diff --git "a/checkpoint-630/trainer_state.json" "b/checkpoint-630/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-630/trainer_state.json"
@@ -0,0 +1,4443 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.965699208443271,
+  "eval_steps": 500,
+  "global_step": 630,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0079155672823219,
+      "grad_norm": 30.869014739990234,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 3.7012,
+      "step": 1
+    },
+    {
+      "epoch": 0.0158311345646438,
+      "grad_norm": 31.733173370361328,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.7487,
+      "step": 2
+    },
+    {
+      "epoch": 0.023746701846965697,
+      "grad_norm": 31.742353439331055,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 3.7018,
+      "step": 3
+    },
+    {
+      "epoch": 0.0316622691292876,
+      "grad_norm": 31.144001007080078,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.6754,
+      "step": 4
+    },
+    {
+      "epoch": 0.0395778364116095,
+      "grad_norm": 30.949337005615234,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 3.7139,
+      "step": 5
+    },
+    {
+      "epoch": 0.047493403693931395,
+      "grad_norm": 31.45248031616211,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.7776,
+      "step": 6
+    },
+    {
+      "epoch": 0.055408970976253295,
+      "grad_norm": 31.660465240478516,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 3.7274,
+      "step": 7
+    },
+    {
+      "epoch": 0.0633245382585752,
+      "grad_norm": 30.728031158447266,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.6926,
+      "step": 8
+    },
+    {
+      "epoch": 0.0712401055408971,
+      "grad_norm": 31.589900970458984,
+      "learning_rate": 4.5000000000000003e-07,
+      "loss": 3.7384,
+      "step": 9
+    },
+    {
+      "epoch": 0.079155672823219,
+      "grad_norm": 30.93056297302246,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.6975,
+      "step": 10
+    },
+    {
+      "epoch": 0.0870712401055409,
+      "grad_norm": 29.612712860107422,
+      "learning_rate": 5.5e-07,
+      "loss": 3.6154,
+      "step": 11
+    },
+    {
+      "epoch": 0.09498680738786279,
+      "grad_norm": 29.427888870239258,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.5904,
+      "step": 12
+    },
+    {
+      "epoch": 0.10290237467018469,
+      "grad_norm": 28.683610916137695,
+      "learning_rate": 6.5e-07,
+      "loss": 3.5048,
+      "step": 13
+    },
+    {
+      "epoch": 0.11081794195250659,
+      "grad_norm": 27.198017120361328,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.3737,
+      "step": 14
+    },
+    {
+      "epoch": 0.11873350923482849,
+      "grad_norm": 26.540634155273438,
+      "learning_rate": 7.5e-07,
+      "loss": 3.4435,
+      "step": 15
+    },
+    {
+      "epoch": 0.1266490765171504,
+      "grad_norm": 25.825363159179688,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.3308,
+      "step": 16
+    },
+    {
+      "epoch": 0.1345646437994723,
+      "grad_norm": 24.061580657958984,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 3.1701,
+      "step": 17
+    },
+    {
+      "epoch": 0.1424802110817942,
+      "grad_norm": 22.808284759521484,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.1343,
+      "step": 18
+    },
+    {
+      "epoch": 0.1503957783641161,
+      "grad_norm": 21.435670852661133,
+      "learning_rate": 9.500000000000001e-07,
+      "loss": 3.0035,
+      "step": 19
+    },
+    {
+      "epoch": 0.158311345646438,
+      "grad_norm": 20.890090942382812,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.9045,
+      "step": 20
+    },
+    {
+      "epoch": 0.1662269129287599,
+      "grad_norm": 19.962282180786133,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 2.7882,
+      "step": 21
+    },
+    {
+      "epoch": 0.1741424802110818,
+      "grad_norm": 19.279138565063477,
+      "learning_rate": 1.1e-06,
+      "loss": 2.6107,
+      "step": 22
+    },
+    {
+      "epoch": 0.1820580474934037,
+      "grad_norm": 18.44759178161621,
+      "learning_rate": 1.1500000000000002e-06,
+      "loss": 2.4873,
+      "step": 23
+    },
+    {
+      "epoch": 0.18997361477572558,
+      "grad_norm": 17.96116065979004,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.2947,
+      "step": 24
+    },
+    {
+      "epoch": 0.19788918205804748,
+      "grad_norm": 18.503511428833008,
+      "learning_rate": 1.25e-06,
+      "loss": 2.2086,
+      "step": 25
+    },
+    {
+      "epoch": 0.20580474934036938,
+      "grad_norm": 17.768098831176758,
+      "learning_rate": 1.3e-06,
+      "loss": 2.018,
+      "step": 26
+    },
+    {
+      "epoch": 0.21372031662269128,
+      "grad_norm": 17.546539306640625,
+      "learning_rate": 1.3500000000000002e-06,
+      "loss": 1.9271,
+      "step": 27
+    },
+    {
+      "epoch": 0.22163588390501318,
+      "grad_norm": 16.827245712280273,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.7777,
+      "step": 28
+    },
+    {
+      "epoch": 0.22955145118733508,
+      "grad_norm": 15.83143138885498,
+      "learning_rate": 1.45e-06,
+      "loss": 1.6259,
+      "step": 29
+    },
+    {
+      "epoch": 0.23746701846965698,
+      "grad_norm": 15.121193885803223,
+      "learning_rate": 1.5e-06,
+      "loss": 1.4957,
+      "step": 30
+    },
+    {
+      "epoch": 0.24538258575197888,
+      "grad_norm": 13.953106880187988,
+      "learning_rate": 1.5500000000000002e-06,
+      "loss": 1.314,
+      "step": 31
+    },
+    {
+      "epoch": 0.2532981530343008,
+      "grad_norm": 13.442436218261719,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.2131,
+      "step": 32
+    },
+    {
+      "epoch": 0.2612137203166227,
+      "grad_norm": 12.913166046142578,
+      "learning_rate": 1.6500000000000003e-06,
+      "loss": 1.0614,
+      "step": 33
+    },
+    {
+      "epoch": 0.2691292875989446,
+      "grad_norm": 12.016983985900879,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 0.926,
+      "step": 34
+    },
+    {
+      "epoch": 0.2770448548812665,
+      "grad_norm": 12.453398704528809,
+      "learning_rate": 1.75e-06,
+      "loss": 0.8099,
+      "step": 35
+    },
+    {
+      "epoch": 0.2849604221635884,
+      "grad_norm": 11.902524948120117,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.7201,
+      "step": 36
+    },
+    {
+      "epoch": 0.2928759894459103,
+      "grad_norm": 11.01728343963623,
+      "learning_rate": 1.85e-06,
+      "loss": 0.6013,
+      "step": 37
+    },
+    {
+      "epoch": 0.3007915567282322,
+      "grad_norm": 9.88390064239502,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 0.4793,
+      "step": 38
+    },
+    {
+      "epoch": 0.3087071240105541,
+      "grad_norm": 8.726274490356445,
+      "learning_rate": 1.9500000000000004e-06,
+      "loss": 0.3859,
+      "step": 39
+    },
+    {
+      "epoch": 0.316622691292876,
+      "grad_norm": 6.780221462249756,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.3218,
+      "step": 40
+    },
+    {
+      "epoch": 0.3245382585751979,
+      "grad_norm": 6.255209445953369,
+      "learning_rate": 2.05e-06,
+      "loss": 0.2498,
+      "step": 41
+    },
+    {
+      "epoch": 0.3324538258575198,
+      "grad_norm": 7.047099590301514,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 0.2299,
+      "step": 42
+    },
+    {
+      "epoch": 0.3403693931398417,
+      "grad_norm": 6.4962944984436035,
+      "learning_rate": 2.15e-06,
+      "loss": 0.1926,
+      "step": 43
+    },
+    {
+      "epoch": 0.3482849604221636,
+      "grad_norm": 5.178858757019043,
+      "learning_rate": 2.2e-06,
+      "loss": 0.1659,
+      "step": 44
+    },
+    {
+      "epoch": 0.3562005277044855,
+      "grad_norm": 3.723464012145996,
+      "learning_rate": 2.25e-06,
+      "loss": 0.1344,
+      "step": 45
+    },
+    {
+      "epoch": 0.3641160949868074,
+      "grad_norm": 2.3385989665985107,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 0.1221,
+      "step": 46
+    },
+    {
+      "epoch": 0.3720316622691293,
+      "grad_norm": 1.5498709678649902,
+      "learning_rate": 2.35e-06,
+      "loss": 0.0951,
+      "step": 47
+    },
+    {
+      "epoch": 0.37994722955145116,
+      "grad_norm": 1.1670784950256348,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.091,
+      "step": 48
+    },
+    {
+      "epoch": 0.38786279683377306,
+      "grad_norm": 1.0624521970748901,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 0.0888,
+      "step": 49
+    },
+    {
+      "epoch": 0.39577836411609496,
+      "grad_norm": 0.9028095006942749,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0832,
+      "step": 50
+    },
+    {
+      "epoch": 0.40369393139841686,
+      "grad_norm": 0.8736869692802429,
+      "learning_rate": 2.55e-06,
+      "loss": 0.0865,
+      "step": 51
+    },
+    {
+      "epoch": 0.41160949868073876,
+      "grad_norm": 0.8399665355682373,
+      "learning_rate": 2.6e-06,
+      "loss": 0.0791,
+      "step": 52
+    },
+    {
+      "epoch": 0.41952506596306066,
+      "grad_norm": 0.7182458639144897,
+      "learning_rate": 2.6500000000000005e-06,
+      "loss": 0.0725,
+      "step": 53
+    },
+    {
+      "epoch": 0.42744063324538256,
+      "grad_norm": 0.5676689147949219,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.077,
+      "step": 54
+    },
+    {
+      "epoch": 0.43535620052770446,
+      "grad_norm": 0.5275952219963074,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 0.0682,
+      "step": 55
+    },
+    {
+      "epoch": 0.44327176781002636,
+      "grad_norm": 0.5252273082733154,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.077,
+      "step": 56
+    },
+    {
+      "epoch": 0.45118733509234826,
+      "grad_norm": 0.5784285664558411,
+      "learning_rate": 2.85e-06,
+      "loss": 0.0727,
+      "step": 57
+    },
+    {
+      "epoch": 0.45910290237467016,
+      "grad_norm": 0.5059791207313538,
+      "learning_rate": 2.9e-06,
+      "loss": 0.0738,
+      "step": 58
+    },
+    {
+      "epoch": 0.46701846965699206,
+      "grad_norm": 0.6042116284370422,
+      "learning_rate": 2.95e-06,
+      "loss": 0.0657,
+      "step": 59
+    },
+    {
+      "epoch": 0.47493403693931396,
+      "grad_norm": 0.7056423425674438,
+      "learning_rate": 3e-06,
+      "loss": 0.0735,
+      "step": 60
+    },
+    {
+      "epoch": 0.48284960422163586,
+      "grad_norm": 0.5071430802345276,
+      "learning_rate": 3.05e-06,
+      "loss": 0.068,
+      "step": 61
+    },
+    {
+      "epoch": 0.49076517150395776,
+      "grad_norm": 0.6120741367340088,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.0663,
+      "step": 62
+    },
+    {
+      "epoch": 0.49868073878627966,
+      "grad_norm": 0.5747594237327576,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 0.0702,
+      "step": 63
+    },
+    {
+      "epoch": 0.5065963060686016,
+      "grad_norm": 0.3565518260002136,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.0602,
+      "step": 64
+    },
+    {
+      "epoch": 0.5145118733509235,
+      "grad_norm": 0.5836294293403625,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 0.07,
+      "step": 65
+    },
+    {
+      "epoch": 0.5224274406332454,
+      "grad_norm": 0.42942124605178833,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.0578,
+      "step": 66
+    },
+    {
+      "epoch": 0.5303430079155673,
+      "grad_norm": 0.3953992426395416,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 0.06,
+      "step": 67
+    },
+    {
+      "epoch": 0.5382585751978892,
+      "grad_norm": 0.525784969329834,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.0544,
+      "step": 68
+    },
+    {
+      "epoch": 0.5461741424802111,
+      "grad_norm": 0.5336060523986816,
+      "learning_rate": 3.45e-06,
+      "loss": 0.0621,
+      "step": 69
+    },
+    {
+      "epoch": 0.554089709762533,
+      "grad_norm": 0.5048466920852661,
+      "learning_rate": 3.5e-06,
+      "loss": 0.0629,
+      "step": 70
+    },
+    {
+      "epoch": 0.5620052770448549,
+      "grad_norm": 0.54691481590271,
+      "learning_rate": 3.5500000000000003e-06,
+      "loss": 0.0685,
+      "step": 71
+    },
+    {
+      "epoch": 0.5699208443271768,
+      "grad_norm": 0.3295685350894928,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.0601,
+      "step": 72
+    },
+    {
+      "epoch": 0.5778364116094987,
+      "grad_norm": 0.44020742177963257,
+      "learning_rate": 3.65e-06,
+      "loss": 0.0569,
+      "step": 73
+    },
+    {
+      "epoch": 0.5857519788918206,
+      "grad_norm": 0.2974744141101837,
+      "learning_rate": 3.7e-06,
+      "loss": 0.0612,
+      "step": 74
+    },
+    {
+      "epoch": 0.5936675461741425,
+      "grad_norm": 0.37300822138786316,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.0554,
+      "step": 75
+    },
+    {
+      "epoch": 0.6015831134564644,
+      "grad_norm": 0.37561413645744324,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.0545,
+      "step": 76
+    },
+    {
+      "epoch": 0.6094986807387863,
+      "grad_norm": 0.4859388768672943,
+      "learning_rate": 3.85e-06,
+      "loss": 0.0578,
+      "step": 77
+    },
+    {
+      "epoch": 0.6174142480211082,
+      "grad_norm": 0.46191921830177307,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.0567,
+      "step": 78
+    },
+    {
+      "epoch": 0.6253298153034301,
+      "grad_norm": 0.4128071963787079,
+      "learning_rate": 3.95e-06,
+      "loss": 0.0602,
+      "step": 79
+    },
+    {
+      "epoch": 0.633245382585752,
+      "grad_norm": 0.31622228026390076,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.0567,
+      "step": 80
+    },
+    {
+      "epoch": 0.6411609498680739,
+      "grad_norm": 0.617358386516571,
+      "learning_rate": 4.05e-06,
+      "loss": 0.0656,
+      "step": 81
+    },
+    {
+      "epoch": 0.6490765171503958,
+      "grad_norm": 0.4259068965911865,
+      "learning_rate": 4.1e-06,
+      "loss": 0.0518,
