diff --git "a/checkpoint-603/trainer_state.json" "b/checkpoint-603/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-603/trainer_state.json"
@@ -0,0 +1,4242 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.4998963730569948,
+  "eval_steps": 500,
+  "global_step": 603,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 27.81778461909011,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.7993,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 28.63833175363421,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.9056,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 25.646828828014854,
+      "learning_rate": 1.5e-06,
+      "loss": 0.8473,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.834124771941388,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.8192,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.558095859980105,
+      "learning_rate": 2.5e-06,
+      "loss": 0.7943,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.905789045775758,
+      "learning_rate": 3e-06,
+      "loss": 0.7075,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.259519170268483,
+      "learning_rate": 3.5e-06,
+      "loss": 0.7537,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.639042051048664,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7471,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.515070932390074,
+      "learning_rate": 4.5e-06,
+      "loss": 0.7689,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.916410424632533,
+      "learning_rate": 5e-06,
+      "loss": 0.7194,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.835046497413255,
+      "learning_rate": 4.9999978617243506e-06,
+      "loss": 0.6949,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.065648500649479,
+      "learning_rate": 4.9999914469010585e-06,
+      "loss": 0.7039,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.299372887839679,
+      "learning_rate": 4.999980755541098e-06,
+      "loss": 0.7067,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.693110837094718,
+      "learning_rate": 4.999965787662758e-06,
+      "loss": 0.7126,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.983869635716314,
+      "learning_rate": 4.999946543291642e-06,
+      "loss": 0.6496,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.2561193962441175,
+      "learning_rate": 4.999923022460671e-06,
+      "loss": 0.7036,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.011772824968437,
+      "learning_rate": 4.999895225210079e-06,
+      "loss": 0.7009,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.386638415717137,
+      "learning_rate": 4.9998631515874165e-06,
+      "loss": 0.6624,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.764658092125165,
+      "learning_rate": 4.999826801647551e-06,
+      "loss": 0.6687,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3982096117966614,
+      "learning_rate": 4.999786175452662e-06,
+      "loss": 0.706,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8051633678260193,
+      "learning_rate": 4.999741273072246e-06,
+      "loss": 0.7031,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1177784624332614,
+      "learning_rate": 4.999692094583114e-06,
+      "loss": 0.7525,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2533819675617806,
+      "learning_rate": 4.9996386400693906e-06,
+      "loss": 0.6767,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.61893793162573,
+      "learning_rate": 4.999580909622518e-06,
+      "loss": 0.6432,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.76057623723569,
+      "learning_rate": 4.999518903341251e-06,
+      "loss": 0.6809,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.27983032069553,
+      "learning_rate": 4.999452621331657e-06,
+      "loss": 0.6798,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.501904568120582,
+      "learning_rate": 4.99938206370712e-06,
+      "loss": 0.6412,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.819229290729669,
+      "learning_rate": 4.999307230588338e-06,
+      "loss": 0.6188,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1233212322022212,
+      "learning_rate": 4.9992281221033224e-06,
+      "loss": 0.6378,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7806911906686755,
+      "learning_rate": 4.999144738387396e-06,
+      "loss": 0.6653,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4045490257014563,
+      "learning_rate": 4.999057079583199e-06,
+      "loss": 0.6377,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3803717769210446,
+      "learning_rate": 4.998965145840681e-06,
+      "loss": 0.6855,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3976652879633473,
+      "learning_rate": 4.998868937317106e-06,
+      "loss": 0.6284,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2958541157119727,
+      "learning_rate": 4.998768454177051e-06,
+      "loss": 0.6521,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1925196833696154,
+      "learning_rate": 4.998663696592403e-06,
+      "loss": 0.6619,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.361006042901851,
+      "learning_rate": 4.998554664742362e-06,
+      "loss": 0.6155,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1577758143653614,
+      "learning_rate": 4.998441358813443e-06,
+      "loss": 0.6398,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.219872074512664,
+      "learning_rate": 4.998323778999467e-06,
+      "loss": 0.6051,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2907501521408546,
+      "learning_rate": 4.9982019255015705e-06,
+      "loss": 0.6337,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1769862324666183,
+      "learning_rate": 4.9980757985281955e-06,
+      "loss": 0.6606,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4252479779661607,
+      "learning_rate": 4.997945398295101e-06,
+      "loss": 0.6685,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3929541982084657,
+      "learning_rate": 4.99781072502535e-06,
+      "loss": 0.6084,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.932539969840091,
+      "learning_rate": 4.997671778949318e-06,
+      "loss": 0.6123,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.191742541327873,
+      "learning_rate": 4.997528560304688e-06,
+      "loss": 0.6247,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.423376784566499,
+      "learning_rate": 4.997381069336455e-06,
+      "loss": 0.7024,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0599055392481076,
+      "learning_rate": 4.997229306296918e-06,
+      "loss": 0.6612,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.16832922087532,
+      "learning_rate": 4.997073271445686e-06,
+      "loss": 0.5949,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0483598654319453,
+      "learning_rate": 4.9969129650496775e-06,
+      "loss": 0.6406,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.963056609139284,
+      "learning_rate": 4.996748387383113e-06,
+      "loss": 0.6361,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2094923844269307,
+      "learning_rate": 4.996579538727527e-06,
+      "loss": 0.5901,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1088153449411857,
+      "learning_rate": 4.996406419371749e-06,
+      "loss": 0.6458,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.093448940617732,
+      "learning_rate": 4.996229029611926e-06,
+      "loss": 0.6509,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.075116207412987,
+      "learning_rate": 4.996047369751502e-06,
+      "loss": 0.6295,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.138141165277684,
+      "learning_rate": 4.995861440101229e-06,
+      "loss": 0.6088,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.186316382848445,
+      "learning_rate": 4.995671240979161e-06,
+      "loss": 0.6307,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2513741083982195,
+      "learning_rate": 4.995476772710657e-06,
+      "loss": 0.6175,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0827167336870596,
+      "learning_rate": 4.995278035628379e-06,
+      "loss": 0.5935,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.117977588574442,
+      "learning_rate": 4.995075030072291e-06,
+      "loss": 0.5998,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0996940200235485,
+      "learning_rate": 4.994867756389658e-06,
+      "loss": 0.6159,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.141096165691323,
+      "learning_rate": 4.994656214935045e-06,
+      "loss": 0.6294,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.022748830058395,
+      "learning_rate": 4.994440406070323e-06,
+      "loss": 0.6315,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.209132168720991,
+      "learning_rate": 4.994220330164654e-06,
+      "loss": 0.5645,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0994557317862674,
+      "learning_rate": 4.993995987594509e-06,
+      "loss": 0.6272,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.204220831053169,
+      "learning_rate": 4.99376737874365e-06,
+      "loss": 0.6379,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.127733932186697,
+      "learning_rate": 4.993534504003141e-06,
+      "loss": 0.622,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1338506582034316,
+      "learning_rate": 4.993297363771342e-06,
+      "loss": 0.6259,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.104802764460729,
+      "learning_rate": 4.993055958453912e-06,
+      "loss": 0.6414,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0889535347771675,
+      "learning_rate": 4.9928102884638004e-06,
+      "loss": 0.6466,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.252225316694296,
+      "learning_rate": 4.992560354221258e-06,
+      "loss": 0.6167,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.015392533516649,
+      "learning_rate": 4.992306156153827e-06,
+      "loss": 0.5958,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.151741408948778,
+      "learning_rate": 4.992047694696343e-06,
+      "loss": 0.5875,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0351299117412696,
+      "learning_rate": 4.991784970290935e-06,
+      "loss": 0.5935,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0000962363827983,
+      "learning_rate": 4.991517983387026e-06,
+      "loss": 0.6091,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.202881736102415,
+      "learning_rate": 4.99124673444133e-06,
+      "loss": 0.6122,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.015074773396151,
+      "learning_rate": 4.990971223917848e-06,
+      "loss": 0.6134,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.009305960567766,
+      "learning_rate": 4.990691452287877e-06,
+      "loss": 0.6308,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9967884756310221,
+      "learning_rate": 4.990407420029999e-06,
+      "loss": 0.6098,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0858738033925905,
+      "learning_rate": 4.990119127630085e-06,
+      "loss": 0.6344,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9427707561903895,
+      "learning_rate": 4.989826575581295e-06,
+      "loss": 0.6049,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.157150584766853,
+      "learning_rate": 4.989529764384073e-06,
+      "loss": 0.5965,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0303527419352583,
