diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4869 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 690,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004347826086956522,
+      "learning_rate": 0,
+      "loss": 2.2149,
+      "step": 1
+    },
+    {
+      "epoch": 0.008695652173913044,
+      "learning_rate": 0,
+      "loss": 2.2161,
+      "step": 2
+    },
+    {
+      "epoch": 0.013043478260869565,
+      "learning_rate": 0,
+      "loss": 2.3121,
+      "step": 3
+    },
+    {
+      "epoch": 0.017391304347826087,
+      "grad_norm": 3.31545352935791,
+      "learning_rate": 0.0,
+      "loss": 2.2694,
+      "step": 4
+    },
+    {
+      "epoch": 0.021739130434782608,
+      "grad_norm": 3.31545352935791,
+      "learning_rate": 0.0,
+      "loss": 2.298,
+      "step": 5
+    },
+    {
+      "epoch": 0.02608695652173913,
+      "grad_norm": 3.31545352935791,
+      "learning_rate": 0.0,
+      "loss": 2.2501,
+      "step": 6
+    },
+    {
+      "epoch": 0.030434782608695653,
+      "grad_norm": 3.31545352935791,
+      "learning_rate": 0.0,
+      "loss": 2.2571,
+      "step": 7
+    },
+    {
+      "epoch": 0.034782608695652174,
+      "grad_norm": 3.273796796798706,
+      "learning_rate": 3.2741131089043125e-06,
+      "loss": 2.286,
+      "step": 8
+    },
+    {
+      "epoch": 0.0391304347826087,
+      "grad_norm": 3.273796796798706,
+      "learning_rate": 3.2741131089043125e-06,
+      "loss": 2.2707,
+      "step": 9
+    },
+    {
+      "epoch": 0.043478260869565216,
+      "grad_norm": 3.273796796798706,
+      "learning_rate": 3.2741131089043125e-06,
+      "loss": 2.3606,
+      "step": 10
+    },
+    {
+      "epoch": 0.04782608695652174,
+      "grad_norm": 3.273796796798706,
+      "learning_rate": 3.2741131089043125e-06,
+      "loss": 2.306,
+      "step": 11
+    },
+    {
+      "epoch": 0.05217391304347826,
+      "grad_norm": 3.272700071334839,
+      "learning_rate": 5.189346500732899e-06,
+      "loss": 2.3623,
+      "step": 12
+    },
+    {
+      "epoch": 0.05652173913043478,
+      "grad_norm": 3.272700071334839,
+      "learning_rate": 5.189346500732899e-06,
+      "loss": 2.3027,
+      "step": 13
+    },
+    {
+      "epoch": 0.06086956521739131,
+      "grad_norm": 3.272700071334839,
+      "learning_rate": 5.189346500732899e-06,
+      "loss": 2.3053,
+      "step": 14
+    },
+    {
+      "epoch": 0.06521739130434782,
+      "grad_norm": 3.272700071334839,
+      "learning_rate": 5.189346500732899e-06,
+      "loss": 2.2793,
+      "step": 15
+    },
+    {
+      "epoch": 0.06956521739130435,
+      "grad_norm": 3.2200253009796143,
+      "learning_rate": 6.548226217808625e-06,
+      "loss": 2.2608,
+      "step": 16
+    },
+    {
+      "epoch": 0.07391304347826087,
+      "grad_norm": 3.2200253009796143,
+      "learning_rate": 6.548226217808625e-06,
+      "loss": 2.2949,
+      "step": 17
+    },
+    {
+      "epoch": 0.0782608695652174,
+      "grad_norm": 3.2200253009796143,
+      "learning_rate": 6.548226217808625e-06,
+      "loss": 2.2639,
+      "step": 18
+    },
+    {
+      "epoch": 0.08260869565217391,
+      "grad_norm": 3.2200253009796143,
+      "learning_rate": 6.548226217808625e-06,
+      "loss": 2.2472,
+      "step": 19
+    },
+    {
+      "epoch": 0.08695652173913043,
+      "grad_norm": 3.144207239151001,
+      "learning_rate": 7.60225521340393e-06,
+      "loss": 2.1505,
+      "step": 20
+    },
+    {
+      "epoch": 0.09130434782608696,
+      "grad_norm": 3.144207239151001,
+      "learning_rate": 7.60225521340393e-06,
+      "loss": 2.2222,
+      "step": 21
+    },
+    {
+      "epoch": 0.09565217391304348,
+      "grad_norm": 3.144207239151001,
+      "learning_rate": 7.60225521340393e-06,
+      "loss": 2.3522,
+      "step": 22
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.144207239151001,
+      "learning_rate": 7.60225521340393e-06,
+      "loss": 2.275,
+      "step": 23
+    },
+    {
+      "epoch": 0.10434782608695652,
+      "grad_norm": 3.0425093173980713,
+      "learning_rate": 8.463459609637211e-06,
+      "loss": 2.3877,
+      "step": 24
+    },
+    {
+      "epoch": 0.10869565217391304,
+      "grad_norm": 3.0425093173980713,
+      "learning_rate": 8.463459609637211e-06,
+      "loss": 2.2595,
+      "step": 25
+    },
+    {
+      "epoch": 0.11304347826086956,
+      "grad_norm": 3.0425093173980713,
+      "learning_rate": 8.463459609637211e-06,
+      "loss": 2.213,
+      "step": 26
+    },
+    {
+      "epoch": 0.11739130434782609,
+      "grad_norm": 3.0425093173980713,
+      "learning_rate": 8.463459609637211e-06,
+      "loss": 2.2204,
+      "step": 27
+    },
+    {
+      "epoch": 0.12173913043478261,
+      "grad_norm": 2.7785863876342773,
+      "learning_rate": 9.191597551655847e-06,
+      "loss": 2.2942,
+      "step": 28
+    },
+    {
+      "epoch": 0.12608695652173912,
+      "grad_norm": 2.7785863876342773,
+      "learning_rate": 9.191597551655847e-06,
+      "loss": 2.1182,
+      "step": 29
+    },
+    {
+      "epoch": 0.13043478260869565,
+      "grad_norm": 2.7785863876342773,
+      "learning_rate": 9.191597551655847e-06,
+      "loss": 2.0758,
+      "step": 30
+    },
+    {
+      "epoch": 0.13478260869565217,
+      "grad_norm": 2.7785863876342773,
+      "learning_rate": 9.191597551655847e-06,
+      "loss": 1.9933,
+      "step": 31
+    },
+    {
+      "epoch": 0.1391304347826087,
+      "grad_norm": 2.2910876274108887,
+      "learning_rate": 9.822339326712938e-06,
+      "loss": 2.1012,
+      "step": 32
+    },
+    {
+      "epoch": 0.14347826086956522,
+      "grad_norm": 2.2910876274108887,
+      "learning_rate": 9.822339326712938e-06,
+      "loss": 1.953,
+      "step": 33
+    },
+    {
+      "epoch": 0.14782608695652175,
+      "grad_norm": 2.2910876274108887,
+      "learning_rate": 9.822339326712938e-06,
+      "loss": 2.0373,
+      "step": 34
+    },
+    {
+      "epoch": 0.15217391304347827,
+      "grad_norm": 2.2910876274108887,
+      "learning_rate": 9.822339326712938e-06,
+      "loss": 1.9755,
+      "step": 35
+    },
+    {
+      "epoch": 0.1565217391304348,
+      "grad_norm": 1.7529128789901733,
+      "learning_rate": 1.0378693001465798e-05,
+      "loss": 2.0704,
+      "step": 36
+    },
+    {
+      "epoch": 0.1608695652173913,
+      "grad_norm": 1.7529128789901733,
+      "learning_rate": 1.0378693001465798e-05,
+      "loss": 1.9861,
+      "step": 37
+    },
+    {
+      "epoch": 0.16521739130434782,
+      "grad_norm": 1.7529128789901733,
+      "learning_rate": 1.0378693001465798e-05,
+      "loss": 2.0054,
+      "step": 38
+    },
+    {
+      "epoch": 0.16956521739130434,
+      "grad_norm": 1.7529128789901733,
+      "learning_rate": 1.0378693001465798e-05,
+      "loss": 2.098,
+      "step": 39
+    },
+    {
+      "epoch": 0.17391304347826086,
+      "grad_norm": 1.4156919717788696,
+      "learning_rate": 1.0876368322308244e-05,
+      "loss": 1.9489,
+      "step": 40
+    },
+    {
+      "epoch": 0.1782608695652174,
+      "grad_norm": 1.4156919717788696,
+      "learning_rate": 1.0876368322308244e-05,
+      "loss": 1.9384,
+      "step": 41
+    },
+    {
+      "epoch": 0.1826086956521739,
+      "grad_norm": 1.4156919717788696,
+      "learning_rate": 1.0876368322308244e-05,
+      "loss": 1.9856,
+      "step": 42
+    },
+    {
+      "epoch": 0.18695652173913044,
+      "grad_norm": 1.4156919717788696,
+      "learning_rate": 1.0876368322308244e-05,
+      "loss": 2.0604,
+      "step": 43
+    },
+    {
+      "epoch": 0.19130434782608696,
+      "grad_norm": 1.382755160331726,
+      "learning_rate": 1.1326570411938442e-05,
+      "loss": 1.9795,
+      "step": 44
+    },
+    {
+      "epoch": 0.1956521739130435,
+      "grad_norm": 1.382755160331726,
+      "learning_rate": 1.1326570411938442e-05,
+      "loss": 2.0085,
+      "step": 45
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.382755160331726,
+      "learning_rate": 1.1326570411938442e-05,
+      "loss": 1.8687,
+      "step": 46
+    },
+    {
+      "epoch": 0.20434782608695654,
+      "grad_norm": 1.382755160331726,
+      "learning_rate": 1.1326570411938442e-05,
+      "loss": 1.8855,
+      "step": 47
+    },
+    {
+      "epoch": 0.20869565217391303,
+      "grad_norm": 1.4150187969207764,
+      "learning_rate": 1.1737572718541526e-05,
+      "loss": 1.9876,
+      "step": 48
+    },
+    {
+      "epoch": 0.21304347826086956,
+      "grad_norm": 1.4150187969207764,
+      "learning_rate": 1.1737572718541526e-05,
+      "loss": 1.8712,
+      "step": 49
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "grad_norm": 1.4150187969207764,
+      "learning_rate": 1.1737572718541526e-05,
+      "loss": 1.8736,
+      "step": 50
+    },
+    {
+      "epoch": 0.2217391304347826,
+      "grad_norm": 1.4150187969207764,
+      "learning_rate": 1.1737572718541526e-05,
+      "loss": 1.9638,
+      "step": 51
+    },
+    {
+      "epoch": 0.22608695652173913,
+      "grad_norm": 1.1287221908569336,
+      "learning_rate": 1.2115658189875932e-05,
+      "loss": 1.9343,
+      "step": 52
+    },
+    {
+      "epoch": 0.23043478260869565,
+      "grad_norm": 1.1287221908569336,
+      "learning_rate": 1.2115658189875932e-05,
+      "loss": 2.0118,
+      "step": 53
+    },
+    {
+      "epoch": 0.23478260869565218,
+      "grad_norm": 1.1287221908569336,
+      "learning_rate": 1.2115658189875932e-05,
+      "loss": 1.9,
+      "step": 54
+    },
+    {
+      "epoch": 0.2391304347826087,
+      "grad_norm": 1.1287221908569336,
+      "learning_rate": 1.2115658189875932e-05,
+      "loss": 1.9386,
+      "step": 55
+    },
+    {
+      "epoch": 0.24347826086956523,
+      "grad_norm": 1.0455442667007446,
+      "learning_rate": 1.2465710660560159e-05,
+      "loss": 1.9017,
+      "step": 56
+    },
+    {
+      "epoch": 0.24782608695652175,
+      "grad_norm": 1.0455442667007446,
+      "learning_rate": 1.2465710660560159e-05,
+      "loss": 1.884,
+      "step": 57
+    },
+    {
+      "epoch": 0.25217391304347825,
+      "grad_norm": 1.0455442667007446,
+      "learning_rate": 1.2465710660560159e-05,
+      "loss": 2.0199,
+      "step": 58
+    },
+    {
+      "epoch": 0.2565217391304348,
+      "grad_norm": 1.0455442667007446,
+      "learning_rate": 1.2465710660560159e-05,
+      "loss": 1.9413,
+      "step": 59
+    },
+    {
+      "epoch": 0.2608695652173913,
+      "grad_norm": 0.9044898152351379,
+      "learning_rate": 1.279160171413683e-05,
+      "loss": 1.8714,
+      "step": 60
+    },
+    {
+      "epoch": 0.26521739130434785,
+      "grad_norm": 0.9044898152351379,
+      "learning_rate": 1.279160171413683e-05,
+      "loss": 1.9318,
+      "step": 61
+    },
+    {
+      "epoch": 0.26956521739130435,
+      "grad_norm": 0.9044898152351379,
+      "learning_rate": 1.279160171413683e-05,
+      "loss": 1.9003,
+      "step": 62
+    },
+    {
+      "epoch": 0.27391304347826084,
+      "grad_norm": 0.9044898152351379,
+      "learning_rate": 1.279160171413683e-05,
+      "loss": 1.8814,
+      "step": 63
+    },
+    {
+      "epoch": 0.2782608695652174,
+      "grad_norm": 0.7885456681251526,
+      "learning_rate": 1.309645243561725e-05,
+      "loss": 1.9581,
+      "step": 64
+    },
+    {
+      "epoch": 0.2826086956521739,
+      "grad_norm": 0.7885456681251526,
+      "learning_rate": 1.309645243561725e-05,
+      "loss": 1.818,
+      "step": 65
+    },
+    {
+      "epoch": 0.28695652173913044,
+      "grad_norm": 0.7885456681251526,
+      "learning_rate": 1.309645243561725e-05,
+      "loss": 1.796,
+      "step": 66
+    },
+    {
+      "epoch": 0.29130434782608694,
+      "grad_norm": 0.7885456681251526,
+      "learning_rate": 1.309645243561725e-05,
+      "loss": 1.8588,
+      "step": 67
+    },
+    {
+      "epoch": 0.2956521739130435,
+      "grad_norm": 0.6783104538917542,
+      "learning_rate": 1.3382815670697004e-05,
+      "loss": 1.8261,
+      "step": 68
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6783104538917542,
+      "learning_rate": 1.3382815670697004e-05,
+      "loss": 1.8421,
+      "step": 69
+    },
+    {
+      "epoch": 0.30434782608695654,
+      "grad_norm": 0.6783104538917542,
+      "learning_rate": 1.3382815670697004e-05,
+      "loss": 1.7374,
+      "step": 70
+    },
+    {
+      "epoch": 0.30869565217391304,
+      "grad_norm": 0.6783104538917542,
+      "learning_rate": 1.3382815670697004e-05,
+      "loss": 1.8355,
+      "step": 71
+    },
+    {
+      "epoch": 0.3130434782608696,
+      "grad_norm": 0.6615838408470154,
+      "learning_rate": 1.365280611037011e-05,
+      "loss": 1.877,
+      "step": 72
+    },
+    {
+      "epoch": 0.3173913043478261,
+      "grad_norm": 0.6615838408470154,
+      "learning_rate": 1.365280611037011e-05,
+      "loss": 1.7588,
+      "step": 73
+    },
+    {
+      "epoch": 0.3217391304347826,
+      "grad_norm": 0.6615838408470154,
+      "learning_rate": 1.365280611037011e-05,
+      "loss": 1.859,
+      "step": 74
+    },
+    {
+      "epoch": 0.32608695652173914,
+      "grad_norm": 0.6615838408470154,
+      "learning_rate": 1.365280611037011e-05,
+      "loss": 1.8263,
+      "step": 75
+    },
+    {
+      "epoch": 0.33043478260869563,
+      "grad_norm": 0.7432522773742676,
+      "learning_rate": 1.3908195157440944e-05,
+      "loss": 1.864,
+      "step": 76
+    },
+    {
+      "epoch": 0.3347826086956522,
+      "grad_norm": 0.7432522773742676,
+      "learning_rate": 1.3908195157440944e-05,
+      "loss": 1.8922,
+      "step": 77
+    },
+    {
+      "epoch": 0.3391304347826087,
+      "grad_norm": 0.7432522773742676,
+      "learning_rate": 1.3908195157440944e-05,
+      "loss": 1.8468,
+      "step": 78
+    },
+    {
+      "epoch": 0.34347826086956523,
+      "grad_norm": 0.7432522773742676,
+      "learning_rate": 1.3908195157440944e-05,
+      "loss": 1.7736,
+      "step": 79
+    },
+    {
+      "epoch": 0.34782608695652173,
+      "grad_norm": 0.6891193389892578,
+      "learning_rate": 1.4150481431212555e-05,
+      "loss": 1.8292,
+      "step": 80
+    },
+    {
+      "epoch": 0.3521739130434783,
+      "grad_norm": 0.6891193389892578,
+      "learning_rate": 1.4150481431212555e-05,
+      "loss": 1.7703,
+      "step": 81
+    },
+    {
+      "epoch": 0.3565217391304348,
+      "grad_norm": 0.6891193389892578,
+      "learning_rate": 1.4150481431212555e-05,
+      "loss": 1.812,
+      "step": 82
+    },
+    {
+      "epoch": 0.36086956521739133,
+      "grad_norm": 0.6891193389892578,
+      "learning_rate": 1.4150481431212555e-05,