+      "step": 82
+    },
+    {
+      "epoch": 0.6569920844327177,
+      "grad_norm": 0.276052862405777,
+      "learning_rate": 4.15e-06,
+      "loss": 0.0622,
+      "step": 83
+    },
+    {
+      "epoch": 0.6649076517150396,
+      "grad_norm": 0.3911490738391876,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.059,
+      "step": 84
+    },
+    {
+      "epoch": 0.6728232189973615,
+      "grad_norm": 0.39195016026496887,
+      "learning_rate": 4.25e-06,
+      "loss": 0.0631,
+      "step": 85
+    },
+    {
+      "epoch": 0.6807387862796834,
+      "grad_norm": 0.267223060131073,
+      "learning_rate": 4.3e-06,
+      "loss": 0.0509,
+      "step": 86
+    },
+    {
+      "epoch": 0.6886543535620053,
+      "grad_norm": 0.6315395832061768,
+      "learning_rate": 4.350000000000001e-06,
+      "loss": 0.0644,
+      "step": 87
+    },
+    {
+      "epoch": 0.6965699208443272,
+      "grad_norm": 0.36506301164627075,
+      "learning_rate": 4.4e-06,
+      "loss": 0.0575,
+      "step": 88
+    },
+    {
+      "epoch": 0.7044854881266491,
+      "grad_norm": 0.30107879638671875,
+      "learning_rate": 4.450000000000001e-06,
+      "loss": 0.0563,
+      "step": 89
+    },
+    {
+      "epoch": 0.712401055408971,
+      "grad_norm": 0.3180297613143921,
+      "learning_rate": 4.5e-06,
+      "loss": 0.0537,
+      "step": 90
+    },
+    {
+      "epoch": 0.7203166226912929,
+      "grad_norm": 0.2672456204891205,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 0.0582,
+      "step": 91
+    },
+    {
+      "epoch": 0.7282321899736148,
+      "grad_norm": 0.3606305420398712,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.0506,
+      "step": 92
+    },
+    {
+      "epoch": 0.7361477572559367,
+      "grad_norm": 0.335498183965683,
+      "learning_rate": 4.65e-06,
+      "loss": 0.0583,
+      "step": 93
+    },
+    {
+      "epoch": 0.7440633245382586,
+      "grad_norm": 0.4400036334991455,
+      "learning_rate": 4.7e-06,
+      "loss": 0.0641,
+      "step": 94
+    },
+    {
+      "epoch": 0.7519788918205804,
+      "grad_norm": 0.31314533948898315,
+      "learning_rate": 4.75e-06,
+      "loss": 0.0549,
+      "step": 95
+    },
+    {
+      "epoch": 0.7598944591029023,
+      "grad_norm": 0.28006380796432495,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.0541,
+      "step": 96
+    },
+    {
+      "epoch": 0.7678100263852242,
+      "grad_norm": 0.3578348755836487,
+      "learning_rate": 4.85e-06,
+      "loss": 0.0513,
+      "step": 97
+    },
+    {
+      "epoch": 0.7757255936675461,
+      "grad_norm": 0.2923860549926758,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.0565,
+      "step": 98
+    },
+    {
+      "epoch": 0.783641160949868,
+      "grad_norm": 0.2959270775318146,
+      "learning_rate": 4.95e-06,
+      "loss": 0.0584,
+      "step": 99
+    },
+    {
+      "epoch": 0.7915567282321899,
+      "grad_norm": 0.3138754367828369,
+      "learning_rate": 5e-06,
+      "loss": 0.0532,
+      "step": 100
+    },
+    {
+      "epoch": 0.7994722955145118,
+      "grad_norm": 0.34502485394477844,
+      "learning_rate": 4.9999713317456065e-06,
+      "loss": 0.0582,
+      "step": 101
+    },
+    {
+      "epoch": 0.8073878627968337,
+      "grad_norm": 0.27575260400772095,
+      "learning_rate": 4.9998853276399215e-06,
+      "loss": 0.05,
+      "step": 102
+    },
+    {
+      "epoch": 0.8153034300791556,
+      "grad_norm": 0.40034064650535583,
+      "learning_rate": 4.999741989655415e-06,
+      "loss": 0.0568,
+      "step": 103
+    },
+    {
+      "epoch": 0.8232189973614775,
+      "grad_norm": 0.463155061006546,
+      "learning_rate": 4.999541321079486e-06,
+      "loss": 0.0584,
+      "step": 104
+    },
+    {
+      "epoch": 0.8311345646437994,
+      "grad_norm": 0.3311242163181305,
+      "learning_rate": 4.99928332651439e-06,
+      "loss": 0.0547,
+      "step": 105
+    },
+    {
+      "epoch": 0.8390501319261213,
+      "grad_norm": 0.3222312927246094,
+      "learning_rate": 4.998968011877129e-06,
+      "loss": 0.0569,
+      "step": 106
+    },
+    {
+      "epoch": 0.8469656992084432,
+      "grad_norm": 0.4004589021205902,
+      "learning_rate": 4.998595384399319e-06,
+      "loss": 0.0509,
+      "step": 107
+    },
+    {
+      "epoch": 0.8548812664907651,
+      "grad_norm": 0.25930100679397583,
+      "learning_rate": 4.998165452627025e-06,
+      "loss": 0.056,
+      "step": 108
+    },
+    {
+      "epoch": 0.862796833773087,
+      "grad_norm": 0.2832070291042328,
+      "learning_rate": 4.997678226420561e-06,
+      "loss": 0.0599,
+      "step": 109
+    },
+    {
+      "epoch": 0.8707124010554089,
+      "grad_norm": 0.27685272693634033,
+      "learning_rate": 4.997133716954266e-06,
+      "loss": 0.0526,
+      "step": 110
+    },
+    {
+      "epoch": 0.8786279683377308,
+      "grad_norm": 0.3196943998336792,
+      "learning_rate": 4.99653193671625e-06,
+      "loss": 0.0538,
+      "step": 111
+    },
+    {
+      "epoch": 0.8865435356200527,
+      "grad_norm": 0.32398006319999695,
+      "learning_rate": 4.995872899508103e-06,
+      "loss": 0.0582,
+      "step": 112
+    },
+    {
+      "epoch": 0.8944591029023746,
+      "grad_norm": 0.264309287071228,
+      "learning_rate": 4.995156620444584e-06,
+      "loss": 0.0545,
+      "step": 113
+    },
+    {
+      "epoch": 0.9023746701846965,
+      "grad_norm": 0.2664014995098114,
+      "learning_rate": 4.994383115953266e-06,
+      "loss": 0.0559,
+      "step": 114
+    },
+    {
+      "epoch": 0.9102902374670184,
+      "grad_norm": 0.30734142661094666,
+      "learning_rate": 4.9935524037741705e-06,
+      "loss": 0.0527,
+      "step": 115
+    },
+    {
+      "epoch": 0.9182058047493403,
+      "grad_norm": 0.35413816571235657,
+      "learning_rate": 4.992664502959351e-06,
+      "loss": 0.0476,
+      "step": 116
+    },
+    {
+      "epoch": 0.9261213720316622,
+      "grad_norm": 0.24693645536899567,
+      "learning_rate": 4.991719433872461e-06,
+      "loss": 0.0442,
+      "step": 117
+    },
+    {
+      "epoch": 0.9340369393139841,
+      "grad_norm": 0.31462687253952026,
+      "learning_rate": 4.990717218188286e-06,
+      "loss": 0.047,
+      "step": 118
+    },
+    {
+      "epoch": 0.941952506596306,
+      "grad_norm": 0.5714662671089172,
+      "learning_rate": 4.989657878892245e-06,
+      "loss": 0.0567,
+      "step": 119
+    },
+    {
+      "epoch": 0.9498680738786279,
+      "grad_norm": 0.36190715432167053,
+      "learning_rate": 4.988541440279862e-06,
+      "loss": 0.0483,
+      "step": 120
+    },
+    {
+      "epoch": 0.9577836411609498,
+      "grad_norm": 0.27186068892478943,
+      "learning_rate": 4.987367927956218e-06,
+      "loss": 0.0499,
+      "step": 121
+    },
+    {
+      "epoch": 0.9656992084432717,
+      "grad_norm": 0.45251426100730896,
+      "learning_rate": 4.986137368835351e-06,
+      "loss": 0.0537,
+      "step": 122
+    },
+    {
+      "epoch": 0.9736147757255936,
+      "grad_norm": 0.2997976541519165,
+      "learning_rate": 4.984849791139647e-06,
+      "loss": 0.0508,
+      "step": 123
+    },
+    {
+      "epoch": 0.9815303430079155,
+      "grad_norm": 0.3122728168964386,
+      "learning_rate": 4.983505224399188e-06,
+      "loss": 0.0497,
+      "step": 124
+    },
+    {
+      "epoch": 0.9894459102902374,
+      "grad_norm": 0.41572335362434387,
+      "learning_rate": 4.9821036994510816e-06,
+      "loss": 0.0493,
+      "step": 125
+    },
+    {
+      "epoch": 0.9973614775725593,
+      "grad_norm": 0.2922910153865814,
+      "learning_rate": 4.980645248438746e-06,
+      "loss": 0.0515,
+      "step": 126
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2922910153865814,
+      "learning_rate": 4.979129904811177e-06,
+      "loss": 0.0512,
+      "step": 127
+    },
+    {
+      "epoch": 1.007915567282322,
+      "grad_norm": 0.9297466278076172,
+      "learning_rate": 4.977557703322178e-06,
+      "loss": 0.0432,
+      "step": 128
+    },
+    {
+      "epoch": 1.0158311345646438,
+      "grad_norm": 0.3013469874858856,
+      "learning_rate": 4.975928680029571e-06,
+      "loss": 0.0421,
+      "step": 129
+    },
+    {
+      "epoch": 1.0237467018469657,
+      "grad_norm": 0.3427578806877136,
+      "learning_rate": 4.974242872294354e-06,
+      "loss": 0.0499,
+      "step": 130
+    },
+    {
+      "epoch": 1.0316622691292876,
+      "grad_norm": 0.623426079750061,
+      "learning_rate": 4.972500318779864e-06,
+      "loss": 0.0529,
+      "step": 131
+    },
+    {
+      "epoch": 1.0395778364116095,
+      "grad_norm": 0.4214429259300232,
+      "learning_rate": 4.970701059450872e-06,
+      "loss": 0.0503,
+      "step": 132
+    },
+    {
+      "epoch": 1.0474934036939314,
+      "grad_norm": 0.45160651206970215,
+      "learning_rate": 4.968845135572678e-06,
+      "loss": 0.0478,
+      "step": 133
+    },
+    {
+      "epoch": 1.0554089709762533,
+      "grad_norm": 0.28806400299072266,
+      "learning_rate": 4.96693258971016e-06,
+      "loss": 0.043,
+      "step": 134
+    },
+    {
+      "epoch": 1.0633245382585752,
+      "grad_norm": 0.302489697933197,
+      "learning_rate": 4.9649634657268e-06,
+      "loss": 0.0457,
+      "step": 135
+    },
+    {
+      "epoch": 1.071240105540897,
+      "grad_norm": 0.3145967125892639,
+      "learning_rate": 4.962937808783675e-06,
+      "loss": 0.0433,
+      "step": 136
+    },
+    {
+      "epoch": 1.079155672823219,
+      "grad_norm": 0.3302624821662903,
+      "learning_rate": 4.960855665338424e-06,
+      "loss": 0.0502,
+      "step": 137
+    },
+    {
+      "epoch": 1.087071240105541,
+      "grad_norm": 0.23032866418361664,
+      "learning_rate": 4.958717083144182e-06,
+      "loss": 0.0436,
+      "step": 138
+    },
+    {
+      "epoch": 1.0949868073878628,
+      "grad_norm": 0.2256026566028595,
+      "learning_rate": 4.956522111248483e-06,
+      "loss": 0.0508,
+      "step": 139
+    },
+    {
+      "epoch": 1.1029023746701847,
+      "grad_norm": 0.311922550201416,
+      "learning_rate": 4.954270799992138e-06,
+      "loss": 0.0483,
+      "step": 140
+    },
+    {
+      "epoch": 1.1108179419525066,
+      "grad_norm": 0.25219956040382385,
+      "learning_rate": 4.9519632010080765e-06,
+      "loss": 0.0455,
+      "step": 141
+    },
+    {
+      "epoch": 1.1187335092348285,
+      "grad_norm": 0.2478555589914322,
+      "learning_rate": 4.9495993672201675e-06,
+      "loss": 0.044,
+      "step": 142
+    },
+    {
+      "epoch": 1.1266490765171504,
+      "grad_norm": 0.30613765120506287,
+      "learning_rate": 4.947179352842001e-06,
+      "loss": 0.047,
+      "step": 143
+    },
+    {
+      "epoch": 1.1345646437994723,
+      "grad_norm": 0.2842668890953064,
+      "learning_rate": 4.944703213375648e-06,
+      "loss": 0.0456,
+      "step": 144
+    },
+    {
+      "epoch": 1.1424802110817942,
+      "grad_norm": 0.2848612070083618,
+      "learning_rate": 4.942171005610385e-06,
+      "loss": 0.0437,
+      "step": 145
+    },
+    {
+      "epoch": 1.150395778364116,
+      "grad_norm": 0.2956187129020691,
+      "learning_rate": 4.939582787621394e-06,
+      "loss": 0.0448,
+      "step": 146
+    },
+    {
+      "epoch": 1.158311345646438,
+      "grad_norm": 0.3415001332759857,
+      "learning_rate": 4.936938618768427e-06,
+      "loss": 0.0479,
+      "step": 147
+    },
+    {
+      "epoch": 1.16622691292876,
+      "grad_norm": 0.2903468608856201,
+      "learning_rate": 4.934238559694448e-06,
+      "loss": 0.0462,
+      "step": 148
+    },
+    {
+      "epoch": 1.1741424802110818,
+      "grad_norm": 0.285067081451416,
+      "learning_rate": 4.9314826723242425e-06,
+      "loss": 0.0437,
+      "step": 149
+    },
+    {
+      "epoch": 1.1820580474934037,
+      "grad_norm": 0.3022693395614624,
+      "learning_rate": 4.928671019862995e-06,
+      "loss": 0.0398,
+      "step": 150
+    },
+    {
+      "epoch": 1.1899736147757256,
+      "grad_norm": 0.2898755669593811,
+      "learning_rate": 4.925803666794839e-06,
+      "loss": 0.0433,
+      "step": 151
+    },
+    {
+      "epoch": 1.1978891820580475,
+      "grad_norm": 0.29884910583496094,
+      "learning_rate": 4.92288067888138e-06,
+      "loss": 0.0417,
+      "step": 152
+    },
+    {
+      "epoch": 1.2058047493403694,
+      "grad_norm": 0.31269946694374084,
+      "learning_rate": 4.919902123160187e-06,
+      "loss": 0.0421,
+      "step": 153
+    },
+    {
+      "epoch": 1.2137203166226913,
+      "grad_norm": 0.2689797580242157,
+      "learning_rate": 4.9168680679432565e-06,
+      "loss": 0.0403,
+      "step": 154
+    },
+    {
+      "epoch": 1.2216358839050132,
+      "grad_norm": 0.2575155198574066,
+      "learning_rate": 4.913778582815439e-06,
+      "loss": 0.04,
+      "step": 155
+    },
+    {
+      "epoch": 1.229551451187335,
+      "grad_norm": 0.3549853265285492,
+      "learning_rate": 4.9106337386328524e-06,
+      "loss": 0.048,
+      "step": 156
+    },
+    {
+      "epoch": 1.237467018469657,
+      "grad_norm": 0.2768900990486145,
+      "learning_rate": 4.907433607521252e-06,
+      "loss": 0.0413,
+      "step": 157
+    },
+    {
+      "epoch": 1.245382585751979,
+      "grad_norm": 0.3248242139816284,
+      "learning_rate": 4.904178262874374e-06,
+      "loss": 0.0423,
+      "step": 158
+    },
+    {
+      "epoch": 1.2532981530343008,
+      "grad_norm": 0.34395328164100647,
+      "learning_rate": 4.900867779352258e-06,
+      "loss": 0.0447,
+      "step": 159
+    },
+    {
+      "epoch": 1.2612137203166227,
+      "grad_norm": 0.31361910700798035,
+      "learning_rate": 4.897502232879533e-06,
+      "loss": 0.0377,
+      "step": 160
+    },
+    {
+      "epoch": 1.2691292875989446,
+      "grad_norm": 0.2977491617202759,
+      "learning_rate": 4.89408170064367e-06,
+      "loss": 0.0406,
+      "step": 161
+    },
+    {
+      "epoch": 1.2770448548812665,
+      "grad_norm": 0.2638033926486969,
+      "learning_rate": 4.890606261093221e-06,
