },
+    {
+      "epoch": 0.23806777310461477,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0001969328649308291,
+      "loss": 4.5255,
+      "step": 2296
+    },
+    {
+      "epoch": 0.23817146115910284,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0001969301947534028,
+      "loss": 4.5547,
+      "step": 2297
+    },
+    {
+      "epoch": 0.23827514921359091,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019692752343230264,
+      "loss": 4.5426,
+      "step": 2298
+    },
+    {
+      "epoch": 0.23837883726807899,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019692485096756016,
+      "loss": 4.551,
+      "step": 2299
+    },
+    {
+      "epoch": 0.23848252532256706,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001969221773592069,
+      "loss": 4.5627,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23858621337705513,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019691950260727437,
+      "loss": 4.5379,
+      "step": 2301
+    },
+    {
+      "epoch": 0.2386899014315432,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019691682671179415,
+      "loss": 4.5418,
+      "step": 2302
+    },
+    {
+      "epoch": 0.23879358948603127,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019691414967279786,
+      "loss": 4.5495,
+      "step": 2303
+    },
+    {
+      "epoch": 0.23889727754051934,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019691147149031703,
+      "loss": 4.557,
+      "step": 2304
+    },
+    {
+      "epoch": 0.23900096559500741,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019690879216438325,
+      "loss": 4.5889,
+      "step": 2305
+    },
+    {
+      "epoch": 0.23910465364949549,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0001969061116950282,
+      "loss": 4.539,
+      "step": 2306
+    },
+    {
+      "epoch": 0.23920834170398356,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019690343008228343,
+      "loss": 4.5705,
+      "step": 2307
+    },
+    {
+      "epoch": 0.23931202975847163,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019690074732618066,
+      "loss": 4.5375,
+      "step": 2308
+    },
+    {
+      "epoch": 0.2394157178129597,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019689806342675147,
+      "loss": 4.5483,
+      "step": 2309
+    },
+    {
+      "epoch": 0.23951940586744777,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019689537838402758,
+      "loss": 4.5793,
+      "step": 2310
+    },
+    {
+      "epoch": 0.23962309392193584,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001968926921980406,
+      "loss": 4.5669,
+      "step": 2311
+    },
+    {
+      "epoch": 0.23972678197642394,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019689000486882235,
+      "loss": 4.5548,
+      "step": 2312
+    },
+    {
+      "epoch": 0.239830470030912,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019688731639640438,
+      "loss": 4.5066,
+      "step": 2313
+    },
+    {
+      "epoch": 0.23993415808540008,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019688462678081852,
+      "loss": 4.5549,
+      "step": 2314
+    },
+    {
+      "epoch": 0.24003784613988816,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019688193602209652,
+      "loss": 4.5162,
+      "step": 2315
+    },
+    {
+      "epoch": 0.24014153419437623,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019687924412027004,
+      "loss": 4.5942,
+      "step": 2316
+    },
+    {
+      "epoch": 0.2402452222488643,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019687655107537087,
+      "loss": 4.5106,
+      "step": 2317
+    },
+    {
+      "epoch": 0.24034891030335237,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019687385688743083,
+      "loss": 4.5898,
+      "step": 2318
+    },
+    {
+      "epoch": 0.24045259835784044,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019687116155648167,
+      "loss": 4.5338,
+      "step": 2319
+    },
+    {
+      "epoch": 0.2405562864123285,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019686846508255518,
+      "loss": 4.5561,
+      "step": 2320
+    },
+    {
+      "epoch": 0.24065997446681658,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019686576746568321,
+      "loss": 4.5322,
+      "step": 2321
+    },
+    {
+      "epoch": 0.24076366252130466,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001968630687058976,
+      "loss": 4.556,
+      "step": 2322
+    },
+    {
+      "epoch": 0.24086735057579273,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00019686036880323012,
+      "loss": 4.5641,
+      "step": 2323
+    },
+    {
+      "epoch": 0.2409710386302808,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019685766775771272,
+      "loss": 4.541,
+      "step": 2324
+    },
+    {
+      "epoch": 0.24107472668476887,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019685496556937722,