+      "loss": 1.8104,
+      "step": 83
+    },
+    {
+      "epoch": 0.3652173913043478,
+      "grad_norm": 0.5785617232322693,
+      "learning_rate": 1.4380944052388746e-05,
+      "loss": 1.8025,
+      "step": 84
+    },
+    {
+      "epoch": 0.3695652173913043,
+      "grad_norm": 0.5785617232322693,
+      "learning_rate": 1.4380944052388746e-05,
+      "loss": 1.8167,
+      "step": 85
+    },
+    {
+      "epoch": 0.3739130434782609,
+      "grad_norm": 0.5785617232322693,
+      "learning_rate": 1.4380944052388746e-05,
+      "loss": 1.8114,
+      "step": 86
+    },
+    {
+      "epoch": 0.3782608695652174,
+      "grad_norm": 0.5785617232322693,
+      "learning_rate": 1.4380944052388746e-05,
+      "loss": 1.7922,
+      "step": 87
+    },
+    {
+      "epoch": 0.3826086956521739,
+      "grad_norm": 0.5551931262016296,
+      "learning_rate": 1.4600683520842756e-05,
+      "loss": 1.7535,
+      "step": 88
+    },
+    {
+      "epoch": 0.3869565217391304,
+      "grad_norm": 0.5551931262016296,
+      "learning_rate": 1.4600683520842756e-05,
+      "loss": 1.8034,
+      "step": 89
+    },
+    {
+      "epoch": 0.391304347826087,
+      "grad_norm": 0.5551931262016296,
+      "learning_rate": 1.4600683520842756e-05,
+      "loss": 1.799,
+      "step": 90
+    },
+    {
+      "epoch": 0.39565217391304347,
+      "grad_norm": 0.5551931262016296,
+      "learning_rate": 1.4600683520842756e-05,
+      "loss": 1.7228,
+      "step": 91
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.548203706741333,
+      "learning_rate": 1.48106534992671e-05,
+      "loss": 1.826,
+      "step": 92
+    },
+    {
+      "epoch": 0.4043478260869565,
+      "grad_norm": 0.548203706741333,
+      "learning_rate": 1.48106534992671e-05,
+      "loss": 1.8175,
+      "step": 93
+    },
+    {
+      "epoch": 0.40869565217391307,
+      "grad_norm": 0.548203706741333,
+      "learning_rate": 1.48106534992671e-05,
+      "loss": 1.7583,
+      "step": 94
+    },
+    {
+      "epoch": 0.41304347826086957,
+      "grad_norm": 0.548203706741333,
+      "learning_rate": 1.48106534992671e-05,
+      "loss": 1.7545,
+      "step": 95
+    },
+    {
+      "epoch": 0.41739130434782606,
+      "grad_norm": 0.5352587699890137,
+      "learning_rate": 1.5011685827445838e-05,
+      "loss": 1.8338,
+      "step": 96
+    },
+    {
+      "epoch": 0.4217391304347826,
+      "grad_norm": 0.5352587699890137,
+      "learning_rate": 1.5011685827445838e-05,
+      "loss": 1.7581,
+      "step": 97
+    },
+    {
+      "epoch": 0.4260869565217391,
+      "grad_norm": 0.5352587699890137,
+      "learning_rate": 1.5011685827445838e-05,
+      "loss": 1.738,
+      "step": 98
+    },
+    {
+      "epoch": 0.43043478260869567,
+      "grad_norm": 0.5352587699890137,
+      "learning_rate": 1.5011685827445838e-05,
+      "loss": 1.7772,
+      "step": 99
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 0.516083836555481,
+      "learning_rate": 1.520451042680786e-05,
+      "loss": 1.7126,
+      "step": 100
+    },
+    {
+      "epoch": 0.4391304347826087,
+      "grad_norm": 0.516083836555481,
+      "learning_rate": 1.520451042680786e-05,
+      "loss": 1.7873,
+      "step": 101
+    },
+    {
+      "epoch": 0.4434782608695652,
+      "grad_norm": 0.516083836555481,
+      "learning_rate": 1.520451042680786e-05,
+      "loss": 1.7673,
+      "step": 102
+    },
+    {
+      "epoch": 0.44782608695652176,
+      "grad_norm": 0.516083836555481,
+      "learning_rate": 1.520451042680786e-05,
+      "loss": 1.7459,
+      "step": 103
+    },
+    {
+      "epoch": 0.45217391304347826,
+      "grad_norm": 0.5135278701782227,
+      "learning_rate": 1.5389771298780244e-05,
+      "loss": 1.7853,
+      "step": 104
+    },
+    {
+      "epoch": 0.45652173913043476,
+      "grad_norm": 0.5135278701782227,
+      "learning_rate": 1.5389771298780244e-05,
+      "loss": 1.7792,
+      "step": 105
+    },
+    {
+      "epoch": 0.4608695652173913,
+      "grad_norm": 0.5135278701782227,
+      "learning_rate": 1.5389771298780244e-05,
+      "loss": 1.8095,
+      "step": 106
+    },
+    {
+      "epoch": 0.4652173913043478,
+      "grad_norm": 0.5135278701782227,
+      "learning_rate": 1.5389771298780244e-05,
+      "loss": 1.714,
+      "step": 107
+    },
+    {
+      "epoch": 0.46956521739130436,
+      "grad_norm": 0.47637420892715454,
+      "learning_rate": 1.5568039502198696e-05,
+      "loss": 1.757,
+      "step": 108
+    },
+    {
+      "epoch": 0.47391304347826085,
+      "grad_norm": 0.47637420892715454,
+      "learning_rate": 1.5568039502198696e-05,
+      "loss": 1.715,
+      "step": 109
+    },
+    {
+      "epoch": 0.4782608695652174,
+      "grad_norm": 0.47637420892715454,
+      "learning_rate": 1.5568039502198696e-05,
+      "loss": 1.7535,
+      "step": 110
+    },
+    {
+      "epoch": 0.4826086956521739,
+      "grad_norm": 0.47637420892715454,
+      "learning_rate": 1.5568039502198696e-05,
+      "loss": 1.7446,
+      "step": 111
+    },
+    {
+      "epoch": 0.48695652173913045,
+      "grad_norm": 0.46542423963546753,
+      "learning_rate": 1.5739823769464473e-05,
+      "loss": 1.7352,
+      "step": 112
+    },
+    {
+      "epoch": 0.49130434782608695,
+      "grad_norm": 0.46542423963546753,
+      "learning_rate": 1.5739823769464473e-05,
+      "loss": 1.7561,
+      "step": 113
+    },
+    {
+      "epoch": 0.4956521739130435,
+      "grad_norm": 0.46542423963546753,
+      "learning_rate": 1.5739823769464473e-05,
+      "loss": 1.7318,
+      "step": 114
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.46542423963546753,
+      "learning_rate": 1.5739823769464473e-05,
+      "loss": 1.7682,
+      "step": 115
+    },
+    {
+      "epoch": 0.5043478260869565,
+      "grad_norm": 0.4574941396713257,
+      "learning_rate": 1.5905579258955202e-05,
+      "loss": 1.7042,
+      "step": 116
+    },
+    {
+      "epoch": 0.508695652173913,
+      "grad_norm": 0.4574941396713257,
+      "learning_rate": 1.5905579258955202e-05,
+      "loss": 1.7712,
+      "step": 117
+    },
+    {
+      "epoch": 0.5130434782608696,
+      "grad_norm": 0.4574941396713257,
+      "learning_rate": 1.5905579258955202e-05,
+      "loss": 1.7179,
+      "step": 118
+    },
+    {
+      "epoch": 0.5173913043478261,
+      "grad_norm": 0.4574941396713257,
+      "learning_rate": 1.5905579258955202e-05,
+      "loss": 1.7505,
+      "step": 119
+    },
+    {
+      "epoch": 0.5217391304347826,
+      "grad_norm": 0.45183202624320984,
+      "learning_rate": 1.606571482304114e-05,
+      "loss": 1.75,
+      "step": 120
+    },
+    {
+      "epoch": 0.5260869565217391,
+      "grad_norm": 0.45183202624320984,
+      "learning_rate": 1.606571482304114e-05,
+      "loss": 1.7725,
+      "step": 121
+    },
+    {
+      "epoch": 0.5304347826086957,
+      "grad_norm": 0.45183202624320984,
+      "learning_rate": 1.606571482304114e-05,
+      "loss": 1.7267,
+      "step": 122
+    },
+    {
+      "epoch": 0.5347826086956522,
+      "grad_norm": 0.45183202624320984,
+      "learning_rate": 1.606571482304114e-05,
+      "loss": 1.6849,
+      "step": 123
+    },
+    {
+      "epoch": 0.5391304347826087,
+      "grad_norm": 0.46187731623649597,
+      "learning_rate": 1.6220599083923048e-05,
+      "loss": 1.731,
+      "step": 124
+    },
+    {
+      "epoch": 0.5434782608695652,
+      "grad_norm": 0.46187731623649597,
+      "learning_rate": 1.6220599083923048e-05,
+      "loss": 1.695,
+      "step": 125
+    },
+    {
+      "epoch": 0.5478260869565217,
+      "grad_norm": 0.46187731623649597,
+      "learning_rate": 1.6220599083923048e-05,
+      "loss": 1.7525,
+      "step": 126
+    },
+    {
+      "epoch": 0.5521739130434783,
+      "grad_norm": 0.46187731623649597,
+      "learning_rate": 1.6220599083923048e-05,
+      "loss": 1.7797,
+      "step": 127
+    },
+    {
+      "epoch": 0.5565217391304348,
+      "grad_norm": 0.44319385290145874,
+      "learning_rate": 1.6370565544521564e-05,
+      "loss": 1.7984,
+      "step": 128
+    },
+    {
+      "epoch": 0.5608695652173913,
+      "grad_norm": 0.44319385290145874,
+      "learning_rate": 1.6370565544521564e-05,
+      "loss": 1.6682,
+      "step": 129
+    },
+    {
+      "epoch": 0.5652173913043478,
+      "grad_norm": 0.44319385290145874,
+      "learning_rate": 1.6370565544521564e-05,
+      "loss": 1.6129,
+      "step": 130
+    },
+    {
+      "epoch": 0.5695652173913044,
+      "grad_norm": 0.44319385290145874,
+      "learning_rate": 1.6370565544521564e-05,
+      "loss": 1.7295,
+      "step": 131
+    },
+    {
+      "epoch": 0.5739130434782609,
+      "grad_norm": 0.43330496549606323,
+      "learning_rate": 1.651591691267134e-05,
+      "loss": 1.7513,
+      "step": 132
+    },
+    {
+      "epoch": 0.5782608695652174,
+      "grad_norm": 0.43330496549606323,
+      "learning_rate": 1.651591691267134e-05,
+      "loss": 1.7653,
+      "step": 133
+    },
+    {
+      "epoch": 0.5826086956521739,
+      "grad_norm": 0.43330496549606323,
+      "learning_rate": 1.651591691267134e-05,
+      "loss": 1.6724,
+      "step": 134
+    },
+    {
+      "epoch": 0.5869565217391305,
+      "grad_norm": 0.43330496549606323,
+      "learning_rate": 1.651591691267134e-05,
+      "loss": 1.7435,
+      "step": 135
+    },
+    {
+      "epoch": 0.591304347826087,
+      "grad_norm": 0.4281637966632843,
+      "learning_rate": 1.6656928779601318e-05,
+      "loss": 1.7297,
+      "step": 136
+    },
+    {
+      "epoch": 0.5956521739130435,
+      "grad_norm": 0.4281637966632843,
+      "learning_rate": 1.6656928779601318e-05,
+      "loss": 1.7354,
+      "step": 137
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4281637966632843,
+      "learning_rate": 1.6656928779601318e-05,
+      "loss": 1.806,
+      "step": 138
+    },
+    {
+      "epoch": 0.6043478260869565,
+      "grad_norm": 0.4281637966632843,
+      "learning_rate": 1.6656928779601318e-05,
+      "loss": 1.7033,
+      "step": 139
+    },
+    {
+      "epoch": 0.6086956521739131,
+      "grad_norm": 0.459945410490036,
+      "learning_rate": 1.6793852765059776e-05,
+      "loss": 1.6826,
+      "step": 140
+    },
+    {
+      "epoch": 0.6130434782608696,
+      "grad_norm": 0.459945410490036,
+      "learning_rate": 1.6793852765059776e-05,
+      "loss": 1.6544,
+      "step": 141
+    },
+    {
+      "epoch": 0.6173913043478261,
+      "grad_norm": 0.459945410490036,
+      "learning_rate": 1.6793852765059776e-05,
+      "loss": 1.7139,
+      "step": 142
+    },
+    {
+      "epoch": 0.6217391304347826,
+      "grad_norm": 0.459945410490036,
+      "learning_rate": 1.6793852765059776e-05,
+      "loss": 1.7592,
+      "step": 143
+    },
+    {
+      "epoch": 0.6260869565217392,
+      "grad_norm": 0.41728129982948303,
+      "learning_rate": 1.6926919219274422e-05,
+      "loss": 1.7499,
+      "step": 144
+    },
+    {
+      "epoch": 0.6304347826086957,
+      "grad_norm": 0.41728129982948303,
+      "learning_rate": 1.6926919219274422e-05,
+      "loss": 1.7006,
+      "step": 145
+    },
+    {
+      "epoch": 0.6347826086956522,
+      "grad_norm": 0.41728129982948303,
+      "learning_rate": 1.6926919219274422e-05,
+      "loss": 1.6757,
+      "step": 146
+    },
+    {
+      "epoch": 0.6391304347826087,
+      "grad_norm": 0.41728129982948303,
+      "learning_rate": 1.6926919219274422e-05,
+      "loss": 1.7343,
+      "step": 147
+    },
+    {
+      "epoch": 0.6434782608695652,
+      "grad_norm": 0.436886727809906,
+      "learning_rate": 1.7056339554631436e-05,
+      "loss": 1.716,
+      "step": 148
+    },
+    {
+      "epoch": 0.6478260869565218,
+      "grad_norm": 0.436886727809906,
+      "learning_rate": 1.7056339554631436e-05,
+      "loss": 1.6438,
+      "step": 149
+    },
+    {
+      "epoch": 0.6521739130434783,
+      "grad_norm": 0.436886727809906,
+      "learning_rate": 1.7056339554631436e-05,
+      "loss": 1.6398,
+      "step": 150
+    },
+    {
+      "epoch": 0.6565217391304348,
+      "grad_norm": 0.436886727809906,
+      "learning_rate": 1.7056339554631436e-05,
+      "loss": 1.65,
+      "step": 151
+    },
+    {
+      "epoch": 0.6608695652173913,
+      "grad_norm": 0.42069295048713684,
+      "learning_rate": 1.7182308266345256e-05,
+      "loss": 1.7152,
+      "step": 152
+    },
+    {
+      "epoch": 0.6652173913043479,
+      "grad_norm": 0.42069295048713684,
+      "learning_rate": 1.7182308266345256e-05,
+      "loss": 1.7178,
+      "step": 153
+    },
+    {
+      "epoch": 0.6695652173913044,
+      "grad_norm": 0.42069295048713684,
+      "learning_rate": 1.7182308266345256e-05,
+      "loss": 1.6735,
+      "step": 154
+    },
+    {
+      "epoch": 0.6739130434782609,
+      "grad_norm": 0.42069295048713684,
+      "learning_rate": 1.7182308266345256e-05,
+      "loss": 1.6999,
+      "step": 155
+    },
+    {
+      "epoch": 0.6782608695652174,
+      "grad_norm": 0.44798803329467773,
+      "learning_rate": 1.7305004690608827e-05,
+      "loss": 1.6549,
+      "step": 156
+    },
+    {
+      "epoch": 0.6826086956521739,
+      "grad_norm": 0.44798803329467773,
+      "learning_rate": 1.7305004690608827e-05,
+      "loss": 1.6681,
+      "step": 157
+    },
+    {
+      "epoch": 0.6869565217391305,
+      "grad_norm": 0.44798803329467773,
+      "learning_rate": 1.7305004690608827e-05,
+      "loss": 1.5953,
+      "step": 158
+    },
+    {
+      "epoch": 0.691304347826087,
+      "grad_norm": 0.44798803329467773,
+      "learning_rate": 1.7305004690608827e-05,
+      "loss": 1.6615,
+      "step": 159
+    },
+    {
+      "epoch": 0.6956521739130435,
+      "grad_norm": 0.4351099133491516,
+      "learning_rate": 1.7424594540116867e-05,
+      "loss": 1.6599,
+      "step": 160
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4351099133491516,
+      "learning_rate": 1.7424594540116867e-05,
+      "loss": 1.6897,
+      "step": 161
+    },
+    {
+      "epoch": 0.7043478260869566,
+      "grad_norm": 0.4351099133491516,
+      "learning_rate": 1.7424594540116867e-05,
+      "loss": 1.5914,
+      "step": 162
+    },
+    {
+      "epoch": 0.7086956521739131,
+      "grad_norm": 0.4351099133491516,
+      "learning_rate": 1.7424594540116867e-05,
+      "loss": 1.6879,
+      "step": 163
+    },
+    {
+      "epoch": 0.7130434782608696,
+      "grad_norm": 0.41112539172172546,
+      "learning_rate": 1.754123124995665e-05,
+      "loss": 1.6539,
+      "step": 164
+    },
+    {
+      "epoch": 0.717391304347826,
+      "grad_norm": 0.41112539172172546,
