diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/last-checkpoint/trainer_state.json"
@@ -0,0 +1,5278 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9993346640053227,
+  "eval_steps": 500,
+  "global_step": 751,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.1463,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.2151,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.5e-06,
+      "loss": 1.1345,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.15625,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.1139,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.375,
+      "learning_rate": 2.5e-06,
+      "loss": 1.0874,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.28125,
+      "learning_rate": 3e-06,
+      "loss": 1.1087,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.34375,
+      "learning_rate": 3.5e-06,
+      "loss": 1.0746,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.0507,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.375,
+      "learning_rate": 4.5e-06,
+      "loss": 1.0417,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5,
+      "learning_rate": 5e-06,
+      "loss": 1.0395,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.75,
+      "learning_rate": 4.999998623722131e-06,
+      "loss": 1.0696,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.390625,
+      "learning_rate": 4.999994494890037e-06,
+      "loss": 1.0175,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.999987613508266e-06,
+      "loss": 1.0273,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.265625,
+      "learning_rate": 4.999977979584393e-06,
+      "loss": 1.0296,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.999965593129025e-06,
+      "loss": 1.0156,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.999950454155801e-06,
+      "loss": 1.0328,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.999932562681389e-06,
+      "loss": 0.9897,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.9999119187254875e-06,
+      "loss": 1.008,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.999888522310826e-06,
+      "loss": 1.0089,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.999862373463166e-06,
+      "loss": 1.0183,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.75,
+      "learning_rate": 4.999833472211295e-06,
+      "loss": 1.0398,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.999801818587036e-06,
+      "loss": 1.0246,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.99976741262524e-06,
+      "loss": 1.0319,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.999730254363788e-06,
+      "loss": 0.9883,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.999690343843594e-06,
+      "loss": 1.0165,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.999647681108598e-06,
+      "loss": 0.9859,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.999602266205774e-06,
+      "loss": 0.9986,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.999554099185124e-06,
+      "loss": 1.0136,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.999503180099683e-06,
+      "loss": 0.9967,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5,
+      "learning_rate": 4.999449509005511e-06,
+      "loss": 0.9814,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.999393085961704e-06,
+      "loss": 1.0176,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.625,
+      "learning_rate": 4.999333911030382e-06,
+      "loss": 0.9849,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.9992719842767e-06,
+      "loss": 0.9645,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.999207305768841e-06,
+      "loss": 0.9765,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.999139875578015e-06,
+      "loss": 0.9966,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.9990696937784675e-06,
+      "loss": 0.9941,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.998996760447469e-06,
+      "loss": 1.0011,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.99892107566532e-06,
+      "loss": 0.9851,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.99884263951535e-06,
+      "loss": 0.9769,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.998761452083922e-06,
+      "loss": 0.9735,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.998677513460423e-06,
+      "loss": 0.9647,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.998590823737272e-06,
+      "loss": 0.9532,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.998501383009916e-06,
+      "loss": 0.9697,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.625,
+      "learning_rate": 4.998409191376832e-06,
+      "loss": 0.9847,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.998314248939524e-06,
+      "loss": 0.9446,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.998216555802526e-06,
+      "loss": 0.9734,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.625,
+      "learning_rate": 4.998116112073401e-06,
+      "loss": 0.9812,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.99801291786274e-06,
+      "loss": 0.9718,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.99790697328416e-06,
+      "loss": 0.9361,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.997798278454311e-06,
+      "loss": 0.9479,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.997686833492866e-06,
+      "loss": 0.9559,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.997572638522531e-06,
+      "loss": 0.9877,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.997455693669036e-06,
+      "loss": 0.9724,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.997335999061138e-06,
+      "loss": 0.9715,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.997213554830627e-06,
+      "loss": 0.9499,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.997088361112314e-06,
+      "loss": 0.9779,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.9969604180440414e-06,
+      "loss": 1.0073,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.996829725766676e-06,
+      "loss": 0.9339,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.996696284424115e-06,
+      "loss": 0.9566,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.996560094163278e-06,
+      "loss": 0.9728,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.9964211551341166e-06,
+      "loss": 0.9584,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.996279467489603e-06,
+      "loss": 0.9692,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.99613503138574e-06,
+      "loss": 0.9662,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.995987846981554e-06,
+      "loss": 0.9427,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5,
+      "learning_rate": 4.995837914439097e-06,
+      "loss": 0.9443,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.995685233923452e-06,
+      "loss": 0.9491,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.99552980560272e-06,
+      "loss": 0.9641,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.9953716296480335e-06,
+      "loss": 0.9572,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.995210706233546e-06,
+      "loss": 0.9752,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.995047035536439e-06,
+      "loss": 0.9664,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.994880617736917e-06,
+      "loss": 0.9298,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.994711453018208e-06,
+      "loss": 0.9727,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.99453954156657e-06,
+      "loss": 0.9432,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.994364883571279e-06,
+      "loss": 0.9561,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.994187479224636e-06,
+      "loss": 0.962,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.9940073287219705e-06,
+      "loss": 0.9644,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.993824432261629e-06,
+      "loss": 0.9744,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.993638790044987e-06,
+      "loss": 0.9822,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.993450402276439e-06,
+      "loss": 0.9704,
+      "step": 79
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5,
+      "learning_rate": 4.993259269163405e-06,
+      "loss": 0.9615,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.9930653909163276e-06,
+      "loss": 0.946,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.992868767748669e-06,
+      "loss": 0.9507,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.375,
+      "learning_rate": 4.992669399876917e-06,
+      "loss": 0.9424,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.625,
+      "learning_rate": 4.99246728752058e-06,
+      "loss": 0.9538,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.992262430902188e-06,
+      "loss": 0.9844,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.992054830247294e-06,