+      "learning_rate": 4.989228694546151e-06,
+      "loss": 0.6524,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.128799919475717,
+      "learning_rate": 4.988923366582546e-06,
+      "loss": 0.5524,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0122786280510696,
+      "learning_rate": 4.988613781015557e-06,
+      "loss": 0.6268,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.104580177719229,
+      "learning_rate": 4.988299938374769e-06,
+      "loss": 0.6229,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3894843860356834,
+      "learning_rate": 4.9879818391970455e-06,
+      "loss": 0.6194,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9615211372441477,
+      "learning_rate": 4.9876594840265355e-06,
+      "loss": 0.6355,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4509852093141937,
+      "learning_rate": 4.987332873414666e-06,
+      "loss": 0.6405,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.178942375285086,
+      "learning_rate": 4.987002007920142e-06,
+      "loss": 0.5593,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2625634345900445,
+      "learning_rate": 4.9866668881089515e-06,
+      "loss": 0.6133,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.363092638811143,
+      "learning_rate": 4.986327514554356e-06,
+      "loss": 0.6298,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0401982492138546,
+      "learning_rate": 4.985983887836894e-06,
+      "loss": 0.6276,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.276956647922478,
+      "learning_rate": 4.985636008544381e-06,
+      "loss": 0.5691,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1072762844110233,
+      "learning_rate": 4.985283877271908e-06,
+      "loss": 0.6175,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2931866879442637,
+      "learning_rate": 4.984927494621836e-06,
+      "loss": 0.6419,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.112474101166308,
+      "learning_rate": 4.984566861203801e-06,
+      "loss": 0.607,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1816059679212634,
+      "learning_rate": 4.984201977634711e-06,
+      "loss": 0.6136,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0620776369966554,
+      "learning_rate": 4.9838328445387415e-06,
+      "loss": 0.6372,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.147592836641578,
+      "learning_rate": 4.983459462547341e-06,
+      "loss": 0.606,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1808001877062453,
+      "learning_rate": 4.983081832299224e-06,
+      "loss": 0.6019,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3751999527114087,
+      "learning_rate": 4.98269995444037e-06,
+      "loss": 0.6021,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8769470206406913,
+      "learning_rate": 4.98231382962403e-06,
+      "loss": 0.6082,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3060925784921347,
+      "learning_rate": 4.981923458510717e-06,
+      "loss": 0.6174,
+      "step": 102
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1543176832473683,
+      "learning_rate": 4.981528841768206e-06,
+      "loss": 0.6092,
+      "step": 103
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1558689520522547,
+      "learning_rate": 4.981129980071538e-06,
+      "loss": 0.587,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3830532005188383,
+      "learning_rate": 4.980726874103014e-06,
+      "loss": 0.6518,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3333119576634767,
+      "learning_rate": 4.980319524552195e-06,
+      "loss": 0.6096,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1135146855324214,
+      "learning_rate": 4.9799079321159e-06,
+      "loss": 0.5728,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2300463384326394,
+      "learning_rate": 4.9794920974982095e-06,
+      "loss": 0.6563,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1745234017525443,
+      "learning_rate": 4.979072021410458e-06,
+      "loss": 0.5968,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1536586182562334,
+      "learning_rate": 4.978647704571237e-06,
+      "loss": 0.6189,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.193809374687326,
+      "learning_rate": 4.97821914770639e-06,
+      "loss": 0.5864,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0525896373682047,
+      "learning_rate": 4.977786351549017e-06,
+      "loss": 0.6101,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.216099286618384,
+      "learning_rate": 4.977349316839467e-06,
+      "loss": 0.5984,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.155122255962579,
+      "learning_rate": 4.97690804432534e-06,
+      "loss": 0.6311,
+      "step": 114
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2972101190291374,
+      "learning_rate": 4.976462534761487e-06,
+      "loss": 0.5813,
+      "step": 115
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9925413745245948,
+      "learning_rate": 4.9760127889100044e-06,
+      "loss": 0.6157,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2802548684036568,
+      "learning_rate": 4.975558807540238e-06,
+      "loss": 0.6079,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.048888007394621,
+      "learning_rate": 4.9751005914287775e-06,
+      "loss": 0.6467,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.28661640438254,
+      "learning_rate": 4.974638141359456e-06,
+      "loss": 0.6029,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.004056683755783,
+      "learning_rate": 4.974171458123351e-06,
+      "loss": 0.6289,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1628470048067667,
+      "learning_rate": 4.97370054251878e-06,
+      "loss": 0.6139,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.056119895466544,
+      "learning_rate": 4.9732253953513e-06,
+      "loss": 0.5798,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1716513163164275,
+      "learning_rate": 4.972746017433709e-06,
+      "loss": 0.6085,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.255856676525811,
+      "learning_rate": 4.97226240958604e-06,
+      "loss": 0.6342,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1049280498075373,
+      "learning_rate": 4.971774572635563e-06,
+      "loss": 0.6197,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.133349390995361,
+      "learning_rate": 4.97128250741678e-06,
+      "loss": 0.5751,
+      "step": 126
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2044887467317578,
+      "learning_rate": 4.97078621477143e-06,
+      "loss": 0.6611,
+      "step": 127
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1413863795698145,
+      "learning_rate": 4.970285695548481e-06,
+      "loss": 0.625,
+      "step": 128
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0229587336296615,
+      "learning_rate": 4.969780950604132e-06,
+      "loss": 0.5989,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0983599595244247,
+      "learning_rate": 4.969271980801808e-06,
+      "loss": 0.5747,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1059041140010786,
+      "learning_rate": 4.9687587870121645e-06,
+      "loss": 0.5869,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8967441614595046,
+      "learning_rate": 4.9682413701130815e-06,
+      "loss": 0.6272,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9976164993621088,
+      "learning_rate": 4.967719730989663e-06,
+      "loss": 0.6282,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8719131324952145,
+      "learning_rate": 4.967193870534235e-06,
+      "loss": 0.6052,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.071702997476533,
+      "learning_rate": 4.9666637896463455e-06,
+      "loss": 0.5785,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9549455320048341,
+      "learning_rate": 4.966129489232762e-06,
+      "loss": 0.5739,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0656898626759315,
+      "learning_rate": 4.9655909702074684e-06,
+      "loss": 0.6651,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1185948604203038,
+      "learning_rate": 4.965048233491669e-06,
+      "loss": 0.5759,
+      "step": 138
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.08566019272993,
+      "learning_rate": 4.964501280013777e-06,
+      "loss": 0.6271,
+      "step": 139
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.117420903965419,
+      "learning_rate": 4.963950110709425e-06,
+      "loss": 0.5968,
+      "step": 140
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9784944143818486,
+      "learning_rate": 4.963394726521453e-06,
+      "loss": 0.6112,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.077292948039572,
+      "learning_rate": 4.9628351283999144e-06,
+      "loss": 0.5636,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.223803520245629,
+      "learning_rate": 4.962271317302068e-06,
+      "loss": 0.6658,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.039369072186367,
+      "learning_rate": 4.9617032941923796e-06,
+      "loss": 0.5853,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.071470113085907,
+      "learning_rate": 4.961131060042522e-06,
+      "loss": 0.601,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.437470272347474,
+      "learning_rate": 4.960554615831372e-06,
+      "loss": 0.6593,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.178684122927139,
+      "learning_rate": 4.959973962545005e-06,
+      "loss": 0.607,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.097006749956471,
+      "learning_rate": 4.9593891011767e-06,
+      "loss": 0.5873,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9801202541822784,
+      "learning_rate": 4.958800032726931e-06,
+      "loss": 0.5877,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.30001951085656,
+      "learning_rate": 4.958206758203373e-06,
+      "loss": 0.6368,
+      "step": 150
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.990094260131078,
+      "learning_rate": 4.957609278620891e-06,
+      "loss": 0.59,
+      "step": 151
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.262163752076628,
+      "learning_rate": 4.957007595001548e-06,
+      "loss": 0.5779,
+      "step": 152
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1970152093220983,
+      "learning_rate": 4.956401708374595e-06,
+      "loss": 0.5894,
+      "step": 153
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.220825872684071,
+      "learning_rate": 4.9557916197764745e-06,
+      "loss": 0.6528,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.099472677591387,
+      "learning_rate": 4.955177330250817e-06,
+      "loss": 0.5798,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.159203936881569,
+      "learning_rate": 4.954558840848437e-06,
+      "loss": 0.6206,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.185152414039555,
+      "learning_rate": 4.953936152627338e-06,
+      "loss": 0.5624,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0679748168992624,
+      "learning_rate": 4.953309266652701e-06,
+      "loss": 0.5859,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.327237187255128,
+      "learning_rate": 4.952678183996891e-06,
+      "loss": 0.5632,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2865519679977417,
+      "learning_rate": 4.952042905739451e-06,
+      "loss": 0.6965,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.523435408018699,