+      "loss": 4.5167,
+      "step": 2325
+    },
+    {
+      "epoch": 0.24117841473925694,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001968522622382555,
+      "loss": 4.5524,
+      "step": 2326
+    },
+    {
+      "epoch": 0.241282102793745,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00019684955776437947,
+      "loss": 4.5259,
+      "step": 2327
+    },
+    {
+      "epoch": 0.24138579084823308,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.000196846852147781,
+      "loss": 4.5491,
+      "step": 2328
+    },
+    {
+      "epoch": 0.24148947890272116,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019684414538849207,
+      "loss": 4.5132,
+      "step": 2329
+    },
+    {
+      "epoch": 0.24159316695720923,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019684143748654458,
+      "loss": 4.5794,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2416968550116973,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019683872844197052,
+      "loss": 4.5889,
+      "step": 2331
+    },
+    {
+      "epoch": 0.24180054306618537,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001968360182548018,
+      "loss": 4.5713,
+      "step": 2332
+    },
+    {
+      "epoch": 0.24190423112067344,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019683330692507042,
+      "loss": 4.5571,
+      "step": 2333
+    },
+    {
+      "epoch": 0.24200791917516154,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019683059445280837,
+      "loss": 4.5484,
+      "step": 2334
+    },
+    {
+      "epoch": 0.2421116072296496,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019682788083804771,
+      "loss": 4.538,
+      "step": 2335
+    },
+    {
+      "epoch": 0.24221529528413768,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019682516608082037,
+      "loss": 4.5748,
+      "step": 2336
+    },
+    {
+      "epoch": 0.24231898333862575,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019682245018115842,
+      "loss": 4.5281,
+      "step": 2337
+    },
+    {
+      "epoch": 0.24242267139311383,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0001968197331390939,
+      "loss": 4.5424,
+      "step": 2338
+    },
+    {
+      "epoch": 0.2425263594476019,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019681701495465889,
+      "loss": 4.5263,
+      "step": 2339
+    },
+    {
+      "epoch": 0.24263004750208997,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019681429562788542,
+      "loss": 4.5765,
+      "step": 2340
+    },
+    {
+      "epoch": 0.24273373555657804,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019681157515880564,
+      "loss": 4.4784,
+      "step": 2341
+    },
+    {
+      "epoch": 0.2428374236110661,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019680885354745158,
+      "loss": 4.542,
+      "step": 2342
+    },
+    {
+      "epoch": 0.24294111166555418,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019680613079385537,
+      "loss": 4.5242,
+      "step": 2343
+    },
+    {
+      "epoch": 0.24304479972004225,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019680340689804914,
+      "loss": 4.5195,
+      "step": 2344
+    },
+    {
+      "epoch": 0.24314848777453033,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019680068186006506,
+      "loss": 4.5474,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2432521758290184,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00019679795567993527,
+      "loss": 4.5701,
+      "step": 2346
+    },
+    {
+      "epoch": 0.24335586388350647,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019679522835769188,
+      "loss": 4.5311,
+      "step": 2347
+    },
+    {
+      "epoch": 0.24345955193799454,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019679249989336715,
+      "loss": 4.5522,
+      "step": 2348
+    },
+    {
+      "epoch": 0.2435632399924826,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019678977028699318,
+      "loss": 4.5616,
+      "step": 2349
+    },
+    {
+      "epoch": 0.24366692804697068,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001967870395386023,
+      "loss": 4.5,
+      "step": 2350
+    },
+    {
+      "epoch": 0.24377061610145875,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019678430764822661,
+      "loss": 4.5487,
+      "step": 2351
+    },
+    {
+      "epoch": 0.24387430415594682,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00019678157461589844,
+      "loss": 4.522,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2439779922104349,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019677884044164997,
+      "loss": 4.5293,
+      "step": 2353
+    },
+    {
+      "epoch": 0.24408168026492297,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019677610512551348,
+      "loss": 4.5629,
+      "step": 2354
+    },
+    {