+      "learning_rate": 1.754123124995665e-05,
+      "loss": 1.6907,
+      "step": 165
+    },
+    {
+      "epoch": 0.7217391304347827,
+      "grad_norm": 0.41112539172172546,
+      "learning_rate": 1.754123124995665e-05,
+      "loss": 1.6745,
+      "step": 166
+    },
+    {
+      "epoch": 0.7260869565217392,
+      "grad_norm": 0.41112539172172546,
+      "learning_rate": 1.754123124995665e-05,
+      "loss": 1.6728,
+      "step": 167
+    },
+    {
+      "epoch": 0.7304347826086957,
+      "grad_norm": 0.4084008038043976,
+      "learning_rate": 1.765505716129306e-05,
+      "loss": 1.6311,
+      "step": 168
+    },
+    {
+      "epoch": 0.7347826086956522,
+      "grad_norm": 0.4084008038043976,
+      "learning_rate": 1.765505716129306e-05,
+      "loss": 1.6012,
+      "step": 169
+    },
+    {
+      "epoch": 0.7391304347826086,
+      "grad_norm": 0.4084008038043976,
+      "learning_rate": 1.765505716129306e-05,
+      "loss": 1.7379,
+      "step": 170
+    },
+    {
+      "epoch": 0.7434782608695653,
+      "grad_norm": 0.4084008038043976,
+      "learning_rate": 1.765505716129306e-05,
+      "loss": 1.6965,
+      "step": 171
+    },
+    {
+      "epoch": 0.7478260869565218,
+      "grad_norm": 0.3991777002811432,
+      "learning_rate": 1.7766204565755586e-05,
+      "loss": 1.6599,
+      "step": 172
+    },
+    {
+      "epoch": 0.7521739130434782,
+      "grad_norm": 0.3991777002811432,
+      "learning_rate": 1.7766204565755586e-05,
+      "loss": 1.6769,
+      "step": 173
+    },
+    {
+      "epoch": 0.7565217391304347,
+      "grad_norm": 0.3991777002811432,
+      "learning_rate": 1.7766204565755586e-05,
+      "loss": 1.6354,
+      "step": 174
+    },
+    {
+      "epoch": 0.7608695652173914,
+      "grad_norm": 0.3991777002811432,
+      "learning_rate": 1.7766204565755586e-05,
+      "loss": 1.6373,
+      "step": 175
+    },
+    {
+      "epoch": 0.7652173913043478,
+      "grad_norm": 0.42373913526535034,
+      "learning_rate": 1.7874796629747068e-05,
+      "loss": 1.6901,
+      "step": 176
+    },
+    {
+      "epoch": 0.7695652173913043,
+      "grad_norm": 0.42373913526535034,
+      "learning_rate": 1.7874796629747068e-05,
+      "loss": 1.6162,
+      "step": 177
+    },
+    {
+      "epoch": 0.7739130434782608,
+      "grad_norm": 0.42373913526535034,
+      "learning_rate": 1.7874796629747068e-05,
+      "loss": 1.6333,
+      "step": 178
+    },
+    {
+      "epoch": 0.7782608695652173,
+      "grad_norm": 0.42373913526535034,
+      "learning_rate": 1.7874796629747068e-05,
+      "loss": 1.7013,
+      "step": 179
+    },
+    {
+      "epoch": 0.782608695652174,
+      "grad_norm": 0.3955826461315155,
+      "learning_rate": 1.7980948214869728e-05,
+      "loss": 1.6865,
+      "step": 180
+    },
+    {
+      "epoch": 0.7869565217391304,
+      "grad_norm": 0.3955826461315155,
+      "learning_rate": 1.7980948214869728e-05,
+      "loss": 1.6776,
+      "step": 181
+    },
+    {
+      "epoch": 0.7913043478260869,
+      "grad_norm": 0.3955826461315155,
+      "learning_rate": 1.7980948214869728e-05,
+      "loss": 1.6903,
+      "step": 182
+    },
+    {
+      "epoch": 0.7956521739130434,
+      "grad_norm": 0.3955826461315155,
+      "learning_rate": 1.7980948214869728e-05,
+      "loss": 1.6803,
+      "step": 183
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4143446087837219,
+      "learning_rate": 1.8084766608171415e-05,
+      "loss": 1.6619,
+      "step": 184
+    },
+    {
+      "epoch": 0.8043478260869565,
+      "grad_norm": 0.4143446087837219,
+      "learning_rate": 1.8084766608171415e-05,
+      "loss": 1.6797,
+      "step": 185
+    },
+    {
+      "epoch": 0.808695652173913,
+      "grad_norm": 0.4143446087837219,
+      "learning_rate": 1.8084766608171415e-05,
+      "loss": 1.6324,
+      "step": 186
+    },
+    {
+      "epoch": 0.8130434782608695,
+      "grad_norm": 0.4143446087837219,
+      "learning_rate": 1.8084766608171415e-05,
+      "loss": 1.6539,
+      "step": 187
+    },
+    {
+      "epoch": 0.8173913043478261,
+      "grad_norm": 0.41547316312789917,
+      "learning_rate": 1.8186352173851508e-05,
+      "loss": 1.7622,
+      "step": 188
+    },
+    {
+      "epoch": 0.8217391304347826,
+      "grad_norm": 0.41547316312789917,
+      "learning_rate": 1.8186352173851508e-05,
+      "loss": 1.6468,
+      "step": 189
+    },
+    {
+      "epoch": 0.8260869565217391,
+      "grad_norm": 0.41547316312789917,
+      "learning_rate": 1.8186352173851508e-05,
+      "loss": 1.6644,
+      "step": 190
+    },
+    {
+      "epoch": 0.8304347826086956,
+      "grad_norm": 0.41547316312789917,
+      "learning_rate": 1.8186352173851508e-05,
+      "loss": 1.6822,
+      "step": 191
+    },
+    {
+      "epoch": 0.8347826086956521,
+      "grad_norm": 0.42321857810020447,
+      "learning_rate": 1.828579893635015e-05,
+      "loss": 1.6321,
+      "step": 192
+    },
+    {
+      "epoch": 0.8391304347826087,
+      "grad_norm": 0.42321857810020447,
+      "learning_rate": 1.828579893635015e-05,
+      "loss": 1.6843,
+      "step": 193
+    },
+    {
+      "epoch": 0.8434782608695652,
+      "grad_norm": 0.42321857810020447,
+      "learning_rate": 1.828579893635015e-05,
+      "loss": 1.63,
+      "step": 194
+    },
+    {
+      "epoch": 0.8478260869565217,
+      "grad_norm": 0.42321857810020447,
+      "learning_rate": 1.828579893635015e-05,
+      "loss": 1.6028,
+      "step": 195
+    },
+    {
+      "epoch": 0.8521739130434782,
+      "grad_norm": 0.41478654742240906,
+      "learning_rate": 1.8383195103311694e-05,
+      "loss": 1.7511,
+      "step": 196
+    },
+    {
+      "epoch": 0.8565217391304348,
+      "grad_norm": 0.41478654742240906,
+      "learning_rate": 1.8383195103311694e-05,
+      "loss": 1.644,
+      "step": 197
+    },
+    {
+      "epoch": 0.8608695652173913,
+      "grad_norm": 0.41478654742240906,
+      "learning_rate": 1.8383195103311694e-05,
+      "loss": 1.6338,
+      "step": 198
+    },
+    {
+      "epoch": 0.8652173913043478,
+      "grad_norm": 0.41478654742240906,
+      "learning_rate": 1.8383195103311694e-05,
+      "loss": 1.6912,
+      "step": 199
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 0.41812795400619507,
+      "learning_rate": 1.8478623535712173e-05,
+      "loss": 1.6322,
+      "step": 200
+    },
+    {
+      "epoch": 0.8739130434782608,
+      "grad_norm": 0.41812795400619507,
+      "learning_rate": 1.8478623535712173e-05,
+      "loss": 1.7339,
+      "step": 201
+    },
+    {
+      "epoch": 0.8782608695652174,
+      "grad_norm": 0.41812795400619507,
+      "learning_rate": 1.8478623535712173e-05,
+      "loss": 1.617,
+      "step": 202
+    },
+    {
+      "epoch": 0.8826086956521739,
+      "grad_norm": 0.41812795400619507,
+      "learning_rate": 1.8478623535712173e-05,
+      "loss": 1.6637,
+      "step": 203
+    },
+    {
+      "epoch": 0.8869565217391304,
+      "grad_norm": 0.3970027565956116,
+      "learning_rate": 1.8572162171429905e-05,
+      "loss": 1.6093,
+      "step": 204
+    },
+    {
+      "epoch": 0.8913043478260869,
+      "grad_norm": 0.3970027565956116,
+      "learning_rate": 1.8572162171429905e-05,
+      "loss": 1.6211,
+      "step": 205
+    },
+    {
+      "epoch": 0.8956521739130435,
+      "grad_norm": 0.3970027565956116,
+      "learning_rate": 1.8572162171429905e-05,
+      "loss": 1.7275,
+      "step": 206
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3970027565956116,
+      "learning_rate": 1.8572162171429905e-05,
+      "loss": 1.6095,
+      "step": 207
+    },
+    {
+      "epoch": 0.9043478260869565,
+      "grad_norm": 0.4158819913864136,
+      "learning_rate": 1.866388440768456e-05,
+      "loss": 1.6403,
+      "step": 208
+    },
+    {
+      "epoch": 0.908695652173913,
+      "grad_norm": 0.4158819913864136,
+      "learning_rate": 1.866388440768456e-05,
+      "loss": 1.5544,
+      "step": 209
+    },
+    {
+      "epoch": 0.9130434782608695,
+      "grad_norm": 0.4158819913864136,
+      "learning_rate": 1.866388440768456e-05,
+      "loss": 1.7182,
+      "step": 210
+    },
+    {
+      "epoch": 0.9173913043478261,
+      "grad_norm": 0.4158819913864136,
+      "learning_rate": 1.866388440768456e-05,
+      "loss": 1.6324,
+      "step": 211
+    },
+    {
+      "epoch": 0.9217391304347826,
+      "grad_norm": 0.41026684641838074,
+      "learning_rate": 1.875385944704652e-05,
+      "loss": 1.6029,
+      "step": 212
+    },
+    {
+      "epoch": 0.9260869565217391,
+      "grad_norm": 0.41026684641838074,
+      "learning_rate": 1.875385944704652e-05,
+      "loss": 1.6479,
+      "step": 213
+    },
+    {
+      "epoch": 0.9304347826086956,
+      "grad_norm": 0.41026684641838074,
+      "learning_rate": 1.875385944704652e-05,
+      "loss": 1.5916,
+      "step": 214
+    },
+    {
+      "epoch": 0.9347826086956522,
+      "grad_norm": 0.41026684641838074,
+      "learning_rate": 1.875385944704652e-05,
+      "loss": 1.6131,
+      "step": 215
+    },
+    {
+      "epoch": 0.9391304347826087,
+      "grad_norm": 0.3943828046321869,
+      "learning_rate": 1.8842152611103012e-05,
+      "loss": 1.653,
+      "step": 216
+    },
+    {
+      "epoch": 0.9434782608695652,
+      "grad_norm": 0.3943828046321869,
+      "learning_rate": 1.8842152611103012e-05,
+      "loss": 1.6321,
+      "step": 217
+    },
+    {
+      "epoch": 0.9478260869565217,
+      "grad_norm": 0.3943828046321869,
+      "learning_rate": 1.8842152611103012e-05,
+      "loss": 1.65,
+      "step": 218
+    },
+    {
+      "epoch": 0.9521739130434783,
+      "grad_norm": 0.3943828046321869,
+      "learning_rate": 1.8842152611103012e-05,
+      "loss": 1.7162,
+      "step": 219
+    },
+    {
+      "epoch": 0.9565217391304348,
+      "grad_norm": 0.4221249520778656,
+      "learning_rate": 1.8928825625342374e-05,
+      "loss": 1.6494,
+      "step": 220
+    },
+    {
+      "epoch": 0.9608695652173913,
+      "grad_norm": 0.4221249520778656,
+      "learning_rate": 1.8928825625342374e-05,
+      "loss": 1.6674,
+      "step": 221
+    },
+    {
+      "epoch": 0.9652173913043478,
+      "grad_norm": 0.4221249520778656,
+      "learning_rate": 1.8928825625342374e-05,
+      "loss": 1.6672,
+      "step": 222
+    },
+    {
+      "epoch": 0.9695652173913043,
+      "grad_norm": 0.4221249520778656,
+      "learning_rate": 1.8928825625342374e-05,
+      "loss": 1.5908,
+      "step": 223
+    },
+    {
+      "epoch": 0.9739130434782609,
+      "grad_norm": 0.43016910552978516,
+      "learning_rate": 1.901393687836879e-05,
+      "loss": 1.6331,
+      "step": 224
+    },
+    {
+      "epoch": 0.9782608695652174,
+      "grad_norm": 0.43016910552978516,
+      "learning_rate": 1.901393687836879e-05,
+      "loss": 1.6466,
+      "step": 225
+    },
+    {
+      "epoch": 0.9826086956521739,
+      "grad_norm": 0.43016910552978516,
+      "learning_rate": 1.901393687836879e-05,
+      "loss": 1.5822,
+      "step": 226
+    },
+    {
+      "epoch": 0.9869565217391304,
+      "grad_norm": 0.43016910552978516,
+      "learning_rate": 1.901393687836879e-05,
+      "loss": 1.6973,
+      "step": 227
+    },
+    {
+      "epoch": 0.991304347826087,
+      "grad_norm": 0.4303857982158661,
+      "learning_rate": 1.9097541658173843e-05,
+      "loss": 1.6486,
+      "step": 228
+    },
+    {
+      "epoch": 0.9956521739130435,
+      "grad_norm": 0.4303857982158661,
+      "learning_rate": 1.9097541658173843e-05,
+      "loss": 1.6024,
+      "step": 229
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4303857982158661,
+      "learning_rate": 1.9097541658173843e-05,
+      "loss": 1.7058,
+      "step": 230
+    },
+    {
+      "epoch": 1.0043478260869565,
+      "grad_norm": 0.4303857982158661,
+      "learning_rate": 1.9097541658173843e-05,
+      "loss": 1.5432,
+      "step": 231
+    },
+    {
+      "epoch": 1.008695652173913,
+      "grad_norm": 0.422577828168869,
+      "learning_rate": 1.9179692367859514e-05,
+      "loss": 1.5811,
+      "step": 232
+    },
+    {
+      "epoch": 1.0130434782608695,
+      "grad_norm": 0.422577828168869,
+      "learning_rate": 1.9179692367859514e-05,
+      "loss": 1.5742,
+      "step": 233
+    },
+    {
+      "epoch": 1.017391304347826,
+      "grad_norm": 0.422577828168869,
+      "learning_rate": 1.9179692367859514e-05,
+      "loss": 1.5679,
+      "step": 234
+    },
+    {
+      "epoch": 1.0217391304347827,
+      "grad_norm": 0.422577828168869,
+      "learning_rate": 1.9179692367859514e-05,
+      "loss": 1.5029,
+      "step": 235
+    },
+    {
+      "epoch": 1.0260869565217392,
+      "grad_norm": 0.4109610915184021,
+      "learning_rate": 1.926043872292045e-05,
+      "loss": 1.6383,
+      "step": 236
+    },
+    {
+      "epoch": 1.0304347826086957,
+      "grad_norm": 0.4109610915184021,
+      "learning_rate": 1.926043872292045e-05,
+      "loss": 1.6379,
+      "step": 237
+    },
+    {
+      "epoch": 1.0347826086956522,
+      "grad_norm": 0.4109610915184021,
+      "learning_rate": 1.926043872292045e-05,
+      "loss": 1.5574,
+      "step": 238
+    },
+    {
+      "epoch": 1.0391304347826087,
+      "grad_norm": 0.4109610915184021,
+      "learning_rate": 1.926043872292045e-05,
+      "loss": 1.5899,
+      "step": 239
+    },
+    {
+      "epoch": 1.0434782608695652,
+      "grad_norm": 0.4028691351413727,
+      "learning_rate": 1.9339827931945454e-05,
+      "loss": 1.6503,
+      "step": 240
+    },
+    {
+      "epoch": 1.0478260869565217,
+      "grad_norm": 0.4028691351413727,
+      "learning_rate": 1.9339827931945454e-05,
+      "loss": 1.5943,
+      "step": 241
+    },
+    {
+      "epoch": 1.0521739130434782,
+      "grad_norm": 0.4028691351413727,
+      "learning_rate": 1.9339827931945454e-05,
+      "loss": 1.5263,
+      "step": 242
+    },
+    {
+      "epoch": 1.0565217391304347,
+      "grad_norm": 0.4028691351413727,
+      "learning_rate": 1.9339827931945454e-05,
+      "loss": 1.6053,
+      "step": 243
+    },
+    {
+      "epoch": 1.0608695652173914,
+      "grad_norm": 0.415353387594223,
+      "learning_rate": 1.941790486238291e-05,
+      "loss": 1.5856,
+      "step": 244
+    },
+    {
+      "epoch": 1.065217391304348,
+      "grad_norm": 0.415353387594223,
+      "learning_rate": 1.941790486238291e-05,
+      "loss": 1.6749,
+      "step": 245
+    },
+    {
+      "epoch": 1.0695652173913044,
+      "grad_norm": 0.415353387594223,
+      "learning_rate": 1.941790486238291e-05,
+      "loss": 1.5961,
+      "step": 246