+      "learning_rate": 4.9514034329671e-06,
+      "loss": 0.6217,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4992653226709636,
+      "learning_rate": 4.950759766773734e-06,
+      "loss": 0.6175,
+      "step": 162
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.432752824777114,
+      "learning_rate": 4.950111908260423e-06,
+      "loss": 0.5862,
+      "step": 163
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.137500912204061,
+      "learning_rate": 4.949459858535404e-06,
+      "loss": 0.6124,
+      "step": 164
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2226376224120474,
+      "learning_rate": 4.94880361871409e-06,
+      "loss": 0.5891,
+      "step": 165
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3821839805775165,
+      "learning_rate": 4.9481431899190544e-06,
+      "loss": 0.6008,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.306242834684614,
+      "learning_rate": 4.947478573280044e-06,
+      "loss": 0.6159,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3298092236851518,
+      "learning_rate": 4.946809769933963e-06,
+      "loss": 0.5809,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.364296499621558,
+      "learning_rate": 4.946136781024883e-06,
+      "loss": 0.5895,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.237241095609228,
+      "learning_rate": 4.945459607704029e-06,
+      "loss": 0.6144,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4027419761972264,
+      "learning_rate": 4.9447782511297905e-06,
+      "loss": 0.5985,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1547059182244284,
+      "learning_rate": 4.944092712467709e-06,
+      "loss": 0.5763,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1530221667047984,
+      "learning_rate": 4.9434029928904805e-06,
+      "loss": 0.5692,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.228588593294869,
+      "learning_rate": 4.942709093577954e-06,
+      "loss": 0.5896,
+      "step": 174
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1597295307130198,
+      "learning_rate": 4.942011015717129e-06,
+      "loss": 0.5864,
+      "step": 175
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.321140955498194,
+      "learning_rate": 4.941308760502149e-06,
+      "loss": 0.6089,
+      "step": 176
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.220124736460707,
+      "learning_rate": 4.940602329134309e-06,
+      "loss": 0.5786,
+      "step": 177
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1698038563080417,
+      "learning_rate": 4.939891722822043e-06,
+      "loss": 0.5749,
+      "step": 178
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.244425969121411,
+      "learning_rate": 4.93917694278093e-06,
+      "loss": 0.5877,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.143920008069458,
+      "learning_rate": 4.938457990233687e-06,
+      "loss": 0.6024,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1786040820345813,
+      "learning_rate": 4.937734866410169e-06,
+      "loss": 0.5845,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.301832824481007,
+      "learning_rate": 4.9370075725473665e-06,
+      "loss": 0.6182,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3748033727083997,
+      "learning_rate": 4.936276109889403e-06,
+      "loss": 0.6073,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.476334487382023,
+      "learning_rate": 4.935540479687534e-06,
+      "loss": 0.5793,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2509466352322494,
+      "learning_rate": 4.934800683200143e-06,
+      "loss": 0.6133,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8391697547684873,
+      "learning_rate": 4.934056721692742e-06,
+      "loss": 0.5967,
+      "step": 186
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4492364225391765,
+      "learning_rate": 4.933308596437965e-06,
+      "loss": 0.5676,
+      "step": 187
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.685548141821295,
+      "learning_rate": 4.932556308715573e-06,
+      "loss": 0.6069,
+      "step": 188
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.261217637824808,
+      "learning_rate": 4.931799859812443e-06,
+      "loss": 0.6411,
+      "step": 189
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3838284395200966,
+      "learning_rate": 4.931039251022573e-06,
+      "loss": 0.5745,
+      "step": 190
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2550921344466164,
+      "learning_rate": 4.930274483647074e-06,
+      "loss": 0.5989,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.078406234527636,
+      "learning_rate": 4.929505558994175e-06,
+      "loss": 0.5998,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.592864566091496,
+      "learning_rate": 4.928732478379214e-06,
+      "loss": 0.5842,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.092752299259724,
+      "learning_rate": 4.927955243124638e-06,
+      "loss": 0.5789,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3799311595696966,
+      "learning_rate": 4.927173854560002e-06,
+      "loss": 0.6265,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.246876688010602,
+      "learning_rate": 4.926388314021964e-06,
+      "loss": 0.6126,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1409898276704578,
+      "learning_rate": 4.925598622854287e-06,
+      "loss": 0.6073,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5946158421875385,
+      "learning_rate": 4.924804782407834e-06,
+      "loss": 0.6154,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1225494320427982,
+      "learning_rate": 4.924006794040562e-06,
+      "loss": 0.583,
+      "step": 199
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1971323526291338,
+      "learning_rate": 4.923204659117528e-06,
+      "loss": 0.6078,
+      "step": 200
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.289185506404785,
+      "learning_rate": 4.92239837901088e-06,
+      "loss": 0.6127,
+      "step": 201
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0071007751625354,
+      "learning_rate": 4.921587955099858e-06,
+      "loss": 0.5804,
+      "step": 202
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2981840149068247,
+      "learning_rate": 4.920773388770789e-06,
+      "loss": 0.6027,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.236179116886702,
+      "learning_rate": 4.919954681417087e-06,
+      "loss": 0.6179,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.007422589251611,
+      "learning_rate": 4.91913183443925e-06,
+      "loss": 0.5647,
+      "step": 205
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1402813555735483,
+      "learning_rate": 4.918304849244857e-06,
+      "loss": 0.5841,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0456415785177104,
+      "learning_rate": 4.917473727248565e-06,
+      "loss": 0.5524,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9673558126020942,
+      "learning_rate": 4.916638469872109e-06,
+      "loss": 0.5698,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.015111672496819,
+      "learning_rate": 4.9157990785442964e-06,
+      "loss": 0.5957,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9502065547578398,
+      "learning_rate": 4.9149555547010086e-06,
+      "loss": 0.5592,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.167936522558899,
+      "learning_rate": 4.9141078997851945e-06,
+      "loss": 0.5705,
+      "step": 211
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2066587458997935,
+      "learning_rate": 4.91325611524687e-06,
+      "loss": 0.5526,
+      "step": 212
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9132995625903553,
+      "learning_rate": 4.9124002025431136e-06,
+      "loss": 0.5767,
+      "step": 213
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0097281107801277,
+      "learning_rate": 4.91154016313807e-06,
+      "loss": 0.6185,
+      "step": 214
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.023532008241332,
+      "learning_rate": 4.910675998502938e-06,
+      "loss": 0.6005,
+      "step": 215
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9253831001776973,
+      "learning_rate": 4.909807710115977e-06,
+      "loss": 0.5769,
+      "step": 216
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.066862408842564,
+      "learning_rate": 4.908935299462497e-06,
+      "loss": 0.5671,
+      "step": 217
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9412704290792853,
+      "learning_rate": 4.908058768034862e-06,
+      "loss": 0.5568,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.185994457097553,
+      "learning_rate": 4.907178117332487e-06,
+      "loss": 0.5621,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.021517127546353,
+      "learning_rate": 4.906293348861829e-06,
+      "loss": 0.5672,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.099703967072734,
+      "learning_rate": 4.905404464136391e-06,
+      "loss": 0.5366,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.030197056583618,
+      "learning_rate": 4.904511464676718e-06,
+      "loss": 0.6064,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4170102988954896,
+      "learning_rate": 4.903614352010393e-06,
+      "loss": 0.5919,
+      "step": 223
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0819468873015476,
+      "learning_rate": 4.9027131276720355e-06,
+      "loss": 0.5366,
+      "step": 224
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.148008018153629,
+      "learning_rate": 4.901807793203299e-06,
+      "loss": 0.597,
+      "step": 225
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0303725862017186,
+      "learning_rate": 4.900898350152866e-06,
+      "loss": 0.6394,
+      "step": 226
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1598989214704334,
+      "learning_rate": 4.899984800076449e-06,
+      "loss": 0.5932,
+      "step": 227
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0816312637185255,
+      "learning_rate": 4.899067144536786e-06,
+      "loss": 0.5909,
+      "step": 228
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9024067197329315,
+      "learning_rate": 4.8981453851036365e-06,
+      "loss": 0.5463,
+      "step": 229
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1830926868871043,
+      "learning_rate": 4.897219523353781e-06,
+      "loss": 0.5821,
+      "step": 230
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1156269612794016,
+      "learning_rate": 4.8962895608710195e-06,
+      "loss": 0.5993,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9653407654210864,
+      "learning_rate": 4.895355499246162e-06,
+      "loss": 0.5525,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.367769051061897,
+      "learning_rate": 4.894417340077036e-06,
+      "loss": 0.5683,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.078327064466567,
+      "learning_rate": 4.893475084968474e-06,
+      "loss": 0.6184,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1661882731589475,
+      "learning_rate": 4.8925287355323195e-06,
+      "loss": 0.6321,
+      "step": 235
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.182760952002799,
+      "learning_rate": 4.891578293387413e-06,
+      "loss": 0.6254,
+      "step": 236
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.998723579962691,
+      "learning_rate": 4.890623760159605e-06,
+      "loss": 0.5371,