+      "loss": 0.9446,
+      "step": 86
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.991844485784468e-06,
+      "loss": 0.9878,
+      "step": 87
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.991631397745307e-06,
+      "loss": 0.9299,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.991415566364423e-06,
+      "loss": 0.9384,
+      "step": 89
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.991196991879454e-06,
+      "loss": 0.931,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.990975674531053e-06,
+      "loss": 0.9589,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5,
+      "learning_rate": 4.990751614562897e-06,
+      "loss": 0.9807,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.99052481222168e-06,
+      "loss": 0.942,
+      "step": 93
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.990295267757117e-06,
+      "loss": 0.9683,
+      "step": 94
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.990062981421941e-06,
+      "loss": 0.9652,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.625,
+      "learning_rate": 4.989827953471905e-06,
+      "loss": 0.9643,
+      "step": 96
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.625,
+      "learning_rate": 4.989590184165779e-06,
+      "loss": 0.9868,
+      "step": 97
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.989349673765354e-06,
+      "loss": 0.9581,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.989106422535436e-06,
+      "loss": 0.9451,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5,
+      "learning_rate": 4.98886043074385e-06,
+      "loss": 0.9653,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.988611698661439e-06,
+      "loss": 0.959,
+      "step": 101
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.988360226562063e-06,
+      "loss": 0.9543,
+      "step": 102
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.988106014722598e-06,
+      "loss": 0.9911,
+      "step": 103
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.987849063422935e-06,
+      "loss": 0.9606,
+      "step": 104
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.987589372945987e-06,
+      "loss": 0.9378,
+      "step": 105
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.987326943577675e-06,
+      "loss": 0.9392,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.987061775606942e-06,
+      "loss": 0.9475,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.986793869325743e-06,
+      "loss": 0.9331,
+      "step": 108
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.98652322502905e-06,
+      "loss": 0.9634,
+      "step": 109
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.986249843014847e-06,
+      "loss": 0.9308,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.985973723584134e-06,
+      "loss": 1.0091,
+      "step": 111
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.985694867040924e-06,
+      "loss": 0.9875,
+      "step": 112
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.625,
+      "learning_rate": 4.985413273692247e-06,
+      "loss": 0.939,
+      "step": 113
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.98512894384814e-06,
+      "loss": 0.9955,
+      "step": 114
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.984841877821659e-06,
+      "loss": 0.9223,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.984552075928869e-06,
+      "loss": 0.9479,
+      "step": 116
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.984259538488848e-06,
+      "loss": 0.9505,
+      "step": 117
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.983964265823687e-06,
+      "loss": 0.9247,
+      "step": 118
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.983666258258488e-06,
+      "loss": 0.9277,
+      "step": 119
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.983365516121363e-06,
+      "loss": 0.9235,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.983062039743436e-06,
+      "loss": 0.9546,
+      "step": 121
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.9827558294588425e-06,
+      "loss": 0.9584,
+      "step": 122
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.982446885604726e-06,
+      "loss": 0.9023,
+      "step": 123
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.98213520852124e-06,
+      "loss": 0.9454,
+      "step": 124
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.981820798551549e-06,
+      "loss": 0.9794,
+      "step": 125
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.981503656041824e-06,
+      "loss": 0.9458,
+      "step": 126
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.981183781341248e-06,
+      "loss": 0.9751,
+      "step": 127
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.980861174802009e-06,
+      "loss": 0.9478,
+      "step": 128
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.9805358367793025e-06,
+      "loss": 0.9768,
+      "step": 129
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.980207767631335e-06,
+      "loss": 0.9539,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.9798769677193185e-06,
+      "loss": 0.9388,
+      "step": 131
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5,
+      "learning_rate": 4.979543437407469e-06,
+      "loss": 0.9674,
+      "step": 132
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.979207177063011e-06,
+      "loss": 0.9331,
+      "step": 133
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.978868187056176e-06,
+      "loss": 0.9582,
+      "step": 134
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.9785264677602e-06,
+      "loss": 0.9441,
+      "step": 135
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5,
+      "learning_rate": 4.978182019551321e-06,
+      "loss": 0.9655,
+      "step": 136
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977834842808786e-06,
+      "loss": 0.9449,
+      "step": 137
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.977484937914844e-06,
+      "loss": 0.9333,
+      "step": 138
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.9771323052547485e-06,
+      "loss": 0.9374,
+      "step": 139
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.976776945216755e-06,
+      "loss": 0.9386,
+      "step": 140
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.976418858192124e-06,
+      "loss": 0.9615,
+      "step": 141
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.976058044575116e-06,
+      "loss": 0.9246,
+      "step": 142
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.375,
+      "learning_rate": 4.9756945047629955e-06,
+      "loss": 0.9142,
+      "step": 143
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.975328239156029e-06,
+      "loss": 0.9302,
+      "step": 144
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.9749592481574806e-06,
+      "loss": 0.915,
+      "step": 145
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.97458753217362e-06,
+      "loss": 0.9329,
+      "step": 146
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.9742130916137125e-06,
+      "loss": 0.9297,
+      "step": 147
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.973835926890027e-06,
+      "loss": 0.9239,
+      "step": 148
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.973456038417831e-06,
+      "loss": 0.9158,
+      "step": 149
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.973073426615389e-06,
+      "loss": 0.968,
+      "step": 150
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.972688091903965e-06,
+      "loss": 0.9523,
+      "step": 151
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.972300034707822e-06,
+      "loss": 0.9265,
+      "step": 152
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.971909255454219e-06,
+      "loss": 0.9421,
+      "step": 153
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.75,
+      "learning_rate": 4.9715157545734124e-06,
+      "loss": 0.9445,
+      "step": 154
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.971119532498656e-06,
+      "loss": 0.9364,
+      "step": 155
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.970720589666199e-06,
+      "loss": 0.9755,
+      "step": 156
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.9703189265152865e-06,
+      "loss": 0.9627,
+      "step": 157
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.969914543488158e-06,
+      "loss": 0.9484,
+      "step": 158
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.969507441030048e-06,
+      "loss": 0.9873,
+      "step": 159
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.969097619589187e-06,
+      "loss": 0.9709,
+      "step": 160
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.968685079616796e-06,
+      "loss": 0.9415,
+      "step": 161
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.968269821567092e-06,