+    },
+    {
+      "epoch": 1.0739130434782609,
+      "grad_norm": 0.415353387594223,
+      "learning_rate": 1.941790486238291e-05,
+      "loss": 1.5564,
+      "step": 247
+    },
+    {
+      "epoch": 1.0782608695652174,
+      "grad_norm": 0.43506088852882385,
+      "learning_rate": 1.949471219282736e-05,
+      "loss": 1.5932,
+      "step": 248
+    },
+    {
+      "epoch": 1.0826086956521739,
+      "grad_norm": 0.43506088852882385,
+      "learning_rate": 1.949471219282736e-05,
+      "loss": 1.5567,
+      "step": 249
+    },
+    {
+      "epoch": 1.0869565217391304,
+      "grad_norm": 0.43506088852882385,
+      "learning_rate": 1.949471219282736e-05,
+      "loss": 1.5601,
+      "step": 250
+    },
+    {
+      "epoch": 1.0913043478260869,
+      "grad_norm": 0.43506088852882385,
+      "learning_rate": 1.949471219282736e-05,
+      "loss": 1.6339,
+      "step": 251
+    },
+    {
+      "epoch": 1.0956521739130434,
+      "grad_norm": 0.4195059835910797,
+      "learning_rate": 1.9570290553121646e-05,
+      "loss": 1.5493,
+      "step": 252
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4195059835910797,
+      "learning_rate": 1.9570290553121646e-05,
+      "loss": 1.5483,
+      "step": 253
+    },
+    {
+      "epoch": 1.1043478260869566,
+      "grad_norm": 0.4195059835910797,
+      "learning_rate": 1.9570290553121646e-05,
+      "loss": 1.5828,
+      "step": 254
+    },
+    {
+      "epoch": 1.108695652173913,
+      "grad_norm": 0.4195059835910797,
+      "learning_rate": 1.9570290553121646e-05,
+      "loss": 1.5189,
+      "step": 255
+    },
+    {
+      "epoch": 1.1130434782608696,
+      "grad_norm": 0.41635432839393616,
+      "learning_rate": 1.9644678653425876e-05,
+      "loss": 1.5438,
+      "step": 256
+    },
+    {
+      "epoch": 1.117391304347826,
+      "grad_norm": 0.41635432839393616,
+      "learning_rate": 1.9644678653425876e-05,
+      "loss": 1.6092,
+      "step": 257
+    },
+    {
+      "epoch": 1.1217391304347826,
+      "grad_norm": 0.41635432839393616,
+      "learning_rate": 1.9644678653425876e-05,
+      "loss": 1.5977,
+      "step": 258
+    },
+    {
+      "epoch": 1.126086956521739,
+      "grad_norm": 0.41635432839393616,
+      "learning_rate": 1.9644678653425876e-05,
+      "loss": 1.6079,
+      "step": 259
+    },
+    {
+      "epoch": 1.1304347826086956,
+      "grad_norm": 0.4125351905822754,
+      "learning_rate": 1.971791340327986e-05,
+      "loss": 1.553,
+      "step": 260
+    },
+    {
+      "epoch": 1.134782608695652,
+      "grad_norm": 0.4125351905822754,
+      "learning_rate": 1.971791340327986e-05,
+      "loss": 1.5419,
+      "step": 261
+    },
+    {
+      "epoch": 1.1391304347826088,
+      "grad_norm": 0.4125351905822754,
+      "learning_rate": 1.971791340327986e-05,
+      "loss": 1.5396,
+      "step": 262
+    },
+    {
+      "epoch": 1.1434782608695653,
+      "grad_norm": 0.4125351905822754,
+      "learning_rate": 1.971791340327986e-05,
+      "loss": 1.6408,
+      "step": 263
+    },
+    {
+      "epoch": 1.1478260869565218,
+      "grad_norm": 0.40901288390159607,
+      "learning_rate": 1.979003002157565e-05,
+      "loss": 1.5664,
+      "step": 264
+    },
+    {
+      "epoch": 1.1521739130434783,
+      "grad_norm": 0.40901288390159607,
+      "learning_rate": 1.979003002157565e-05,
+      "loss": 1.5708,
+      "step": 265
+    },
+    {
+      "epoch": 1.1565217391304348,
+      "grad_norm": 0.40901288390159607,
+      "learning_rate": 1.979003002157565e-05,
+      "loss": 1.6379,
+      "step": 266
+    },
+    {
+      "epoch": 1.1608695652173913,
+      "grad_norm": 0.40901288390159607,
+      "learning_rate": 1.979003002157565e-05,
+      "loss": 1.5962,
+      "step": 267
+    },
+    {
+      "epoch": 1.1652173913043478,
+      "grad_norm": 0.4133271872997284,
+      "learning_rate": 1.9861062138260542e-05,
+      "loss": 1.6233,
+      "step": 268
+    },
+    {
+      "epoch": 1.1695652173913043,
+      "grad_norm": 0.4133271872997284,
+      "learning_rate": 1.9861062138260542e-05,
+      "loss": 1.5597,
+      "step": 269
+    },
+    {
+      "epoch": 1.1739130434782608,
+      "grad_norm": 0.4133271872997284,
+      "learning_rate": 1.9861062138260542e-05,
+      "loss": 1.5396,
+      "step": 270
+    },
+    {
+      "epoch": 1.1782608695652175,
+      "grad_norm": 0.4133271872997284,
+      "learning_rate": 1.9861062138260542e-05,
+      "loss": 1.6504,
+      "step": 271
+    },
+    {
+      "epoch": 1.182608695652174,
+      "grad_norm": 0.4083419144153595,
+      "learning_rate": 1.993104188850563e-05,
+      "loss": 1.519,
+      "step": 272
+    },
+    {
+      "epoch": 1.1869565217391305,
+      "grad_norm": 0.4083419144153595,
+      "learning_rate": 1.993104188850563e-05,
+      "loss": 1.6034,
+      "step": 273
+    },
+    {
+      "epoch": 1.191304347826087,
+      "grad_norm": 0.4083419144153595,
+      "learning_rate": 1.993104188850563e-05,
+      "loss": 1.5509,
+      "step": 274
+    },
+    {
+      "epoch": 1.1956521739130435,
+      "grad_norm": 0.4083419144153595,
+      "learning_rate": 1.993104188850563e-05,
+      "loss": 1.4744,
+      "step": 275
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.40321245789527893,
+      "learning_rate": 2e-05,
+      "loss": 1.5611,
+      "step": 276
+    },
+    {
+      "epoch": 1.2043478260869565,
+      "grad_norm": 0.40321245789527893,
+      "learning_rate": 2e-05,
+      "loss": 1.5937,
+      "step": 277
+    },
+    {
+      "epoch": 1.208695652173913,
+      "grad_norm": 0.40321245789527893,
+      "learning_rate": 2e-05,
+      "loss": 1.5801,
+      "step": 278
+    },
+    {
+      "epoch": 1.2130434782608694,
+      "grad_norm": 0.40321245789527893,
+      "learning_rate": 2e-05,
+      "loss": 1.5999,
+      "step": 279
+    },
+    {
+      "epoch": 1.2173913043478262,
+      "grad_norm": 0.4187948703765869,
+      "learning_rate": 2e-05,
+      "loss": 1.5488,
+      "step": 280
+    },
+    {
+      "epoch": 1.2217391304347827,
+      "grad_norm": 0.4187948703765869,
+      "learning_rate": 2e-05,
+      "loss": 1.4954,
+      "step": 281
+    },
+    {
+      "epoch": 1.2260869565217392,
+      "grad_norm": 0.4187948703765869,
+      "learning_rate": 2e-05,
+      "loss": 1.4704,
+      "step": 282
+    },
+    {
+      "epoch": 1.2304347826086957,
+      "grad_norm": 0.4187948703765869,
+      "learning_rate": 2e-05,
+      "loss": 1.5532,
+      "step": 283
+    },
+    {
+      "epoch": 1.2347826086956522,
+      "grad_norm": 0.39265695214271545,
+      "learning_rate": 2e-05,
+      "loss": 1.6158,
+      "step": 284
+    },
+    {
+      "epoch": 1.2391304347826086,
+      "grad_norm": 0.39265695214271545,
+      "learning_rate": 2e-05,
+      "loss": 1.5282,
+      "step": 285
+    },
+    {
+      "epoch": 1.2434782608695651,
+      "grad_norm": 0.39265695214271545,
+      "learning_rate": 2e-05,
+      "loss": 1.6288,
+      "step": 286
+    },
+    {
+      "epoch": 1.2478260869565219,
+      "grad_norm": 0.39265695214271545,
+      "learning_rate": 2e-05,
+      "loss": 1.6232,
+      "step": 287
+    },
+    {
+      "epoch": 1.2521739130434781,
+      "grad_norm": 0.4242652654647827,
+      "learning_rate": 2e-05,
+      "loss": 1.5431,
+      "step": 288
+    },
+    {
+      "epoch": 1.2565217391304349,
+      "grad_norm": 0.4242652654647827,
+      "learning_rate": 2e-05,
+      "loss": 1.5737,
+      "step": 289
+    },
+    {
+      "epoch": 1.2608695652173914,
+      "grad_norm": 0.4242652654647827,
+      "learning_rate": 2e-05,
+      "loss": 1.5198,
+      "step": 290
+    },
+    {
+      "epoch": 1.2652173913043478,
+      "grad_norm": 0.4242652654647827,
+      "learning_rate": 2e-05,
+      "loss": 1.5479,
+      "step": 291
+    },
+    {
+      "epoch": 1.2695652173913043,
+      "grad_norm": 0.4110427796840668,
+      "learning_rate": 2e-05,
+      "loss": 1.6079,
+      "step": 292
+    },
+    {
+      "epoch": 1.2739130434782608,
+      "grad_norm": 0.4110427796840668,
+      "learning_rate": 2e-05,
+      "loss": 1.6189,
+      "step": 293
+    },
+    {
+      "epoch": 1.2782608695652173,
+      "grad_norm": 0.4110427796840668,
+      "learning_rate": 2e-05,
+      "loss": 1.5868,
+      "step": 294
+    },
+    {
+      "epoch": 1.2826086956521738,
+      "grad_norm": 0.4110427796840668,
+      "learning_rate": 2e-05,
+      "loss": 1.5536,
+      "step": 295
+    },
+    {
+      "epoch": 1.2869565217391306,
+      "grad_norm": 0.4077237546443939,
+      "learning_rate": 2e-05,
+      "loss": 1.6191,
+      "step": 296
+    },
+    {
+      "epoch": 1.2913043478260868,
+      "grad_norm": 0.4077237546443939,
+      "learning_rate": 2e-05,
+      "loss": 1.5256,
+      "step": 297
+    },
+    {
+      "epoch": 1.2956521739130435,
+      "grad_norm": 0.4077237546443939,
+      "learning_rate": 2e-05,
+      "loss": 1.4858,
+      "step": 298
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4077237546443939,
+      "learning_rate": 2e-05,
+      "loss": 1.6217,
+      "step": 299
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 0.4294397234916687,
+      "learning_rate": 2e-05,
+      "loss": 1.5448,
+      "step": 300
+    },
+    {
+      "epoch": 1.308695652173913,
+      "grad_norm": 0.4294397234916687,
+      "learning_rate": 2e-05,
+      "loss": 1.5482,
+      "step": 301
+    },
+    {
+      "epoch": 1.3130434782608695,
+      "grad_norm": 0.4294397234916687,
+      "learning_rate": 2e-05,
+      "loss": 1.5587,
+      "step": 302
+    },
+    {
+      "epoch": 1.317391304347826,
+      "grad_norm": 0.4294397234916687,
+      "learning_rate": 2e-05,
+      "loss": 1.5039,
+      "step": 303
+    },
+    {
+      "epoch": 1.3217391304347825,
+      "grad_norm": 0.42842897772789,
+      "learning_rate": 2e-05,
+      "loss": 1.6381,
+      "step": 304
+    },
+    {
+      "epoch": 1.3260869565217392,
+      "grad_norm": 0.42842897772789,
+      "learning_rate": 2e-05,
+      "loss": 1.6137,
+      "step": 305
+    },
+    {
+      "epoch": 1.3304347826086955,
+      "grad_norm": 0.42842897772789,
+      "learning_rate": 2e-05,
+      "loss": 1.5976,
+      "step": 306
+    },
+    {
+      "epoch": 1.3347826086956522,
+      "grad_norm": 0.42842897772789,
+      "learning_rate": 2e-05,
+      "loss": 1.6185,
+      "step": 307
+    },
+    {
+      "epoch": 1.3391304347826087,
+      "grad_norm": 0.41912516951560974,
+      "learning_rate": 2e-05,
+      "loss": 1.5669,
+      "step": 308
+    },
+    {
+      "epoch": 1.3434782608695652,
+      "grad_norm": 0.41912516951560974,
+      "learning_rate": 2e-05,
+      "loss": 1.5439,
+      "step": 309
+    },
+    {
+      "epoch": 1.3478260869565217,
+      "grad_norm": 0.41912516951560974,
+      "learning_rate": 2e-05,
+      "loss": 1.5554,
+      "step": 310
+    },
+    {
+      "epoch": 1.3521739130434782,
+      "grad_norm": 0.41912516951560974,
+      "learning_rate": 2e-05,
+      "loss": 1.5461,
+      "step": 311
+    },
+    {
+      "epoch": 1.3565217391304347,
+      "grad_norm": 0.4236287772655487,
+      "learning_rate": 2e-05,
+      "loss": 1.5871,
+      "step": 312
+    },
+    {
+      "epoch": 1.3608695652173912,
+      "grad_norm": 0.4236287772655487,
+      "learning_rate": 2e-05,
+      "loss": 1.4733,
+      "step": 313
+    },
+    {
+      "epoch": 1.365217391304348,
+      "grad_norm": 0.4236287772655487,
+      "learning_rate": 2e-05,
+      "loss": 1.5409,
+      "step": 314
+    },
+    {
+      "epoch": 1.3695652173913042,
+      "grad_norm": 0.4236287772655487,
+      "learning_rate": 2e-05,
+      "loss": 1.5149,
+      "step": 315
+    },
+    {
+      "epoch": 1.373913043478261,
+      "grad_norm": 0.38583850860595703,
+      "learning_rate": 2e-05,
+      "loss": 1.4956,
+      "step": 316
+    },
+    {
+      "epoch": 1.3782608695652174,
+      "grad_norm": 0.38583850860595703,
+      "learning_rate": 2e-05,
+      "loss": 1.5329,
+      "step": 317
+    },
+    {
+      "epoch": 1.382608695652174,
+      "grad_norm": 0.38583850860595703,
+      "learning_rate": 2e-05,
+      "loss": 1.5038,
+      "step": 318
+    },
+    {
+      "epoch": 1.3869565217391304,
+      "grad_norm": 0.38583850860595703,
+      "learning_rate": 2e-05,
+      "loss": 1.5759,
+      "step": 319
+    },
+    {
+      "epoch": 1.391304347826087,
+      "grad_norm": 0.4223187267780304,
+      "learning_rate": 2e-05,
+      "loss": 1.5672,
+      "step": 320
+    },
+    {
+      "epoch": 1.3956521739130434,
+      "grad_norm": 0.4223187267780304,
+      "learning_rate": 2e-05,
+      "loss": 1.5082,
+      "step": 321
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.4223187267780304,
+      "learning_rate": 2e-05,
+      "loss": 1.5306,
+      "step": 322
+    },
+    {
+      "epoch": 1.4043478260869566,
+      "grad_norm": 0.4223187267780304,
+      "learning_rate": 2e-05,
+      "loss": 1.5861,
+      "step": 323
+    },
+    {
+      "epoch": 1.4086956521739131,
+      "grad_norm": 0.42702987790107727,
+      "learning_rate": 2e-05,
+      "loss": 1.6032,
+      "step": 324
+    },
+    {
+      "epoch": 1.4130434782608696,
+      "grad_norm": 0.42702987790107727,
+      "learning_rate": 2e-05,
+      "loss": 1.5757,
+      "step": 325
+    },
+    {
+      "epoch": 1.4173913043478261,
+      "grad_norm": 0.42702987790107727,
+      "learning_rate": 2e-05,
+      "loss": 1.5847,
+      "step": 326
+    },
+    {
+      "epoch": 1.4217391304347826,
+      "grad_norm": 0.42702987790107727,
+      "learning_rate": 2e-05,
+      "loss": 1.5571,
+      "step": 327
+    },
+    {
+      "epoch": 1.4260869565217391,
+      "grad_norm": 0.4275616705417633,
+      "learning_rate": 2e-05,
+      "loss": 1.4902,
+      "step": 328
+    },
+    {
+      "epoch": 1.4304347826086956,
+      "grad_norm": 0.4275616705417633,
+      "learning_rate": 2e-05,
+      "loss": 1.5888,
+      "step": 329
+    },
+    {
+      "epoch": 1.434782608695652,
+      "grad_norm": 0.4275616705417633,
+      "learning_rate": 2e-05,
+      "loss": 1.5521,
+      "step": 330
+    },
+    {
+      "epoch": 1.4391304347826086,
+      "grad_norm": 0.4275616705417633,
+      "learning_rate": 2e-05,
+      "loss": 1.5895,
+      "step": 331
+    },
+    {
+      "epoch": 1.4434782608695653,
+      "grad_norm": 0.4133172333240509,
+      "learning_rate": 2e-05,
+      "loss": 1.5116,
+      "step": 332
+    },
+    {
+      "epoch": 1.4478260869565218,
+      "grad_norm": 0.4133172333240509,
+      "learning_rate": 2e-05,
+      "loss": 1.5397,
+      "step": 333
+    },
+    {