+      "step": 237
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.319922346931926,
+      "learning_rate": 4.8896651374817365e-06,
+      "loss": 0.5941,
+      "step": 238
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.090735197217999,
+      "learning_rate": 4.888702426993648e-06,
+      "loss": 0.577,
+      "step": 239
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1247199987228558,
+      "learning_rate": 4.887735630342173e-06,
+      "loss": 0.5928,
+      "step": 240
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.33151114429804,
+      "learning_rate": 4.8867647491811315e-06,
+      "loss": 0.5838,
+      "step": 241
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1570026356289147,
+      "learning_rate": 4.885789785171334e-06,
+      "loss": 0.5642,
+      "step": 242
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.049571197047368,
+      "learning_rate": 4.884810739980575e-06,
+      "loss": 0.6684,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9810062424466381,
+      "learning_rate": 4.883827615283626e-06,
+      "loss": 0.5942,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.145869663660159,
+      "learning_rate": 4.882840412762244e-06,
+      "loss": 0.6356,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.19290302186514,
+      "learning_rate": 4.881849134105156e-06,
+      "loss": 0.6189,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0561043419872984,
+      "learning_rate": 4.880853781008062e-06,
+      "loss": 0.5563,
+      "step": 247
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8831183793224635,
+      "learning_rate": 4.879854355173638e-06,
+      "loss": 0.5522,
+      "step": 248
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.020981606684741,
+      "learning_rate": 4.878850858311518e-06,
+      "loss": 0.5548,
+      "step": 249
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.060242570493272,
+      "learning_rate": 4.877843292138307e-06,
+      "loss": 0.5715,
+      "step": 250
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.082455778933014,
+      "learning_rate": 4.8768316583775665e-06,
+      "loss": 0.5959,
+      "step": 251
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9830929719438626,
+      "learning_rate": 4.875815958759819e-06,
+      "loss": 0.5813,
+      "step": 252
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9772267506828567,
+      "learning_rate": 4.8747961950225406e-06,
+      "loss": 0.539,
+      "step": 253
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1492561995002104,
+      "learning_rate": 4.873772368910161e-06,
+      "loss": 0.6059,
+      "step": 254
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.253757247139787,
+      "learning_rate": 4.872744482174058e-06,
+      "loss": 0.5897,
+      "step": 255
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3282624851882496,
+      "learning_rate": 4.8717125365725545e-06,
+      "loss": 0.5675,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.15573581133063,
+      "learning_rate": 4.8706765338709185e-06,
+      "loss": 0.5958,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.073289220218241,
+      "learning_rate": 4.869636475841358e-06,
+      "loss": 0.6052,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.293714090249444,
+      "learning_rate": 4.8685923642630165e-06,
+      "loss": 0.5786,
+      "step": 259
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9496544276539172,
+      "learning_rate": 4.867544200921974e-06,
+      "loss": 0.6163,
+      "step": 260
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5267016753690132,
+      "learning_rate": 4.866491987611239e-06,
+      "loss": 0.6223,
+      "step": 261
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8731249445320794,
+      "learning_rate": 4.865435726130751e-06,
+      "loss": 0.5632,
+      "step": 262
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3586331105798863,
+      "learning_rate": 4.86437541828737e-06,
+      "loss": 0.5769,
+      "step": 263
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0258106914510585,
+      "learning_rate": 4.863311065894883e-06,
+      "loss": 0.6103,
+      "step": 264
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2543614390885955,
+      "learning_rate": 4.862242670773991e-06,
+      "loss": 0.5844,
+      "step": 265
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9440299381244668,
+      "learning_rate": 4.861170234752314e-06,
+      "loss": 0.5559,
+      "step": 266
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.254538268495492,
+      "learning_rate": 4.8600937596643815e-06,
+      "loss": 0.5709,
+      "step": 267
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.007651746385687,
+      "learning_rate": 4.8590132473516346e-06,
+      "loss": 0.573,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0735253118288837,
+      "learning_rate": 4.857928699662421e-06,
+      "loss": 0.5954,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.024775417101569,
+      "learning_rate": 4.856840118451989e-06,
+      "loss": 0.5992,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1043310699945814,
+      "learning_rate": 4.855747505582488e-06,
+      "loss": 0.6507,
+      "step": 271
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0386353328313214,
+      "learning_rate": 4.854650862922965e-06,
+      "loss": 0.5666,
+      "step": 272
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.978698841367705,
+      "learning_rate": 4.853550192349358e-06,
+      "loss": 0.5593,
+      "step": 273
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9386534247633986,
+      "learning_rate": 4.852445495744497e-06,
+      "loss": 0.5735,
+      "step": 274
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.049346245018599,
+      "learning_rate": 4.8513367749981e-06,
+      "loss": 0.5415,
+      "step": 275
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1051969521216605,
+      "learning_rate": 4.850224032006765e-06,
+      "loss": 0.5532,
+      "step": 276
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2006792558872315,
+      "learning_rate": 4.849107268673975e-06,
+      "loss": 0.5696,
+      "step": 277
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0460787736353647,
+      "learning_rate": 4.847986486910088e-06,
+      "loss": 0.5658,
+      "step": 278
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1161843259225406,
+      "learning_rate": 4.846861688632336e-06,
+      "loss": 0.583,
+      "step": 279
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8882198480393542,
+      "learning_rate": 4.8457328757648224e-06,
+      "loss": 0.5693,
+      "step": 280
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1578413701109596,
+      "learning_rate": 4.844600050238517e-06,
+      "loss": 0.5409,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.03912467778954,
+      "learning_rate": 4.843463213991255e-06,
+      "loss": 0.5908,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2333462480826247,
+      "learning_rate": 4.842322368967731e-06,
+      "loss": 0.6088,
+      "step": 283
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.06698702157327,
+      "learning_rate": 4.8411775171194986e-06,
+      "loss": 0.5953,
+      "step": 284
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1433923121572045,
+      "learning_rate": 4.840028660404964e-06,
+      "loss": 0.5851,
+      "step": 285
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.214858780835041,
+      "learning_rate": 4.838875800789386e-06,
+      "loss": 0.5913,
+      "step": 286
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.038128612492624,
+      "learning_rate": 4.837718940244871e-06,
+      "loss": 0.5827,
+      "step": 287
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9894065096959768,
+      "learning_rate": 4.836558080750365e-06,
+      "loss": 0.5769,
+      "step": 288
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1711590153285822,
+      "learning_rate": 4.835393224291662e-06,
+      "loss": 0.654,
+      "step": 289
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.105004451988696,
+      "learning_rate": 4.834224372861386e-06,
+      "loss": 0.6158,
+      "step": 290
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9554568023729102,
+      "learning_rate": 4.833051528459001e-06,
+      "loss": 0.5807,
+      "step": 291
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2693917834500312,
+      "learning_rate": 4.831874693090797e-06,
+      "loss": 0.5557,
+      "step": 292
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9081391627126192,
+      "learning_rate": 4.830693868769892e-06,
+      "loss": 0.6057,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2133664110768585,
+      "learning_rate": 4.82950905751623e-06,
+      "loss": 0.6103,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.015392814211589,
+      "learning_rate": 4.8283202613565735e-06,
+      "loss": 0.5578,
+      "step": 295
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.142124020349717,
+      "learning_rate": 4.8271274823245e-06,
+      "loss": 0.5675,
+      "step": 296
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.981611826462286,
+      "learning_rate": 4.825930722460405e-06,
+      "loss": 0.5696,
+      "step": 297
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.966759748348117,
+      "learning_rate": 4.824729983811486e-06,
+      "loss": 0.58,
+      "step": 298
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0117040369769397,
+      "learning_rate": 4.823525268431754e-06,
+      "loss": 0.6005,
+      "step": 299
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9579664917991193,
+      "learning_rate": 4.822316578382019e-06,
+      "loss": 0.5472,
+      "step": 300
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9075723479635032,
+      "learning_rate": 4.821103915729892e-06,
+      "loss": 0.5834,
+      "step": 301
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.289340229011896,
+      "learning_rate": 4.819887282549777e-06,
+      "loss": 0.6088,
+      "step": 302
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0410700553735235,
+      "learning_rate": 4.818666680922874e-06,
+      "loss": 0.5449,
+      "step": 303
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.074434792511819,
+      "learning_rate": 4.8174421129371675e-06,
+      "loss": 0.5826,
+      "step": 304
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1377170527698865,
+      "learning_rate": 4.816213580687428e-06,
+      "loss": 0.6262,
+      "step": 305
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.060340839248083,
+      "learning_rate": 4.814981086275209e-06,
+      "loss": 0.5479,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.007036467413588,
+      "learning_rate": 4.813744631808841e-06,
+      "loss": 0.5642,
+      "step": 307
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.016779606220332,
+      "learning_rate": 4.8125042194034285e-06,
+      "loss": 0.5503,
+      "step": 308
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.930004252757651,
+      "learning_rate": 4.811259851180845e-06,
+      "loss": 0.582,
+      "step": 309
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9179477992752856,
+      "learning_rate": 4.810011529269734e-06,
+      "loss": 0.5678,
+      "step": 310
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.023430757276848,