+      "loss": 0.0428,
+      "step": 162
+    },
+    {
+      "epoch": 1.2849604221635884,
+      "grad_norm": 0.25792911648750305,
+      "learning_rate": 4.887075993936014e-06,
+      "loss": 0.0358,
+      "step": 163
+    },
+    {
+      "epoch": 1.2928759894459103,
+      "grad_norm": 0.2919572591781616,
+      "learning_rate": 4.883490980137327e-06,
+      "loss": 0.0352,
+      "step": 164
+    },
+    {
+      "epoch": 1.3007915567282322,
+      "grad_norm": 0.27602723240852356,
+      "learning_rate": 4.8798513019180295e-06,
+      "loss": 0.0388,
+      "step": 165
+    },
+    {
+      "epoch": 1.3087071240105541,
+      "grad_norm": 0.3702069818973541,
+      "learning_rate": 4.876157042752698e-06,
+      "loss": 0.0445,
+      "step": 166
+    },
+    {
+      "epoch": 1.316622691292876,
+      "grad_norm": 0.2848120927810669,
+      "learning_rate": 4.872408287367702e-06,
+      "loss": 0.0385,
+      "step": 167
+    },
+    {
+      "epoch": 1.324538258575198,
+      "grad_norm": 0.30758577585220337,
+      "learning_rate": 4.868605121739261e-06,
+      "loss": 0.0368,
+      "step": 168
+    },
+    {
+      "epoch": 1.3324538258575198,
+      "grad_norm": 0.2721364200115204,
+      "learning_rate": 4.86474763309147e-06,
+      "loss": 0.0361,
+      "step": 169
+    },
+    {
+      "epoch": 1.3403693931398417,
+      "grad_norm": 0.2803274095058441,
+      "learning_rate": 4.8608359098943014e-06,
+      "loss": 0.0385,
+      "step": 170
+    },
+    {
+      "epoch": 1.3482849604221636,
+      "grad_norm": 0.28101253509521484,
+      "learning_rate": 4.856870041861576e-06,
+      "loss": 0.0404,
+      "step": 171
+    },
+    {
+      "epoch": 1.3562005277044855,
+      "grad_norm": 0.30860188603401184,
+      "learning_rate": 4.8528501199489045e-06,
+      "loss": 0.0361,
+      "step": 172
+    },
+    {
+      "epoch": 1.3641160949868074,
+      "grad_norm": 0.28227391839027405,
+      "learning_rate": 4.848776236351602e-06,
+      "loss": 0.0361,
+      "step": 173
+    },
+    {
+      "epoch": 1.3720316622691293,
+      "grad_norm": 0.2740943133831024,
+      "learning_rate": 4.8446484845025754e-06,
+      "loss": 0.0337,
+      "step": 174
+    },
+    {
+      "epoch": 1.3799472295514512,
+      "grad_norm": 0.2890925407409668,
+      "learning_rate": 4.840466959070174e-06,
+      "loss": 0.0429,
+      "step": 175
+    },
+    {
+      "epoch": 1.3878627968337731,
+      "grad_norm": 0.308539479970932,
+      "learning_rate": 4.836231755956028e-06,
+      "loss": 0.0385,
+      "step": 176
+    },
+    {
+      "epoch": 1.395778364116095,
+      "grad_norm": 0.279610276222229,
+      "learning_rate": 4.83194297229284e-06,
+      "loss": 0.0397,
+      "step": 177
+    },
+    {
+      "epoch": 1.403693931398417,
+      "grad_norm": 0.26512083411216736,
+      "learning_rate": 4.827600706442164e-06,
+      "loss": 0.0421,
+      "step": 178
+    },
+    {
+      "epoch": 1.4116094986807388,
+      "grad_norm": 0.2635437548160553,
+      "learning_rate": 4.823205057992145e-06,
+      "loss": 0.0386,
+      "step": 179
+    },
+    {
+      "epoch": 1.4195250659630607,
+      "grad_norm": 0.2850892245769501,
+      "learning_rate": 4.8187561277552376e-06,
+      "loss": 0.0338,
+      "step": 180
+    },
+    {
+      "epoch": 1.4274406332453826,
+      "grad_norm": 0.2996429204940796,
+      "learning_rate": 4.8142540177658925e-06,
+      "loss": 0.0362,
+      "step": 181
+    },
+    {
+      "epoch": 1.4353562005277045,
+      "grad_norm": 0.2694872319698334,
+      "learning_rate": 4.809698831278217e-06,
+      "loss": 0.0376,
+      "step": 182
+    },
+    {
+      "epoch": 1.4432717678100264,
+      "grad_norm": 0.2809225916862488,
+      "learning_rate": 4.805090672763609e-06,
+      "loss": 0.0397,
+      "step": 183
+    },
+    {
+      "epoch": 1.4511873350923483,
+      "grad_norm": 0.4314921498298645,
+      "learning_rate": 4.800429647908354e-06,
+      "loss": 0.0415,
+      "step": 184
+    },
+    {
+      "epoch": 1.4591029023746702,
+      "grad_norm": 0.3187998831272125,
+      "learning_rate": 4.795715863611212e-06,
+      "loss": 0.0332,
+      "step": 185
+    },
+    {
+      "epoch": 1.4670184696569921,
+      "grad_norm": 0.32976770401000977,
+      "learning_rate": 4.790949427980956e-06,
+      "loss": 0.0335,
+      "step": 186
+    },
+    {
+      "epoch": 1.474934036939314,
+      "grad_norm": 0.31698155403137207,
+      "learning_rate": 4.786130450333897e-06,
+      "loss": 0.0365,
+      "step": 187
+    },
+    {
+      "epoch": 1.482849604221636,
+      "grad_norm": 0.30868053436279297,
+      "learning_rate": 4.7812590411913755e-06,
+      "loss": 0.0342,
+      "step": 188
+    },
+    {
+      "epoch": 1.4907651715039578,
+      "grad_norm": 0.31149864196777344,
+      "learning_rate": 4.77633531227723e-06,
+      "loss": 0.036,
+      "step": 189
+    },
+    {
+      "epoch": 1.4986807387862797,
+      "grad_norm": 0.3089791238307953,
+      "learning_rate": 4.771359376515231e-06,
+      "loss": 0.0288,
+      "step": 190
+    },
+    {
+      "epoch": 1.5065963060686016,
+      "grad_norm": 0.3037573993206024,
+      "learning_rate": 4.766331348026493e-06,
+      "loss": 0.0298,
+      "step": 191
+    },
+    {
+      "epoch": 1.5145118733509235,
+      "grad_norm": 0.34367772936820984,
+      "learning_rate": 4.7612513421268546e-06,
+      "loss": 0.0334,
+      "step": 192
+    },
+    {
+      "epoch": 1.5224274406332454,
+      "grad_norm": 0.34849783778190613,
+      "learning_rate": 4.756119475324237e-06,
+      "loss": 0.0311,
+      "step": 193
+    },
+    {
+      "epoch": 1.5303430079155673,
+      "grad_norm": 0.35077860951423645,
+      "learning_rate": 4.750935865315972e-06,
+      "loss": 0.0378,
+      "step": 194
+    },
+    {
+      "epoch": 1.5382585751978892,
+      "grad_norm": 0.33701786398887634,
+      "learning_rate": 4.745700630986097e-06,
+      "loss": 0.0285,
+      "step": 195
+    },
+    {
+      "epoch": 1.5461741424802111,
+      "grad_norm": 0.27960076928138733,
+      "learning_rate": 4.740413892402639e-06,
+      "loss": 0.0279,
+      "step": 196
+    },
+    {
+      "epoch": 1.554089709762533,
+      "grad_norm": 0.33046165108680725,
+      "learning_rate": 4.73507577081485e-06,
+      "loss": 0.0315,
+      "step": 197
+    },
+    {
+      "epoch": 1.562005277044855,
+      "grad_norm": 0.33765262365341187,
+      "learning_rate": 4.7296863886504315e-06,
+      "loss": 0.0316,
+      "step": 198
+    },
+    {
+      "epoch": 1.5699208443271768,
+      "grad_norm": 0.3312811255455017,
+      "learning_rate": 4.7242458695127275e-06,
+      "loss": 0.0296,
+      "step": 199
+    },
+    {
+      "epoch": 1.5778364116094987,
+      "grad_norm": 0.2937123477458954,
+      "learning_rate": 4.718754338177887e-06,
+      "loss": 0.0245,
+      "step": 200
+    },
+    {
+      "epoch": 1.5857519788918206,
+      "grad_norm": 0.32295259833335876,
+      "learning_rate": 4.713211920592003e-06,
+      "loss": 0.0249,
+      "step": 201
+    },
+    {
+      "epoch": 1.5936675461741425,
+      "grad_norm": 0.3463473320007324,
+      "learning_rate": 4.707618743868226e-06,
+      "loss": 0.0341,
+      "step": 202
+    },
+    {
+      "epoch": 1.6015831134564644,
+      "grad_norm": 0.4024774134159088,
+      "learning_rate": 4.701974936283848e-06,
+      "loss": 0.0261,
+      "step": 203
+    },
+    {
+      "epoch": 1.6094986807387863,
+      "grad_norm": 0.41717198491096497,
+      "learning_rate": 4.696280627277356e-06,
+      "loss": 0.0303,
+      "step": 204
+    },
+    {
+      "epoch": 1.6174142480211082,
+      "grad_norm": 0.44161510467529297,
+      "learning_rate": 4.690535947445471e-06,
+      "loss": 0.0297,
+      "step": 205
+    },
+    {
+      "epoch": 1.6253298153034301,
+      "grad_norm": 0.3284972906112671,
+      "learning_rate": 4.6847410285401465e-06,
+      "loss": 0.0268,
+      "step": 206
+    },
+    {
+      "epoch": 1.633245382585752,
+      "grad_norm": 0.3295843005180359,
+      "learning_rate": 4.67889600346555e-06,
+      "loss": 0.0298,
+      "step": 207
+    },
+    {
+      "epoch": 1.641160949868074,
+      "grad_norm": 0.341462641954422,
+      "learning_rate": 4.673001006275013e-06,
+      "loss": 0.0218,
+      "step": 208
+    },
+    {
+      "epoch": 1.6490765171503958,
+      "grad_norm": 0.3291555941104889,
+      "learning_rate": 4.667056172167962e-06,
+      "loss": 0.0318,
+      "step": 209
+    },
+    {
+      "epoch": 1.6569920844327177,
+      "grad_norm": 0.3117462992668152,
+      "learning_rate": 4.6610616374868066e-06,
+      "loss": 0.0239,
+      "step": 210
+    },
+    {
+      "epoch": 1.6649076517150396,
+      "grad_norm": 0.38252758979797363,
+      "learning_rate": 4.655017539713826e-06,
+      "loss": 0.03,
+      "step": 211
+    },
+    {
+      "epoch": 1.6728232189973615,
+      "grad_norm": 0.36440685391426086,
+      "learning_rate": 4.648924017468003e-06,
+      "loss": 0.0281,
+      "step": 212
+    },
+    {
+      "epoch": 1.6807387862796834,
+      "grad_norm": 0.3495568037033081,
+      "learning_rate": 4.642781210501858e-06,
+      "loss": 0.026,
+      "step": 213
+    },
+    {
+      "epoch": 1.6886543535620053,
+      "grad_norm": 0.3864191174507141,
+      "learning_rate": 4.6365892596982295e-06,
+      "loss": 0.0293,
+      "step": 214
+    },
+    {
+      "epoch": 1.6965699208443272,
+      "grad_norm": 0.36920300126075745,
+      "learning_rate": 4.6303483070670574e-06,
+      "loss": 0.0305,
+      "step": 215
+    },
+    {
+      "epoch": 1.7044854881266491,
+      "grad_norm": 0.34929153323173523,
+      "learning_rate": 4.624058495742115e-06,
+      "loss": 0.0232,
+      "step": 216
+    },
+    {
+      "epoch": 1.712401055408971,
+      "grad_norm": 0.33653852343559265,
+      "learning_rate": 4.617719969977729e-06,
+      "loss": 0.029,
+      "step": 217
+    },
+    {
+      "epoch": 1.720316622691293,
+      "grad_norm": 0.3103637993335724,
+      "learning_rate": 4.611332875145476e-06,
+      "loss": 0.0277,
+      "step": 218
+    },
+    {
+      "epoch": 1.7282321899736148,
+      "grad_norm": 0.3294495642185211,
+      "learning_rate": 4.604897357730846e-06,
+      "loss": 0.0303,
+      "step": 219
+    },
+    {
+      "epoch": 1.7361477572559367,
+      "grad_norm": 0.35194551944732666,
+      "learning_rate": 4.598413565329876e-06,
+      "loss": 0.0231,
+      "step": 220
+    },
+    {
+      "epoch": 1.7440633245382586,
+      "grad_norm": 0.30809342861175537,
+      "learning_rate": 4.591881646645775e-06,
+      "loss": 0.0231,
+      "step": 221
+    },
+    {
+      "epoch": 1.7519788918205803,
+      "grad_norm": 0.3537861704826355,
+      "learning_rate": 4.585301751485508e-06,
+      "loss": 0.0218,
+      "step": 222
+    },
+    {
+      "epoch": 1.7598944591029024,
+      "grad_norm": 0.38943609595298767,
+      "learning_rate": 4.578674030756364e-06,
+      "loss": 0.0227,
+      "step": 223
+    },
+    {
+      "epoch": 1.767810026385224,
+      "grad_norm": 0.5681232810020447,
+      "learning_rate": 4.571998636462487e-06,
+      "loss": 0.0271,
+      "step": 224
+    },
+    {
+      "epoch": 1.7757255936675462,
+      "grad_norm": 0.49491041898727417,
+      "learning_rate": 4.5652757217014e-06,
+      "loss": 0.0223,
+      "step": 225
+    },
+    {
+      "epoch": 1.783641160949868,
+      "grad_norm": 0.38186731934547424,
+      "learning_rate": 4.5585054406604865e-06,
+      "loss": 0.0245,
+      "step": 226
+    },
+    {
+      "epoch": 1.79155672823219,
+      "grad_norm": 0.38631126284599304,
+      "learning_rate": 4.551687948613459e-06,
+      "loss": 0.0264,
+      "step": 227
+    },
+    {
+      "epoch": 1.7994722955145117,
+      "grad_norm": 0.3324311375617981,
+      "learning_rate": 4.544823401916794e-06,
+      "loss": 0.0197,
+      "step": 228
+    },
+    {
+      "epoch": 1.8073878627968338,
+      "grad_norm": 0.3379819989204407,
+      "learning_rate": 4.537911958006149e-06,
+      "loss": 0.0227,
+      "step": 229
+    },
+    {
+      "epoch": 1.8153034300791555,
+      "grad_norm": 0.29030483961105347,
+      "learning_rate": 4.530953775392749e-06,
+      "loss": 0.0232,
+      "step": 230
+    },
+    {
+      "epoch": 1.8232189973614776,
+      "grad_norm": 0.2910752296447754,
+      "learning_rate": 4.523949013659754e-06,
+      "loss": 0.0213,
+      "step": 231
+    },
+    {
+      "epoch": 1.8311345646437993,
+      "grad_norm": 0.3083966076374054,
+      "learning_rate": 4.5168978334585955e-06,
+      "loss": 0.0258,
+      "step": 232
+    },
+    {
+      "epoch": 1.8390501319261214,
+      "grad_norm": 0.36828356981277466,
+      "learning_rate": 4.509800396505298e-06,
+      "loss": 0.0241,
+      "step": 233
+    },
+    {
+      "epoch": 1.8469656992084431,
+      "grad_norm": 0.26139211654663086,
+      "learning_rate": 4.502656865576762e-06,
+      "loss": 0.0191,
+      "step": 234
+    },
+    {
+      "epoch": 1.8548812664907652,
+      "grad_norm": 0.34473150968551636,
+      "learning_rate": 4.495467404507039e-06,
+      "loss": 0.0218,
+      "step": 235
+    },
+    {
+      "epoch": 1.862796833773087,
+      "grad_norm": 0.3284291625022888,
+      "learning_rate": 4.4882321781835666e-06,
+      "loss": 0.023,
+      "step": 236
+    },
+    {
+      "epoch": 1.870712401055409,
+      "grad_norm": 0.339542955160141,
+      "learning_rate": 4.4809513525433925e-06,
+      "loss": 0.0156,
+      "step": 237
+    },
+    {
+      "epoch": 1.8786279683377307,
+      "grad_norm": 0.3365022540092468,
+      "learning_rate": 4.473625094569366e-06,
+      "loss": 0.0232,
+      "step": 238
+    },
+    {
+      "epoch": 1.8865435356200528,
+      "grad_norm": 0.34620219469070435,
+      "learning_rate": 4.466253572286308e-06,
+      "loss": 0.0169,
+      "step": 239
+    },
+    {
+      "epoch": 1.8944591029023745,
+      "grad_norm": 0.38080206513404846,
+      "learning_rate": 4.458836954757161e-06,
+      "loss": 0.0204,
+      "step": 240
+    },