+      "epoch": 0.24418536831941104,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019677336866752123,
+      "loss": 4.5529,
+      "step": 2355
+    },
+    {
+      "epoch": 0.24428905637389914,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019677063106770555,
+      "loss": 4.5488,
+      "step": 2356
+    },
+    {
+      "epoch": 0.2443927444283872,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019676789232609868,
+      "loss": 4.5474,
+      "step": 2357
+    },
+    {
+      "epoch": 0.24449643248287528,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.000196765152442733,
+      "loss": 4.5178,
+      "step": 2358
+    },
+    {
+      "epoch": 0.24460012053736335,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001967624114176408,
+      "loss": 4.5141,
+      "step": 2359
+    },
+    {
+      "epoch": 0.24470380859185142,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019675966925085443,
+      "loss": 4.4834,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2448074966463395,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00019675692594240624,
+      "loss": 4.5927,
+      "step": 2361
+    },
+    {
+      "epoch": 0.24491118470082757,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001967541814923286,
+      "loss": 4.5491,
+      "step": 2362
+    },
+    {
+      "epoch": 0.24501487275531564,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00019675143590065387,
+      "loss": 4.5038,
+      "step": 2363
+    },
+    {
+      "epoch": 0.2451185608098037,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00019674868916741452,
+      "loss": 4.5872,
+      "step": 2364
+    },
+    {
+      "epoch": 0.24522224886429178,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019674594129264286,
+      "loss": 4.5441,
+      "step": 2365
+    },
+    {
+      "epoch": 0.24532593691877985,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.0001967431922763714,
+      "loss": 4.5074,
+      "step": 2366
+    },
+    {
+      "epoch": 0.24542962497326792,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019674044211863247,
+      "loss": 4.4877,
+      "step": 2367
+    },
+    {
+      "epoch": 0.245533313027756,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019673769081945863,
+      "loss": 4.527,
+      "step": 2368
+    },
+    {
+      "epoch": 0.24563700108224407,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019673493837888228,
+      "loss": 4.5584,
+      "step": 2369
+    },
+    {
+      "epoch": 0.24574068913673214,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019673218479693592,
+      "loss": 4.5505,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2458443771912202,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019672943007365202,
+      "loss": 4.5845,
+      "step": 2371
+    },
+    {
+      "epoch": 0.24594806524570828,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019672667420906308,
+      "loss": 4.5538,
+      "step": 2372
+    },
+    {
+      "epoch": 0.24605175330019635,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019672391720320165,
+      "loss": 4.523,
+      "step": 2373
+    },
+    {
+      "epoch": 0.24615544135468442,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019672115905610023,
+      "loss": 4.5312,
+      "step": 2374
+    },
+    {
+      "epoch": 0.2462591294091725,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019671839976779138,
+      "loss": 4.5019,
+      "step": 2375
+    },
+    {
+      "epoch": 0.24636281746366057,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019671563933830767,
+      "loss": 4.5387,
+      "step": 2376
+    },
+    {
+      "epoch": 0.24646650551814864,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001967128777676816,
+      "loss": 4.5441,
+      "step": 2377
+    },
+    {
+      "epoch": 0.2465701935726367,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019671011505594581,
+      "loss": 4.5023,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2466738816271248,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001967073512031329,
+      "loss": 4.5643,
+      "step": 2379
+    },
+    {
+      "epoch": 0.24677756968161288,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019670458620927548,
+      "loss": 4.5538,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24688125773610095,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019670182007440614,
+      "loss": 4.583,
+      "step": 2381
+    },
+    {
+      "epoch": 0.24698494579058902,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0001966990527985576,
+      "loss": 4.5283,
+      "step": 2382
+    },
+    {
+      "epoch": 0.2470886338450771,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001966962843817624,
+      "loss": 4.5397,
+      "step": 2383
+    },
+    {
+      "epoch": 0.24719232189956516,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019669351482405324,
+      "loss": 4.5366,
+      "step": 2384
+    },
+    {