+      "learning_rate": 4.808759255805498e-06,
+      "loss": 0.614,
+      "step": 311
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8334738409404936,
+      "learning_rate": 4.807503032930306e-06,
+      "loss": 0.5742,
+      "step": 312
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.937332706274502,
+      "learning_rate": 4.806242862793075e-06,
+      "loss": 0.6257,
+      "step": 313
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0265383045700363,
+      "learning_rate": 4.8049787475494786e-06,
+      "loss": 0.5733,
+      "step": 314
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.056444039073761,
+      "learning_rate": 4.803710689361939e-06,
+      "loss": 0.578,
+      "step": 315
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.411132719183335,
+      "learning_rate": 4.802438690399622e-06,
+      "loss": 0.5778,
+      "step": 316
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0233969242222853,
+      "learning_rate": 4.801162752838436e-06,
+      "loss": 0.5649,
+      "step": 317
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2809121915132815,
+      "learning_rate": 4.799882878861025e-06,
+      "loss": 0.5589,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9806834041020271,
+      "learning_rate": 4.798599070656768e-06,
+      "loss": 0.5753,
+      "step": 319
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.095099671577702,
+      "learning_rate": 4.797311330421773e-06,
+      "loss": 0.5644,
+      "step": 320
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1697606190375764,
+      "learning_rate": 4.796019660358877e-06,
+      "loss": 0.6009,
+      "step": 321
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9549416103216173,
+      "learning_rate": 4.794724062677635e-06,
+      "loss": 0.5429,
+      "step": 322
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9986949357292838,
+      "learning_rate": 4.793424539594323e-06,
+      "loss": 0.5456,
+      "step": 323
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9414831957796765,
+      "learning_rate": 4.792121093331935e-06,
+      "loss": 0.5468,
+      "step": 324
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.100702188933012,
+      "learning_rate": 4.7908137261201685e-06,
+      "loss": 0.5763,
+      "step": 325
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2747471285831025,
+      "learning_rate": 4.789502440195436e-06,
+      "loss": 0.5637,
+      "step": 326
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8996382919319124,
+      "learning_rate": 4.788187237800849e-06,
+      "loss": 0.5285,
+      "step": 327
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3451495174978847,
+      "learning_rate": 4.786868121186218e-06,
+      "loss": 0.5638,
+      "step": 328
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0437536068229565,
+      "learning_rate": 4.7855450926080535e-06,
+      "loss": 0.5282,
+      "step": 329
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1185488514745554,
+      "learning_rate": 4.784218154329555e-06,
+      "loss": 0.5689,
+      "step": 330
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.08745956731504,
+      "learning_rate": 4.78288730862061e-06,
+      "loss": 0.5772,
+      "step": 331
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9479507156354359,
+      "learning_rate": 4.781552557757789e-06,
+      "loss": 0.5419,
+      "step": 332
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0211480847937255,
+      "learning_rate": 4.780213904024346e-06,
+      "loss": 0.5757,
+      "step": 333
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9075335749936069,
+      "learning_rate": 4.7788713497102094e-06,
+      "loss": 0.5693,
+      "step": 334
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9590727137410602,
+      "learning_rate": 4.777524897111979e-06,
+      "loss": 0.5501,
+      "step": 335
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0328480247612752,
+      "learning_rate": 4.776174548532926e-06,
+      "loss": 0.587,
+      "step": 336
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.062540517496736,
+      "learning_rate": 4.774820306282982e-06,
+      "loss": 0.5819,
+      "step": 337
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0054452800156195,
+      "learning_rate": 4.773462172678744e-06,
+      "loss": 0.5529,
+      "step": 338
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9641125644599562,
+      "learning_rate": 4.772100150043462e-06,
+      "loss": 0.5895,
+      "step": 339
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9196744569285298,
+      "learning_rate": 4.77073424070704e-06,
+      "loss": 0.5504,
+      "step": 340
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0002752186146484,
+      "learning_rate": 4.76936444700603e-06,
+      "loss": 0.5307,
+      "step": 341
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1068919823054344,
+      "learning_rate": 4.76799077128363e-06,
+      "loss": 0.5908,
+      "step": 342
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.919597745459612,
+      "learning_rate": 4.766613215889678e-06,
+      "loss": 0.5423,
+      "step": 343
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0670928578728716,
+      "learning_rate": 4.765231783180648e-06,
+      "loss": 0.5901,
+      "step": 344
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.906116148793229,
+      "learning_rate": 4.763846475519648e-06,
+      "loss": 0.5919,
+      "step": 345
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9133575268702454,
+      "learning_rate": 4.762457295276413e-06,
+      "loss": 0.585,
+      "step": 346
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.133902651855379,
+      "learning_rate": 4.7610642448273025e-06,
+      "loss": 0.5444,
+      "step": 347
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.95222194640397,
+      "learning_rate": 4.7596673265552985e-06,
+      "loss": 0.5941,
+      "step": 348
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.095010268380277,
+      "learning_rate": 4.758266542849997e-06,
+      "loss": 0.6045,
+      "step": 349
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0493864712059655,
+      "learning_rate": 4.756861896107609e-06,
+      "loss": 0.6011,
+      "step": 350
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9222198823064967,
+      "learning_rate": 4.755453388730949e-06,
+      "loss": 0.5521,
+      "step": 351
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.368147154955994,
+      "learning_rate": 4.754041023129442e-06,
+      "loss": 0.6117,
+      "step": 352
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9734596786106697,
+      "learning_rate": 4.752624801719108e-06,
+      "loss": 0.5727,
+      "step": 353
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.151510566977991,
+      "learning_rate": 4.751204726922564e-06,
+      "loss": 0.6085,
+      "step": 354
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9291219072892685,
+      "learning_rate": 4.74978080116902e-06,
+      "loss": 0.5655,
+      "step": 355
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.838592559018919,
+      "learning_rate": 4.748353026894273e-06,
+      "loss": 0.5508,
+      "step": 356
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.069156589116884,
+      "learning_rate": 4.7469214065407e-06,
+      "loss": 0.5942,
+      "step": 357
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8960817746615841,
+      "learning_rate": 4.745485942557264e-06,
+      "loss": 0.5902,
+      "step": 358
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0606557307859634,
+      "learning_rate": 4.744046637399497e-06,
+      "loss": 0.556,
+      "step": 359
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9660065879130573,
+      "learning_rate": 4.742603493529505e-06,
+      "loss": 0.5364,
+      "step": 360
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9647921383638112,
+      "learning_rate": 4.741156513415958e-06,
+      "loss": 0.5601,
+      "step": 361
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.049074688423064,
+      "learning_rate": 4.739705699534092e-06,
+      "loss": 0.556,
+      "step": 362
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.962593945802751,
+      "learning_rate": 4.738251054365697e-06,
+      "loss": 0.5609,
+      "step": 363
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.059675349950347,
+      "learning_rate": 4.736792580399119e-06,
+      "loss": 0.5499,
+      "step": 364
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8479566025134508,
+      "learning_rate": 4.7353302801292555e-06,
+      "loss": 0.5621,
+      "step": 365
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9405450724813613,
+      "learning_rate": 4.733864156057545e-06,
+      "loss": 0.5437,
+      "step": 366
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.122487864033456,
+      "learning_rate": 4.7323942106919715e-06,
+      "loss": 0.5984,
+      "step": 367
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6822841144123046,
+      "learning_rate": 4.730920446547052e-06,
+      "loss": 0.5951,
+      "step": 368
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.001405394086718,
+      "learning_rate": 4.729442866143838e-06,
+      "loss": 0.5552,
+      "step": 369
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.081154186949651,
+      "learning_rate": 4.72796147200991e-06,
+      "loss": 0.587,
+      "step": 370
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1196544292473236,
+      "learning_rate": 4.72647626667937e-06,
+      "loss": 0.5882,
+      "step": 371
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.107445583509131,
+      "learning_rate": 4.724987252692841e-06,
+      "loss": 0.5389,
+      "step": 372
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9529785007256542,
+      "learning_rate": 4.723494432597462e-06,
+      "loss": 0.6439,
+      "step": 373
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.11513441515607,
+      "learning_rate": 4.72199780894688e-06,
+      "loss": 0.6089,
+      "step": 374
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9769899713721226,
+      "learning_rate": 4.7204973843012504e-06,
+      "loss": 0.5393,
+      "step": 375
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.063749623036316,
+      "learning_rate": 4.718993161227231e-06,
+      "loss": 0.5987,
+      "step": 376
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0515862288253883,
+      "learning_rate": 4.717485142297977e-06,
+      "loss": 0.5772,
+      "step": 377
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8962297741946081,
+      "learning_rate": 4.715973330093135e-06,
+      "loss": 0.5424,
+      "step": 378
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2210958340400087,
+      "learning_rate": 4.7144577271988435e-06,
+      "loss": 0.6072,
+      "step": 379
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.067113337475314,
+      "learning_rate": 4.712938336207724e-06,
+      "loss": 0.5482,
+      "step": 380
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8985489253954526,
+      "learning_rate": 4.711415159718876e-06,
+      "loss": 0.5593,
+      "step": 381
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.085236381118245,
+      "learning_rate": 4.709888200337879e-06,
+      "loss": 0.5704,
+      "step": 382