+    {
+      "epoch": 1.9023746701846966,
+      "grad_norm": 0.3532657325267792,
+      "learning_rate": 4.4513754120791065e-06,
+      "loss": 0.0203,
+      "step": 241
+    },
+    {
+      "epoch": 1.9102902374670183,
+      "grad_norm": 0.3338775336742401,
+      "learning_rate": 4.443869115379667e-06,
+      "loss": 0.0216,
+      "step": 242
+    },
+    {
+      "epoch": 1.9182058047493404,
+      "grad_norm": 0.43685808777809143,
+      "learning_rate": 4.436318236812782e-06,
+      "loss": 0.0192,
+      "step": 243
+    },
+    {
+      "epoch": 1.9261213720316621,
+      "grad_norm": 0.3716251850128174,
+      "learning_rate": 4.428722949554858e-06,
+      "loss": 0.0216,
+      "step": 244
+    },
+    {
+      "epoch": 1.9340369393139842,
+      "grad_norm": 0.34019848704338074,
+      "learning_rate": 4.421083427800795e-06,
+      "loss": 0.0247,
+      "step": 245
+    },
+    {
+      "epoch": 1.941952506596306,
+      "grad_norm": 0.28350546956062317,
+      "learning_rate": 4.413399846759998e-06,
+      "loss": 0.0111,
+      "step": 246
+    },
+    {
+      "epoch": 1.949868073878628,
+      "grad_norm": 0.3894903361797333,
+      "learning_rate": 4.405672382652349e-06,
+      "loss": 0.0281,
+      "step": 247
+    },
+    {
+      "epoch": 1.9577836411609497,
+      "grad_norm": 0.32462528347969055,
+      "learning_rate": 4.397901212704176e-06,
+      "loss": 0.0194,
+      "step": 248
+    },
+    {
+      "epoch": 1.9656992084432718,
+      "grad_norm": 0.3865855634212494,
+      "learning_rate": 4.390086515144179e-06,
+      "loss": 0.0168,
+      "step": 249
+    },
+    {
+      "epoch": 1.9736147757255935,
+      "grad_norm": 0.32022252678871155,
+      "learning_rate": 4.38222846919935e-06,
+      "loss": 0.0179,
+      "step": 250
+    },
+    {
+      "epoch": 1.9815303430079156,
+      "grad_norm": 0.33319395780563354,
+      "learning_rate": 4.3743272550908545e-06,
+      "loss": 0.0151,
+      "step": 251
+    },
+    {
+      "epoch": 1.9894459102902373,
+      "grad_norm": 0.2932108938694,
+      "learning_rate": 4.366383054029907e-06,
+      "loss": 0.0129,
+      "step": 252
+    },
+    {
+      "epoch": 1.9973614775725594,
+      "grad_norm": 0.31495171785354614,
+      "learning_rate": 4.358396048213609e-06,
+      "loss": 0.0138,
+      "step": 253
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.31495171785354614,
+      "learning_rate": 4.350366420820771e-06,
+      "loss": 0.0195,
+      "step": 254
+    },
+    {
+      "epoch": 2.007915567282322,
+      "grad_norm": 2.109926223754883,
+      "learning_rate": 4.342294356007715e-06,
+      "loss": 0.0178,
+      "step": 255
+    },
+    {
+      "epoch": 2.015831134564644,
+      "grad_norm": 0.2790742814540863,
+      "learning_rate": 4.3341800389040465e-06,
+      "loss": 0.0132,
+      "step": 256
+    },
+    {
+      "epoch": 2.0237467018469655,
+      "grad_norm": 0.28284987807273865,
+      "learning_rate": 4.326023655608412e-06,
+      "loss": 0.01,
+      "step": 257
+    },
+    {
+      "epoch": 2.0316622691292876,
+      "grad_norm": 0.28543591499328613,
+      "learning_rate": 4.317825393184226e-06,
+      "loss": 0.0123,
+      "step": 258
+    },
+    {
+      "epoch": 2.0395778364116097,
+      "grad_norm": 0.2991085946559906,
+      "learning_rate": 4.30958543965539e-06,
+      "loss": 0.0095,
+      "step": 259
+    },
+    {
+      "epoch": 2.0474934036939314,
+      "grad_norm": 0.28775736689567566,
+      "learning_rate": 4.3013039840019675e-06,
+      "loss": 0.0097,
+      "step": 260
+    },
+    {
+      "epoch": 2.055408970976253,
+      "grad_norm": 0.24827070534229279,
+      "learning_rate": 4.292981216155864e-06,
+      "loss": 0.0088,
+      "step": 261
+    },
+    {
+      "epoch": 2.063324538258575,
+      "grad_norm": 0.44231149554252625,
+      "learning_rate": 4.284617326996458e-06,
+      "loss": 0.0095,
+      "step": 262
+    },
+    {
+      "epoch": 2.0712401055408973,
+      "grad_norm": 0.3509388267993927,
+      "learning_rate": 4.276212508346232e-06,
+      "loss": 0.0065,
+      "step": 263
+    },
+    {
+      "epoch": 2.079155672823219,
+      "grad_norm": 0.3389664590358734,
+      "learning_rate": 4.267766952966369e-06,
+      "loss": 0.0096,
+      "step": 264
+    },
+    {
+      "epoch": 2.0870712401055407,
+      "grad_norm": 0.3581394851207733,
+      "learning_rate": 4.2592808545523335e-06,
+      "loss": 0.0135,
+      "step": 265
+    },
+    {
+      "epoch": 2.094986807387863,
+      "grad_norm": 0.40085369348526,
+      "learning_rate": 4.250754407729428e-06,
+      "loss": 0.0097,
+      "step": 266
+    },
+    {
+      "epoch": 2.1029023746701845,
+      "grad_norm": 0.3983692228794098,
+      "learning_rate": 4.242187808048329e-06,
+      "loss": 0.0068,
+      "step": 267
+    },
+    {
+      "epoch": 2.1108179419525066,
+      "grad_norm": 0.36511439085006714,
+      "learning_rate": 4.233581251980604e-06,
+      "loss": 0.012,
+      "step": 268
+    },
+    {
+      "epoch": 2.1187335092348283,
+      "grad_norm": 0.3559989035129547,
+      "learning_rate": 4.224934936914206e-06,
+      "loss": 0.0111,
+      "step": 269
+    },
+    {
+      "epoch": 2.1266490765171504,
+      "grad_norm": 0.40258750319480896,
+      "learning_rate": 4.216249061148939e-06,
+      "loss": 0.0093,
+      "step": 270
+    },
+    {
+      "epoch": 2.1345646437994725,
+      "grad_norm": 0.4325791299343109,
+      "learning_rate": 4.207523823891924e-06,
+      "loss": 0.007,
+      "step": 271
+    },
+    {
+      "epoch": 2.142480211081794,
+      "grad_norm": 0.2487863153219223,
+      "learning_rate": 4.198759425253015e-06,
+      "loss": 0.0073,
+      "step": 272
+    },
+    {
+      "epoch": 2.150395778364116,
+      "grad_norm": 0.27570807933807373,
+      "learning_rate": 4.1899560662402204e-06,
+      "loss": 0.0125,
+      "step": 273
+    },
+    {
+      "epoch": 2.158311345646438,
+      "grad_norm": 0.3402951955795288,
+      "learning_rate": 4.18111394875509e-06,
+      "loss": 0.0083,
+      "step": 274
+    },
+    {
+      "epoch": 2.16622691292876,
+      "grad_norm": 0.21437712013721466,
+      "learning_rate": 4.172233275588082e-06,
+      "loss": 0.0086,
+      "step": 275
+    },
+    {
+      "epoch": 2.174142480211082,
+      "grad_norm": 0.439169704914093,
+      "learning_rate": 4.163314250413913e-06,
+      "loss": 0.0093,
+      "step": 276
+    },
+    {
+      "epoch": 2.1820580474934035,
+      "grad_norm": 0.3987133502960205,
+      "learning_rate": 4.154357077786892e-06,
+      "loss": 0.011,
+      "step": 277
+    },
+    {
+      "epoch": 2.1899736147757256,
+      "grad_norm": 0.36065250635147095,
+      "learning_rate": 4.145361963136219e-06,
+      "loss": 0.0091,
+      "step": 278
+    },
+    {
+      "epoch": 2.1978891820580473,
+      "grad_norm": 0.22470036149024963,
+      "learning_rate": 4.136329112761285e-06,
+      "loss": 0.0072,
+      "step": 279
+    },
+    {
+      "epoch": 2.2058047493403694,
+      "grad_norm": 0.28544190526008606,
+      "learning_rate": 4.127258733826929e-06,
+      "loss": 0.0066,
+      "step": 280
+    },
+    {
+      "epoch": 2.213720316622691,
+      "grad_norm": 0.22354596853256226,
+      "learning_rate": 4.118151034358696e-06,
+      "loss": 0.008,
+      "step": 281
+    },
+    {
+      "epoch": 2.221635883905013,
+      "grad_norm": 0.29122433066368103,
+      "learning_rate": 4.109006223238064e-06,
+      "loss": 0.0074,
+      "step": 282
+    },
+    {
+      "epoch": 2.229551451187335,
+      "grad_norm": 0.3470163941383362,
+      "learning_rate": 4.0998245101976495e-06,
+      "loss": 0.0025,
+      "step": 283
+    },
+    {
+      "epoch": 2.237467018469657,
+      "grad_norm": 0.26265788078308105,
+      "learning_rate": 4.0906061058164e-06,
+      "loss": 0.0097,
+      "step": 284
+    },
+    {
+      "epoch": 2.2453825857519787,
+      "grad_norm": 0.24661079049110413,
+      "learning_rate": 4.081351221514765e-06,
+      "loss": 0.0086,
+      "step": 285
+    },
+    {
+      "epoch": 2.253298153034301,
+      "grad_norm": 0.39738425612449646,
+      "learning_rate": 4.072060069549848e-06,
+      "loss": 0.0084,
+      "step": 286
+    },
+    {
+      "epoch": 2.261213720316623,
+      "grad_norm": 0.25185173749923706,
+      "learning_rate": 4.062732863010534e-06,
+      "loss": 0.0071,
+      "step": 287
+    },
+    {
+      "epoch": 2.2691292875989446,
+      "grad_norm": 0.3392907977104187,
+      "learning_rate": 4.053369815812608e-06,
+      "loss": 0.0073,
+      "step": 288
+    },
+    {
+      "epoch": 2.2770448548812663,
+      "grad_norm": 0.2747718095779419,
+      "learning_rate": 4.043971142693845e-06,
+      "loss": 0.0074,
+      "step": 289
+    },
+    {
+      "epoch": 2.2849604221635884,
+      "grad_norm": 0.4787997901439667,
+      "learning_rate": 4.034537059209085e-06,
+      "loss": 0.0105,
+      "step": 290
+    },
+    {
+      "epoch": 2.2928759894459105,
+      "grad_norm": 0.4995233416557312,
+      "learning_rate": 4.025067781725294e-06,
+      "loss": 0.0086,
+      "step": 291
+    },
+    {
+      "epoch": 2.300791556728232,
+      "grad_norm": 0.3541993200778961,
+      "learning_rate": 4.015563527416596e-06,
+      "loss": 0.0121,
+      "step": 292
+    },
+    {
+      "epoch": 2.308707124010554,
+      "grad_norm": 0.41221192479133606,
+      "learning_rate": 4.006024514259295e-06,
+      "loss": 0.0063,
+      "step": 293
+    },
+    {
+      "epoch": 2.316622691292876,
+      "grad_norm": 0.2869221270084381,
+      "learning_rate": 3.996450961026876e-06,
+      "loss": 0.0103,
+      "step": 294
+    },
+    {
+      "epoch": 2.324538258575198,
+      "grad_norm": 0.2938433885574341,
+      "learning_rate": 3.986843087284986e-06,
+      "loss": 0.0074,
+      "step": 295
+    },
+    {
+      "epoch": 2.33245382585752,
+      "grad_norm": 0.3887174427509308,
+      "learning_rate": 3.977201113386402e-06,
+      "loss": 0.011,
+      "step": 296
+    },
+    {
+      "epoch": 2.3403693931398415,
+      "grad_norm": 0.40513908863067627,
+      "learning_rate": 3.96752526046597e-06,
+      "loss": 0.009,
+      "step": 297
+    },
+    {
+      "epoch": 2.3482849604221636,
+      "grad_norm": 0.23019923269748688,
+      "learning_rate": 3.957815750435542e-06,
+      "loss": 0.0082,
+      "step": 298
+    },
+    {
+      "epoch": 2.3562005277044857,
+      "grad_norm": 0.24421247839927673,
+      "learning_rate": 3.948072805978879e-06,
+      "loss": 0.0065,
+      "step": 299
+    },
+    {
+      "epoch": 2.3641160949868074,
+      "grad_norm": 0.5507193803787231,
+      "learning_rate": 3.938296650546552e-06,
+      "loss": 0.0125,
+      "step": 300
+    },
+    {
+      "epoch": 2.372031662269129,
+      "grad_norm": 0.32125774025917053,
+      "learning_rate": 3.928487508350808e-06,
+      "loss": 0.007,
+      "step": 301
+    },
+    {
+      "epoch": 2.379947229551451,
+      "grad_norm": 0.2759876251220703,
+      "learning_rate": 3.918645604360434e-06,
+      "loss": 0.005,
+      "step": 302
+    },
+    {
+      "epoch": 2.387862796833773,
+      "grad_norm": 0.26057979464530945,
+      "learning_rate": 3.908771164295595e-06,
+      "loss": 0.0056,
+      "step": 303
+    },
+    {
+      "epoch": 2.395778364116095,
+      "grad_norm": 0.3396565020084381,
+      "learning_rate": 3.898864414622661e-06,
+      "loss": 0.0065,
+      "step": 304
+    },
+    {
+      "epoch": 2.4036939313984167,
+      "grad_norm": 0.32410189509391785,
+      "learning_rate": 3.888925582549006e-06,
+      "loss": 0.0105,
+      "step": 305
+    },
+    {
+      "epoch": 2.411609498680739,
+      "grad_norm": 0.48274368047714233,
+      "learning_rate": 3.878954896017804e-06,
+      "loss": 0.0113,
+      "step": 306
+    },
+    {
+      "epoch": 2.4195250659630605,
+      "grad_norm": 0.4283212721347809,
+      "learning_rate": 3.868952583702798e-06,
+      "loss": 0.005,
+      "step": 307
+    },
+    {
+      "epoch": 2.4274406332453826,
+      "grad_norm": 0.32722166180610657,
+      "learning_rate": 3.858918875003053e-06,
+      "loss": 0.0079,
+      "step": 308
+    },
+    {
+      "epoch": 2.4353562005277043,
+      "grad_norm": 0.24039319157600403,
+      "learning_rate": 3.848854000037702e-06,
+      "loss": 0.0067,
+      "step": 309
+    },
+    {
+      "epoch": 2.4432717678100264,
+      "grad_norm": 0.2368592470884323,
+      "learning_rate": 3.83875818964066e-06,
+      "loss": 0.0064,
+      "step": 310
+    },
+    {
+      "epoch": 2.451187335092348,
+      "grad_norm": 0.28300940990448,
+      "learning_rate": 3.828631675355338e-06,
+      "loss": 0.0079,
+      "step": 311
+    },
+    {
+      "epoch": 2.45910290237467,
+      "grad_norm": 0.2424677461385727,
+      "learning_rate": 3.818474689429324e-06,
+      "loss": 0.0057,
+      "step": 312
+    },
+    {
+      "epoch": 2.467018469656992,
+      "grad_norm": 0.2779932916164398,
+      "learning_rate": 3.808287464809063e-06,
+      "loss": 0.0091,
+      "step": 313
+    },
+    {
+      "epoch": 2.474934036939314,
+      "grad_norm": 0.39993351697921753,
+      "learning_rate": 3.7980702351345146e-06,
+      "loss": 0.0061,
+      "step": 314
+    },
+    {
+      "epoch": 2.4828496042216357,
+      "grad_norm": 0.3384133279323578,
+      "learning_rate": 3.787823234733788e-06,
+      "loss": 0.0068,
+      "step": 315
+    },
+    {
+      "epoch": 2.490765171503958,
+      "grad_norm": 0.27047693729400635,
+      "learning_rate": 3.7775466986177763e-06,
+      "loss": 0.0044,
+      "step": 316
+    },
+    {
+      "epoch": 2.4986807387862795,
+      "grad_norm": 0.23102645576000214,
+      "learning_rate": 3.7672408624747598e-06,
+      "loss": 0.0068,
+      "step": 317
+    },
+    {
+      "epoch": 2.5065963060686016,
+      "grad_norm": 0.2386091947555542,
+      "learning_rate": 3.756905962665005e-06,
+      "loss": 0.0082,
+      "step": 318
+    },
+    {
+      "epoch": 2.5145118733509237,
+      "grad_norm": 0.41824275255203247,
+      "learning_rate": 3.7465422362153416e-06,