+      "epoch": 0.24729600995405324,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019669074412546284,
+      "loss": 4.5615,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2473996980085413,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001966879722860239,
+      "loss": 4.5026,
+      "step": 2386
+    },
+    {
+      "epoch": 0.24750338606302938,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019668519930576904,
+      "loss": 4.5545,
+      "step": 2387
+    },
+    {
+      "epoch": 0.24760707411751745,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019668242518473106,
+      "loss": 4.5279,
+      "step": 2388
+    },
+    {
+      "epoch": 0.24771076217200552,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019667964992294264,
+      "loss": 4.5575,
+      "step": 2389
+    },
+    {
+      "epoch": 0.2478144502264936,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019667687352043655,
+      "loss": 4.5308,
+      "step": 2390
+    },
+    {
+      "epoch": 0.24791813828098166,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019667409597724553,
+      "loss": 4.5585,
+      "step": 2391
+    },
+    {
+      "epoch": 0.24802182633546974,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001966713172934024,
+      "loss": 4.5285,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2481255143899578,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019666853746893987,
+      "loss": 4.4994,
+      "step": 2393
+    },
+    {
+      "epoch": 0.24822920244444588,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019666575650389084,
+      "loss": 4.491,
+      "step": 2394
+    },
+    {
+      "epoch": 0.24833289049893395,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.000196662974398288,
+      "loss": 4.4838,
+      "step": 2395
+    },
+    {
+      "epoch": 0.24843657855342202,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0001966601911521643,
+      "loss": 4.5466,
+      "step": 2396
+    },
+    {
+      "epoch": 0.2485402666079101,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019665740676555246,
+      "loss": 4.5693,
+      "step": 2397
+    },
+    {
+      "epoch": 0.24864395466239816,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019665462123848545,
+      "loss": 4.5713,
+      "step": 2398
+    },
+    {
+      "epoch": 0.24874764271688624,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019665183457099602,
+      "loss": 4.5254,
+      "step": 2399
+    },
+    {
+      "epoch": 0.2488513307713743,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019664904676311716,
+      "loss": 4.4666,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2489550188258624,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019664625781488167,
+      "loss": 4.4776,
+      "step": 2401
+    },
+    {
+      "epoch": 0.24905870688035048,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019664346772632252,
+      "loss": 4.5156,
+      "step": 2402
+    },
+    {
+      "epoch": 0.24916239493483855,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001966406764974726,
+      "loss": 4.5436,
+      "step": 2403
+    },
+    {
+      "epoch": 0.24926608298932662,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019663788412836483,
+      "loss": 4.4748,
+      "step": 2404
+    },
+    {
+      "epoch": 0.2493697710438147,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0001966350906190322,
+      "loss": 4.5189,
+      "step": 2405
+    },
+    {
+      "epoch": 0.24947345909830276,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019663229596950766,
+      "loss": 4.4902,
+      "step": 2406
+    },
+    {
+      "epoch": 0.24957714715279083,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019662950017982416,
+      "loss": 4.5363,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2496808352072789,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019662670325001468,
+      "loss": 4.5453,
+      "step": 2408
+    },
+    {
+      "epoch": 0.24978452326176698,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019662390518011228,
+      "loss": 4.5477,
+      "step": 2409
+    },
+    {
+      "epoch": 0.24988821131625505,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.0001966211059701499,
+      "loss": 4.5409,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24999189937074312,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0001966183056201606,
+      "loss": 4.525,
+      "step": 2411
+    },
+    {
+      "epoch": 0.24999189937074312,
+      "eval_loss": 4.54582405090332,
+      "eval_runtime": 0.4393,
+      "eval_samples_per_second": 339.196,
+      "eval_steps_per_second": 15.935,
+      "step": 2411
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 28932,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 2411,
+  "total_flos": 3.4729528980175585e+18,
+  "train_batch_size": 3,
+  "trial_name": null,
+  "trial_params": null
+}