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0967664183909784,
+      "learning_rate": 4.708357460676779e-06,
+      "loss": 0.5997,
+      "step": 383
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0454278026009645,
+      "learning_rate": 4.706822943354092e-06,
+      "loss": 0.5669,
+      "step": 384
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9171673309342674,
+      "learning_rate": 4.705284650994793e-06,
+      "loss": 0.517,
+      "step": 385
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2003223432761287,
+      "learning_rate": 4.70374258623032e-06,
+      "loss": 0.5957,
+      "step": 386
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.936392519491186,
+      "learning_rate": 4.702196751698557e-06,
+      "loss": 0.5767,
+      "step": 387
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.354272003403086,
+      "learning_rate": 4.700647150043841e-06,
+      "loss": 0.6515,
+      "step": 388
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9115059027323418,
+      "learning_rate": 4.699093783916955e-06,
+      "loss": 0.5579,
+      "step": 389
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9878827587010002,
+      "learning_rate": 4.697536655975115e-06,
+      "loss": 0.572,
+      "step": 390
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9729552535473858,
+      "learning_rate": 4.69597576888198e-06,
+      "loss": 0.5665,
+      "step": 391
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.177634366499155,
+      "learning_rate": 4.694411125307632e-06,
+      "loss": 0.6363,
+      "step": 392
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8955146664976508,
+      "learning_rate": 4.692842727928584e-06,
+      "loss": 0.5682,
+      "step": 393
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.175305874476245,
+      "learning_rate": 4.691270579427769e-06,
+      "loss": 0.5943,
+      "step": 394
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.068140527232831,
+      "learning_rate": 4.689694682494537e-06,
+      "loss": 0.5659,
+      "step": 395
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9112960694448755,
+      "learning_rate": 4.688115039824648e-06,
+      "loss": 0.6048,
+      "step": 396
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9778305624626604,
+      "learning_rate": 4.686531654120272e-06,
+      "loss": 0.5695,
+      "step": 397
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.096904163204813,
+      "learning_rate": 4.684944528089981e-06,
+      "loss": 0.6113,
+      "step": 398
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0011934144948516,
+      "learning_rate": 4.683353664448745e-06,
+      "loss": 0.5568,
+      "step": 399
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8562851971757464,
+      "learning_rate": 4.681759065917929e-06,
+      "loss": 0.5474,
+      "step": 400
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8190547574166316,
+      "learning_rate": 4.680160735225285e-06,
+      "loss": 0.5315,
+      "step": 401
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9247862956929132,
+      "learning_rate": 4.6785586751049505e-06,
+      "loss": 0.5568,
+      "step": 402
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8469793674077621,
+      "learning_rate": 4.676952888297442e-06,
+      "loss": 0.5811,
+      "step": 403
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.946943145198674,
+      "learning_rate": 4.675343377549653e-06,
+      "loss": 0.5475,
+      "step": 404
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.991304422730463,
+      "learning_rate": 4.6737301456148445e-06,
+      "loss": 0.5856,
+      "step": 405
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9168241989446437,
+      "learning_rate": 4.672113195252644e-06,
+      "loss": 0.6069,
+      "step": 406
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9305433665377905,
+      "learning_rate": 4.670492529229039e-06,
+      "loss": 0.5536,
+      "step": 407
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8441008898830742,
+      "learning_rate": 4.668868150316377e-06,
+      "loss": 0.5859,
+      "step": 408
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8879301596961315,
+      "learning_rate": 4.667240061293351e-06,
+      "loss": 0.5483,
+      "step": 409
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.024767417636281,
+      "learning_rate": 4.665608264945004e-06,
+      "loss": 0.5414,
+      "step": 410
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1331610141797395,
+      "learning_rate": 4.663972764062722e-06,
+      "loss": 0.5811,
+      "step": 411
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8132480265817386,
+      "learning_rate": 4.662333561444226e-06,
+      "loss": 0.5573,
+      "step": 412
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9795813972027145,
+      "learning_rate": 4.6606906598935675e-06,
+      "loss": 0.5814,
+      "step": 413
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8782931074297053,
+      "learning_rate": 4.6590440622211295e-06,
+      "loss": 0.569,
+      "step": 414
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8219945335518706,
+      "learning_rate": 4.657393771243614e-06,
+      "loss": 0.5669,
+      "step": 415
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4047268604371306,
+      "learning_rate": 4.6557397897840454e-06,
+      "loss": 0.5602,
+      "step": 416
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.064501780523946,
+      "learning_rate": 4.654082120671757e-06,
+      "loss": 0.5699,
+      "step": 417
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9183128854940252,
+      "learning_rate": 4.65242076674239e-06,
+      "loss": 0.6112,
+      "step": 418
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9315698971629633,
+      "learning_rate": 4.650755730837894e-06,
+      "loss": 0.5537,
+      "step": 419
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9527809333659218,
+      "learning_rate": 4.649087015806509e-06,
+      "loss": 0.5423,
+      "step": 420
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8940523915995442,
+      "learning_rate": 4.647414624502777e-06,
+      "loss": 0.5708,
+      "step": 421
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9976964785548623,
+      "learning_rate": 4.645738559787524e-06,
+      "loss": 0.6006,
+      "step": 422
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9098681403283917,
+      "learning_rate": 4.64405882452786e-06,
+      "loss": 0.5591,
+      "step": 423
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8695612182804557,
+      "learning_rate": 4.642375421597175e-06,
+      "loss": 0.5219,
+      "step": 424
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8912077704810082,
+      "learning_rate": 4.6406883538751315e-06,
+      "loss": 0.5224,
+      "step": 425
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9390714726978922,
+      "learning_rate": 4.638997624247664e-06,
+      "loss": 0.5359,
+      "step": 426
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.051545992296337,
+      "learning_rate": 4.637303235606968e-06,
+      "loss": 0.544,
+      "step": 427
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0657109136265914,
+      "learning_rate": 4.6356051908515e-06,
+      "loss": 0.5429,
+      "step": 428
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0301022307984793,
+      "learning_rate": 4.63390349288597e-06,
+      "loss": 0.5787,
+      "step": 429
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.052515756169346,
+      "learning_rate": 4.632198144621338e-06,
+      "loss": 0.5778,
+      "step": 430
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9741370495474897,
+      "learning_rate": 4.630489148974807e-06,
+      "loss": 0.5142,
+      "step": 431
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9713229498863698,
+      "learning_rate": 4.62877650886982e-06,
+      "loss": 0.6127,
+      "step": 432
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1609440121306007,
+      "learning_rate": 4.627060227236055e-06,
+      "loss": 0.5886,
+      "step": 433
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.944966445355139,
+      "learning_rate": 4.625340307009418e-06,
+      "loss": 0.5657,
+      "step": 434
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.031003925680835,
+      "learning_rate": 4.623616751132041e-06,
+      "loss": 0.5628,
+      "step": 435
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8774113373137704,
+      "learning_rate": 4.621889562552272e-06,
+      "loss": 0.6068,
+      "step": 436
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0385201543401785,
+      "learning_rate": 4.620158744224677e-06,
+      "loss": 0.5511,
+      "step": 437
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8440750841938207,
+      "learning_rate": 4.618424299110028e-06,
+      "loss": 0.5261,
+      "step": 438
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8978691755923442,
+      "learning_rate": 4.616686230175303e-06,
+      "loss": 0.5862,
+      "step": 439
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8120850246861446,
+      "learning_rate": 4.614944540393679e-06,
+      "loss": 0.5652,
+      "step": 440
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1821084695714914,
+      "learning_rate": 4.613199232744525e-06,
+      "loss": 0.5598,
+      "step": 441
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9626422737625222,
+      "learning_rate": 4.611450310213401e-06,
+      "loss": 0.5267,
+      "step": 442
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9714913234889215,
+      "learning_rate": 4.6096977757920505e-06,
+      "loss": 0.5658,
+      "step": 443
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0179324078198233,
+      "learning_rate": 4.607941632478393e-06,
+      "loss": 0.582,
+      "step": 444
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8565193856331161,
+      "learning_rate": 4.6061818832765246e-06,
+      "loss": 0.5715,
+      "step": 445
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9798501479599246,
+      "learning_rate": 4.604418531196708e-06,
+      "loss": 0.6007,
+      "step": 446
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0095846956468257,
+      "learning_rate": 4.602651579255369e-06,
+      "loss": 0.5947,
+      "step": 447
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9316541079988245,
+      "learning_rate": 4.600881030475093e-06,
+      "loss": 0.5501,
+      "step": 448
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.080069353365406,
+      "learning_rate": 4.599106887884616e-06,
+      "loss": 0.5631,
+      "step": 449
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.965973137652201,
+      "learning_rate": 4.5973291545188235e-06,
+      "loss": 0.5267,
+      "step": 450
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1082225966704087,
+      "learning_rate": 4.595547833418741e-06,
+      "loss": 0.6418,
+      "step": 451
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0359312594194083,
+      "learning_rate": 4.593762927631536e-06,
+      "loss": 0.5644,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1254892914109433,
+      "learning_rate": 4.591974440210502e-06,
+      "loss": 0.5693,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9121188587334927,
+      "learning_rate": 4.590182374215064e-06,