+      "epoch": 1.4521739130434783,
+      "grad_norm": 0.4133172333240509,
+      "learning_rate": 2e-05,
+      "loss": 1.5809,
+      "step": 334
+    },
+    {
+      "epoch": 1.4565217391304348,
+      "grad_norm": 0.4133172333240509,
+      "learning_rate": 2e-05,
+      "loss": 1.4922,
+      "step": 335
+    },
+    {
+      "epoch": 1.4608695652173913,
+      "grad_norm": 0.4177166819572449,
+      "learning_rate": 2e-05,
+      "loss": 1.5604,
+      "step": 336
+    },
+    {
+      "epoch": 1.4652173913043478,
+      "grad_norm": 0.4177166819572449,
+      "learning_rate": 2e-05,
+      "loss": 1.5536,
+      "step": 337
+    },
+    {
+      "epoch": 1.4695652173913043,
+      "grad_norm": 0.4177166819572449,
+      "learning_rate": 2e-05,
+      "loss": 1.5712,
+      "step": 338
+    },
+    {
+      "epoch": 1.4739130434782608,
+      "grad_norm": 0.4177166819572449,
+      "learning_rate": 2e-05,
+      "loss": 1.4619,
+      "step": 339
+    },
+    {
+      "epoch": 1.4782608695652173,
+      "grad_norm": 0.4101729393005371,
+      "learning_rate": 2e-05,
+      "loss": 1.5609,
+      "step": 340
+    },
+    {
+      "epoch": 1.482608695652174,
+      "grad_norm": 0.4101729393005371,
+      "learning_rate": 2e-05,
+      "loss": 1.5474,
+      "step": 341
+    },
+    {
+      "epoch": 1.4869565217391305,
+      "grad_norm": 0.4101729393005371,
+      "learning_rate": 2e-05,
+      "loss": 1.5178,
+      "step": 342
+    },
+    {
+      "epoch": 1.491304347826087,
+      "grad_norm": 0.4101729393005371,
+      "learning_rate": 2e-05,
+      "loss": 1.4915,
+      "step": 343
+    },
+    {
+      "epoch": 1.4956521739130435,
+      "grad_norm": 0.41781285405158997,
+      "learning_rate": 2e-05,
+      "loss": 1.5977,
+      "step": 344
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.41781285405158997,
+      "learning_rate": 2e-05,
+      "loss": 1.5584,
+      "step": 345
+    },
+    {
+      "epoch": 1.5043478260869565,
+      "grad_norm": 0.41781285405158997,
+      "learning_rate": 2e-05,
+      "loss": 1.5904,
+      "step": 346
+    },
+    {
+      "epoch": 1.508695652173913,
+      "grad_norm": 0.41781285405158997,
+      "learning_rate": 2e-05,
+      "loss": 1.6015,
+      "step": 347
+    },
+    {
+      "epoch": 1.5130434782608697,
+      "grad_norm": 0.4429299235343933,
+      "learning_rate": 2e-05,
+      "loss": 1.5282,
+      "step": 348
+    },
+    {
+      "epoch": 1.517391304347826,
+      "grad_norm": 0.4429299235343933,
+      "learning_rate": 2e-05,
+      "loss": 1.5595,
+      "step": 349
+    },
+    {
+      "epoch": 1.5217391304347827,
+      "grad_norm": 0.4429299235343933,
+      "learning_rate": 2e-05,
+      "loss": 1.5522,
+      "step": 350
+    },
+    {
+      "epoch": 1.526086956521739,
+      "grad_norm": 0.4429299235343933,
+      "learning_rate": 2e-05,
+      "loss": 1.5934,
+      "step": 351
+    },
+    {
+      "epoch": 1.5304347826086957,
+      "grad_norm": 0.4120587706565857,
+      "learning_rate": 2e-05,
+      "loss": 1.4523,
+      "step": 352
+    },
+    {
+      "epoch": 1.5347826086956522,
+      "grad_norm": 0.4120587706565857,
+      "learning_rate": 2e-05,
+      "loss": 1.5408,
+      "step": 353
+    },
+    {
+      "epoch": 1.5391304347826087,
+      "grad_norm": 0.4120587706565857,
+      "learning_rate": 2e-05,
+      "loss": 1.6324,
+      "step": 354
+    },
+    {
+      "epoch": 1.5434782608695652,
+      "grad_norm": 0.4120587706565857,
+      "learning_rate": 2e-05,
+      "loss": 1.4724,
+      "step": 355
+    },
+    {
+      "epoch": 1.5478260869565217,
+      "grad_norm": 0.4269024431705475,
+      "learning_rate": 2e-05,
+      "loss": 1.5582,
+      "step": 356
+    },
+    {
+      "epoch": 1.5521739130434784,
+      "grad_norm": 0.4269024431705475,
+      "learning_rate": 2e-05,
+      "loss": 1.4954,
+      "step": 357
+    },
+    {
+      "epoch": 1.5565217391304347,
+      "grad_norm": 0.4269024431705475,
+      "learning_rate": 2e-05,
+      "loss": 1.5452,
+      "step": 358
+    },
+    {
+      "epoch": 1.5608695652173914,
+      "grad_norm": 0.4269024431705475,
+      "learning_rate": 2e-05,
+      "loss": 1.4923,
+      "step": 359
+    },
+    {
+      "epoch": 1.5652173913043477,
+      "grad_norm": 0.4282481074333191,
+      "learning_rate": 2e-05,
+      "loss": 1.5833,
+      "step": 360
+    },
+    {
+      "epoch": 1.5695652173913044,
+      "grad_norm": 0.4282481074333191,
+      "learning_rate": 2e-05,
+      "loss": 1.589,
+      "step": 361
+    },
+    {
+      "epoch": 1.5739130434782609,
+      "grad_norm": 0.4282481074333191,
+      "learning_rate": 2e-05,
+      "loss": 1.5912,
+      "step": 362
+    },
+    {
+      "epoch": 1.5782608695652174,
+      "grad_norm": 0.4282481074333191,
+      "learning_rate": 2e-05,
+      "loss": 1.5138,
+      "step": 363
+    },
+    {
+      "epoch": 1.5826086956521739,
+      "grad_norm": 0.40982702374458313,
+      "learning_rate": 2e-05,
+      "loss": 1.5423,
+      "step": 364
+    },
+    {
+      "epoch": 1.5869565217391304,
+      "grad_norm": 0.40982702374458313,
+      "learning_rate": 2e-05,
+      "loss": 1.5006,
+      "step": 365
+    },
+    {
+      "epoch": 1.591304347826087,
+      "grad_norm": 0.40982702374458313,
+      "learning_rate": 2e-05,
+      "loss": 1.5404,
+      "step": 366
+    },
+    {
+      "epoch": 1.5956521739130434,
+      "grad_norm": 0.40982702374458313,
+      "learning_rate": 2e-05,
+      "loss": 1.5456,
+      "step": 367
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.46088093519210815,
+      "learning_rate": 2e-05,
+      "loss": 1.5764,
+      "step": 368
+    },
+    {
+      "epoch": 1.6043478260869564,
+      "grad_norm": 0.46088093519210815,
+      "learning_rate": 2e-05,
+      "loss": 1.5396,
+      "step": 369
+    },
+    {
+      "epoch": 1.608695652173913,
+      "grad_norm": 0.46088093519210815,
+      "learning_rate": 2e-05,
+      "loss": 1.5228,
+      "step": 370
+    },
+    {
+      "epoch": 1.6130434782608696,
+      "grad_norm": 0.46088093519210815,
+      "learning_rate": 2e-05,
+      "loss": 1.5328,
+      "step": 371
+    },
+    {
+      "epoch": 1.617391304347826,
+      "grad_norm": 0.43352362513542175,
+      "learning_rate": 2e-05,
+      "loss": 1.5284,
+      "step": 372
+    },
+    {
+      "epoch": 1.6217391304347826,
+      "grad_norm": 0.43352362513542175,
+      "learning_rate": 2e-05,
+      "loss": 1.5102,
+      "step": 373
+    },
+    {
+      "epoch": 1.626086956521739,
+      "grad_norm": 0.43352362513542175,
+      "learning_rate": 2e-05,
+      "loss": 1.5272,
+      "step": 374
+    },
+    {
+      "epoch": 1.6304347826086958,
+      "grad_norm": 0.43352362513542175,
+      "learning_rate": 2e-05,
+      "loss": 1.5409,
+      "step": 375
+    },
+    {
+      "epoch": 1.634782608695652,
+      "grad_norm": 0.436289519071579,
+      "learning_rate": 2e-05,
+      "loss": 1.495,
+      "step": 376
+    },
+    {
+      "epoch": 1.6391304347826088,
+      "grad_norm": 0.436289519071579,
+      "learning_rate": 2e-05,
+      "loss": 1.5289,
+      "step": 377
+    },
+    {
+      "epoch": 1.643478260869565,
+      "grad_norm": 0.436289519071579,
+      "learning_rate": 2e-05,
+      "loss": 1.5114,
+      "step": 378
+    },
+    {
+      "epoch": 1.6478260869565218,
+      "grad_norm": 0.436289519071579,
+      "learning_rate": 2e-05,
+      "loss": 1.5157,
+      "step": 379
+    },
+    {
+      "epoch": 1.6521739130434783,
+      "grad_norm": 0.41269221901893616,
+      "learning_rate": 2e-05,
+      "loss": 1.5147,
+      "step": 380
+    },
+    {
+      "epoch": 1.6565217391304348,
+      "grad_norm": 0.41269221901893616,
+      "learning_rate": 2e-05,
+      "loss": 1.4763,
+      "step": 381
+    },
+    {
+      "epoch": 1.6608695652173913,
+      "grad_norm": 0.41269221901893616,
+      "learning_rate": 2e-05,
+      "loss": 1.4898,
+      "step": 382
+    },
+    {
+      "epoch": 1.6652173913043478,
+      "grad_norm": 0.41269221901893616,
+      "learning_rate": 2e-05,
+      "loss": 1.5237,
+      "step": 383
+    },
+    {
+      "epoch": 1.6695652173913045,
+      "grad_norm": 0.42367058992385864,
+      "learning_rate": 2e-05,
+      "loss": 1.5789,
+      "step": 384
+    },
+    {
+      "epoch": 1.6739130434782608,
+      "grad_norm": 0.42367058992385864,
+      "learning_rate": 2e-05,
+      "loss": 1.5807,
+      "step": 385
+    },
+    {
+      "epoch": 1.6782608695652175,
+      "grad_norm": 0.42367058992385864,
+      "learning_rate": 2e-05,
+      "loss": 1.5238,
+      "step": 386
+    },
+    {
+      "epoch": 1.6826086956521737,
+      "grad_norm": 0.42367058992385864,
+      "learning_rate": 2e-05,
+      "loss": 1.5146,
+      "step": 387
+    },
+    {
+      "epoch": 1.6869565217391305,
+      "grad_norm": 0.4358891546726227,
+      "learning_rate": 2e-05,
+      "loss": 1.5674,
+      "step": 388
+    },
+    {
+      "epoch": 1.691304347826087,
+      "grad_norm": 0.4358891546726227,
+      "learning_rate": 2e-05,
+      "loss": 1.4984,
+      "step": 389
+    },
+    {
+      "epoch": 1.6956521739130435,
+      "grad_norm": 0.4358891546726227,
+      "learning_rate": 2e-05,
+      "loss": 1.5361,
+      "step": 390
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.4358891546726227,
+      "learning_rate": 2e-05,
+      "loss": 1.4934,
+      "step": 391
+    },
+    {
+      "epoch": 1.7043478260869565,
+      "grad_norm": 0.42739376425743103,
+      "learning_rate": 2e-05,
+      "loss": 1.5489,
+      "step": 392
+    },
+    {
+      "epoch": 1.7086956521739132,
+      "grad_norm": 0.42739376425743103,
+      "learning_rate": 2e-05,
+      "loss": 1.5327,
+      "step": 393
+    },
+    {
+      "epoch": 1.7130434782608694,
+      "grad_norm": 0.42739376425743103,
+      "learning_rate": 2e-05,
+      "loss": 1.5521,
+      "step": 394
+    },
+    {
+      "epoch": 1.7173913043478262,
+      "grad_norm": 0.42739376425743103,
+      "learning_rate": 2e-05,
+      "loss": 1.586,
+      "step": 395
+    },
+    {
+      "epoch": 1.7217391304347827,
+      "grad_norm": 0.43897366523742676,
+      "learning_rate": 2e-05,
+      "loss": 1.5053,
+      "step": 396
+    },
+    {
+      "epoch": 1.7260869565217392,
+      "grad_norm": 0.43897366523742676,
+      "learning_rate": 2e-05,
+      "loss": 1.5414,
+      "step": 397
+    },
+    {
+      "epoch": 1.7304347826086957,
+      "grad_norm": 0.43897366523742676,
+      "learning_rate": 2e-05,
+      "loss": 1.5572,
+      "step": 398
+    },
+    {
+      "epoch": 1.7347826086956522,
+      "grad_norm": 0.43897366523742676,
+      "learning_rate": 2e-05,
+      "loss": 1.565,
+      "step": 399
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 0.4224741756916046,
+      "learning_rate": 2e-05,
+      "loss": 1.5816,
+      "step": 400
+    },
+    {
+      "epoch": 1.7434782608695651,
+      "grad_norm": 0.4224741756916046,
+      "learning_rate": 2e-05,
+      "loss": 1.5445,
+      "step": 401
+    },
+    {
+      "epoch": 1.7478260869565219,
+      "grad_norm": 0.4224741756916046,
+      "learning_rate": 2e-05,
+      "loss": 1.5533,
+      "step": 402
+    },
+    {
+      "epoch": 1.7521739130434781,
+      "grad_norm": 0.4224741756916046,
+      "learning_rate": 2e-05,
+      "loss": 1.558,
+      "step": 403
+    },
+    {
+      "epoch": 1.7565217391304349,
+      "grad_norm": 0.4469275176525116,
+      "learning_rate": 2e-05,
+      "loss": 1.5241,
+      "step": 404
+    },
+    {
+      "epoch": 1.7608695652173914,
+      "grad_norm": 0.4469275176525116,
+      "learning_rate": 2e-05,
+      "loss": 1.5289,
+      "step": 405
+    },
+    {
+      "epoch": 1.7652173913043478,
+      "grad_norm": 0.4469275176525116,
+      "learning_rate": 2e-05,
+      "loss": 1.5669,
+      "step": 406
+    },
+    {
+      "epoch": 1.7695652173913043,
+      "grad_norm": 0.4469275176525116,
+      "learning_rate": 2e-05,
+      "loss": 1.5317,
+      "step": 407
+    },
+    {
+      "epoch": 1.7739130434782608,
+      "grad_norm": 0.44230249524116516,
+      "learning_rate": 2e-05,
+      "loss": 1.5305,
+      "step": 408
+    },
+    {
+      "epoch": 1.7782608695652173,
+      "grad_norm": 0.44230249524116516,
+      "learning_rate": 2e-05,
+      "loss": 1.5701,
+      "step": 409
+    },
+    {
+      "epoch": 1.7826086956521738,
+      "grad_norm": 0.44230249524116516,
+      "learning_rate": 2e-05,
+      "loss": 1.5086,
+      "step": 410
+    },
+    {
+      "epoch": 1.7869565217391306,
+      "grad_norm": 0.44230249524116516,
+      "learning_rate": 2e-05,
+      "loss": 1.4514,
+      "step": 411
+    },
+    {
+      "epoch": 1.7913043478260868,
+      "grad_norm": 0.42857682704925537,
+      "learning_rate": 2e-05,
+      "loss": 1.5295,
+      "step": 412
+    },
+    {
+      "epoch": 1.7956521739130435,
+      "grad_norm": 0.42857682704925537,
+      "learning_rate": 2e-05,
+      "loss": 1.4738,
+      "step": 413
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.42857682704925537,
+      "learning_rate": 2e-05,
+      "loss": 1.5406,
+      "step": 414
+    },
+    {
+      "epoch": 1.8043478260869565,
+      "grad_norm": 0.42857682704925537,
+      "learning_rate": 2e-05,
+      "loss": 1.533,
+      "step": 415
+    },
+    {
+      "epoch": 1.808695652173913,
+      "grad_norm": 0.44212475419044495,
+      "learning_rate": 2e-05,
+      "loss": 1.5497,
+      "step": 416
+    },
+    {
+      "epoch": 1.8130434782608695,
+      "grad_norm": 0.44212475419044495,
+      "learning_rate": 2e-05,
+      "loss": 1.558,
+      "step": 417
+    },
+    {
+      "epoch": 1.8173913043478263,
+      "grad_norm": 0.44212475419044495,
+      "learning_rate": 2e-05,
+      "loss": 1.557,
+      "step": 418
+    },
+    {
+      "epoch": 1.8217391304347825,
+      "grad_norm": 0.44212475419044495,
+      "learning_rate": 2e-05,
+      "loss": 1.5076,
+      "step": 419
+    },
+    {
+      "epoch": 1.8260869565217392,
+      "grad_norm": 0.42618614435195923,
+      "learning_rate": 2e-05,
+      "loss": 1.5906,
+      "step": 420
+    },
+    {
+      "epoch": 1.8304347826086955,
+      "grad_norm": 0.42618614435195923,
+      "learning_rate": 2e-05,
+      "loss": 1.5247,
+      "step": 421
+    },
+    {
+      "epoch": 1.8347826086956522,
+      "grad_norm": 0.42618614435195923,
+      "learning_rate": 2e-05,
+      "loss": 1.4649,
+      "step": 422
+    },
+    {
+      "epoch": 1.8391304347826087,
+      "grad_norm": 0.42618614435195923,
+      "learning_rate": 2e-05,
+      "loss": 1.5779,
+      "step": 423
+    },