+      "loss": 0.9673,
+      "step": 162
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.967851845897282e-06,
+      "loss": 0.9395,
+      "step": 163
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.967431153067568e-06,
+      "loss": 0.912,
+      "step": 164
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.96700774354114e-06,
+      "loss": 0.933,
+      "step": 165
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.9665816177841845e-06,
+      "loss": 0.941,
+      "step": 166
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.75,
+      "learning_rate": 4.966152776265872e-06,
+      "loss": 0.929,
+      "step": 167
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.625,
+      "learning_rate": 4.965721219458369e-06,
+      "loss": 0.9401,
+      "step": 168
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.965286947836828e-06,
+      "loss": 0.9366,
+      "step": 169
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.964849961879392e-06,
+      "loss": 0.9318,
+      "step": 170
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.964410262067193e-06,
+      "loss": 0.9184,
+      "step": 171
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.963967848884349e-06,
+      "loss": 0.9049,
+      "step": 172
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.963522722817968e-06,
+      "loss": 0.9519,
+      "step": 173
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.963074884358143e-06,
+      "loss": 0.9106,
+      "step": 174
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.962624333997954e-06,
+      "loss": 0.9483,
+      "step": 175
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.962171072233468e-06,
+      "loss": 0.9023,
+      "step": 176
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.961715099563735e-06,
+      "loss": 0.9313,
+      "step": 177
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.961256416490793e-06,
+      "loss": 0.903,
+      "step": 178
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.960795023519659e-06,
+      "loss": 0.9346,
+      "step": 179
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.96033092115834e-06,
+      "loss": 0.9435,
+      "step": 180
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.375,
+      "learning_rate": 4.959864109917822e-06,
+      "loss": 0.919,
+      "step": 181
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.9593945903120744e-06,
+      "loss": 0.9309,
+      "step": 182
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.958922362858048e-06,
+      "loss": 0.9322,
+      "step": 183
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.9584474280756785e-06,
+      "loss": 0.9052,
+      "step": 184
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.957969786487877e-06,
+      "loss": 0.9712,
+      "step": 185
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.9574894386205384e-06,
+      "loss": 0.9478,
+      "step": 186
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.625,
+      "learning_rate": 4.957006385002536e-06,
+      "loss": 0.9682,
+      "step": 187
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.956520626165724e-06,
+      "loss": 0.9612,
+      "step": 188
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.956032162644932e-06,
+      "loss": 0.9473,
+      "step": 189
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.625,
+      "learning_rate": 4.9555409949779695e-06,
+      "loss": 0.9271,
+      "step": 190
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.955047123705624e-06,
+      "loss": 0.9448,
+      "step": 191
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.954550549371659e-06,
+      "loss": 0.9144,
+      "step": 192
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.9540512725228124e-06,
+      "loss": 0.9308,
+      "step": 193
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.953549293708801e-06,
+      "loss": 0.9361,
+      "step": 194
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.953044613482312e-06,
+      "loss": 0.9268,
+      "step": 195
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.952537232399012e-06,
+      "loss": 0.9317,
+      "step": 196
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.952027151017538e-06,
+      "loss": 0.9393,
+      "step": 197
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.9515143698995006e-06,
+      "loss": 0.9591,
+      "step": 198
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.9509988896094845e-06,
+      "loss": 0.9402,
+      "step": 199
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.950480710715043e-06,
+      "loss": 0.9749,
+      "step": 200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.949959833786705e-06,
+      "loss": 0.945,
+      "step": 201
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.9494362593979665e-06,
+      "loss": 0.9298,
+      "step": 202
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.625,
+      "learning_rate": 4.9489099881252935e-06,
+      "loss": 0.9616,
+      "step": 203
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.9483810205481244e-06,
+      "loss": 0.9294,
+      "step": 204
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5,
+      "learning_rate": 4.947849357248862e-06,
+      "loss": 0.9418,
+      "step": 205
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.947314998812882e-06,
+      "loss": 0.9473,
+      "step": 206
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.946777945828523e-06,
+      "loss": 0.938,
+      "step": 207
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.946238198887093e-06,
+      "loss": 0.9142,
+      "step": 208
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.945695758582865e-06,
+      "loss": 0.9513,
+      "step": 209
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.94515062551308e-06,
+      "loss": 0.9449,
+      "step": 210
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5,
+      "learning_rate": 4.944602800277939e-06,
+      "loss": 0.9239,
+      "step": 211
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.94405228348061e-06,
+      "loss": 0.9238,
+      "step": 212
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.625,
+      "learning_rate": 4.9434990757272255e-06,
+      "loss": 0.9205,
+      "step": 213
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.942943177626879e-06,
+      "loss": 0.9312,
+      "step": 214
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.942384589791627e-06,
+      "loss": 0.9689,
+      "step": 215
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.941823312836487e-06,
+      "loss": 0.9055,
+      "step": 216
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.941259347379437e-06,
+      "loss": 0.9175,
+      "step": 217
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.9406926940414165e-06,
+      "loss": 0.9503,
+      "step": 218
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.940123353446322e-06,
+      "loss": 0.934,
+      "step": 219
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.625,
+      "learning_rate": 4.939551326221012e-06,
+      "loss": 0.9551,
+      "step": 220
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.9389766129953e-06,
+      "loss": 0.9428,
+      "step": 221
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.9383992144019586e-06,
+      "loss": 0.9544,
+      "step": 222
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.937819131076716e-06,
+      "loss": 0.9369,
+      "step": 223
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.937236363658257e-06,
+      "loss": 0.9151,
+      "step": 224
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.936650912788222e-06,
+      "loss": 0.9303,
+      "step": 225
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5,
+      "learning_rate": 4.936062779111205e-06,
+      "loss": 0.9535,
+      "step": 226
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.935471963274755e-06,
+      "loss": 0.9413,
+      "step": 227
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.934878465929372e-06,
+      "loss": 0.9249,
+      "step": 228
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.934282287728511e-06,
+      "loss": 0.9139,
+      "step": 229
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.9336834293285774e-06,
+      "loss": 0.9246,
+      "step": 230
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.933081891388927e-06,
+      "loss": 0.9264,
+      "step": 231
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.932477674571867e-06,
+      "loss": 0.9087,
+      "step": 232
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.931870779542654e-06,
+      "loss": 0.9263,
+      "step": 233
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.625,
+      "learning_rate": 4.931261206969492e-06,
+      "loss": 0.9539,
+      "step": 234