+      "loss": 0.0043,
+      "step": 319
+    },
+    {
+      "epoch": 2.5224274406332454,
+      "grad_norm": 0.3015156388282776,
+      "learning_rate": 3.736149920813726e-06,
+      "loss": 0.0044,
+      "step": 320
+    },
+    {
+      "epoch": 2.530343007915567,
+      "grad_norm": 0.18920938670635223,
+      "learning_rate": 3.7257292548037917e-06,
+      "loss": 0.0046,
+      "step": 321
+    },
+    {
+      "epoch": 2.538258575197889,
+      "grad_norm": 0.5243077874183655,
+      "learning_rate": 3.715280477179382e-06,
+      "loss": 0.0044,
+      "step": 322
+    },
+    {
+      "epoch": 2.5461741424802113,
+      "grad_norm": 0.262588769197464,
+      "learning_rate": 3.7048038275790695e-06,
+      "loss": 0.0032,
+      "step": 323
+    },
+    {
+      "epoch": 2.554089709762533,
+      "grad_norm": 0.22281767427921295,
+      "learning_rate": 3.6942995462806574e-06,
+      "loss": 0.0042,
+      "step": 324
+    },
+    {
+      "epoch": 2.5620052770448547,
+      "grad_norm": 0.374078631401062,
+      "learning_rate": 3.6837678741956747e-06,
+      "loss": 0.0031,
+      "step": 325
+    },
+    {
+      "epoch": 2.569920844327177,
+      "grad_norm": 0.34161674976348877,
+      "learning_rate": 3.6732090528638432e-06,
+      "loss": 0.0098,
+      "step": 326
+    },
+    {
+      "epoch": 2.577836411609499,
+      "grad_norm": 0.8478104472160339,
+      "learning_rate": 3.6626233244475445e-06,
+      "loss": 0.003,
+      "step": 327
+    },
+    {
+      "epoch": 2.5857519788918206,
+      "grad_norm": 0.3653620481491089,
+      "learning_rate": 3.6520109317262624e-06,
+      "loss": 0.0053,
+      "step": 328
+    },
+    {
+      "epoch": 2.5936675461741423,
+      "grad_norm": 0.19305464625358582,
+      "learning_rate": 3.6413721180910165e-06,
+      "loss": 0.0017,
+      "step": 329
+    },
+    {
+      "epoch": 2.6015831134564644,
+      "grad_norm": 0.1898661106824875,
+      "learning_rate": 3.6307071275387807e-06,
+      "loss": 0.0024,
+      "step": 330
+    },
+    {
+      "epoch": 2.6094986807387865,
+      "grad_norm": 0.16431643068790436,
+      "learning_rate": 3.6200162046668826e-06,
+      "loss": 0.0058,
+      "step": 331
+    },
+    {
+      "epoch": 2.6174142480211082,
+      "grad_norm": 0.38869982957839966,
+      "learning_rate": 3.6092995946673996e-06,
+      "loss": 0.0065,
+      "step": 332
+    },
+    {
+      "epoch": 2.62532981530343,
+      "grad_norm": 0.344003289937973,
+      "learning_rate": 3.5985575433215345e-06,
+      "loss": 0.0059,
+      "step": 333
+    },
+    {
+      "epoch": 2.633245382585752,
+      "grad_norm": 1.8430119752883911,
+      "learning_rate": 3.587790296993976e-06,
+      "loss": 0.0045,
+      "step": 334
+    },
+    {
+      "epoch": 2.641160949868074,
+      "grad_norm": 0.22104960680007935,
+      "learning_rate": 3.5769981026272477e-06,
+      "loss": 0.0036,
+      "step": 335
+    },
+    {
+      "epoch": 2.649076517150396,
+      "grad_norm": 0.27808675169944763,
+      "learning_rate": 3.5661812077360496e-06,
+      "loss": 0.0055,
+      "step": 336
+    },
+    {
+      "epoch": 2.6569920844327175,
+      "grad_norm": 0.3757489323616028,
+      "learning_rate": 3.5553398604015777e-06,
+      "loss": 0.0086,
+      "step": 337
+    },
+    {
+      "epoch": 2.6649076517150396,
+      "grad_norm": 0.2972381114959717,
+      "learning_rate": 3.544474309265834e-06,
+      "loss": 0.0037,
+      "step": 338
+    },
+    {
+      "epoch": 2.6728232189973617,
+      "grad_norm": 0.2020997554063797,
+      "learning_rate": 3.5335848035259257e-06,
+      "loss": 0.0052,
+      "step": 339
+    },
+    {
+      "epoch": 2.6807387862796834,
+      "grad_norm": 0.13826540112495422,
+      "learning_rate": 3.5226715929283507e-06,
+      "loss": 0.003,
+      "step": 340
+    },
+    {
+      "epoch": 2.688654353562005,
+      "grad_norm": 0.2204088717699051,
+      "learning_rate": 3.511734927763265e-06,
+      "loss": 0.0036,
+      "step": 341
+    },
+    {
+      "epoch": 2.6965699208443272,
+      "grad_norm": 0.17721427977085114,
+      "learning_rate": 3.5007750588587495e-06,
+      "loss": 0.0053,
+      "step": 342
+    },
+    {
+      "epoch": 2.7044854881266494,
+      "grad_norm": 0.3896600902080536,
+      "learning_rate": 3.4897922375750517e-06,
+      "loss": 0.0051,
+      "step": 343
+    },
+    {
+      "epoch": 2.712401055408971,
+      "grad_norm": 0.13132958114147186,
+      "learning_rate": 3.478786715798823e-06,
+      "loss": 0.0012,
+      "step": 344
+    },
+    {
+      "epoch": 2.7203166226912927,
+      "grad_norm": 0.14952410757541656,
+      "learning_rate": 3.4677587459373417e-06,
+      "loss": 0.0015,
+      "step": 345
+    },
+    {
+      "epoch": 2.728232189973615,
+      "grad_norm": 0.19524404406547546,
+      "learning_rate": 3.4567085809127247e-06,
+      "loss": 0.0051,
+      "step": 346
+    },
+    {
+      "epoch": 2.736147757255937,
+      "grad_norm": 0.31595954298973083,
+      "learning_rate": 3.4456364741561256e-06,
+      "loss": 0.0045,
+      "step": 347
+    },
+    {
+      "epoch": 2.7440633245382586,
+      "grad_norm": 0.3021059036254883,
+      "learning_rate": 3.434542679601922e-06,
+      "loss": 0.0052,
+      "step": 348
+    },
+    {
+      "epoch": 2.7519788918205803,
+      "grad_norm": 0.38778263330459595,
+      "learning_rate": 3.423427451681895e-06,
+      "loss": 0.0086,
+      "step": 349
+    },
+    {
+      "epoch": 2.7598944591029024,
+      "grad_norm": 0.3644554913043976,
+      "learning_rate": 3.4122910453193885e-06,
+      "loss": 0.0009,
+      "step": 350
+    },
+    {
+      "epoch": 2.767810026385224,
+      "grad_norm": 0.18448686599731445,
+      "learning_rate": 3.4011337159234674e-06,
+      "loss": 0.0056,
+      "step": 351
+    },
+    {
+      "epoch": 2.7757255936675462,
+      "grad_norm": 0.2891385853290558,
+      "learning_rate": 3.3899557193830585e-06,
+      "loss": 0.0057,
+      "step": 352
+    },
+    {
+      "epoch": 2.783641160949868,
+      "grad_norm": 0.24937017261981964,
+      "learning_rate": 3.3787573120610794e-06,
+      "loss": 0.0022,
+      "step": 353
+    },
+    {
+      "epoch": 2.79155672823219,
+      "grad_norm": 0.3462439477443695,
+      "learning_rate": 3.367538750788563e-06,
+      "loss": 0.0018,
+      "step": 354
+    },
+    {
+      "epoch": 2.7994722955145117,
+      "grad_norm": 0.1728816032409668,
+      "learning_rate": 3.356300292858763e-06,
+      "loss": 0.0025,
+      "step": 355
+    },
+    {
+      "epoch": 2.807387862796834,
+      "grad_norm": 0.137979194521904,
+      "learning_rate": 3.345042196021257e-06,
+      "loss": 0.0065,
+      "step": 356
+    },
+    {
+      "epoch": 2.8153034300791555,
+      "grad_norm": 0.21424880623817444,
+      "learning_rate": 3.333764718476032e-06,
+      "loss": 0.0031,
+      "step": 357
+    },
+    {
+      "epoch": 2.8232189973614776,
+      "grad_norm": 0.29475298523902893,
+      "learning_rate": 3.3224681188675643e-06,
+      "loss": 0.0038,
+      "step": 358
+    },
+    {
+      "epoch": 2.8311345646437993,
+      "grad_norm": 0.26818305253982544,
+      "learning_rate": 3.3111526562788864e-06,
+      "loss": 0.0022,
+      "step": 359
+    },
+    {
+      "epoch": 2.8390501319261214,
+      "grad_norm": 0.25431951880455017,
+      "learning_rate": 3.2998185902256475e-06,
+      "loss": 0.0082,
+      "step": 360
+    },
+    {
+      "epoch": 2.846965699208443,
+      "grad_norm": 0.20935852825641632,
+      "learning_rate": 3.2884661806501576e-06,
+      "loss": 0.0034,
+      "step": 361
+    },
+    {
+      "epoch": 2.8548812664907652,
+      "grad_norm": 0.21449363231658936,
+      "learning_rate": 3.2770956879154305e-06,
+      "loss": 0.0029,
+      "step": 362
+    },
+    {
+      "epoch": 2.862796833773087,
+      "grad_norm": 0.11124449968338013,
+      "learning_rate": 3.2657073727992078e-06,
+      "loss": 0.0041,
+      "step": 363
+    },
+    {
+      "epoch": 2.870712401055409,
+      "grad_norm": 0.27211007475852966,
+      "learning_rate": 3.2543014964879814e-06,
+      "loss": 0.004,
+      "step": 364
+    },
+    {
+      "epoch": 2.8786279683377307,
+      "grad_norm": 0.18816880881786346,
+      "learning_rate": 3.2428783205710023e-06,
+      "loss": 0.0012,
+      "step": 365
+    },
+    {
+      "epoch": 2.886543535620053,
+      "grad_norm": 0.1929738074541092,
+      "learning_rate": 3.2314381070342815e-06,
+      "loss": 0.0012,
+      "step": 366
+    },
+    {
+      "epoch": 2.8944591029023745,
+      "grad_norm": 0.15621940791606903,
+      "learning_rate": 3.21998111825458e-06,
+      "loss": 0.0032,
+      "step": 367
+    },
+    {
+      "epoch": 2.9023746701846966,
+      "grad_norm": 0.2255018949508667,
+      "learning_rate": 3.208507616993393e-06,
+      "loss": 0.0047,
+      "step": 368
+    },
+    {
+      "epoch": 2.9102902374670183,
+      "grad_norm": 0.29748937487602234,
+      "learning_rate": 3.1970178663909233e-06,
+      "loss": 0.002,
+      "step": 369
+    },
+    {
+      "epoch": 2.9182058047493404,
+      "grad_norm": 0.35048818588256836,
+      "learning_rate": 3.1855121299600454e-06,
+      "loss": 0.0025,
+      "step": 370
+    },
+    {
+      "epoch": 2.926121372031662,
+      "grad_norm": 0.1799587458372116,
+      "learning_rate": 3.173990671580263e-06,
+      "loss": 0.0007,
+      "step": 371
+    },
+    {
+      "epoch": 2.9340369393139842,
+      "grad_norm": 0.15089461207389832,
+      "learning_rate": 3.162453755491655e-06,
+      "loss": 0.0035,
+      "step": 372
+    },
+    {
+      "epoch": 2.941952506596306,
+      "grad_norm": 0.20923982560634613,
+      "learning_rate": 3.1509016462888175e-06,
+      "loss": 0.0006,
+      "step": 373
+    },
+    {
+      "epoch": 2.949868073878628,
+      "grad_norm": 0.15215936303138733,
+      "learning_rate": 3.139334608914795e-06,
+      "loss": 0.0022,
+      "step": 374
+    },
+    {
+      "epoch": 2.9577836411609497,
+      "grad_norm": 0.054921459406614304,
+      "learning_rate": 3.1277529086550044e-06,
+      "loss": 0.0016,
+      "step": 375
+    },
+    {
+      "epoch": 2.965699208443272,
+      "grad_norm": 0.4386288523674011,
+      "learning_rate": 3.1161568111311487e-06,
+      "loss": 0.0036,
+      "step": 376
+    },
+    {
+      "epoch": 2.9736147757255935,
+      "grad_norm": 0.24071349203586578,
+      "learning_rate": 3.1045465822951265e-06,
+      "loss": 0.0024,
+      "step": 377
+    },
+    {
+      "epoch": 2.9815303430079156,
+      "grad_norm": 0.18303240835666656,
+      "learning_rate": 3.092922488422933e-06,
+      "loss": 0.0026,
+      "step": 378
+    },
+    {
+      "epoch": 2.9894459102902373,
+      "grad_norm": 0.24064595997333527,
+      "learning_rate": 3.0812847961085527e-06,
+      "loss": 0.0026,
+      "step": 379
+    },
+    {
+      "epoch": 2.9973614775725594,
+      "grad_norm": 0.18229585886001587,
+      "learning_rate": 3.0696337722578444e-06,
+      "loss": 0.0067,
+      "step": 380
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.3703795373439789,
+      "learning_rate": 3.057969684082421e-06,
+      "loss": 0.0002,
+      "step": 381
+    },
+    {
+      "epoch": 3.007915567282322,
+      "grad_norm": 0.1413760781288147,
+      "learning_rate": 3.04629279909352e-06,
+      "loss": 0.002,
+      "step": 382
+    },
+    {
+      "epoch": 3.015831134564644,
+      "grad_norm": 0.11547141522169113,
+      "learning_rate": 3.0346033850958685e-06,
+      "loss": 0.0011,
+      "step": 383
+    },
+    {
+      "epoch": 3.0237467018469655,
+      "grad_norm": 0.13543327152729034,
+      "learning_rate": 3.0229017101815424e-06,
+      "loss": 0.0009,
+      "step": 384
+    },
+    {
+      "epoch": 3.0316622691292876,
+      "grad_norm": 0.17435476183891296,
+      "learning_rate": 3.011188042723816e-06,
+      "loss": 0.0019,
+      "step": 385
+    },
+    {
+      "epoch": 3.0395778364116097,
+      "grad_norm": 0.38076433539390564,
+      "learning_rate": 2.9994626513710085e-06,
+      "loss": 0.0015,
+      "step": 386
+    },
+    {
+      "epoch": 3.0474934036939314,
+      "grad_norm": 0.30369430780410767,
+      "learning_rate": 2.9877258050403214e-06,
+      "loss": 0.0035,
+      "step": 387
+    },
+    {
+      "epoch": 3.055408970976253,
+      "grad_norm": 0.08789186179637909,
+      "learning_rate": 2.975977772911671e-06,
+      "loss": 0.0005,
+      "step": 388
+    },
+    {
+      "epoch": 3.063324538258575,
+      "grad_norm": 0.10484655946493149,
+      "learning_rate": 2.964218824421518e-06,
+      "loss": 0.001,
+      "step": 389
+    },
+    {
+      "epoch": 3.0712401055408973,
+      "grad_norm": 0.1550711691379547,
+      "learning_rate": 2.9524492292566824e-06,
+      "loss": 0.0019,
+      "step": 390
+    },
+    {
+      "epoch": 3.079155672823219,
+      "grad_norm": 0.1652969866991043,
+      "learning_rate": 2.9406692573481634e-06,
+      "loss": 0.0011,
+      "step": 391
+    },
+    {
+      "epoch": 3.0870712401055407,
+      "grad_norm": 0.4653196632862091,
+      "learning_rate": 2.928879178864946e-06,
+      "loss": 0.0033,
+      "step": 392
+    },
+    {
+      "epoch": 3.094986807387863,
+      "grad_norm": 0.3251839280128479,
+      "learning_rate": 2.9170792642078057e-06,
+      "loss": 0.0027,
+      "step": 393
+    },
+    {
+      "epoch": 3.1029023746701845,
+      "grad_norm": 0.23033401370048523,
+      "learning_rate": 2.9052697840031065e-06,
+      "loss": 0.0016,
+      "step": 394
+    },
+    {
+      "epoch": 3.1108179419525066,
+      "grad_norm": 0.2523248791694641,
+      "learning_rate": 2.8934510090965943e-06,
+      "loss": 0.003,
+      "step": 395
+    },
+    {
+      "epoch": 3.1187335092348283,
+      "grad_norm": 0.11632515490055084,
+      "learning_rate": 2.8816232105471864e-06,
+      "loss": 0.0026,
+      "step": 396
+    },
+    {
+      "epoch": 3.1266490765171504,
+      "grad_norm": 0.06233569234609604,
+      "learning_rate": 2.8697866596207524e-06,
+      "loss": 0.0003,