+    {
+      "epoch": 1.8434782608695652,
+      "grad_norm": 0.44451162219047546,
+      "learning_rate": 2e-05,
+      "loss": 1.5232,
+      "step": 424
+    },
+    {
+      "epoch": 1.8478260869565217,
+      "grad_norm": 0.44451162219047546,
+      "learning_rate": 2e-05,
+      "loss": 1.5158,
+      "step": 425
+    },
+    {
+      "epoch": 1.8521739130434782,
+      "grad_norm": 0.44451162219047546,
+      "learning_rate": 2e-05,
+      "loss": 1.4625,
+      "step": 426
+    },
+    {
+      "epoch": 1.856521739130435,
+      "grad_norm": 0.44451162219047546,
+      "learning_rate": 2e-05,
+      "loss": 1.5394,
+      "step": 427
+    },
+    {
+      "epoch": 1.8608695652173912,
+      "grad_norm": 0.41748350858688354,
+      "learning_rate": 2e-05,
+      "loss": 1.5336,
+      "step": 428
+    },
+    {
+      "epoch": 1.865217391304348,
+      "grad_norm": 0.41748350858688354,
+      "learning_rate": 2e-05,
+      "loss": 1.5231,
+      "step": 429
+    },
+    {
+      "epoch": 1.8695652173913042,
+      "grad_norm": 0.41748350858688354,
+      "learning_rate": 2e-05,
+      "loss": 1.498,
+      "step": 430
+    },
+    {
+      "epoch": 1.873913043478261,
+      "grad_norm": 0.41748350858688354,
+      "learning_rate": 2e-05,
+      "loss": 1.4854,
+      "step": 431
+    },
+    {
+      "epoch": 1.8782608695652174,
+      "grad_norm": 0.426869660615921,
+      "learning_rate": 2e-05,
+      "loss": 1.5253,
+      "step": 432
+    },
+    {
+      "epoch": 1.882608695652174,
+      "grad_norm": 0.426869660615921,
+      "learning_rate": 2e-05,
+      "loss": 1.5273,
+      "step": 433
+    },
+    {
+      "epoch": 1.8869565217391304,
+      "grad_norm": 0.426869660615921,
+      "learning_rate": 2e-05,
+      "loss": 1.5221,
+      "step": 434
+    },
+    {
+      "epoch": 1.891304347826087,
+      "grad_norm": 0.426869660615921,
+      "learning_rate": 2e-05,
+      "loss": 1.5335,
+      "step": 435
+    },
+    {
+      "epoch": 1.8956521739130436,
+      "grad_norm": 0.45192039012908936,
+      "learning_rate": 2e-05,
+      "loss": 1.425,
+      "step": 436
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.45192039012908936,
+      "learning_rate": 2e-05,
+      "loss": 1.5333,
+      "step": 437
+    },
+    {
+      "epoch": 1.9043478260869566,
+      "grad_norm": 0.45192039012908936,
+      "learning_rate": 2e-05,
+      "loss": 1.5895,
+      "step": 438
+    },
+    {
+      "epoch": 1.908695652173913,
+      "grad_norm": 0.45192039012908936,
+      "learning_rate": 2e-05,
+      "loss": 1.5203,
+      "step": 439
+    },
+    {
+      "epoch": 1.9130434782608696,
+      "grad_norm": 0.42814260721206665,
+      "learning_rate": 2e-05,
+      "loss": 1.5387,
+      "step": 440
+    },
+    {
+      "epoch": 1.9173913043478261,
+      "grad_norm": 0.42814260721206665,
+      "learning_rate": 2e-05,
+      "loss": 1.492,
+      "step": 441
+    },
+    {
+      "epoch": 1.9217391304347826,
+      "grad_norm": 0.42814260721206665,
+      "learning_rate": 2e-05,
+      "loss": 1.4469,
+      "step": 442
+    },
+    {
+      "epoch": 1.9260869565217391,
+      "grad_norm": 0.42814260721206665,
+      "learning_rate": 2e-05,
+      "loss": 1.4633,
+      "step": 443
+    },
+    {
+      "epoch": 1.9304347826086956,
+      "grad_norm": 0.42902347445487976,
+      "learning_rate": 2e-05,
+      "loss": 1.5168,
+      "step": 444
+    },
+    {
+      "epoch": 1.9347826086956523,
+      "grad_norm": 0.42902347445487976,
+      "learning_rate": 2e-05,
+      "loss": 1.5096,
+      "step": 445
+    },
+    {
+      "epoch": 1.9391304347826086,
+      "grad_norm": 0.42902347445487976,
+      "learning_rate": 2e-05,
+      "loss": 1.5049,
+      "step": 446
+    },
+    {
+      "epoch": 1.9434782608695653,
+      "grad_norm": 0.42902347445487976,
+      "learning_rate": 2e-05,
+      "loss": 1.5848,
+      "step": 447
+    },
+    {
+      "epoch": 1.9478260869565216,
+      "grad_norm": 0.4723300337791443,
+      "learning_rate": 2e-05,
+      "loss": 1.4657,
+      "step": 448
+    },
+    {
+      "epoch": 1.9521739130434783,
+      "grad_norm": 0.4723300337791443,
+      "learning_rate": 2e-05,
+      "loss": 1.5034,
+      "step": 449
+    },
+    {
+      "epoch": 1.9565217391304348,
+      "grad_norm": 0.4723300337791443,
+      "learning_rate": 2e-05,
+      "loss": 1.5047,
+      "step": 450
+    },
+    {
+      "epoch": 1.9608695652173913,
+      "grad_norm": 0.4723300337791443,
+      "learning_rate": 2e-05,
+      "loss": 1.4795,
+      "step": 451
+    },
+    {
+      "epoch": 1.9652173913043478,
+      "grad_norm": 0.4308329224586487,
+      "learning_rate": 2e-05,
+      "loss": 1.5861,
+      "step": 452
+    },
+    {
+      "epoch": 1.9695652173913043,
+      "grad_norm": 0.4308329224586487,
+      "learning_rate": 2e-05,
+      "loss": 1.532,
+      "step": 453
+    },
+    {
+      "epoch": 1.973913043478261,
+      "grad_norm": 0.4308329224586487,
+      "learning_rate": 2e-05,
+      "loss": 1.4708,
+      "step": 454
+    },
+    {
+      "epoch": 1.9782608695652173,
+      "grad_norm": 0.4308329224586487,
+      "learning_rate": 2e-05,
+      "loss": 1.5533,
+      "step": 455
+    },
+    {
+      "epoch": 1.982608695652174,
+      "grad_norm": 0.4359206557273865,
+      "learning_rate": 2e-05,
+      "loss": 1.4912,
+      "step": 456
+    },
+    {
+      "epoch": 1.9869565217391303,
+      "grad_norm": 0.4359206557273865,
+      "learning_rate": 2e-05,
+      "loss": 1.5453,
+      "step": 457
+    },
+    {
+      "epoch": 1.991304347826087,
+      "grad_norm": 0.4359206557273865,
+      "learning_rate": 2e-05,
+      "loss": 1.5074,
+      "step": 458
+    },
+    {
+      "epoch": 1.9956521739130435,
+      "grad_norm": 0.4359206557273865,
+      "learning_rate": 2e-05,
+      "loss": 1.6286,
+      "step": 459
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.43849724531173706,
+      "learning_rate": 2e-05,
+      "loss": 1.4783,
+      "step": 460
+    },
+    {
+      "epoch": 2.0043478260869567,
+      "grad_norm": 0.43849724531173706,
+      "learning_rate": 2e-05,
+      "loss": 1.4826,
+      "step": 461
+    },
+    {
+      "epoch": 2.008695652173913,
+      "grad_norm": 0.43849724531173706,
+      "learning_rate": 2e-05,
+      "loss": 1.4406,
+      "step": 462
+    },
+    {
+      "epoch": 2.0130434782608697,
+      "grad_norm": 0.43849724531173706,
+      "learning_rate": 2e-05,
+      "loss": 1.4177,
+      "step": 463
+    },
+    {
+      "epoch": 2.017391304347826,
+      "grad_norm": 0.4389805495738983,
+      "learning_rate": 2e-05,
+      "loss": 1.51,
+      "step": 464
+    },
+    {
+      "epoch": 2.0217391304347827,
+      "grad_norm": 0.4389805495738983,
+      "learning_rate": 2e-05,
+      "loss": 1.4149,
+      "step": 465
+    },
+    {
+      "epoch": 2.026086956521739,
+      "grad_norm": 0.4389805495738983,
+      "learning_rate": 2e-05,
+      "loss": 1.4876,
+      "step": 466
+    },
+    {
+      "epoch": 2.0304347826086957,
+      "grad_norm": 0.4389805495738983,
+      "learning_rate": 2e-05,
+      "loss": 1.5165,
+      "step": 467
+    },
+    {
+      "epoch": 2.034782608695652,
+      "grad_norm": 0.4780814051628113,
+      "learning_rate": 2e-05,
+      "loss": 1.4525,
+      "step": 468
+    },
+    {
+      "epoch": 2.0391304347826087,
+      "grad_norm": 0.4780814051628113,
+      "learning_rate": 2e-05,
+      "loss": 1.4272,
+      "step": 469
+    },
+    {
+      "epoch": 2.0434782608695654,
+      "grad_norm": 0.4780814051628113,
+      "learning_rate": 2e-05,
+      "loss": 1.4653,
+      "step": 470
+    },
+    {
+      "epoch": 2.0478260869565217,
+      "grad_norm": 0.4780814051628113,
+      "learning_rate": 2e-05,
+      "loss": 1.4531,
+      "step": 471
+    },
+    {
+      "epoch": 2.0521739130434784,
+      "grad_norm": 0.4610339105129242,
+      "learning_rate": 2e-05,
+      "loss": 1.3922,
+      "step": 472
+    },
+    {
+      "epoch": 2.0565217391304347,
+      "grad_norm": 0.4610339105129242,
+      "learning_rate": 2e-05,
+      "loss": 1.4538,
+      "step": 473
+    },
+    {
+      "epoch": 2.0608695652173914,
+      "grad_norm": 0.4610339105129242,
+      "learning_rate": 2e-05,
+      "loss": 1.5089,
+      "step": 474
+    },
+    {
+      "epoch": 2.0652173913043477,
+      "grad_norm": 0.4610339105129242,
+      "learning_rate": 2e-05,
+      "loss": 1.4111,
+      "step": 475
+    },
+    {
+      "epoch": 2.0695652173913044,
+      "grad_norm": 0.4437248706817627,
+      "learning_rate": 2e-05,
+      "loss": 1.4582,
+      "step": 476
+    },
+    {
+      "epoch": 2.0739130434782607,
+      "grad_norm": 0.4437248706817627,
+      "learning_rate": 2e-05,
+      "loss": 1.4614,
+      "step": 477
+    },
+    {
+      "epoch": 2.0782608695652174,
+      "grad_norm": 0.4437248706817627,
+      "learning_rate": 2e-05,
+      "loss": 1.4708,
+      "step": 478
+    },
+    {
+      "epoch": 2.082608695652174,
+      "grad_norm": 0.4437248706817627,
+      "learning_rate": 2e-05,
+      "loss": 1.4156,
+      "step": 479
+    },
+    {
+      "epoch": 2.0869565217391304,
+      "grad_norm": 0.41777855157852173,
+      "learning_rate": 2e-05,
+      "loss": 1.4923,
+      "step": 480
+    },
+    {
+      "epoch": 2.091304347826087,
+      "grad_norm": 0.41777855157852173,
+      "learning_rate": 2e-05,
+      "loss": 1.4219,
+      "step": 481
+    },
+    {
+      "epoch": 2.0956521739130434,
+      "grad_norm": 0.41777855157852173,
+      "learning_rate": 2e-05,
+      "loss": 1.4431,
+      "step": 482
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.41777855157852173,
+      "learning_rate": 2e-05,
+      "loss": 1.3807,
+      "step": 483
+    },
+    {
+      "epoch": 2.1043478260869564,
+      "grad_norm": 0.43318524956703186,
+      "learning_rate": 2e-05,
+      "loss": 1.4227,
+      "step": 484
+    },
+    {
+      "epoch": 2.108695652173913,
+      "grad_norm": 0.43318524956703186,
+      "learning_rate": 2e-05,
+      "loss": 1.4109,
+      "step": 485
+    },
+    {
+      "epoch": 2.1130434782608694,
+      "grad_norm": 0.43318524956703186,
+      "learning_rate": 2e-05,
+      "loss": 1.4565,
+      "step": 486
+    },
+    {
+      "epoch": 2.117391304347826,
+      "grad_norm": 0.43318524956703186,
+      "learning_rate": 2e-05,
+      "loss": 1.4458,
+      "step": 487
+    },
+    {
+      "epoch": 2.121739130434783,
+      "grad_norm": 0.46397483348846436,
+      "learning_rate": 2e-05,
+      "loss": 1.4003,
+      "step": 488
+    },
+    {
+      "epoch": 2.126086956521739,
+      "grad_norm": 0.46397483348846436,
+      "learning_rate": 2e-05,
+      "loss": 1.4473,
+      "step": 489
+    },
+    {
+      "epoch": 2.130434782608696,
+      "grad_norm": 0.46397483348846436,
+      "learning_rate": 2e-05,
+      "loss": 1.4112,
+      "step": 490
+    },
+    {
+      "epoch": 2.134782608695652,
+      "grad_norm": 0.46397483348846436,
+      "learning_rate": 2e-05,
+      "loss": 1.4627,
+      "step": 491
+    },
+    {
+      "epoch": 2.139130434782609,
+      "grad_norm": 0.43796172738075256,
+      "learning_rate": 2e-05,
+      "loss": 1.4559,
+      "step": 492
+    },
+    {
+      "epoch": 2.143478260869565,
+      "grad_norm": 0.43796172738075256,
+      "learning_rate": 2e-05,
+      "loss": 1.456,
+      "step": 493
+    },
+    {
+      "epoch": 2.1478260869565218,
+      "grad_norm": 0.43796172738075256,
+      "learning_rate": 2e-05,
+      "loss": 1.4903,
+      "step": 494
+    },
+    {
+      "epoch": 2.1521739130434785,
+      "grad_norm": 0.43796172738075256,
+      "learning_rate": 2e-05,
+      "loss": 1.4208,
+      "step": 495
+    },
+    {
+      "epoch": 2.1565217391304348,
+      "grad_norm": 0.4605541229248047,
+      "learning_rate": 2e-05,
+      "loss": 1.4443,
+      "step": 496
+    },
+    {
+      "epoch": 2.1608695652173915,
+      "grad_norm": 0.4605541229248047,
+      "learning_rate": 2e-05,
+      "loss": 1.4738,
+      "step": 497
+    },
+    {
+      "epoch": 2.1652173913043478,
+      "grad_norm": 0.4605541229248047,
+      "learning_rate": 2e-05,
+      "loss": 1.4085,
+      "step": 498
+    },
+    {
+      "epoch": 2.1695652173913045,
+      "grad_norm": 0.4605541229248047,
+      "learning_rate": 2e-05,
+      "loss": 1.5024,
+      "step": 499
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "grad_norm": 0.4581138491630554,
+      "learning_rate": 2e-05,
+      "loss": 1.4846,
+      "step": 500
+    },
+    {
+      "epoch": 2.1782608695652175,
+      "grad_norm": 0.4581138491630554,
+      "learning_rate": 2e-05,
+      "loss": 1.473,
+      "step": 501
+    },
+    {
+      "epoch": 2.1826086956521737,
+      "grad_norm": 0.4581138491630554,
+      "learning_rate": 2e-05,
+      "loss": 1.4521,
+      "step": 502
+    },
+    {
+      "epoch": 2.1869565217391305,
+      "grad_norm": 0.4581138491630554,
+      "learning_rate": 2e-05,
+      "loss": 1.443,
+      "step": 503
+    },
+    {
+      "epoch": 2.1913043478260867,
+      "grad_norm": 0.429681658744812,
+      "learning_rate": 2e-05,
+      "loss": 1.4399,
+      "step": 504
+    },
+    {
+      "epoch": 2.1956521739130435,
+      "grad_norm": 0.429681658744812,
+      "learning_rate": 2e-05,
+      "loss": 1.469,
+      "step": 505
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.429681658744812,
+      "learning_rate": 2e-05,
+      "loss": 1.4104,
+      "step": 506
+    },
+    {
+      "epoch": 2.2043478260869565,
+      "grad_norm": 0.429681658744812,
+      "learning_rate": 2e-05,
+      "loss": 1.4559,
+      "step": 507
+    },
+    {
+      "epoch": 2.208695652173913,
+      "grad_norm": 0.4941965937614441,
+      "learning_rate": 2e-05,
+      "loss": 1.4777,
+      "step": 508
+    },
+    {
+      "epoch": 2.2130434782608694,
+      "grad_norm": 0.4941965937614441,
+      "learning_rate": 2e-05,
+      "loss": 1.4692,
+      "step": 509
+    },
+    {
+      "epoch": 2.217391304347826,
+      "grad_norm": 0.4941965937614441,
+      "learning_rate": 2e-05,
+      "loss": 1.48,
+      "step": 510
+    },
+    {
+      "epoch": 2.2217391304347824,
+      "grad_norm": 0.4941965937614441,
+      "learning_rate": 2e-05,
+      "loss": 1.335,
+      "step": 511
+    },
+    {
+      "epoch": 2.226086956521739,
+      "grad_norm": 0.439325213432312,
+      "learning_rate": 2e-05,
+      "loss": 1.4571,
+      "step": 512
+    },
+    {
+      "epoch": 2.230434782608696,
+      "grad_norm": 0.439325213432312,
+      "learning_rate": 2e-05,
+      "loss": 1.3929,
+      "step": 513
+    },