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.9306489575235344e-06,
+      "loss": 0.9408,
+      "step": 235
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.930034031878882e-06,
+      "loss": 0.9176,
+      "step": 236
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.92941643071258e-06,
+      "loss": 0.9464,
+      "step": 237
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.928796154704623e-06,
+      "loss": 0.9155,
+      "step": 238
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.928173204537948e-06,
+      "loss": 0.9032,
+      "step": 239
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.927547580898438e-06,
+      "loss": 0.9316,
+      "step": 240
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.9269192844749156e-06,
+      "loss": 0.9308,
+      "step": 241
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.926288315959152e-06,
+      "loss": 0.9319,
+      "step": 242
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.925654676045857e-06,
+      "loss": 0.9296,
+      "step": 243
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.925018365432681e-06,
+      "loss": 0.9086,
+      "step": 244
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.924379384820218e-06,
+      "loss": 0.9198,
+      "step": 245
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.923737734911999e-06,
+      "loss": 0.9411,
+      "step": 246
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.923093416414495e-06,
+      "loss": 0.9412,
+      "step": 247
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.922446430037115e-06,
+      "loss": 0.9282,
+      "step": 248
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.921796776492205e-06,
+      "loss": 0.9155,
+      "step": 249
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.921144456495048e-06,
+      "loss": 0.9376,
+      "step": 250
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5,
+      "learning_rate": 4.920489470763864e-06,
+      "loss": 0.9182,
+      "step": 251
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.919831820019806e-06,
+      "loss": 0.9316,
+      "step": 252
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.919171504986963e-06,
+      "loss": 0.9456,
+      "step": 253
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.918508526392355e-06,
+      "loss": 0.9165,
+      "step": 254
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.917842884965937e-06,
+      "loss": 0.9365,
+      "step": 255
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.9171745814405945e-06,
+      "loss": 0.9493,
+      "step": 256
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.375,
+      "learning_rate": 4.916503616552146e-06,
+      "loss": 0.9179,
+      "step": 257
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.915829991039338e-06,
+      "loss": 0.9334,
+      "step": 258
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.9151537056438484e-06,
+      "loss": 0.952,
+      "step": 259
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.91447476111028e-06,
+      "loss": 0.9441,
+      "step": 260
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.625,
+      "learning_rate": 4.913793158186167e-06,
+      "loss": 0.92,
+      "step": 261
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.9131088976219695e-06,
+      "loss": 0.9439,
+      "step": 262
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.912421980171075e-06,
+      "loss": 0.9368,
+      "step": 263
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.911732406589792e-06,
+      "loss": 0.9176,
+      "step": 264
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.91104017763736e-06,
+      "loss": 0.9065,
+      "step": 265
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5,
+      "learning_rate": 4.910345294075935e-06,
+      "loss": 0.9171,
+      "step": 266
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.9096477566706015e-06,
+      "loss": 0.9234,
+      "step": 267
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.908947566189362e-06,
+      "loss": 0.8894,
+      "step": 268
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.908244723403145e-06,
+      "loss": 0.9232,
+      "step": 269
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.9075392290857925e-06,
+      "loss": 0.9025,
+      "step": 270
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.9068310840140725e-06,
+      "loss": 0.9252,
+      "step": 271
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5,
+      "learning_rate": 4.906120288967665e-06,
+      "loss": 0.9382,
+      "step": 272
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.905406844729175e-06,
+      "loss": 0.9253,
+      "step": 273
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5,
+      "learning_rate": 4.904690752084117e-06,
+      "loss": 0.9071,
+      "step": 274
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.903972011820928e-06,
+      "loss": 0.9108,
+      "step": 275
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.9032506247309545e-06,
+      "loss": 0.9391,
+      "step": 276
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.9025265916084615e-06,
+      "loss": 0.9208,
+      "step": 277
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.9017999132506254e-06,
+      "loss": 0.9219,
+      "step": 278
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.901070590457535e-06,
+      "loss": 0.8975,
+      "step": 279
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.900338624032191e-06,
+      "loss": 0.9385,
+      "step": 280
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.8996040147805045e-06,
+      "loss": 0.9384,
+      "step": 281
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.898866763511297e-06,
+      "loss": 0.8997,
+      "step": 282
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.898126871036298e-06,
+      "loss": 0.9267,
+      "step": 283
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.8973843381701465e-06,
+      "loss": 0.9006,
+      "step": 284
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5,
+      "learning_rate": 4.8966391657303875e-06,
+      "loss": 0.9391,
+      "step": 285
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.895891354537472e-06,
+      "loss": 0.9408,
+      "step": 286
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.895140905414757e-06,
+      "loss": 0.9378,
+      "step": 287
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5,
+      "learning_rate": 4.894387819188504e-06,
+      "loss": 0.9567,
+      "step": 288
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.893632096687877e-06,
+      "loss": 0.958,
+      "step": 289
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.892873738744944e-06,
+      "loss": 0.9024,
+      "step": 290
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.625,
+      "learning_rate": 4.892112746194674e-06,
+      "loss": 0.9276,
+      "step": 291
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.891349119874936e-06,
+      "loss": 0.9269,
+      "step": 292
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.890582860626501e-06,
+      "loss": 0.9452,
+      "step": 293
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.8898139692930365e-06,
+      "loss": 0.9617,
+      "step": 294
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.889042446721109e-06,
+      "loss": 0.9369,
+      "step": 295
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.888268293760182e-06,
+      "loss": 0.911,
+      "step": 296
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.887491511262616e-06,
+      "loss": 0.9277,
+      "step": 297
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.886712100083664e-06,
+      "loss": 0.9266,
+      "step": 298
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.885930061081478e-06,
+      "loss": 0.9333,
+      "step": 299
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.885145395117099e-06,
+      "loss": 0.9078,
+      "step": 300
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.88435810305446e-06,
+      "loss": 0.9177,
+      "step": 301
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.88356818576039e-06,
+      "loss": 0.9046,
+      "step": 302
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.882775644104604e-06,
+      "loss": 0.9371,
+      "step": 303
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.881980478959707e-06,
+      "loss": 0.9134,
+      "step": 304
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.881182691201195e-06,
+      "loss": 0.9257,
+      "step": 305
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.880382281707451e-06,
+      "loss": 0.9289,
+      "step": 306
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.879579251359741e-06,
+      "loss": 0.9026,
+      "step": 307
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.878773601042221e-06,
+      "loss": 0.8994,