+      "step": 397
+    },
+    {
+      "epoch": 3.1345646437994725,
+      "grad_norm": 0.160960853099823,
+      "learning_rate": 2.8579416277838952e-06,
+      "loss": 0.0012,
+      "step": 398
+    },
+    {
+      "epoch": 3.142480211081794,
+      "grad_norm": 0.13908636569976807,
+      "learning_rate": 2.846088386697723e-06,
+      "loss": 0.0014,
+      "step": 399
+    },
+    {
+      "epoch": 3.150395778364116,
+      "grad_norm": 0.2525525987148285,
+      "learning_rate": 2.8342272082116214e-06,
+      "loss": 0.0011,
+      "step": 400
+    },
+    {
+      "epoch": 3.158311345646438,
+      "grad_norm": 0.11086013168096542,
+      "learning_rate": 2.822358364357015e-06,
+      "loss": 0.001,
+      "step": 401
+    },
+    {
+      "epoch": 3.16622691292876,
+      "grad_norm": 0.10374631732702255,
+      "learning_rate": 2.8104821273411333e-06,
+      "loss": 0.001,
+      "step": 402
+    },
+    {
+      "epoch": 3.174142480211082,
+      "grad_norm": 0.09266223013401031,
+      "learning_rate": 2.7985987695407618e-06,
+      "loss": 0.0011,
+      "step": 403
+    },
+    {
+      "epoch": 3.1820580474934035,
+      "grad_norm": 0.05143650993704796,
+      "learning_rate": 2.786708563496002e-06,
+      "loss": 0.0003,
+      "step": 404
+    },
+    {
+      "epoch": 3.1899736147757256,
+      "grad_norm": 0.1214371770620346,
+      "learning_rate": 2.774811781904013e-06,
+      "loss": 0.0005,
+      "step": 405
+    },
+    {
+      "epoch": 3.1978891820580473,
+      "grad_norm": 0.02491622418165207,
+      "learning_rate": 2.762908697612765e-06,
+      "loss": 0.0002,
+      "step": 406
+    },
+    {
+      "epoch": 3.2058047493403694,
+      "grad_norm": 0.15371961891651154,
+      "learning_rate": 2.750999583614777e-06,
+      "loss": 0.0016,
+      "step": 407
+    },
+    {
+      "epoch": 3.213720316622691,
+      "grad_norm": 0.04354507476091385,
+      "learning_rate": 2.739084713040856e-06,
+      "loss": 0.0003,
+      "step": 408
+    },
+    {
+      "epoch": 3.221635883905013,
+      "grad_norm": 0.12696519494056702,
+      "learning_rate": 2.7271643591538355e-06,
+      "loss": 0.002,
+      "step": 409
+    },
+    {
+      "epoch": 3.229551451187335,
+      "grad_norm": 0.3985283374786377,
+      "learning_rate": 2.7152387953423047e-06,
+      "loss": 0.001,
+      "step": 410
+    },
+    {
+      "epoch": 3.237467018469657,
+      "grad_norm": 0.15001602470874786,
+      "learning_rate": 2.703308295114342e-06,
+      "loss": 0.001,
+      "step": 411
+    },
+    {
+      "epoch": 3.2453825857519787,
+      "grad_norm": 0.024628320708870888,
+      "learning_rate": 2.69137313209124e-06,
+      "loss": 0.0001,
+      "step": 412
+    },
+    {
+      "epoch": 3.253298153034301,
+      "grad_norm": 0.08502338081598282,
+      "learning_rate": 2.6794335800012294e-06,
+      "loss": 0.0004,
+      "step": 413
+    },
+    {
+      "epoch": 3.261213720316623,
+      "grad_norm": 0.09296593815088272,
+      "learning_rate": 2.6674899126732045e-06,
+      "loss": 0.0002,
+      "step": 414
+    },
+    {
+      "epoch": 3.2691292875989446,
+      "grad_norm": 0.08725863695144653,
+      "learning_rate": 2.65554240403044e-06,
+      "loss": 0.0005,
+      "step": 415
+    },
+    {
+      "epoch": 3.2770448548812663,
+      "grad_norm": 0.19437682628631592,
+      "learning_rate": 2.643591328084309e-06,
+      "loss": 0.0009,
+      "step": 416
+    },
+    {
+      "epoch": 3.2849604221635884,
+      "grad_norm": 0.4695538580417633,
+      "learning_rate": 2.631636958928e-06,
+      "loss": 0.0068,
+      "step": 417
+    },
+    {
+      "epoch": 3.2928759894459105,
+      "grad_norm": 0.09818229079246521,
+      "learning_rate": 2.6196795707302304e-06,
+      "loss": 0.0008,
+      "step": 418
+    },
+    {
+      "epoch": 3.300791556728232,
+      "grad_norm": 0.012197593227028847,
+      "learning_rate": 2.607719437728957e-06,
+      "loss": 0.0001,
+      "step": 419
+    },
+    {
+      "epoch": 3.308707124010554,
+      "grad_norm": 0.04225357249379158,
+      "learning_rate": 2.595756834225089e-06,
+      "loss": 0.0002,
+      "step": 420
+    },
+    {
+      "epoch": 3.316622691292876,
+      "grad_norm": 0.11543109267950058,
+      "learning_rate": 2.583792034576194e-06,
+      "loss": 0.0004,
+      "step": 421
+    },
+    {
+      "epoch": 3.324538258575198,
+      "grad_norm": 0.3622385263442993,
+      "learning_rate": 2.5718253131902084e-06,
+      "loss": 0.0013,
+      "step": 422
+    },
+    {
+      "epoch": 3.33245382585752,
+      "grad_norm": 0.10956711322069168,
+      "learning_rate": 2.5598569445191418e-06,
+      "loss": 0.0003,
+      "step": 423
+    },
+    {
+      "epoch": 3.3403693931398415,
+      "grad_norm": 0.05142383649945259,
+      "learning_rate": 2.547887203052786e-06,
+      "loss": 0.0003,
+      "step": 424
+    },
+    {
+      "epoch": 3.3482849604221636,
+      "grad_norm": 0.13771450519561768,
+      "learning_rate": 2.535916363312414e-06,
+      "loss": 0.0011,
+      "step": 425
+    },
+    {
+      "epoch": 3.3562005277044857,
+      "grad_norm": 0.1246413066983223,
+      "learning_rate": 2.52394469984449e-06,
+      "loss": 0.0011,
+      "step": 426
+    },
+    {
+      "epoch": 3.3641160949868074,
+      "grad_norm": 0.33866506814956665,
+      "learning_rate": 2.5119724872143693e-06,
+      "loss": 0.0013,
+      "step": 427
+    },
+    {
+      "epoch": 3.372031662269129,
+      "grad_norm": 0.24942219257354736,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0016,
+      "step": 428
+    },
+    {
+      "epoch": 3.379947229551451,
+      "grad_norm": 0.09010064601898193,
+      "learning_rate": 2.4880275127856324e-06,
+      "loss": 0.0003,
+      "step": 429
+    },
+    {
+      "epoch": 3.387862796833773,
+      "grad_norm": 0.1901097297668457,
+      "learning_rate": 2.4760553001555105e-06,
+      "loss": 0.002,
+      "step": 430
+    },
+    {
+      "epoch": 3.395778364116095,
+      "grad_norm": 0.16615132987499237,
+      "learning_rate": 2.4640836366875872e-06,
+      "loss": 0.0011,
+      "step": 431
+    },
+    {
+      "epoch": 3.4036939313984167,
+      "grad_norm": 0.1473286896944046,
+      "learning_rate": 2.452112796947215e-06,
+      "loss": 0.0017,
+      "step": 432
+    },
+    {
+      "epoch": 3.411609498680739,
+      "grad_norm": 0.07820681482553482,
+      "learning_rate": 2.440143055480859e-06,
+      "loss": 0.0003,
+      "step": 433
+    },
+    {
+      "epoch": 3.4195250659630605,
+      "grad_norm": 0.32423466444015503,
+      "learning_rate": 2.428174686809793e-06,
+      "loss": 0.0025,
+      "step": 434
+    },
+    {
+      "epoch": 3.4274406332453826,
+      "grad_norm": 0.02954992465674877,
+      "learning_rate": 2.4162079654238073e-06,
+      "loss": 0.0001,
+      "step": 435
+    },
+    {
+      "epoch": 3.4353562005277043,
+      "grad_norm": 0.04201340302824974,
+      "learning_rate": 2.404243165774912e-06,
+      "loss": 0.0003,
+      "step": 436
+    },
+    {
+      "epoch": 3.4432717678100264,
+      "grad_norm": 0.03119390271604061,
+      "learning_rate": 2.392280562271044e-06,
+      "loss": 0.0001,
+      "step": 437
+    },
+    {
+      "epoch": 3.451187335092348,
+      "grad_norm": 0.03694690391421318,
+      "learning_rate": 2.3803204292697705e-06,
+      "loss": 0.0002,
+      "step": 438
+    },
+    {
+      "epoch": 3.45910290237467,
+      "grad_norm": 0.08830390870571136,
+      "learning_rate": 2.3683630410720013e-06,
+      "loss": 0.0006,
+      "step": 439
+    },
+    {
+      "epoch": 3.467018469656992,
+      "grad_norm": 1.7088481187820435,
+      "learning_rate": 2.356408671915692e-06,
+      "loss": 0.001,
+      "step": 440
+    },
+    {
+      "epoch": 3.474934036939314,
+      "grad_norm": 0.13228784501552582,
+      "learning_rate": 2.3444575959695615e-06,
+      "loss": 0.0007,
+      "step": 441
+    },
+    {
+      "epoch": 3.4828496042216357,
+      "grad_norm": 0.01566927880048752,
+      "learning_rate": 2.3325100873267963e-06,
+      "loss": 0.0001,
+      "step": 442
+    },
+    {
+      "epoch": 3.490765171503958,
+      "grad_norm": 0.03420136868953705,
+      "learning_rate": 2.320566419998772e-06,
+      "loss": 0.0002,
+      "step": 443
+    },
+    {
+      "epoch": 3.4986807387862795,
+      "grad_norm": 0.2430853545665741,
+      "learning_rate": 2.308626867908761e-06,
+      "loss": 0.0008,
+      "step": 444
+    },
+    {
+      "epoch": 3.5065963060686016,
+      "grad_norm": 0.1316990852355957,
+      "learning_rate": 2.296691704885659e-06,
+      "loss": 0.0012,
+      "step": 445
+    },
+    {
+      "epoch": 3.5145118733509237,
+      "grad_norm": 0.3267118036746979,
+      "learning_rate": 2.284761204657696e-06,
+      "loss": 0.0015,
+      "step": 446
+    },
+    {
+      "epoch": 3.5224274406332454,
+      "grad_norm": 0.20723529160022736,
+      "learning_rate": 2.2728356408461653e-06,
+      "loss": 0.003,
+      "step": 447
+    },
+    {
+      "epoch": 3.530343007915567,
+      "grad_norm": 0.03687990829348564,
+      "learning_rate": 2.2609152869591445e-06,
+      "loss": 0.0002,
+      "step": 448
+    },
+    {
+      "epoch": 3.538258575197889,
+      "grad_norm": 0.03033611737191677,
+      "learning_rate": 2.249000416385224e-06,
+      "loss": 0.0002,
+      "step": 449
+    },
+    {
+      "epoch": 3.5461741424802113,
+      "grad_norm": 0.0805787518620491,
+      "learning_rate": 2.2370913023872357e-06,
+      "loss": 0.0008,
+      "step": 450
+    },
+    {
+      "epoch": 3.554089709762533,
+      "grad_norm": 0.11575227975845337,
+      "learning_rate": 2.2251882180959876e-06,
+      "loss": 0.0009,
+      "step": 451
+    },
+    {
+      "epoch": 3.5620052770448547,
+      "grad_norm": 0.022496838122606277,
+      "learning_rate": 2.2132914365039993e-06,
+      "loss": 0.0002,
+      "step": 452
+    },
+    {
+      "epoch": 3.569920844327177,
+      "grad_norm": 0.12225953489542007,
+      "learning_rate": 2.201401230459239e-06,
+      "loss": 0.0007,
+      "step": 453
+    },
+    {
+      "epoch": 3.577836411609499,
+      "grad_norm": 0.08851709961891174,
+      "learning_rate": 2.189517872658867e-06,
+      "loss": 0.0007,
+      "step": 454
+    },
+    {
+      "epoch": 3.5857519788918206,
+      "grad_norm": 0.15894952416419983,
+      "learning_rate": 2.1776416356429857e-06,
+      "loss": 0.0016,
+      "step": 455
+    },
+    {
+      "epoch": 3.5936675461741423,
+      "grad_norm": 0.06427901983261108,
+      "learning_rate": 2.165772791788379e-06,
+      "loss": 0.0004,
+      "step": 456
+    },
+    {
+      "epoch": 3.6015831134564644,
+      "grad_norm": 0.4027339816093445,
+      "learning_rate": 2.1539116133022776e-06,
+      "loss": 0.0011,
+      "step": 457
+    },
+    {
+      "epoch": 3.6094986807387865,
+      "grad_norm": 0.0348910428583622,
+      "learning_rate": 2.142058372216105e-06,
+      "loss": 0.0002,
+      "step": 458
+    },
+    {
+      "epoch": 3.6174142480211082,
+      "grad_norm": 0.05866856500506401,
+      "learning_rate": 2.130213340379248e-06,
+      "loss": 0.0004,
+      "step": 459
+    },
+    {
+      "epoch": 3.62532981530343,
+      "grad_norm": 0.13465961813926697,
+      "learning_rate": 2.1183767894528135e-06,
+      "loss": 0.0011,
+      "step": 460
+    },
+    {
+      "epoch": 3.633245382585752,
+      "grad_norm": 0.03272956609725952,
+      "learning_rate": 2.1065489909034065e-06,
+      "loss": 0.0002,
+      "step": 461
+    },
+    {
+      "epoch": 3.641160949868074,
+      "grad_norm": 0.06437230855226517,
+      "learning_rate": 2.094730215996894e-06,
+      "loss": 0.0003,
+      "step": 462
+    },
+    {
+      "epoch": 3.649076517150396,
+      "grad_norm": 0.07983466237783432,
+      "learning_rate": 2.082920735792195e-06,
+      "loss": 0.0006,
+      "step": 463
+    },
+    {
+      "epoch": 3.6569920844327175,
+      "grad_norm": 0.09880750626325607,
+      "learning_rate": 2.0711208211350543e-06,
+      "loss": 0.0004,
+      "step": 464
+    },
+    {
+      "epoch": 3.6649076517150396,
+      "grad_norm": 0.09212072193622589,
+      "learning_rate": 2.059330742651837e-06,
+      "loss": 0.0006,
+      "step": 465
+    },
+    {
+      "epoch": 3.6728232189973617,
+      "grad_norm": 0.276640921831131,
+      "learning_rate": 2.047550770743318e-06,
+      "loss": 0.0012,
+      "step": 466
+    },
+    {
+      "epoch": 3.6807387862796834,
+      "grad_norm": 0.11709734052419662,
+      "learning_rate": 2.035781175578483e-06,
+      "loss": 0.0005,
+      "step": 467
+    },
+    {
+      "epoch": 3.688654353562005,
+      "grad_norm": 0.023123467341065407,
+      "learning_rate": 2.024022227088329e-06,
+      "loss": 0.0002,
+      "step": 468
+    },
+    {
+      "epoch": 3.6965699208443272,
+      "grad_norm": 0.014415634796023369,
+      "learning_rate": 2.01227419495968e-06,
+      "loss": 0.0001,
+      "step": 469
+    },
+    {
+      "epoch": 3.7044854881266494,
+      "grad_norm": 0.018584486097097397,
+      "learning_rate": 2.0005373486289932e-06,
+      "loss": 0.0001,
+      "step": 470
+    },
+    {
+      "epoch": 3.712401055408971,
+      "grad_norm": 0.02571115642786026,
+      "learning_rate": 1.988811957276185e-06,
+      "loss": 0.0002,
+      "step": 471
+    },
+    {
+      "epoch": 3.7203166226912927,
+      "grad_norm": 0.02951018139719963,
+      "learning_rate": 1.977098289818459e-06,
+      "loss": 0.0001,
+      "step": 472
+    },
+    {
+      "epoch": 3.728232189973615,
+      "grad_norm": 0.010080009698867798,
+      "learning_rate": 1.9653966149041323e-06,
+      "loss": 0.0001,
+      "step": 473
+    },
+    {
+      "epoch": 3.736147757255937,
+      "grad_norm": 0.008738484233617783,
+      "learning_rate": 1.9537072009064814e-06,
+      "loss": 0.0001,
+      "step": 474
+    },
+    {
+      "epoch": 3.7440633245382586,
+      "grad_norm": 0.0930676981806755,
+      "learning_rate": 1.9420303159175795e-06,
+      "loss": 0.0004,
+      "step": 475
+    },
+    {