+    {
+      "epoch": 2.234782608695652,
+      "grad_norm": 0.439325213432312,
+      "learning_rate": 2e-05,
+      "loss": 1.4184,
+      "step": 514
+    },
+    {
+      "epoch": 2.239130434782609,
+      "grad_norm": 0.439325213432312,
+      "learning_rate": 2e-05,
+      "loss": 1.4146,
+      "step": 515
+    },
+    {
+      "epoch": 2.243478260869565,
+      "grad_norm": 0.4552342891693115,
+      "learning_rate": 2e-05,
+      "loss": 1.5433,
+      "step": 516
+    },
+    {
+      "epoch": 2.247826086956522,
+      "grad_norm": 0.4552342891693115,
+      "learning_rate": 2e-05,
+      "loss": 1.4631,
+      "step": 517
+    },
+    {
+      "epoch": 2.252173913043478,
+      "grad_norm": 0.4552342891693115,
+      "learning_rate": 2e-05,
+      "loss": 1.4267,
+      "step": 518
+    },
+    {
+      "epoch": 2.256521739130435,
+      "grad_norm": 0.4552342891693115,
+      "learning_rate": 2e-05,
+      "loss": 1.4323,
+      "step": 519
+    },
+    {
+      "epoch": 2.260869565217391,
+      "grad_norm": 0.4403064548969269,
+      "learning_rate": 2e-05,
+      "loss": 1.4965,
+      "step": 520
+    },
+    {
+      "epoch": 2.265217391304348,
+      "grad_norm": 0.4403064548969269,
+      "learning_rate": 2e-05,
+      "loss": 1.4342,
+      "step": 521
+    },
+    {
+      "epoch": 2.269565217391304,
+      "grad_norm": 0.4403064548969269,
+      "learning_rate": 2e-05,
+      "loss": 1.3941,
+      "step": 522
+    },
+    {
+      "epoch": 2.273913043478261,
+      "grad_norm": 0.4403064548969269,
+      "learning_rate": 2e-05,
+      "loss": 1.4012,
+      "step": 523
+    },
+    {
+      "epoch": 2.2782608695652176,
+      "grad_norm": 0.4627401530742645,
+      "learning_rate": 2e-05,
+      "loss": 1.4316,
+      "step": 524
+    },
+    {
+      "epoch": 2.282608695652174,
+      "grad_norm": 0.4627401530742645,
+      "learning_rate": 2e-05,
+      "loss": 1.3794,
+      "step": 525
+    },
+    {
+      "epoch": 2.2869565217391306,
+      "grad_norm": 0.4627401530742645,
+      "learning_rate": 2e-05,
+      "loss": 1.4841,
+      "step": 526
+    },
+    {
+      "epoch": 2.291304347826087,
+      "grad_norm": 0.4627401530742645,
+      "learning_rate": 2e-05,
+      "loss": 1.403,
+      "step": 527
+    },
+    {
+      "epoch": 2.2956521739130435,
+      "grad_norm": 0.4262266755104065,
+      "learning_rate": 2e-05,
+      "loss": 1.4839,
+      "step": 528
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4262266755104065,
+      "learning_rate": 2e-05,
+      "loss": 1.4936,
+      "step": 529
+    },
+    {
+      "epoch": 2.3043478260869565,
+      "grad_norm": 0.4262266755104065,
+      "learning_rate": 2e-05,
+      "loss": 1.4565,
+      "step": 530
+    },
+    {
+      "epoch": 2.3086956521739133,
+      "grad_norm": 0.4262266755104065,
+      "learning_rate": 2e-05,
+      "loss": 1.3934,
+      "step": 531
+    },
+    {
+      "epoch": 2.3130434782608695,
+      "grad_norm": 0.41838669776916504,
+      "learning_rate": 2e-05,
+      "loss": 1.4768,
+      "step": 532
+    },
+    {
+      "epoch": 2.3173913043478263,
+      "grad_norm": 0.41838669776916504,
+      "learning_rate": 2e-05,
+      "loss": 1.4235,
+      "step": 533
+    },
+    {
+      "epoch": 2.3217391304347825,
+      "grad_norm": 0.41838669776916504,
+      "learning_rate": 2e-05,
+      "loss": 1.3986,
+      "step": 534
+    },
+    {
+      "epoch": 2.3260869565217392,
+      "grad_norm": 0.41838669776916504,
+      "learning_rate": 2e-05,
+      "loss": 1.4706,
+      "step": 535
+    },
+    {
+      "epoch": 2.3304347826086955,
+      "grad_norm": 0.44334736466407776,
+      "learning_rate": 2e-05,
+      "loss": 1.4141,
+      "step": 536
+    },
+    {
+      "epoch": 2.3347826086956522,
+      "grad_norm": 0.44334736466407776,
+      "learning_rate": 2e-05,
+      "loss": 1.4433,
+      "step": 537
+    },
+    {
+      "epoch": 2.3391304347826085,
+      "grad_norm": 0.44334736466407776,
+      "learning_rate": 2e-05,
+      "loss": 1.4327,
+      "step": 538
+    },
+    {
+      "epoch": 2.3434782608695652,
+      "grad_norm": 0.44334736466407776,
+      "learning_rate": 2e-05,
+      "loss": 1.4567,
+      "step": 539
+    },
+    {
+      "epoch": 2.3478260869565215,
+      "grad_norm": 0.45214343070983887,
+      "learning_rate": 2e-05,
+      "loss": 1.4291,
+      "step": 540
+    },
+    {
+      "epoch": 2.3521739130434782,
+      "grad_norm": 0.45214343070983887,
+      "learning_rate": 2e-05,
+      "loss": 1.4177,
+      "step": 541
+    },
+    {
+      "epoch": 2.356521739130435,
+      "grad_norm": 0.45214343070983887,
+      "learning_rate": 2e-05,
+      "loss": 1.4696,
+      "step": 542
+    },
+    {
+      "epoch": 2.360869565217391,
+      "grad_norm": 0.45214343070983887,
+      "learning_rate": 2e-05,
+      "loss": 1.4217,
+      "step": 543
+    },
+    {
+      "epoch": 2.365217391304348,
+      "grad_norm": 0.4370884299278259,
+      "learning_rate": 2e-05,
+      "loss": 1.447,
+      "step": 544
+    },
+    {
+      "epoch": 2.369565217391304,
+      "grad_norm": 0.4370884299278259,
+      "learning_rate": 2e-05,
+      "loss": 1.4681,
+      "step": 545
+    },
+    {
+      "epoch": 2.373913043478261,
+      "grad_norm": 0.4370884299278259,
+      "learning_rate": 2e-05,
+      "loss": 1.3833,
+      "step": 546
+    },
+    {
+      "epoch": 2.378260869565217,
+      "grad_norm": 0.4370884299278259,
+      "learning_rate": 2e-05,
+      "loss": 1.4899,
+      "step": 547
+    },
+    {
+      "epoch": 2.382608695652174,
+      "grad_norm": 0.45092737674713135,
+      "learning_rate": 2e-05,
+      "loss": 1.4567,
+      "step": 548
+    },
+    {
+      "epoch": 2.3869565217391306,
+      "grad_norm": 0.45092737674713135,
+      "learning_rate": 2e-05,
+      "loss": 1.4025,
+      "step": 549
+    },
+    {
+      "epoch": 2.391304347826087,
+      "grad_norm": 0.45092737674713135,
+      "learning_rate": 2e-05,
+      "loss": 1.4732,
+      "step": 550
+    },
+    {
+      "epoch": 2.3956521739130436,
+      "grad_norm": 0.45092737674713135,
+      "learning_rate": 2e-05,
+      "loss": 1.5067,
+      "step": 551
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.4495592713356018,
+      "learning_rate": 2e-05,
+      "loss": 1.4597,
+      "step": 552
+    },
+    {
+      "epoch": 2.4043478260869566,
+      "grad_norm": 0.4495592713356018,
+      "learning_rate": 2e-05,
+      "loss": 1.3733,
+      "step": 553
+    },
+    {
+      "epoch": 2.408695652173913,
+      "grad_norm": 0.4495592713356018,
+      "learning_rate": 2e-05,
+      "loss": 1.4127,
+      "step": 554
+    },
+    {
+      "epoch": 2.4130434782608696,
+      "grad_norm": 0.4495592713356018,
+      "learning_rate": 2e-05,
+      "loss": 1.4427,
+      "step": 555
+    },
+    {
+      "epoch": 2.417391304347826,
+      "grad_norm": 0.44914448261260986,
+      "learning_rate": 2e-05,
+      "loss": 1.4285,
+      "step": 556
+    },
+    {
+      "epoch": 2.4217391304347826,
+      "grad_norm": 0.44914448261260986,
+      "learning_rate": 2e-05,
+      "loss": 1.3925,
+      "step": 557
+    },
+    {
+      "epoch": 2.426086956521739,
+      "grad_norm": 0.44914448261260986,
+      "learning_rate": 2e-05,
+      "loss": 1.4277,
+      "step": 558
+    },
+    {
+      "epoch": 2.4304347826086956,
+      "grad_norm": 0.44914448261260986,
+      "learning_rate": 2e-05,
+      "loss": 1.449,
+      "step": 559
+    },
+    {
+      "epoch": 2.4347826086956523,
+      "grad_norm": 0.4478745758533478,
+      "learning_rate": 2e-05,
+      "loss": 1.4359,
+      "step": 560
+    },
+    {
+      "epoch": 2.4391304347826086,
+      "grad_norm": 0.4478745758533478,
+      "learning_rate": 2e-05,
+      "loss": 1.4546,
+      "step": 561
+    },
+    {
+      "epoch": 2.4434782608695653,
+      "grad_norm": 0.4478745758533478,
+      "learning_rate": 2e-05,
+      "loss": 1.3997,
+      "step": 562
+    },
+    {
+      "epoch": 2.4478260869565216,
+      "grad_norm": 0.4478745758533478,
+      "learning_rate": 2e-05,
+      "loss": 1.4591,
+      "step": 563
+    },
+    {
+      "epoch": 2.4521739130434783,
+      "grad_norm": 0.43071508407592773,
+      "learning_rate": 2e-05,
+      "loss": 1.3532,
+      "step": 564
+    },
+    {
+      "epoch": 2.4565217391304346,
+      "grad_norm": 0.43071508407592773,
+      "learning_rate": 2e-05,
+      "loss": 1.343,
+      "step": 565
+    },
+    {
+      "epoch": 2.4608695652173913,
+      "grad_norm": 0.43071508407592773,
+      "learning_rate": 2e-05,
+      "loss": 1.4923,
+      "step": 566
+    },
+    {
+      "epoch": 2.465217391304348,
+      "grad_norm": 0.43071508407592773,
+      "learning_rate": 2e-05,
+      "loss": 1.4101,
+      "step": 567
+    },
+    {
+      "epoch": 2.4695652173913043,
+      "grad_norm": 0.4356819689273834,
+      "learning_rate": 2e-05,
+      "loss": 1.4828,
+      "step": 568
+    },
+    {
+      "epoch": 2.473913043478261,
+      "grad_norm": 0.4356819689273834,
+      "learning_rate": 2e-05,
+      "loss": 1.4608,
+      "step": 569
+    },
+    {
+      "epoch": 2.4782608695652173,
+      "grad_norm": 0.4356819689273834,
+      "learning_rate": 2e-05,
+      "loss": 1.3872,
+      "step": 570
+    },
+    {
+      "epoch": 2.482608695652174,
+      "grad_norm": 0.4356819689273834,
+      "learning_rate": 2e-05,
+      "loss": 1.4487,
+      "step": 571
+    },
+    {
+      "epoch": 2.4869565217391303,
+      "grad_norm": 0.44995924830436707,
+      "learning_rate": 2e-05,
+      "loss": 1.4694,
+      "step": 572
+    },
+    {
+      "epoch": 2.491304347826087,
+      "grad_norm": 0.44995924830436707,
+      "learning_rate": 2e-05,
+      "loss": 1.4029,
+      "step": 573
+    },
+    {
+      "epoch": 2.4956521739130437,
+      "grad_norm": 0.44995924830436707,
+      "learning_rate": 2e-05,
+      "loss": 1.4111,
+      "step": 574
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.44995924830436707,
+      "learning_rate": 2e-05,
+      "loss": 1.4275,
+      "step": 575
+    },
+    {
+      "epoch": 2.5043478260869563,
+      "grad_norm": 0.4352399408817291,
+      "learning_rate": 2e-05,
+      "loss": 1.443,
+      "step": 576
+    },
+    {
+      "epoch": 2.508695652173913,
+      "grad_norm": 0.4352399408817291,
+      "learning_rate": 2e-05,
+      "loss": 1.4107,
+      "step": 577
+    },
+    {
+      "epoch": 2.5130434782608697,
+      "grad_norm": 0.4352399408817291,
+      "learning_rate": 2e-05,
+      "loss": 1.4341,
+      "step": 578
+    },
+    {
+      "epoch": 2.517391304347826,
+      "grad_norm": 0.4352399408817291,
+      "learning_rate": 2e-05,
+      "loss": 1.3694,
+      "step": 579
+    },
+    {
+      "epoch": 2.5217391304347827,
+      "grad_norm": 0.45637837052345276,
+      "learning_rate": 2e-05,
+      "loss": 1.428,
+      "step": 580
+    },
+    {
+      "epoch": 2.526086956521739,
+      "grad_norm": 0.45637837052345276,
+      "learning_rate": 2e-05,
+      "loss": 1.409,
+      "step": 581
+    },
+    {
+      "epoch": 2.5304347826086957,
+      "grad_norm": 0.45637837052345276,
+      "learning_rate": 2e-05,
+      "loss": 1.3807,
+      "step": 582
+    },
+    {
+      "epoch": 2.534782608695652,
+      "grad_norm": 0.45637837052345276,
+      "learning_rate": 2e-05,
+      "loss": 1.3878,
+      "step": 583
+    },
+    {
+      "epoch": 2.5391304347826087,
+      "grad_norm": 0.4560960531234741,
+      "learning_rate": 2e-05,
+      "loss": 1.327,
+      "step": 584
+    },
+    {
+      "epoch": 2.5434782608695654,
+      "grad_norm": 0.4560960531234741,
+      "learning_rate": 2e-05,
+      "loss": 1.4564,
+      "step": 585
+    },
+    {
+      "epoch": 2.5478260869565217,
+      "grad_norm": 0.4560960531234741,
+      "learning_rate": 2e-05,
+      "loss": 1.3758,
+      "step": 586
+    },
+    {
+      "epoch": 2.5521739130434784,
+      "grad_norm": 0.4560960531234741,
+      "learning_rate": 2e-05,
+      "loss": 1.4642,
+      "step": 587
+    },
+    {
+      "epoch": 2.5565217391304347,
+      "grad_norm": 0.458340048789978,
+      "learning_rate": 2e-05,
+      "loss": 1.4277,
+      "step": 588
+    },
+    {
+      "epoch": 2.5608695652173914,
+      "grad_norm": 0.458340048789978,
+      "learning_rate": 2e-05,
+      "loss": 1.423,
+      "step": 589
+    },
+    {
+      "epoch": 2.5652173913043477,
+      "grad_norm": 0.458340048789978,
+      "learning_rate": 2e-05,
+      "loss": 1.4059,
+      "step": 590
+    },
+    {
+      "epoch": 2.5695652173913044,
+      "grad_norm": 0.458340048789978,
+      "learning_rate": 2e-05,
+      "loss": 1.4327,
+      "step": 591
+    },
+    {
+      "epoch": 2.573913043478261,
+      "grad_norm": 0.46771275997161865,
+      "learning_rate": 2e-05,
+      "loss": 1.4411,
+      "step": 592
+    },
+    {
+      "epoch": 2.5782608695652174,
+      "grad_norm": 0.46771275997161865,
+      "learning_rate": 2e-05,
+      "loss": 1.3755,
+      "step": 593
+    },
+    {
+      "epoch": 2.5826086956521737,
+      "grad_norm": 0.46771275997161865,
+      "learning_rate": 2e-05,
+      "loss": 1.474,
+      "step": 594
+    },
+    {
+      "epoch": 2.5869565217391304,
+      "grad_norm": 0.46771275997161865,
+      "learning_rate": 2e-05,
+      "loss": 1.4165,
+      "step": 595
+    },
+    {
+      "epoch": 2.591304347826087,
+      "grad_norm": 0.44320788979530334,
+      "learning_rate": 2e-05,
+      "loss": 1.402,
+      "step": 596
+    },
+    {
+      "epoch": 2.5956521739130434,
+      "grad_norm": 0.44320788979530334,
+      "learning_rate": 2e-05,
+      "loss": 1.3985,
+      "step": 597
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.44320788979530334,
+      "learning_rate": 2e-05,
+      "loss": 1.4689,
+      "step": 598
+    },
+    {
+      "epoch": 2.6043478260869564,
+      "grad_norm": 0.44320788979530334,
+      "learning_rate": 2e-05,
+      "loss": 1.4768,
+      "step": 599
+    },
+    {
+      "epoch": 2.608695652173913,
+      "grad_norm": 0.4548298418521881,
+      "learning_rate": 2e-05,
+      "loss": 1.4092,
+      "step": 600
+    },
+    {
+      "epoch": 2.6130434782608694,
+      "grad_norm": 0.4548298418521881,
+      "learning_rate": 2e-05,
+      "loss": 1.3731,
+      "step": 601
+    },
+    {
+      "epoch": 2.617391304347826,
+      "grad_norm": 0.4548298418521881,
+      "learning_rate": 2e-05,
+      "loss": 1.3681,
+      "step": 602
+    },
+    {
+      "epoch": 2.621739130434783,
+      "grad_norm": 0.4548298418521881,
+      "learning_rate": 2e-05,
+      "loss": 1.3759,
+      "step": 603
+    },
+    {