+      "step": 308
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.877965331641931e-06,
+      "loss": 0.9357,
+      "step": 309
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.877154444048792e-06,
+      "loss": 0.9267,
+      "step": 310
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.876340939155609e-06,
+      "loss": 0.9506,
+      "step": 311
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.8755248178580705e-06,
+      "loss": 0.9568,
+      "step": 312
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.874706081054743e-06,
+      "loss": 0.9095,
+      "step": 313
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.873884729647075e-06,
+      "loss": 0.9262,
+      "step": 314
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.873060764539393e-06,
+      "loss": 0.9181,
+      "step": 315
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.872234186638898e-06,
+      "loss": 0.9279,
+      "step": 316
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.871404996855674e-06,
+      "loss": 0.9206,
+      "step": 317
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.870573196102677e-06,
+      "loss": 0.9132,
+      "step": 318
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.869738785295737e-06,
+      "loss": 0.9195,
+      "step": 319
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.868901765353559e-06,
+      "loss": 0.9508,
+      "step": 320
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.8680621371977214e-06,
+      "loss": 0.9327,
+      "step": 321
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.8672199017526725e-06,
+      "loss": 0.8962,
+      "step": 322
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.866375059945733e-06,
+      "loss": 0.915,
+      "step": 323
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.865527612707094e-06,
+      "loss": 0.9145,
+      "step": 324
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.8646775609698105e-06,
+      "loss": 0.9195,
+      "step": 325
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.863824905669811e-06,
+      "loss": 0.9137,
+      "step": 326
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.8629696477458874e-06,
+      "loss": 0.9165,
+      "step": 327
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.375,
+      "learning_rate": 4.862111788139697e-06,
+      "loss": 0.9248,
+      "step": 328
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.8612513277957626e-06,
+      "loss": 0.9469,
+      "step": 329
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.860388267661472e-06,
+      "loss": 0.9309,
+      "step": 330
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.859522608687071e-06,
+      "loss": 0.907,
+      "step": 331
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5,
+      "learning_rate": 4.8586543518256715e-06,
+      "loss": 0.9261,
+      "step": 332
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.625,
+      "learning_rate": 4.857783498033241e-06,
+      "loss": 0.9309,
+      "step": 333
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.856910048268613e-06,
+      "loss": 0.917,
+      "step": 334
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.856034003493471e-06,
+      "loss": 0.9196,
+      "step": 335
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.855155364672363e-06,
+      "loss": 0.9213,
+      "step": 336
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5,
+      "learning_rate": 4.854274132772688e-06,
+      "loss": 0.9183,
+      "step": 337
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5,
+      "learning_rate": 4.853390308764703e-06,
+      "loss": 0.9113,
+      "step": 338
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.8525038936215165e-06,
+      "loss": 0.9596,
+      "step": 339
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.851614888319093e-06,
+      "loss": 0.9302,
+      "step": 340
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.850723293836247e-06,
+      "loss": 0.9169,
+      "step": 341
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.849829111154643e-06,
+      "loss": 0.933,
+      "step": 342
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.848932341258796e-06,
+      "loss": 0.9312,
+      "step": 343
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.848032985136071e-06,
+      "loss": 0.9061,
+      "step": 344
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.847131043776678e-06,
+      "loss": 0.916,
+      "step": 345
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.846226518173676e-06,
+      "loss": 0.9375,
+      "step": 346
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.625,
+      "learning_rate": 4.845319409322965e-06,
+      "loss": 0.917,
+      "step": 347
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.8444097182232944e-06,
+      "loss": 0.9018,
+      "step": 348
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.843497445876254e-06,
+      "loss": 0.9017,
+      "step": 349
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.842582593286275e-06,
+      "loss": 0.9268,
+      "step": 350
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.841665161460632e-06,
+      "loss": 0.9069,
+      "step": 351
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.840745151409437e-06,
+      "loss": 0.896,
+      "step": 352
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.839822564145642e-06,
+      "loss": 0.9167,
+      "step": 353
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.838897400685036e-06,
+      "loss": 0.9099,
+      "step": 354
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.837969662046243e-06,
+      "loss": 0.9308,
+      "step": 355
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.837039349250726e-06,
+      "loss": 0.9106,
+      "step": 356
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.836106463322781e-06,
+      "loss": 0.887,
+      "step": 357
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.835171005289533e-06,
+      "loss": 0.9053,
+      "step": 358
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.8342329761809444e-06,
+      "loss": 0.9342,
+      "step": 359
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.833292377029806e-06,
+      "loss": 0.9259,
+      "step": 360
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.832349208871738e-06,
+      "loss": 0.9105,
+      "step": 361
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.83140347274519e-06,
+      "loss": 0.9321,
+      "step": 362
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.830455169691437e-06,
+      "loss": 0.8922,
+      "step": 363
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.8295043007545836e-06,
+      "loss": 0.9138,
+      "step": 364
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.828550866981557e-06,
+      "loss": 0.9181,
+      "step": 365
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.625,
+      "learning_rate": 4.82759486942211e-06,
+      "loss": 0.9139,
+      "step": 366
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.826636309128816e-06,
+      "loss": 0.9029,
+      "step": 367
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5,
+      "learning_rate": 4.825675187157071e-06,
+      "loss": 0.8999,
+      "step": 368
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.824711504565093e-06,
+      "loss": 0.9349,
+      "step": 369
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.823745262413917e-06,
+      "loss": 0.9283,
+      "step": 370
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.822776461767398e-06,
+      "loss": 0.9031,
+      "step": 371
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.821805103692206e-06,
+      "loss": 0.9443,
+      "step": 372
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.82083118925783e-06,
+      "loss": 0.9348,
+      "step": 373
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.819854719536568e-06,
+      "loss": 0.9355,
+      "step": 374
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.818875695603539e-06,
+      "loss": 0.9098,
+      "step": 375
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.817894118536667e-06,
+      "loss": 0.881,
+      "step": 376
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.816909989416691e-06,
+      "loss": 0.8945,
+      "step": 377
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.81592330932716e-06,
+      "loss": 0.9028,
+      "step": 378
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.814934079354431e-06,
+      "loss": 0.9018,
+      "step": 379
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5,
+      "learning_rate": 4.813942300587666e-06,
+      "loss": 0.9128,
+      "step": 380
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.812947974118839e-06,
+      "loss": 0.9131,
+      "step": 381
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.811951101042722e-06,