+      "epoch": 3.7519788918205803,
+      "grad_norm": 0.4496051073074341,
+      "learning_rate": 1.930366227742157e-06,
+      "loss": 0.0051,
+      "step": 476
+    },
+    {
+      "epoch": 3.7598944591029024,
+      "grad_norm": 0.005042586475610733,
+      "learning_rate": 1.918715203891448e-06,
+      "loss": 0.0001,
+      "step": 477
+    },
+    {
+      "epoch": 3.767810026385224,
+      "grad_norm": 0.1514541506767273,
+      "learning_rate": 1.907077511577068e-06,
+      "loss": 0.0006,
+      "step": 478
+    },
+    {
+      "epoch": 3.7757255936675462,
+      "grad_norm": 0.08802618831396103,
+      "learning_rate": 1.8954534177048744e-06,
+      "loss": 0.0005,
+      "step": 479
+    },
+    {
+      "epoch": 3.783641160949868,
+      "grad_norm": 0.05380689352750778,
+      "learning_rate": 1.8838431888688528e-06,
+      "loss": 0.0003,
+      "step": 480
+    },
+    {
+      "epoch": 3.79155672823219,
+      "grad_norm": 0.03282235190272331,
+      "learning_rate": 1.8722470913449962e-06,
+      "loss": 0.0001,
+      "step": 481
+    },
+    {
+      "epoch": 3.7994722955145117,
+      "grad_norm": 0.1280638724565506,
+      "learning_rate": 1.8606653910852058e-06,
+      "loss": 0.0004,
+      "step": 482
+    },
+    {
+      "epoch": 3.807387862796834,
+      "grad_norm": 0.025515379384160042,
+      "learning_rate": 1.8490983537111831e-06,
+      "loss": 0.0001,
+      "step": 483
+    },
+    {
+      "epoch": 3.8153034300791555,
+      "grad_norm": 0.2563674747943878,
+      "learning_rate": 1.8375462445083464e-06,
+      "loss": 0.0013,
+      "step": 484
+    },
+    {
+      "epoch": 3.8232189973614776,
+      "grad_norm": 0.14869002997875214,
+      "learning_rate": 1.8260093284197371e-06,
+      "loss": 0.0004,
+      "step": 485
+    },
+    {
+      "epoch": 3.8311345646437993,
+      "grad_norm": 0.028036003932356834,
+      "learning_rate": 1.814487870039955e-06,
+      "loss": 0.0001,
+      "step": 486
+    },
+    {
+      "epoch": 3.8390501319261214,
+      "grad_norm": 0.03774454444646835,
+      "learning_rate": 1.8029821336090769e-06,
+      "loss": 0.0002,
+      "step": 487
+    },
+    {
+      "epoch": 3.846965699208443,
+      "grad_norm": 0.01931951940059662,
+      "learning_rate": 1.7914923830066074e-06,
+      "loss": 0.0001,
+      "step": 488
+    },
+    {
+      "epoch": 3.8548812664907652,
+      "grad_norm": 0.031951338052749634,
+      "learning_rate": 1.7800188817454209e-06,
+      "loss": 0.0002,
+      "step": 489
+    },
+    {
+      "epoch": 3.862796833773087,
+      "grad_norm": 0.14470775425434113,
+      "learning_rate": 1.7685618929657193e-06,
+      "loss": 0.0006,
+      "step": 490
+    },
+    {
+      "epoch": 3.870712401055409,
+      "grad_norm": 0.17831256985664368,
+      "learning_rate": 1.7571216794289985e-06,
+      "loss": 0.0014,
+      "step": 491
+    },
+    {
+      "epoch": 3.8786279683377307,
+      "grad_norm": 0.027009831741452217,
+      "learning_rate": 1.7456985035120194e-06,
+      "loss": 0.0001,
+      "step": 492
+    },
+    {
+      "epoch": 3.886543535620053,
+      "grad_norm": 0.07074262201786041,
+      "learning_rate": 1.734292627200793e-06,
+      "loss": 0.0001,
+      "step": 493
+    },
+    {
+      "epoch": 3.8944591029023745,
+      "grad_norm": 0.04783803969621658,
+      "learning_rate": 1.7229043120845705e-06,
+      "loss": 0.0002,
+      "step": 494
+    },
+    {
+      "epoch": 3.9023746701846966,
+      "grad_norm": 0.1035776138305664,
+      "learning_rate": 1.711533819349842e-06,
+      "loss": 0.0004,
+      "step": 495
+    },
+    {
+      "epoch": 3.9102902374670183,
+      "grad_norm": 0.2424655705690384,
+      "learning_rate": 1.700181409774353e-06,
+      "loss": 0.0003,
+      "step": 496
+    },
+    {
+      "epoch": 3.9182058047493404,
+      "grad_norm": 0.2336961030960083,
+      "learning_rate": 1.6888473437211134e-06,
+      "loss": 0.0016,
+      "step": 497
+    },
+    {
+      "epoch": 3.926121372031662,
+      "grad_norm": 0.04740168899297714,
+      "learning_rate": 1.6775318811324365e-06,
+      "loss": 0.0002,
+      "step": 498
+    },
+    {
+      "epoch": 3.9340369393139842,
+      "grad_norm": 0.01394867803901434,
+      "learning_rate": 1.666235281523968e-06,
+      "loss": 0.0001,
+      "step": 499
+    },
+    {
+      "epoch": 3.941952506596306,
+      "grad_norm": 0.17383810877799988,
+      "learning_rate": 1.6549578039787436e-06,
+      "loss": 0.0009,
+      "step": 500
+    },
+    {
+      "epoch": 3.949868073878628,
+      "grad_norm": 0.07459729164838791,
+      "learning_rate": 1.643699707141237e-06,
+      "loss": 0.0004,
+      "step": 501
+    },
+    {
+      "epoch": 3.9577836411609497,
+      "grad_norm": 0.04102525860071182,
+      "learning_rate": 1.6324612492114378e-06,
+      "loss": 0.0002,
+      "step": 502
+    },
+    {
+      "epoch": 3.965699208443272,
+      "grad_norm": 0.038857750594615936,
+      "learning_rate": 1.6212426879389205e-06,
+      "loss": 0.0002,
+      "step": 503
+    },
+    {
+      "epoch": 3.9736147757255935,
+      "grad_norm": 0.0759626254439354,
+      "learning_rate": 1.6100442806169423e-06,
+      "loss": 0.0002,
+      "step": 504
+    },
+    {
+      "epoch": 3.9815303430079156,
+      "grad_norm": 0.10411451011896133,
+      "learning_rate": 1.5988662840765323e-06,
+      "loss": 0.0005,
+      "step": 505
+    },
+    {
+      "epoch": 3.9894459102902373,
+      "grad_norm": 0.041932545602321625,
+      "learning_rate": 1.5877089546806123e-06,
+      "loss": 0.0001,
+      "step": 506
+    },
+    {
+      "epoch": 3.9973614775725594,
+      "grad_norm": 0.14835326373577118,
+      "learning_rate": 1.5765725483181056e-06,
+      "loss": 0.0004,
+      "step": 507
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.14835326373577118,
+      "learning_rate": 1.5654573203980782e-06,
+      "loss": 0.0002,
+      "step": 508
+    },
+    {
+      "epoch": 4.007915567282322,
+      "grad_norm": 0.09130535274744034,
+      "learning_rate": 1.5543635258438746e-06,
+      "loss": 0.0002,
+      "step": 509
+    },
+    {
+      "epoch": 4.015831134564644,
+      "grad_norm": 0.020162789151072502,
+      "learning_rate": 1.5432914190872757e-06,
+      "loss": 0.0001,
+      "step": 510
+    },
+    {
+      "epoch": 4.0237467018469655,
+      "grad_norm": 0.006909206975251436,
+      "learning_rate": 1.5322412540626593e-06,
+      "loss": 0.0001,
+      "step": 511
+    },
+    {
+      "epoch": 4.031662269129288,
+      "grad_norm": 0.015305164270102978,
+      "learning_rate": 1.5212132842011778e-06,
+      "loss": 0.0001,
+      "step": 512
+    },
+    {
+      "epoch": 4.03957783641161,
+      "grad_norm": 0.021421000361442566,
+      "learning_rate": 1.5102077624249498e-06,
+      "loss": 0.0001,
+      "step": 513
+    },
+    {
+      "epoch": 4.047493403693931,
+      "grad_norm": 0.009534668177366257,
+      "learning_rate": 1.4992249411412514e-06,
+      "loss": 0.0001,
+      "step": 514
+    },
+    {
+      "epoch": 4.055408970976253,
+      "grad_norm": 0.007734082639217377,
+      "learning_rate": 1.4882650722367365e-06,
+      "loss": 0.0001,
+      "step": 515
+    },
+    {
+      "epoch": 4.063324538258575,
+      "grad_norm": 0.028215201571583748,
+      "learning_rate": 1.4773284070716504e-06,
+      "loss": 0.0001,
+      "step": 516
+    },
+    {
+      "epoch": 4.071240105540897,
+      "grad_norm": 0.009350771084427834,
+      "learning_rate": 1.4664151964740754e-06,
+      "loss": 0.0001,
+      "step": 517
+    },
+    {
+      "epoch": 4.0791556728232194,
+      "grad_norm": 0.06418479979038239,
+      "learning_rate": 1.4555256907341668e-06,
+      "loss": 0.0001,
+      "step": 518
+    },
+    {
+      "epoch": 4.087071240105541,
+      "grad_norm": 0.008197133429348469,
+      "learning_rate": 1.4446601395984233e-06,
+      "loss": 0.0001,
+      "step": 519
+    },
+    {
+      "epoch": 4.094986807387863,
+      "grad_norm": 0.021477939561009407,
+      "learning_rate": 1.4338187922639506e-06,
+      "loss": 0.0001,
+      "step": 520
+    },
+    {
+      "epoch": 4.102902374670185,
+      "grad_norm": 0.036561910063028336,
+      "learning_rate": 1.4230018973727535e-06,
+      "loss": 0.0001,
+      "step": 521
+    },
+    {
+      "epoch": 4.110817941952506,
+      "grad_norm": 0.011644291691482067,
+      "learning_rate": 1.412209703006025e-06,
+      "loss": 0.0001,
+      "step": 522
+    },
+    {
+      "epoch": 4.118733509234828,
+      "grad_norm": 0.0057818996720016,
+      "learning_rate": 1.4014424566784663e-06,
+      "loss": 0.0001,
+      "step": 523
+    },
+    {
+      "epoch": 4.12664907651715,
+      "grad_norm": 0.004692242946475744,
+      "learning_rate": 1.3907004053326006e-06,
+      "loss": 0.0001,
+      "step": 524
+    },
+    {
+      "epoch": 4.1345646437994725,
+      "grad_norm": 0.005857614800333977,
+      "learning_rate": 1.3799837953331191e-06,
+      "loss": 0.0012,
+      "step": 525
+    },
+    {
+      "epoch": 4.142480211081795,
+      "grad_norm": 0.22720293700695038,
+      "learning_rate": 1.3692928724612204e-06,
+      "loss": 0.0001,
+      "step": 526
+    },
+    {
+      "epoch": 4.150395778364116,
+      "grad_norm": 0.006255078595131636,
+      "learning_rate": 1.3586278819089837e-06,
+      "loss": 0.0001,
+      "step": 527
+    },
+    {
+      "epoch": 4.158311345646438,
+      "grad_norm": 0.010897025465965271,
+      "learning_rate": 1.347989068273738e-06,
+      "loss": 0.0001,
+      "step": 528
+    },
+    {
+      "epoch": 4.16622691292876,
+      "grad_norm": 0.00591659639030695,
+      "learning_rate": 1.3373766755524564e-06,
+      "loss": 0.0001,
+      "step": 529
+    },
+    {
+      "epoch": 4.174142480211081,
+      "grad_norm": 0.025465618818998337,
+      "learning_rate": 1.3267909471361574e-06,
+      "loss": 0.0001,
+      "step": 530
+    },
+    {
+      "epoch": 4.1820580474934035,
+      "grad_norm": 0.02871805801987648,
+      "learning_rate": 1.3162321258043261e-06,
+      "loss": 0.0001,
+      "step": 531
+    },
+    {
+      "epoch": 4.189973614775726,
+      "grad_norm": 0.008329014293849468,
+      "learning_rate": 1.3057004537193424e-06,
+      "loss": 0.0002,
+      "step": 532
+    },
+    {
+      "epoch": 4.197889182058048,
+      "grad_norm": 0.08574787527322769,
+      "learning_rate": 1.2951961724209317e-06,
+      "loss": 0.0,
+      "step": 533
+    },
+    {
+      "epoch": 4.205804749340369,
+      "grad_norm": 0.008351406082510948,
+      "learning_rate": 1.284719522820618e-06,
+      "loss": 0.0001,
+      "step": 534
+    },
+    {
+      "epoch": 4.213720316622691,
+      "grad_norm": 0.004138413816690445,
+      "learning_rate": 1.274270745196209e-06,
+      "loss": 0.0001,
+      "step": 535
+    },
+    {
+      "epoch": 4.221635883905013,
+      "grad_norm": 0.040472522377967834,
+      "learning_rate": 1.263850079186274e-06,
+      "loss": 0.0001,
+      "step": 536
+    },
+    {
+      "epoch": 4.229551451187335,
+      "grad_norm": 0.034887202084064484,
+      "learning_rate": 1.253457763784659e-06,
+      "loss": 0.0001,
+      "step": 537
+    },
+    {
+      "epoch": 4.237467018469657,
+      "grad_norm": 0.029348624870181084,
+      "learning_rate": 1.2430940373349944e-06,
+      "loss": 0.0002,
+      "step": 538
+    },
+    {
+      "epoch": 4.245382585751979,
+      "grad_norm": 0.302213579416275,
+      "learning_rate": 1.2327591375252404e-06,
+      "loss": 0.0019,
+      "step": 539
+    },
+    {
+      "epoch": 4.253298153034301,
+      "grad_norm": 0.03044111467897892,
+      "learning_rate": 1.2224533013822237e-06,
+      "loss": 0.0001,
+      "step": 540
+    },
+    {
+      "epoch": 4.261213720316623,
+      "grad_norm": 0.03175894170999527,
+      "learning_rate": 1.2121767652662122e-06,
+      "loss": 0.0001,
+      "step": 541
+    },
+    {
+      "epoch": 4.269129287598945,
+      "grad_norm": 0.010581142269074917,
+      "learning_rate": 1.2019297648654856e-06,
+      "loss": 0.0001,
+      "step": 542
+    },
+    {
+      "epoch": 4.277044854881266,
+      "grad_norm": 0.018574964255094528,
+      "learning_rate": 1.191712535190937e-06,
+      "loss": 0.0001,
+      "step": 543
+    },
+    {
+      "epoch": 4.284960422163588,
+      "grad_norm": 0.024211766198277473,
+      "learning_rate": 1.181525310570677e-06,
+      "loss": 0.0001,
+      "step": 544
+    },
+    {
+      "epoch": 4.2928759894459105,
+      "grad_norm": 0.024882890284061432,
+      "learning_rate": 1.1713683246446622e-06,
+      "loss": 0.0001,
+      "step": 545
+    },
+    {
+      "epoch": 4.300791556728232,
+      "grad_norm": 0.15666691958904266,
+      "learning_rate": 1.16124181035934e-06,
+      "loss": 0.001,
+      "step": 546
+    },
+    {
+      "epoch": 4.308707124010554,
+      "grad_norm": 0.08423323184251785,
+      "learning_rate": 1.1511459999622982e-06,
+      "loss": 0.0002,
+      "step": 547
+    },
+    {
+      "epoch": 4.316622691292876,
+      "grad_norm": 0.004608614835888147,
+      "learning_rate": 1.1410811249969475e-06,
+      "loss": 0.0,
+      "step": 548
+    },
+    {
+      "epoch": 4.324538258575198,
+      "grad_norm": 0.0066484021954238415,
+      "learning_rate": 1.1310474162972026e-06,
+      "loss": 0.0001,
+      "step": 549
+    },
+    {
+      "epoch": 4.33245382585752,
+      "grad_norm": 0.012281147763133049,
+      "learning_rate": 1.1210451039821965e-06,
+      "loss": 0.0001,
+      "step": 550
+    },
+    {
+      "epoch": 4.3403693931398415,
+      "grad_norm": 0.012285086326301098,
+      "learning_rate": 1.1110744174509952e-06,
+      "loss": 0.0001,
+      "step": 551
+    },
+    {
+      "epoch": 4.348284960422164,
+      "grad_norm": 0.15158827602863312,
+      "learning_rate": 1.10113558537734e-06,
+      "loss": 0.0014,
+      "step": 552
+    },
+    {
+      "epoch": 4.356200527704486,
+      "grad_norm": 0.09957592934370041,
+      "learning_rate": 1.0912288357044063e-06,
+      "loss": 0.0004,
+      "step": 553
+    },
+    {