+      "epoch": 2.626086956521739,
+      "grad_norm": 0.4255348742008209,
+      "learning_rate": 2e-05,
+      "loss": 1.3581,
+      "step": 604
+    },
+    {
+      "epoch": 2.630434782608696,
+      "grad_norm": 0.4255348742008209,
+      "learning_rate": 2e-05,
+      "loss": 1.389,
+      "step": 605
+    },
+    {
+      "epoch": 2.634782608695652,
+      "grad_norm": 0.4255348742008209,
+      "learning_rate": 2e-05,
+      "loss": 1.4142,
+      "step": 606
+    },
+    {
+      "epoch": 2.639130434782609,
+      "grad_norm": 0.4255348742008209,
+      "learning_rate": 2e-05,
+      "loss": 1.4308,
+      "step": 607
+    },
+    {
+      "epoch": 2.643478260869565,
+      "grad_norm": 0.44347140192985535,
+      "learning_rate": 2e-05,
+      "loss": 1.3694,
+      "step": 608
+    },
+    {
+      "epoch": 2.6478260869565218,
+      "grad_norm": 0.44347140192985535,
+      "learning_rate": 2e-05,
+      "loss": 1.3745,
+      "step": 609
+    },
+    {
+      "epoch": 2.6521739130434785,
+      "grad_norm": 0.44347140192985535,
+      "learning_rate": 2e-05,
+      "loss": 1.3965,
+      "step": 610
+    },
+    {
+      "epoch": 2.6565217391304348,
+      "grad_norm": 0.44347140192985535,
+      "learning_rate": 2e-05,
+      "loss": 1.3552,
+      "step": 611
+    },
+    {
+      "epoch": 2.660869565217391,
+      "grad_norm": 0.43682318925857544,
+      "learning_rate": 2e-05,
+      "loss": 1.3977,
+      "step": 612
+    },
+    {
+      "epoch": 2.6652173913043478,
+      "grad_norm": 0.43682318925857544,
+      "learning_rate": 2e-05,
+      "loss": 1.4557,
+      "step": 613
+    },
+    {
+      "epoch": 2.6695652173913045,
+      "grad_norm": 0.43682318925857544,
+      "learning_rate": 2e-05,
+      "loss": 1.4087,
+      "step": 614
+    },
+    {
+      "epoch": 2.6739130434782608,
+      "grad_norm": 0.43682318925857544,
+      "learning_rate": 2e-05,
+      "loss": 1.3819,
+      "step": 615
+    },
+    {
+      "epoch": 2.6782608695652175,
+      "grad_norm": 0.43287912011146545,
+      "learning_rate": 2e-05,
+      "loss": 1.4954,
+      "step": 616
+    },
+    {
+      "epoch": 2.6826086956521737,
+      "grad_norm": 0.43287912011146545,
+      "learning_rate": 2e-05,
+      "loss": 1.4255,
+      "step": 617
+    },
+    {
+      "epoch": 2.6869565217391305,
+      "grad_norm": 0.43287912011146545,
+      "learning_rate": 2e-05,
+      "loss": 1.4469,
+      "step": 618
+    },
+    {
+      "epoch": 2.6913043478260867,
+      "grad_norm": 0.43287912011146545,
+      "learning_rate": 2e-05,
+      "loss": 1.3364,
+      "step": 619
+    },
+    {
+      "epoch": 2.6956521739130435,
+      "grad_norm": 0.437889039516449,
+      "learning_rate": 2e-05,
+      "loss": 1.4392,
+      "step": 620
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.437889039516449,
+      "learning_rate": 2e-05,
+      "loss": 1.4408,
+      "step": 621
+    },
+    {
+      "epoch": 2.7043478260869565,
+      "grad_norm": 0.437889039516449,
+      "learning_rate": 2e-05,
+      "loss": 1.4241,
+      "step": 622
+    },
+    {
+      "epoch": 2.708695652173913,
+      "grad_norm": 0.437889039516449,
+      "learning_rate": 2e-05,
+      "loss": 1.4606,
+      "step": 623
+    },
+    {
+      "epoch": 2.7130434782608694,
+      "grad_norm": 0.4409219026565552,
+      "learning_rate": 2e-05,
+      "loss": 1.4289,
+      "step": 624
+    },
+    {
+      "epoch": 2.717391304347826,
+      "grad_norm": 0.4409219026565552,
+      "learning_rate": 2e-05,
+      "loss": 1.3626,
+      "step": 625
+    },
+    {
+      "epoch": 2.7217391304347824,
+      "grad_norm": 0.4409219026565552,
+      "learning_rate": 2e-05,
+      "loss": 1.3953,
+      "step": 626
+    },
+    {
+      "epoch": 2.726086956521739,
+      "grad_norm": 0.4409219026565552,
+      "learning_rate": 2e-05,
+      "loss": 1.4515,
+      "step": 627
+    },
+    {
+      "epoch": 2.730434782608696,
+      "grad_norm": 0.4441738724708557,
+      "learning_rate": 2e-05,
+      "loss": 1.3941,
+      "step": 628
+    },
+    {
+      "epoch": 2.734782608695652,
+      "grad_norm": 0.4441738724708557,
+      "learning_rate": 2e-05,
+      "loss": 1.4925,
+      "step": 629
+    },
+    {
+      "epoch": 2.7391304347826084,
+      "grad_norm": 0.4441738724708557,
+      "learning_rate": 2e-05,
+      "loss": 1.4274,
+      "step": 630
+    },
+    {
+      "epoch": 2.743478260869565,
+      "grad_norm": 0.4441738724708557,
+      "learning_rate": 2e-05,
+      "loss": 1.4483,
+      "step": 631
+    },
+    {
+      "epoch": 2.747826086956522,
+      "grad_norm": 0.44980600476264954,
+      "learning_rate": 2e-05,
+      "loss": 1.4143,
+      "step": 632
+    },
+    {
+      "epoch": 2.752173913043478,
+      "grad_norm": 0.44980600476264954,
+      "learning_rate": 2e-05,
+      "loss": 1.4103,
+      "step": 633
+    },
+    {
+      "epoch": 2.756521739130435,
+      "grad_norm": 0.44980600476264954,
+      "learning_rate": 2e-05,
+      "loss": 1.4343,
+      "step": 634
+    },
+    {
+      "epoch": 2.7608695652173916,
+      "grad_norm": 0.44980600476264954,
+      "learning_rate": 2e-05,
+      "loss": 1.4193,
+      "step": 635
+    },
+    {
+      "epoch": 2.765217391304348,
+      "grad_norm": 0.4380491077899933,
+      "learning_rate": 2e-05,
+      "loss": 1.3609,
+      "step": 636
+    },
+    {
+      "epoch": 2.769565217391304,
+      "grad_norm": 0.4380491077899933,
+      "learning_rate": 2e-05,
+      "loss": 1.4065,
+      "step": 637
+    },
+    {
+      "epoch": 2.773913043478261,
+      "grad_norm": 0.4380491077899933,
+      "learning_rate": 2e-05,
+      "loss": 1.3425,
+      "step": 638
+    },
+    {
+      "epoch": 2.7782608695652176,
+      "grad_norm": 0.4380491077899933,
+      "learning_rate": 2e-05,
+      "loss": 1.3964,
+      "step": 639
+    },
+    {
+      "epoch": 2.782608695652174,
+      "grad_norm": 0.42945510149002075,
+      "learning_rate": 2e-05,
+      "loss": 1.3966,
+      "step": 640
+    },
+    {
+      "epoch": 2.7869565217391306,
+      "grad_norm": 0.42945510149002075,
+      "learning_rate": 2e-05,
+      "loss": 1.4238,
+      "step": 641
+    },
+    {
+      "epoch": 2.791304347826087,
+      "grad_norm": 0.42945510149002075,
+      "learning_rate": 2e-05,
+      "loss": 1.3981,
+      "step": 642
+    },
+    {
+      "epoch": 2.7956521739130435,
+      "grad_norm": 0.42945510149002075,
+      "learning_rate": 2e-05,
+      "loss": 1.4368,
+      "step": 643
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4350828528404236,
+      "learning_rate": 2e-05,
+      "loss": 1.3879,
+      "step": 644
+    },
+    {
+      "epoch": 2.8043478260869565,
+      "grad_norm": 0.4350828528404236,
+      "learning_rate": 2e-05,
+      "loss": 1.3954,
+      "step": 645
+    },
+    {
+      "epoch": 2.8086956521739133,
+      "grad_norm": 0.4350828528404236,
+      "learning_rate": 2e-05,
+      "loss": 1.4562,
+      "step": 646
+    },
+    {
+      "epoch": 2.8130434782608695,
+      "grad_norm": 0.4350828528404236,
+      "learning_rate": 2e-05,
+      "loss": 1.3741,
+      "step": 647
+    },
+    {
+      "epoch": 2.8173913043478263,
+      "grad_norm": 0.43145105242729187,
+      "learning_rate": 2e-05,
+      "loss": 1.3833,
+      "step": 648
+    },
+    {
+      "epoch": 2.8217391304347825,
+      "grad_norm": 0.43145105242729187,
+      "learning_rate": 2e-05,
+      "loss": 1.4499,
+      "step": 649
+    },
+    {
+      "epoch": 2.8260869565217392,
+      "grad_norm": 0.43145105242729187,
+      "learning_rate": 2e-05,
+      "loss": 1.4488,
+      "step": 650
+    },
+    {
+      "epoch": 2.8304347826086955,
+      "grad_norm": 0.43145105242729187,
+      "learning_rate": 2e-05,
+      "loss": 1.3758,
+      "step": 651
+    },
+    {
+      "epoch": 2.8347826086956522,
+      "grad_norm": 0.44008344411849976,
+      "learning_rate": 2e-05,
+      "loss": 1.4219,
+      "step": 652
+    },
+    {
+      "epoch": 2.839130434782609,
+      "grad_norm": 0.44008344411849976,
+      "learning_rate": 2e-05,
+      "loss": 1.4525,
+      "step": 653
+    },
+    {
+      "epoch": 2.8434782608695652,
+      "grad_norm": 0.44008344411849976,
+      "learning_rate": 2e-05,
+      "loss": 1.4668,
+      "step": 654
+    },
+    {
+      "epoch": 2.8478260869565215,
+      "grad_norm": 0.44008344411849976,
+      "learning_rate": 2e-05,
+      "loss": 1.4025,
+      "step": 655
+    },
+    {
+      "epoch": 2.8521739130434782,
+      "grad_norm": 0.4309910535812378,
+      "learning_rate": 2e-05,
+      "loss": 1.4007,
+      "step": 656
+    },
+    {
+      "epoch": 2.856521739130435,
+      "grad_norm": 0.4309910535812378,
+      "learning_rate": 2e-05,
+      "loss": 1.3499,
+      "step": 657
+    },
+    {
+      "epoch": 2.860869565217391,
+      "grad_norm": 0.4309910535812378,
+      "learning_rate": 2e-05,
+      "loss": 1.4505,
+      "step": 658
+    },
+    {
+      "epoch": 2.865217391304348,
+      "grad_norm": 0.4309910535812378,
+      "learning_rate": 2e-05,
+      "loss": 1.4449,
+      "step": 659
+    },
+    {
+      "epoch": 2.869565217391304,
+      "grad_norm": 0.44285786151885986,
+      "learning_rate": 2e-05,
+      "loss": 1.3851,
+      "step": 660
+    },
+    {
+      "epoch": 2.873913043478261,
+      "grad_norm": 0.44285786151885986,
+      "learning_rate": 2e-05,
+      "loss": 1.4374,
+      "step": 661
+    },
+    {
+      "epoch": 2.878260869565217,
+      "grad_norm": 0.44285786151885986,
+      "learning_rate": 2e-05,
+      "loss": 1.4486,
+      "step": 662
+    },
+    {
+      "epoch": 2.882608695652174,
+      "grad_norm": 0.44285786151885986,
+      "learning_rate": 2e-05,
+      "loss": 1.4333,
+      "step": 663
+    },
+    {
+      "epoch": 2.8869565217391306,
+      "grad_norm": 0.43612438440322876,
+      "learning_rate": 2e-05,
+      "loss": 1.418,
+      "step": 664
+    },
+    {
+      "epoch": 2.891304347826087,
+      "grad_norm": 0.43612438440322876,
+      "learning_rate": 2e-05,
+      "loss": 1.5115,
+      "step": 665
+    },
+    {
+      "epoch": 2.8956521739130436,
+      "grad_norm": 0.43612438440322876,
+      "learning_rate": 2e-05,
+      "loss": 1.4567,
+      "step": 666
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.43612438440322876,
+      "learning_rate": 2e-05,
+      "loss": 1.4776,
+      "step": 667
+    },
+    {
+      "epoch": 2.9043478260869566,
+      "grad_norm": 0.4316587448120117,
+      "learning_rate": 2e-05,
+      "loss": 1.3331,
+      "step": 668
+    },
+    {
+      "epoch": 2.908695652173913,
+      "grad_norm": 0.4316587448120117,
+      "learning_rate": 2e-05,
+      "loss": 1.4252,
+      "step": 669
+    },
+    {
+      "epoch": 2.9130434782608696,
+      "grad_norm": 0.4316587448120117,
+      "learning_rate": 2e-05,
+      "loss": 1.4541,
+      "step": 670
+    },
+    {
+      "epoch": 2.9173913043478263,
+      "grad_norm": 0.4316587448120117,
+      "learning_rate": 2e-05,
+      "loss": 1.3706,
+      "step": 671
+    },
+    {
+      "epoch": 2.9217391304347826,
+      "grad_norm": 0.45414403080940247,
+      "learning_rate": 2e-05,
+      "loss": 1.4172,
+      "step": 672
+    },
+    {
+      "epoch": 2.926086956521739,
+      "grad_norm": 0.45414403080940247,
+      "learning_rate": 2e-05,
+      "loss": 1.4269,
+      "step": 673
+    },
+    {
+      "epoch": 2.9304347826086956,
+      "grad_norm": 0.45414403080940247,
+      "learning_rate": 2e-05,
+      "loss": 1.4144,
+      "step": 674
+    },
+    {
+      "epoch": 2.9347826086956523,
+      "grad_norm": 0.45414403080940247,
+      "learning_rate": 2e-05,
+      "loss": 1.3762,
+      "step": 675
+    },
+    {
+      "epoch": 2.9391304347826086,
+      "grad_norm": 0.4472872018814087,
+      "learning_rate": 2e-05,
+      "loss": 1.3774,
+      "step": 676
+    },
+    {
+      "epoch": 2.9434782608695653,
+      "grad_norm": 0.4472872018814087,
+      "learning_rate": 2e-05,
+      "loss": 1.4429,
+      "step": 677
+    },
+    {
+      "epoch": 2.9478260869565216,
+      "grad_norm": 0.4472872018814087,
+      "learning_rate": 2e-05,
+      "loss": 1.3774,
+      "step": 678
+    },
+    {
+      "epoch": 2.9521739130434783,
+      "grad_norm": 0.4472872018814087,
+      "learning_rate": 2e-05,
+      "loss": 1.4752,
+      "step": 679
+    },
+    {
+      "epoch": 2.9565217391304346,
+      "grad_norm": 0.4483736753463745,
+      "learning_rate": 2e-05,
+      "loss": 1.3637,
+      "step": 680
+    },
+    {
+      "epoch": 2.9608695652173913,
+      "grad_norm": 0.4483736753463745,
+      "learning_rate": 2e-05,
+      "loss": 1.4605,
+      "step": 681
+    },
+    {
+      "epoch": 2.965217391304348,
+      "grad_norm": 0.4483736753463745,
+      "learning_rate": 2e-05,
+      "loss": 1.369,
+      "step": 682
+    },
+    {
+      "epoch": 2.9695652173913043,
+      "grad_norm": 0.4483736753463745,
+      "learning_rate": 2e-05,
+      "loss": 1.38,
+      "step": 683
+    },
+    {
+      "epoch": 2.973913043478261,
+      "grad_norm": 0.4401063621044159,
+      "learning_rate": 2e-05,
+      "loss": 1.3735,
+      "step": 684
+    },
+    {
+      "epoch": 2.9782608695652173,
+      "grad_norm": 0.4401063621044159,
+      "learning_rate": 2e-05,
+      "loss": 1.4165,
+      "step": 685
+    },
+    {
+      "epoch": 2.982608695652174,
+      "grad_norm": 0.4401063621044159,
+      "learning_rate": 2e-05,
+      "loss": 1.4523,
+      "step": 686
+    },
+    {
+      "epoch": 2.9869565217391303,
+      "grad_norm": 0.4401063621044159,
+      "learning_rate": 2e-05,
+      "loss": 1.3555,
+      "step": 687
+    },
+    {
+      "epoch": 2.991304347826087,
+      "grad_norm": 0.43969812989234924,
+      "learning_rate": 2e-05,
+      "loss": 1.3699,
+      "step": 688
+    },
+    {
+      "epoch": 2.9956521739130437,
+      "grad_norm": 0.43969812989234924,
+      "learning_rate": 2e-05,
+      "loss": 1.3609,
+      "step": 689
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.43969812989234924,
+      "learning_rate": 2e-05,
+      "loss": 1.3846,
+      "step": 690
+    },
+    {
+      "epoch": 3.0,
+      "step": 690,
+      "total_flos": 3.864388766479155e+17,
+      "train_loss": 1.5982977670171987,
+      "train_runtime": 425.6486,
+      "train_samples_per_second": 362.842,
+      "train_steps_per_second": 1.621
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 690,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 3000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.864388766479155e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}