+      "loss": 0.9352,
+      "step": 382
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.625,
+      "learning_rate": 4.810951682456897e-06,
+      "loss": 0.9102,
+      "step": 383
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.625,
+      "learning_rate": 4.809949719461745e-06,
+      "loss": 0.9172,
+      "step": 384
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.8089452131604495e-06,
+      "loss": 0.9358,
+      "step": 385
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.807938164658996e-06,
+      "loss": 0.8908,
+      "step": 386
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.806928575066165e-06,
+      "loss": 0.9267,
+      "step": 387
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.805916445493538e-06,
+      "loss": 0.9408,
+      "step": 388
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.804901777055494e-06,
+      "loss": 0.9034,
+      "step": 389
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.803884570869203e-06,
+      "loss": 0.9193,
+      "step": 390
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5,
+      "learning_rate": 4.802864828054633e-06,
+      "loss": 0.929,
+      "step": 391
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.801842549734542e-06,
+      "loss": 0.9027,
+      "step": 392
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.8008177370344845e-06,
+      "loss": 0.9434,
+      "step": 393
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.799790391082799e-06,
+      "loss": 0.9008,
+      "step": 394
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.7987605130106175e-06,
+      "loss": 0.8791,
+      "step": 395
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.797728103951859e-06,
+      "loss": 0.8861,
+      "step": 396
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.796693165043229e-06,
+      "loss": 0.9106,
+      "step": 397
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.795655697424217e-06,
+      "loss": 0.9297,
+      "step": 398
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.7946157022371e-06,
+      "loss": 0.9272,
+      "step": 399
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.793573180626934e-06,
+      "loss": 0.9147,
+      "step": 400
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.7925281337415584e-06,
+      "loss": 0.9053,
+      "step": 401
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.791480562731595e-06,
+      "loss": 0.9212,
+      "step": 402
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.790430468750442e-06,
+      "loss": 0.8995,
+      "step": 403
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.789377852954276e-06,
+      "loss": 0.9043,
+      "step": 404
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.788322716502051e-06,
+      "loss": 0.9284,
+      "step": 405
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.787265060555495e-06,
+      "loss": 0.8956,
+      "step": 406
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.78620488627911e-06,
+      "loss": 0.8896,
+      "step": 407
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.785142194840174e-06,
+      "loss": 0.9349,
+      "step": 408
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5,
+      "learning_rate": 4.7840769874087325e-06,
+      "loss": 0.934,
+      "step": 409
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.625,
+      "learning_rate": 4.783009265157602e-06,
+      "loss": 0.9395,
+      "step": 410
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.78193902926237e-06,
+      "loss": 0.9383,
+      "step": 411
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.7808662809013895e-06,
+      "loss": 0.8927,
+      "step": 412
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.779791021255779e-06,
+      "loss": 0.927,
+      "step": 413
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.7787132515094255e-06,
+      "loss": 0.8992,
+      "step": 414
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.625,
+      "learning_rate": 4.7776329728489764e-06,
+      "loss": 0.9315,
+      "step": 415
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.776550186463843e-06,
+      "loss": 0.9187,
+      "step": 416
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.775464893546196e-06,
+      "loss": 0.906,
+      "step": 417
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.774377095290969e-06,
+      "loss": 0.9046,
+      "step": 418
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.773286792895852e-06,
+      "loss": 0.8962,
+      "step": 419
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.772193987561291e-06,
+      "loss": 0.889,
+      "step": 420
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.771098680490489e-06,
+      "loss": 0.9171,
+      "step": 421
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.770000872889404e-06,
+      "loss": 0.9073,
+      "step": 422
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.768900565966746e-06,
+      "loss": 0.9485,
+      "step": 423
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.76779776093398e-06,
+      "loss": 0.9146,
+      "step": 424
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.7666924590053154e-06,
+      "loss": 0.9302,
+      "step": 425
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.765584661397717e-06,
+      "loss": 0.9182,
+      "step": 426
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.764474369330893e-06,
+      "loss": 0.9143,
+      "step": 427
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.7633615840273e-06,
+      "loss": 0.8779,
+      "step": 428
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.76224630671214e-06,
+      "loss": 0.9096,
+      "step": 429
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.761128538613359e-06,
+      "loss": 0.9154,
+      "step": 430
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.760008280961642e-06,
+      "loss": 0.9437,
+      "step": 431
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.625,
+      "learning_rate": 4.758885534990419e-06,
+      "loss": 0.8816,
+      "step": 432
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.75776030193586e-06,
+      "loss": 0.9069,
+      "step": 433
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.75663258303687e-06,
+      "loss": 0.9146,
+      "step": 434
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.755502379535092e-06,
+      "loss": 0.9347,
+      "step": 435
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.754369692674906e-06,
+      "loss": 0.9009,
+      "step": 436
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.753234523703426e-06,
+      "loss": 0.9114,
+      "step": 437
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.752096873870499e-06,
+      "loss": 0.879,
+      "step": 438
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.7509567444287e-06,
+      "loss": 0.8999,
+      "step": 439
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7498141366333395e-06,
+      "loss": 0.9447,
+      "step": 440
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.748669051742454e-06,
+      "loss": 0.8989,
+      "step": 441
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.625,
+      "learning_rate": 4.747521491016805e-06,
+      "loss": 0.8934,
+      "step": 442
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.746371455719886e-06,
+      "loss": 0.9109,
+      "step": 443
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.74521894711791e-06,
+      "loss": 0.8961,
+      "step": 444
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.744063966479813e-06,
+      "loss": 0.9153,
+      "step": 445
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.742906515077257e-06,
+      "loss": 0.9015,
+      "step": 446
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.74174659418462e-06,
+      "loss": 0.8945,
+      "step": 447
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5,
+      "learning_rate": 4.740584205079002e-06,
+      "loss": 0.898,
+      "step": 448
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.7394193490402185e-06,
+      "loss": 0.8939,
+      "step": 449
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.375,
+      "learning_rate": 4.738252027350803e-06,
+      "loss": 0.9275,
+      "step": 450
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.875,
+      "learning_rate": 4.737082241296001e-06,
+      "loss": 0.8992,
+      "step": 451
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.735909992163774e-06,
+      "loss": 0.9385,
+      "step": 452
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.734735281244795e-06,
+      "loss": 0.933,
+      "step": 453
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.7335581098324465e-06,
+      "loss": 0.9252,
+      "step": 454
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.73237847922282e-06,