+      "epoch": 4.364116094986807,
+      "grad_norm": 0.0042252130806446075,
+      "learning_rate": 1.0813543956395675e-06,
+      "loss": 0.0001,
+      "step": 554
+    },
+    {
+      "epoch": 4.372031662269129,
+      "grad_norm": 0.0247097946703434,
+      "learning_rate": 1.0715124916491937e-06,
+      "loss": 0.0001,
+      "step": 555
+    },
+    {
+      "epoch": 4.379947229551451,
+      "grad_norm": 0.2519308924674988,
+      "learning_rate": 1.0617033494534486e-06,
+      "loss": 0.0011,
+      "step": 556
+    },
+    {
+      "epoch": 4.387862796833773,
+      "grad_norm": 0.016611002385616302,
+      "learning_rate": 1.0519271940211214e-06,
+      "loss": 0.0001,
+      "step": 557
+    },
+    {
+      "epoch": 4.395778364116095,
+      "grad_norm": 0.031726885586977005,
+      "learning_rate": 1.0421842495644588e-06,
+      "loss": 0.0001,
+      "step": 558
+    },
+    {
+      "epoch": 4.403693931398417,
+      "grad_norm": 0.008313744328916073,
+      "learning_rate": 1.032474739534031e-06,
+      "loss": 0.0001,
+      "step": 559
+    },
+    {
+      "epoch": 4.411609498680739,
+      "grad_norm": 0.018961846828460693,
+      "learning_rate": 1.0227988866135995e-06,
+      "loss": 0.0001,
+      "step": 560
+    },
+    {
+      "epoch": 4.419525065963061,
+      "grad_norm": 0.019846709445118904,
+      "learning_rate": 1.0131569127150143e-06,
+      "loss": 0.0001,
+      "step": 561
+    },
+    {
+      "epoch": 4.427440633245382,
+      "grad_norm": 0.011897215619683266,
+      "learning_rate": 1.0035490389731257e-06,
+      "loss": 0.0003,
+      "step": 562
+    },
+    {
+      "epoch": 4.435356200527704,
+      "grad_norm": 0.07276739925146103,
+      "learning_rate": 9.939754857407064e-07,
+      "loss": 0.0002,
+      "step": 563
+    },
+    {
+      "epoch": 4.443271767810026,
+      "grad_norm": 0.016536539420485497,
+      "learning_rate": 9.844364725834058e-07,
+      "loss": 0.0001,
+      "step": 564
+    },
+    {
+      "epoch": 4.4511873350923485,
+      "grad_norm": 0.01340264268219471,
+      "learning_rate": 9.749322182747074e-07,
+      "loss": 0.0001,
+      "step": 565
+    },
+    {
+      "epoch": 4.45910290237467,
+      "grad_norm": 0.01568949595093727,
+      "learning_rate": 9.654629407909163e-07,
+      "loss": 0.0001,
+      "step": 566
+    },
+    {
+      "epoch": 4.467018469656992,
+      "grad_norm": 0.026973314583301544,
+      "learning_rate": 9.560288573061562e-07,
+      "loss": 0.0001,
+      "step": 567
+    },
+    {
+      "epoch": 4.474934036939314,
+      "grad_norm": 0.005499368533492088,
+      "learning_rate": 9.466301841873929e-07,
+      "loss": 0.0001,
+      "step": 568
+    },
+    {
+      "epoch": 4.482849604221636,
+      "grad_norm": 0.013346527703106403,
+      "learning_rate": 9.372671369894662e-07,
+      "loss": 0.0001,
+      "step": 569
+    },
+    {
+      "epoch": 4.490765171503957,
+      "grad_norm": 0.01682930439710617,
+      "learning_rate": 9.279399304501526e-07,
+      "loss": 0.0001,
+      "step": 570
+    },
+    {
+      "epoch": 4.4986807387862795,
+      "grad_norm": 0.017017221078276634,
+      "learning_rate": 9.186487784852349e-07,
+      "loss": 0.0001,
+      "step": 571
+    },
+    {
+      "epoch": 4.506596306068602,
+      "grad_norm": 0.07894521951675415,
+      "learning_rate": 9.093938941836012e-07,
+      "loss": 0.0002,
+      "step": 572
+    },
+    {
+      "epoch": 4.514511873350924,
+      "grad_norm": 0.010458113625645638,
+      "learning_rate": 9.001754898023512e-07,
+      "loss": 0.0001,
+      "step": 573
+    },
+    {
+      "epoch": 4.522427440633246,
+      "grad_norm": 0.012565977871418,
+      "learning_rate": 8.909937767619369e-07,
+      "loss": 0.0001,
+      "step": 574
+    },
+    {
+      "epoch": 4.530343007915567,
+      "grad_norm": 0.02009965479373932,
+      "learning_rate": 8.818489656413042e-07,
+      "loss": 0.0001,
+      "step": 575
+    },
+    {
+      "epoch": 4.538258575197889,
+      "grad_norm": 0.01608721725642681,
+      "learning_rate": 8.727412661730724e-07,
+      "loss": 0.0001,
+      "step": 576
+    },
+    {
+      "epoch": 4.546174142480211,
+      "grad_norm": 0.008416295051574707,
+      "learning_rate": 8.636708872387162e-07,
+      "loss": 0.0001,
+      "step": 577
+    },
+    {
+      "epoch": 4.554089709762533,
+      "grad_norm": 0.014859945513308048,
+      "learning_rate": 8.546380368637813e-07,
+      "loss": 0.0001,
+      "step": 578
+    },
+    {
+      "epoch": 4.562005277044855,
+      "grad_norm": 0.009466668590903282,
+      "learning_rate": 8.456429222131083e-07,
+      "loss": 0.0001,
+      "step": 579
+    },
+    {
+      "epoch": 4.569920844327177,
+      "grad_norm": 0.012603671289980412,
+      "learning_rate": 8.366857495860869e-07,
+      "loss": 0.0001,
+      "step": 580
+    },
+    {
+      "epoch": 4.577836411609499,
+      "grad_norm": 0.008713091723620892,
+      "learning_rate": 8.277667244119186e-07,
+      "loss": 0.0001,
+      "step": 581
+    },
+    {
+      "epoch": 4.585751978891821,
+      "grad_norm": 0.008174837566912174,
+      "learning_rate": 8.188860512449107e-07,
+      "loss": 0.0001,
+      "step": 582
+    },
+    {
+      "epoch": 4.593667546174142,
+      "grad_norm": 0.029141252860426903,
+      "learning_rate": 8.100439337597799e-07,
+      "loss": 0.0001,
+      "step": 583
+    },
+    {
+      "epoch": 4.601583113456464,
+      "grad_norm": 0.007655627094209194,
+      "learning_rate": 8.012405747469861e-07,
+      "loss": 0.0001,
+      "step": 584
+    },
+    {
+      "epoch": 4.6094986807387865,
+      "grad_norm": 0.016902687028050423,
+      "learning_rate": 7.924761761080769e-07,
+      "loss": 0.0001,
+      "step": 585
+    },
+    {
+      "epoch": 4.617414248021108,
+      "grad_norm": 0.1739710420370102,
+      "learning_rate": 7.837509388510611e-07,
+      "loss": 0.001,
+      "step": 586
+    },
+    {
+      "epoch": 4.62532981530343,
+      "grad_norm": 0.06047099083662033,
+      "learning_rate": 7.750650630857948e-07,
+      "loss": 0.0002,
+      "step": 587
+    },
+    {
+      "epoch": 4.633245382585752,
+      "grad_norm": 0.006141235586255789,
+      "learning_rate": 7.66418748019396e-07,
+      "loss": 0.0001,
+      "step": 588
+    },
+    {
+      "epoch": 4.641160949868074,
+      "grad_norm": 0.030430037528276443,
+      "learning_rate": 7.578121919516712e-07,
+      "loss": 0.0001,
+      "step": 589
+    },
+    {
+      "epoch": 4.649076517150396,
+      "grad_norm": 0.02062329091131687,
+      "learning_rate": 7.492455922705727e-07,
+      "loss": 0.0001,
+      "step": 590
+    },
+    {
+      "epoch": 4.6569920844327175,
+      "grad_norm": 0.047486960887908936,
+      "learning_rate": 7.407191454476667e-07,
+      "loss": 0.0002,
+      "step": 591
+    },
+    {
+      "epoch": 4.66490765171504,
+      "grad_norm": 0.006273901090025902,
+      "learning_rate": 7.322330470336314e-07,
+      "loss": 0.0007,
+      "step": 592
+    },
+    {
+      "epoch": 4.672823218997362,
+      "grad_norm": 0.18134039640426636,
+      "learning_rate": 7.23787491653769e-07,
+      "loss": 0.0001,
+      "step": 593
+    },
+    {
+      "epoch": 4.680738786279683,
+      "grad_norm": 0.02885393425822258,
+      "learning_rate": 7.153826730035424e-07,
+      "loss": 0.0002,
+      "step": 594
+    },
+    {
+      "epoch": 4.688654353562005,
+      "grad_norm": 0.015093500725924969,
+      "learning_rate": 7.07018783844137e-07,
+      "loss": 0.0001,
+      "step": 595
+    },
+    {
+      "epoch": 4.696569920844327,
+      "grad_norm": 0.03887791559100151,
+      "learning_rate": 6.986960159980327e-07,
+      "loss": 0.0,
+      "step": 596
+    },
+    {
+      "epoch": 4.704485488126649,
+      "grad_norm": 0.006633929908275604,
+      "learning_rate": 6.904145603446116e-07,
+      "loss": 0.0001,
+      "step": 597
+    },
+    {
+      "epoch": 4.7124010554089715,
+      "grad_norm": 0.00938256736844778,
+      "learning_rate": 6.821746068157742e-07,
+      "loss": 0.0001,
+      "step": 598
+    },
+    {
+      "epoch": 4.720316622691293,
+      "grad_norm": 0.008557009510695934,
+      "learning_rate": 6.739763443915894e-07,
+      "loss": 0.0001,
+      "step": 599
+    },
+    {
+      "epoch": 4.728232189973615,
+      "grad_norm": 0.008703123778104782,
+      "learning_rate": 6.658199610959537e-07,
+      "loss": 0.0001,
+      "step": 600
+    },
+    {
+      "epoch": 4.736147757255937,
+      "grad_norm": 0.02286776900291443,
+      "learning_rate": 6.577056439922858e-07,
+      "loss": 0.0001,
+      "step": 601
+    },
+    {
+      "epoch": 4.744063324538258,
+      "grad_norm": 0.00852195918560028,
+      "learning_rate": 6.496335791792294e-07,
+      "loss": 0.0001,
+      "step": 602
+    },
+    {
+      "epoch": 4.75197889182058,
+      "grad_norm": 0.010079172439873219,
+      "learning_rate": 6.41603951786392e-07,
+      "loss": 0.0001,
+      "step": 603
+    },
+    {
+      "epoch": 4.759894459102902,
+      "grad_norm": 0.014380019158124924,
+      "learning_rate": 6.336169459700933e-07,
+      "loss": 0.0001,
+      "step": 604
+    },
+    {
+      "epoch": 4.7678100263852246,
+      "grad_norm": 0.004473215434700251,
+      "learning_rate": 6.25672744909146e-07,
+      "loss": 0.0,
+      "step": 605
+    },
+    {
+      "epoch": 4.775725593667546,
+      "grad_norm": 0.009571720845997334,
+      "learning_rate": 6.177715308006505e-07,
+      "loss": 0.0001,
+      "step": 606
+    },
+    {
+      "epoch": 4.783641160949868,
+      "grad_norm": 0.00657674903050065,
+      "learning_rate": 6.099134848558208e-07,
+      "loss": 0.0001,
+      "step": 607
+    },
+    {
+      "epoch": 4.79155672823219,
+      "grad_norm": 0.0122053362429142,
+      "learning_rate": 6.020987872958237e-07,
+      "loss": 0.0001,
+      "step": 608
+    },
+    {
+      "epoch": 4.799472295514512,
+      "grad_norm": 0.005356465466320515,
+      "learning_rate": 5.943276173476509e-07,
+      "loss": 0.0001,
+      "step": 609
+    },
+    {
+      "epoch": 4.807387862796833,
+      "grad_norm": 0.0075048161670565605,
+      "learning_rate": 5.866001532400023e-07,
+      "loss": 0.0001,
+      "step": 610
+    },
+    {
+      "epoch": 4.8153034300791555,
+      "grad_norm": 0.010855237022042274,
+      "learning_rate": 5.789165721992052e-07,
+      "loss": 0.0001,
+      "step": 611
+    },
+    {
+      "epoch": 4.823218997361478,
+      "grad_norm": 0.007174635771661997,
+      "learning_rate": 5.712770504451426e-07,
+      "loss": 0.0001,
+      "step": 612
+    },
+    {
+      "epoch": 4.8311345646438,
+      "grad_norm": 0.017903555184602737,
+      "learning_rate": 5.636817631872185e-07,
+      "loss": 0.0001,
+      "step": 613
+    },
+    {
+      "epoch": 4.839050131926121,
+      "grad_norm": 0.006769450381398201,
+      "learning_rate": 5.561308846203333e-07,
+      "loss": 0.0001,
+      "step": 614
+    },
+    {
+      "epoch": 4.846965699208443,
+      "grad_norm": 0.018178122118115425,
+      "learning_rate": 5.486245879208946e-07,
+      "loss": 0.0001,
+      "step": 615
+    },
+    {
+      "epoch": 4.854881266490765,
+      "grad_norm": 0.016543753445148468,
+      "learning_rate": 5.411630452428396e-07,
+      "loss": 0.0001,
+      "step": 616
+    },
+    {
+      "epoch": 4.862796833773087,
+      "grad_norm": 0.008568588644266129,
+      "learning_rate": 5.337464277136925e-07,
+      "loss": 0.0001,
+      "step": 617
+    },
+    {
+      "epoch": 4.870712401055409,
+      "grad_norm": 0.007094940636307001,
+      "learning_rate": 5.263749054306347e-07,
+      "loss": 0.0001,
+      "step": 618
+    },
+    {
+      "epoch": 4.878627968337731,
+      "grad_norm": 0.00926823541522026,
+      "learning_rate": 5.190486474566084e-07,
+      "loss": 0.0001,
+      "step": 619
+    },
+    {
+      "epoch": 4.886543535620053,
+      "grad_norm": 0.014867834746837616,
+      "learning_rate": 5.117678218164337e-07,
+      "loss": 0.0001,
+      "step": 620
+    },
+    {
+      "epoch": 4.894459102902375,
+      "grad_norm": 0.006251030135899782,
+      "learning_rate": 5.045325954929614e-07,
+      "loss": 0.0001,
+      "step": 621
+    },
+    {
+      "epoch": 4.902374670184696,
+      "grad_norm": 0.006628802511841059,
+      "learning_rate": 4.973431344232377e-07,
+      "loss": 0.0001,
+      "step": 622
+    },
+    {
+      "epoch": 4.910290237467018,
+      "grad_norm": 0.020900681614875793,
+      "learning_rate": 4.901996034947026e-07,
+      "loss": 0.0002,
+      "step": 623
+    },
+    {
+      "epoch": 4.91820580474934,
+      "grad_norm": 0.008598307147622108,
+      "learning_rate": 4.831021665414043e-07,
+      "loss": 0.0001,
+      "step": 624
+    },
+    {
+      "epoch": 4.926121372031663,
+      "grad_norm": 0.013619592413306236,
+      "learning_rate": 4.7605098634024684e-07,
+      "loss": 0.0001,
+      "step": 625
+    },
+    {
+      "epoch": 4.934036939313984,
+      "grad_norm": 0.004765545949339867,
+      "learning_rate": 4.6904622460725163e-07,
+      "loss": 0.0001,
+      "step": 626
+    },
+    {
+      "epoch": 4.941952506596306,
+      "grad_norm": 0.01161926332861185,
+      "learning_rate": 4.6208804199385114e-07,
+      "loss": 0.0001,
+      "step": 627
+    },
+    {
+      "epoch": 4.949868073878628,
+      "grad_norm": 0.008569381199777126,
+      "learning_rate": 4.55176598083206e-07,
+      "loss": 0.0001,
+      "step": 628
+    },
+    {
+      "epoch": 4.95778364116095,
+      "grad_norm": 0.04928499087691307,
+      "learning_rate": 4.4831205138654105e-07,
+      "loss": 0.0001,
+      "step": 629
+    },
+    {
+      "epoch": 4.965699208443271,
+      "grad_norm": 0.016929304227232933,
+      "learning_rate": 4.41494559339514e-07,
+      "loss": 0.0001,
+      "step": 630
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 756,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 126,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4038229492233667e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}