+      "loss": 0.5572,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9348642624953207,
+      "learning_rate": 4.588386732710765e-06,
+      "loss": 0.5446,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8667846547370581,
+      "learning_rate": 4.5865875187692695e-06,
+      "loss": 0.5681,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9219061327454674,
+      "learning_rate": 4.5847847354683465e-06,
+      "loss": 0.5508,
+      "step": 457
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8106132369123122,
+      "learning_rate": 4.5829783858918756e-06,
+      "loss": 0.5626,
+      "step": 458
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7827483964442634,
+      "learning_rate": 4.5811684731298355e-06,
+      "loss": 0.5575,
+      "step": 459
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9284196979863513,
+      "learning_rate": 4.5793550002783e-06,
+      "loss": 0.5363,
+      "step": 460
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.029647468705457,
+      "learning_rate": 4.577537970439433e-06,
+      "loss": 0.5415,
+      "step": 461
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0997127029950087,
+      "learning_rate": 4.575717386721482e-06,
+      "loss": 0.5814,
+      "step": 462
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9589290300656341,
+      "learning_rate": 4.573893252238777e-06,
+      "loss": 0.5156,
+      "step": 463
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.905237143908251,
+      "learning_rate": 4.572065570111717e-06,
+      "loss": 0.5536,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.929519794935609,
+      "learning_rate": 4.570234343466775e-06,
+      "loss": 0.5879,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.096095808886982,
+      "learning_rate": 4.568399575436484e-06,
+      "loss": 0.6241,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9486118894048778,
+      "learning_rate": 4.566561269159437e-06,
+      "loss": 0.6307,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0839490306744586,
+      "learning_rate": 4.564719427780276e-06,
+      "loss": 0.5655,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9439525665822102,
+      "learning_rate": 4.562874054449694e-06,
+      "loss": 0.5437,
+      "step": 469
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9409142791465297,
+      "learning_rate": 4.5610251523244244e-06,
+      "loss": 0.6429,
+      "step": 470
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8664574493795525,
+      "learning_rate": 4.559172724567238e-06,
+      "loss": 0.5826,
+      "step": 471
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.80819349503324,
+      "learning_rate": 4.557316774346934e-06,
+      "loss": 0.5372,
+      "step": 472
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8680097526865296,
+      "learning_rate": 4.555457304838341e-06,
+      "loss": 0.5503,
+      "step": 473
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7466938790815696,
+      "learning_rate": 4.553594319222303e-06,
+      "loss": 0.5425,
+      "step": 474
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9610557658505607,
+      "learning_rate": 4.551727820685684e-06,
+      "loss": 0.5755,
+      "step": 475
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9414839604282412,
+      "learning_rate": 4.549857812421353e-06,
+      "loss": 0.5915,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8484957644576423,
+      "learning_rate": 4.547984297628186e-06,
+      "loss": 0.5676,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.074524028551078,
+      "learning_rate": 4.546107279511055e-06,
+      "loss": 0.6084,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.069692704122282,
+      "learning_rate": 4.544226761280826e-06,
+      "loss": 0.5676,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8975472248317244,
+      "learning_rate": 4.54234274615435e-06,
+      "loss": 0.5904,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0118868982719897,
+      "learning_rate": 4.540455237354466e-06,
+      "loss": 0.5722,
+      "step": 481
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9733105429381828,
+      "learning_rate": 4.5385642381099814e-06,
+      "loss": 0.6112,
+      "step": 482
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.862156914026863,
+      "learning_rate": 4.53666975165568e-06,
+      "loss": 0.5951,
+      "step": 483
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9512940035297868,
+      "learning_rate": 4.53477178123231e-06,
+      "loss": 0.5223,
+      "step": 484
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9202464191558823,
+      "learning_rate": 4.532870330086577e-06,
+      "loss": 0.5638,
+      "step": 485
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9015767656854419,
+      "learning_rate": 4.530965401471143e-06,
+      "loss": 0.5911,
+      "step": 486
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.95190921973106,
+      "learning_rate": 4.529056998644619e-06,
+      "loss": 0.6053,
+      "step": 487
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0058459596081644,
+      "learning_rate": 4.527145124871556e-06,
+      "loss": 0.5466,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8902620959998047,
+      "learning_rate": 4.5252297834224454e-06,
+      "loss": 0.5526,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.985466416169018,
+      "learning_rate": 4.523310977573711e-06,
+      "loss": 0.5958,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1140148957176415,
+      "learning_rate": 4.521388710607699e-06,
+      "loss": 0.613,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9470601192089525,
+      "learning_rate": 4.51946298581268e-06,
+      "loss": 0.5847,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0227057176069603,
+      "learning_rate": 4.51753380648284e-06,
+      "loss": 0.5784,
+      "step": 493
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.05501863673554,
+      "learning_rate": 4.515601175918269e-06,
+      "loss": 0.5501,
+      "step": 494
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0129325402811715,
+      "learning_rate": 4.513665097424967e-06,
+      "loss": 0.5641,
+      "step": 495
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0322333044110468,
+      "learning_rate": 4.51172557431483e-06,
+      "loss": 0.5422,
+      "step": 496
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9573055659958774,
+      "learning_rate": 4.509782609905644e-06,
+      "loss": 0.516,
+      "step": 497
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8223127451485421,
+      "learning_rate": 4.507836207521085e-06,
+      "loss": 0.5714,
+      "step": 498
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9343089861079434,
+      "learning_rate": 4.50588637049071e-06,
+      "loss": 0.5424,
+      "step": 499
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8940990649350729,
+      "learning_rate": 4.503933102149948e-06,
+      "loss": 0.5832,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.908617301933682,
+      "learning_rate": 4.501976405840101e-06,
+      "loss": 0.5399,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8290259512093785,
+      "learning_rate": 4.500016284908334e-06,
+      "loss": 0.5561,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9840280991844164,
+      "learning_rate": 4.49805274270767e-06,
+      "loss": 0.5645,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9864953051636856,
+      "learning_rate": 4.496085782596984e-06,
+      "loss": 0.5369,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.979387839103732,
+      "learning_rate": 4.494115407940999e-06,
+      "loss": 0.6196,
+      "step": 505
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9266869362165981,
+      "learning_rate": 4.492141622110279e-06,
+      "loss": 0.5687,
+      "step": 506
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9887461782376619,
+      "learning_rate": 4.4901644284812205e-06,
+      "loss": 0.5264,
+      "step": 507
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8717867803152208,
+      "learning_rate": 4.488183830436052e-06,
+      "loss": 0.5612,
+      "step": 508
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0044226171493,
+      "learning_rate": 4.486199831362828e-06,
+      "loss": 0.5571,
+      "step": 509
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1075571016617958,
+      "learning_rate": 4.484212434655414e-06,
+      "loss": 0.5642,
+      "step": 510
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8031612547539957,
+      "learning_rate": 4.482221643713494e-06,
+      "loss": 0.5805,
+      "step": 511
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8782516337672304,
+      "learning_rate": 4.480227461942556e-06,
+      "loss": 0.5596,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.075073901596185,
+      "learning_rate": 4.478229892753886e-06,
+      "loss": 0.6124,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0588983460568304,
+      "learning_rate": 4.47622893956457e-06,
+      "loss": 0.5589,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.850248236464706,
+      "learning_rate": 4.474224605797476e-06,
+      "loss": 0.5603,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.932844310652863,
+      "learning_rate": 4.472216894881261e-06,
+      "loss": 0.5571,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.09975454805468,
+      "learning_rate": 4.470205810250357e-06,
+      "loss": 0.5975,
+      "step": 517
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9694087093010304,
+      "learning_rate": 4.468191355344965e-06,
+      "loss": 0.5698,
+      "step": 518
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8794788153917539,
+      "learning_rate": 4.466173533611053e-06,
+      "loss": 0.5559,
+      "step": 519
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0650455557855434,
+      "learning_rate": 4.46415234850035e-06,
+      "loss": 0.5644,
+      "step": 520
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0062649027982022,
+      "learning_rate": 4.462127803470334e-06,
+      "loss": 0.608,
+      "step": 521
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.043267877462657,
+      "learning_rate": 4.460099901984235e-06,
+      "loss": 0.573,
+      "step": 522
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.056372436619027,
+      "learning_rate": 4.4580686475110235e-06,
+      "loss": 0.5748,
+      "step": 523
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8871033520138176,
+      "learning_rate": 4.456034043525404e-06,
+      "loss": 0.5339,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.889474616209236,
+      "learning_rate": 4.45399609350781e-06,
+      "loss": 0.5185,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9767406217632912,
+      "learning_rate": 4.451954800944405e-06,
+      "loss": 0.5758,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9588695861513832,
+      "learning_rate": 4.449910169327062e-06,
+      "loss": 0.5472,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8852210889000718,
+      "learning_rate": 4.447862202153372e-06,
+      "loss": 0.5917,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0103638871993077,
+      "learning_rate": 4.445810902926629e-06,
+      "loss": 0.5761,
+      "step": 529
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.201836945389513,