+      "loss": 0.9365,
+      "step": 455
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.731196390714715e-06,
+      "loss": 0.8771,
+      "step": 456
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.7300118456096385e-06,
+      "loss": 0.9012,
+      "step": 457
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.7288248452118e-06,
+      "loss": 0.9167,
+      "step": 458
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.727635390828114e-06,
+      "loss": 0.9254,
+      "step": 459
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.726443483768195e-06,
+      "loss": 0.9191,
+      "step": 460
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.725249125344361e-06,
+      "loss": 0.9075,
+      "step": 461
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.375,
+      "learning_rate": 4.724052316871626e-06,
+      "loss": 0.9084,
+      "step": 462
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.671875,
+      "learning_rate": 4.722853059667704e-06,
+      "loss": 0.893,
+      "step": 463
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.721651355053003e-06,
+      "loss": 0.9535,
+      "step": 464
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.720447204350627e-06,
+      "loss": 0.9241,
+      "step": 465
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.719240608886372e-06,
+      "loss": 0.9065,
+      "step": 466
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.718031569988727e-06,
+      "loss": 0.9284,
+      "step": 467
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.716820088988871e-06,
+      "loss": 0.8855,
+      "step": 468
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.715606167220672e-06,
+      "loss": 0.9286,
+      "step": 469
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.7143898060206835e-06,
+      "loss": 0.8848,
+      "step": 470
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.713171006728148e-06,
+      "loss": 0.8955,
+      "step": 471
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.625,
+      "learning_rate": 4.711949770684989e-06,
+      "loss": 0.9493,
+      "step": 472
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.710726099235816e-06,
+      "loss": 0.8894,
+      "step": 473
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.709499993727919e-06,
+      "loss": 0.8956,
+      "step": 474
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.7082714555112655e-06,
+      "loss": 0.9407,
+      "step": 475
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.875,
+      "learning_rate": 4.707040485938503e-06,
+      "loss": 0.9198,
+      "step": 476
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.705807086364961e-06,
+      "loss": 0.9193,
+      "step": 477
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.704571258148634e-06,
+      "loss": 0.9127,
+      "step": 478
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.7033330026502015e-06,
+      "loss": 0.8927,
+      "step": 479
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.702092321233008e-06,
+      "loss": 0.8932,
+      "step": 480
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.700849215263071e-06,
+      "loss": 0.8915,
+      "step": 481
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.699603686109078e-06,
+      "loss": 0.9166,
+      "step": 482
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.698355735142386e-06,
+      "loss": 0.8853,
+      "step": 483
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.697105363737015e-06,
+      "loss": 0.9071,
+      "step": 484
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.695852573269652e-06,
+      "loss": 0.9292,
+      "step": 485
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.625,
+      "learning_rate": 4.694597365119649e-06,
+      "loss": 0.9058,
+      "step": 486
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.6933397406690165e-06,
+      "loss": 0.8937,
+      "step": 487
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.692079701302427e-06,
+      "loss": 0.9029,
+      "step": 488
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.690817248407213e-06,
+      "loss": 0.9258,
+      "step": 489
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.689552383373362e-06,
+      "loss": 0.9275,
+      "step": 490
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.688285107593519e-06,
+      "loss": 0.8913,
+      "step": 491
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.6870154224629836e-06,
+      "loss": 0.9133,
+      "step": 492
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.685743329379708e-06,
+      "loss": 0.8976,
+      "step": 493
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.684468829744293e-06,
+      "loss": 0.8927,
+      "step": 494
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.683191924959992e-06,
+      "loss": 0.8796,
+      "step": 495
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.681912616432707e-06,
+      "loss": 0.9008,
+      "step": 496
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.680630905570983e-06,
+      "loss": 0.8976,
+      "step": 497
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.679346793786014e-06,
+      "loss": 0.8863,
+      "step": 498
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.678060282491633e-06,
+      "loss": 0.9357,
+      "step": 499
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.6767713731043215e-06,
+      "loss": 0.9006,
+      "step": 500
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.6754800670431955e-06,
+      "loss": 0.9202,
+      "step": 501
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.674186365730012e-06,
+      "loss": 0.9398,
+      "step": 502
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.672890270589163e-06,
+      "loss": 0.9239,
+      "step": 503
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.671591783047682e-06,
+      "loss": 0.8904,
+      "step": 504
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.67029090453523e-06,
+      "loss": 0.8968,
+      "step": 505
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.668987636484103e-06,
+      "loss": 0.8959,
+      "step": 506
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.66768198032923e-06,
+      "loss": 0.8865,
+      "step": 507
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.666373937508166e-06,
+      "loss": 0.9079,
+      "step": 508
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.665063509461098e-06,
+      "loss": 0.9102,
+      "step": 509
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.663750697630832e-06,
+      "loss": 0.9287,
+      "step": 510
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.662435503462806e-06,
+      "loss": 0.8943,
+      "step": 511
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.661117928405079e-06,
+      "loss": 0.9043,
+      "step": 512
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.659797973908328e-06,
+      "loss": 0.9049,
+      "step": 513
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.658475641425854e-06,
+      "loss": 0.9117,
+      "step": 514
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.6571509324135735e-06,
+      "loss": 0.9125,
+      "step": 515
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.6558238483300215e-06,
+      "loss": 0.8802,
+      "step": 516
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5,
+      "learning_rate": 4.654494390636347e-06,
+      "loss": 0.9083,
+      "step": 517
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.653162560796313e-06,
+      "loss": 0.9107,
+      "step": 518
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.6518283602762925e-06,
+      "loss": 0.8955,
+      "step": 519
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.6504917905452705e-06,
+      "loss": 0.8993,
+      "step": 520
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.649152853074841e-06,
+      "loss": 0.9319,
+      "step": 521
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.647811549339203e-06,
+      "loss": 0.9099,
+      "step": 522
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.646467880815163e-06,
+      "loss": 0.9443,
+      "step": 523
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.645121848982128e-06,
+      "loss": 0.9181,
+      "step": 524
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.643773455322112e-06,
+      "loss": 0.9079,
+      "step": 525
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.6424227013197235e-06,
+      "loss": 0.8995,
+      "step": 526
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.641069588462174e-06,
+      "loss": 0.884,
+      "step": 527
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.639714118239271e-06,
+      "loss": 0.9278,
+      "step": 528
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.75,
+      "learning_rate": 4.638356292143417e-06,