+      "learning_rate": 4.443756275155827e-06,
+      "loss": 0.5614,
+      "step": 530
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.900702305836831,
+      "learning_rate": 4.441698322355656e-06,
+      "loss": 0.5254,
+      "step": 531
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.134694583439314,
+      "learning_rate": 4.4396370480464915e-06,
+      "loss": 0.5607,
+      "step": 532
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8073751630381198,
+      "learning_rate": 4.437572455754391e-06,
+      "loss": 0.536,
+      "step": 533
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9607338020142653,
+      "learning_rate": 4.435504549011088e-06,
+      "loss": 0.59,
+      "step": 534
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0756430867435274,
+      "learning_rate": 4.433433331353988e-06,
+      "loss": 0.5538,
+      "step": 535
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8280570853718465,
+      "learning_rate": 4.431358806326158e-06,
+      "loss": 0.5789,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2005143967434977,
+      "learning_rate": 4.429280977476321e-06,
+      "loss": 0.545,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.896479397543979,
+      "learning_rate": 4.4271998483588565e-06,
+      "loss": 0.5791,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.117773381781195,
+      "learning_rate": 4.425115422533785e-06,
+      "loss": 0.5234,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4438942429566617,
+      "learning_rate": 4.423027703566769e-06,
+      "loss": 0.5692,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.873481152225171,
+      "learning_rate": 4.4209366950291025e-06,
+      "loss": 0.5739,
+      "step": 541
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8655199147974673,
+      "learning_rate": 4.4188424004977085e-06,
+      "loss": 0.5795,
+      "step": 542
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.948840412241188,
+      "learning_rate": 4.416744823555129e-06,
+      "loss": 0.5304,
+      "step": 543
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8389034133315045,
+      "learning_rate": 4.414643967789523e-06,
+      "loss": 0.5076,
+      "step": 544
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8269235720085213,
+      "learning_rate": 4.412539836794657e-06,
+      "loss": 0.5837,
+      "step": 545
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1298715969759505,
+      "learning_rate": 4.410432434169902e-06,
+      "loss": 0.5694,
+      "step": 546
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0057741366005746,
+      "learning_rate": 4.408321763520223e-06,
+      "loss": 0.557,
+      "step": 547
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7901331374893255,
+      "learning_rate": 4.406207828456177e-06,
+      "loss": 0.5746,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1994839889416187,
+      "learning_rate": 4.404090632593904e-06,
+      "loss": 0.5407,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9664921082690268,
+      "learning_rate": 4.401970179555123e-06,
+      "loss": 0.5322,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9933486180243851,
+      "learning_rate": 4.399846472967124e-06,
+      "loss": 0.5798,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.986612256562151,
+      "learning_rate": 4.397719516462765e-06,
+      "loss": 0.5213,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.046550123292336,
+      "learning_rate": 4.395589313680459e-06,
+      "loss": 0.5857,
+      "step": 553
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7902327250340486,
+      "learning_rate": 4.393455868264176e-06,
+      "loss": 0.555,
+      "step": 554
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0203627138517146,
+      "learning_rate": 4.391319183863432e-06,
+      "loss": 0.6329,
+      "step": 555
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9373549045181289,
+      "learning_rate": 4.389179264133281e-06,
+      "loss": 0.566,
+      "step": 556
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8936753353678124,
+      "learning_rate": 4.387036112734316e-06,
+      "loss": 0.5555,
+      "step": 557
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8493817575820743,
+      "learning_rate": 4.3848897333326545e-06,
+      "loss": 0.5427,
+      "step": 558
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9119588677783816,
+      "learning_rate": 4.382740129599937e-06,
+      "loss": 0.5157,
+      "step": 559
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8190137094200924,
+      "learning_rate": 4.380587305213321e-06,
+      "loss": 0.503,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9891332712764953,
+      "learning_rate": 4.37843126385547e-06,
+      "loss": 0.5761,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8620896547461154,
+      "learning_rate": 4.376272009214555e-06,
+      "loss": 0.5259,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8896721756477406,
+      "learning_rate": 4.37410954498424e-06,
+      "loss": 0.5632,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8302281976781984,
+      "learning_rate": 4.37194387486368e-06,
+      "loss": 0.5612,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0721820586440165,
+      "learning_rate": 4.369775002557516e-06,
+      "loss": 0.533,
+      "step": 565
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8259926551813157,
+      "learning_rate": 4.367602931775865e-06,
+      "loss": 0.526,
+      "step": 566
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8096334574000785,
+      "learning_rate": 4.3654276662343155e-06,
+      "loss": 0.5306,
+      "step": 567
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9675637591445598,
+      "learning_rate": 4.363249209653922e-06,
+      "loss": 0.5577,
+      "step": 568
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8800389115841605,
+      "learning_rate": 4.361067565761197e-06,
+      "loss": 0.5553,
+      "step": 569
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.827485496395265,
+      "learning_rate": 4.358882738288105e-06,
+      "loss": 0.5587,
+      "step": 570
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.820954908943235,
+      "learning_rate": 4.356694730972056e-06,
+      "loss": 0.6186,
+      "step": 571
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.952072431699686,
+      "learning_rate": 4.3545035475559025e-06,
+      "loss": 0.5488,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8292648968688423,
+      "learning_rate": 4.352309191787924e-06,
+      "loss": 0.5534,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.826293122529813,
+      "learning_rate": 4.350111667421835e-06,
+      "loss": 0.5872,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9251425791166785,
+      "learning_rate": 4.347910978216763e-06,
+      "loss": 0.5298,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8330818196811385,
+      "learning_rate": 4.345707127937253e-06,
+      "loss": 0.5871,
+      "step": 576
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7842986545873851,
+      "learning_rate": 4.3435001203532555e-06,
+      "loss": 0.4898,
+      "step": 577
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8778666245156521,
+      "learning_rate": 4.341289959240124e-06,
+      "loss": 0.5385,
+      "step": 578
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9300679499181266,
+      "learning_rate": 4.339076648378605e-06,
+      "loss": 0.5698,
+      "step": 579
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9440861965960357,
+      "learning_rate": 4.336860191554833e-06,
+      "loss": 0.5984,
+      "step": 580
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.929951096053947,
+      "learning_rate": 4.3346405925603265e-06,
+      "loss": 0.6222,
+      "step": 581
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9138258400335695,
+      "learning_rate": 4.332417855191974e-06,
+      "loss": 0.5498,
+      "step": 582
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.058548455869675,
+      "learning_rate": 4.330191983252039e-06,
+      "loss": 0.5218,
+      "step": 583
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.243429045583125,
+      "learning_rate": 4.327962980548142e-06,
+      "loss": 0.5768,
+      "step": 584
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9213537104634244,
+      "learning_rate": 4.32573085089326e-06,
+      "loss": 0.5784,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9165291289119128,
+      "learning_rate": 4.32349559810572e-06,
+      "loss": 0.5697,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9674279518735756,
+      "learning_rate": 4.321257226009193e-06,
+      "loss": 0.5104,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9051339015323923,
+      "learning_rate": 4.319015738432683e-06,
+      "loss": 0.5711,
+      "step": 588
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.957357618850765,
+      "learning_rate": 4.3167711392105245e-06,
+      "loss": 0.5854,
+      "step": 589
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9859311708308915,
+      "learning_rate": 4.314523432182376e-06,
+      "loss": 0.547,
+      "step": 590
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.773704456523191,
+      "learning_rate": 4.312272621193209e-06,
+      "loss": 0.5259,
+      "step": 591
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.82988033655793,
+      "learning_rate": 4.31001871009331e-06,
+      "loss": 0.5209,
+      "step": 592
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8925134832060522,
+      "learning_rate": 4.307761702738264e-06,
+      "loss": 0.59,
+      "step": 593
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8477075780641046,
+      "learning_rate": 4.305501602988953e-06,
+      "loss": 0.5714,
+      "step": 594
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8568432886623798,
+      "learning_rate": 4.303238414711552e-06,
+      "loss": 0.5877,
+      "step": 595
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8179798660158206,
+      "learning_rate": 4.3009721417775166e-06,
+      "loss": 0.6029,
+      "step": 596
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8494963193854803,
+      "learning_rate": 4.29870278806358e-06,
+      "loss": 0.5236,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9586017397154731,
+      "learning_rate": 4.296430357451744e-06,
+      "loss": 0.5998,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.926616057974202,
+      "learning_rate": 4.2941548538292765e-06,
+      "loss": 0.5914,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9321738359144827,
+      "learning_rate": 4.291876281088701e-06,
+      "loss": 0.5358,
+      "step": 600
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8229177571361932,
+      "learning_rate": 4.289594643127788e-06,
+      "loss": 0.5284,
+      "step": 601
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.849252449531427,
+      "learning_rate": 4.287309943849558e-06,
+      "loss": 0.5689,
+      "step": 602
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.985343175388319,
+      "learning_rate": 4.285022187162261e-06,
+      "loss": 0.6101,
+      "step": 603
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2412,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 603,
+  "total_flos": 283958071787520.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}