+      "loss": 0.906,
+      "step": 529
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.636996111669609e-06,
+      "loss": 0.9004,
+      "step": 530
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.635633578315437e-06,
+      "loss": 0.9069,
+      "step": 531
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6342686935810795e-06,
+      "loss": 0.8993,
+      "step": 532
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.625,
+      "learning_rate": 4.632901458969306e-06,
+      "loss": 0.9308,
+      "step": 533
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.631531875985471e-06,
+      "loss": 0.9474,
+      "step": 534
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.630159946137517e-06,
+      "loss": 0.9289,
+      "step": 535
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.628785670935968e-06,
+      "loss": 0.8797,
+      "step": 536
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5,
+      "learning_rate": 4.627409051893934e-06,
+      "loss": 0.9047,
+      "step": 537
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5,
+      "learning_rate": 4.6260300905271e-06,
+      "loss": 0.9052,
+      "step": 538
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.624648788353736e-06,
+      "loss": 0.9036,
+      "step": 539
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.623265146894685e-06,
+      "loss": 0.9168,
+      "step": 540
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.6218791676733676e-06,
+      "loss": 0.8978,
+      "step": 541
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.620490852215779e-06,
+      "loss": 0.9158,
+      "step": 542
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5,
+      "learning_rate": 4.619100202050482e-06,
+      "loss": 0.8922,
+      "step": 543
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.617707218708617e-06,
+      "loss": 0.8691,
+      "step": 544
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.6163119037238894e-06,
+      "loss": 0.9018,
+      "step": 545
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.6149142586325705e-06,
+      "loss": 0.9409,
+      "step": 546
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.613514284973498e-06,
+      "loss": 0.908,
+      "step": 547
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.6121119842880765e-06,
+      "loss": 0.929,
+      "step": 548
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.610707358120269e-06,
+      "loss": 0.9262,
+      "step": 549
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.6093004080166e-06,
+      "loss": 0.936,
+      "step": 550
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.607891135526155e-06,
+      "loss": 0.9007,
+      "step": 551
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.606479542200572e-06,
+      "loss": 0.9127,
+      "step": 552
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5,
+      "learning_rate": 4.605065629594048e-06,
+      "loss": 0.8838,
+      "step": 553
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.603649399263331e-06,
+      "loss": 0.9013,
+      "step": 554
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.602230852767724e-06,
+      "loss": 0.9162,
+      "step": 555
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.600809991669076e-06,
+      "loss": 0.8751,
+      "step": 556
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.599386817531789e-06,
+      "loss": 0.8815,
+      "step": 557
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.597961331922809e-06,
+      "loss": 0.8804,
+      "step": 558
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.375,
+      "learning_rate": 4.596533536411626e-06,
+      "loss": 0.8846,
+      "step": 559
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.595103432570277e-06,
+      "loss": 0.876,
+      "step": 560
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.593671021973336e-06,
+      "loss": 0.8982,
+      "step": 561
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.59223630619792e-06,
+      "loss": 0.9138,
+      "step": 562
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.590799286823683e-06,
+      "loss": 0.9065,
+      "step": 563
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5,
+      "learning_rate": 4.589359965432816e-06,
+      "loss": 0.8892,
+      "step": 564
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.587918343610042e-06,
+      "loss": 0.9372,
+      "step": 565
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.58647442294262e-06,
+      "loss": 0.8768,
+      "step": 566
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.585028205020341e-06,
+      "loss": 0.8886,
+      "step": 567
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.5835796914355195e-06,
+      "loss": 0.9188,
+      "step": 568
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.582128883783004e-06,
+      "loss": 0.8921,
+      "step": 569
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.580675783660163e-06,
+      "loss": 0.9031,
+      "step": 570
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.579220392666896e-06,
+      "loss": 0.8932,
+      "step": 571
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.577762712405618e-06,
+      "loss": 0.9398,
+      "step": 572
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.576302744481268e-06,
+      "loss": 0.9324,
+      "step": 573
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.5748404905013045e-06,
+      "loss": 0.8908,
+      "step": 574
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.5733759520757e-06,
+      "loss": 0.9067,
+      "step": 575
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.5719091308169464e-06,
+      "loss": 0.8771,
+      "step": 576
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.570440028340044e-06,
+      "loss": 0.8843,
+      "step": 577
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.568968646262509e-06,
+      "loss": 0.8934,
+      "step": 578
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.567494986204365e-06,
+      "loss": 0.945,
+      "step": 579
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.5660190497881455e-06,
+      "loss": 0.8931,
+      "step": 580
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.5645408386388875e-06,
+      "loss": 0.9008,
+      "step": 581
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.563060354384137e-06,
+      "loss": 0.8965,
+      "step": 582
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.561577598653938e-06,
+      "loss": 0.8701,
+      "step": 583
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.560092573080839e-06,
+      "loss": 0.8996,
+      "step": 584
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.558605279299886e-06,
+      "loss": 0.8921,
+      "step": 585
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.375,
+      "learning_rate": 4.557115718948622e-06,
+      "loss": 0.8921,
+      "step": 586
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.555623893667087e-06,
+      "loss": 0.892,
+      "step": 587
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.375,
+      "learning_rate": 4.554129805097813e-06,
+      "loss": 0.9122,
+      "step": 588
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.552633454885826e-06,
+      "loss": 0.9063,
+      "step": 589
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.55113484467864e-06,
+      "loss": 0.9161,
+      "step": 590
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.375,
+      "learning_rate": 4.549633976126259e-06,
+      "loss": 0.8789,
+      "step": 591
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.548130850881171e-06,
+      "loss": 0.883,
+      "step": 592
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.546625470598353e-06,
+      "loss": 0.8527,
+      "step": 593
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.54511783693526e-06,
+      "loss": 0.9305,
+      "step": 594
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.5436079515518315e-06,
+      "loss": 0.9114,
+      "step": 595
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.542095816110485e-06,
+      "loss": 0.9278,
+      "step": 596
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.540581432276114e-06,
+      "loss": 0.8865,
+      "step": 597
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.53906480171609e-06,
+      "loss": 0.8821,
+      "step": 598
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.5375459261002575e-06,
+      "loss": 0.856,
+      "step": 599
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.536024807100932e-06,
+      "loss": 0.9159,
+      "step": 600
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.534501446392898e-06,
+      "loss": 0.8963,
+      "step": 601
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.532975845653412e-06,
+      "loss": 0.8925,
+      "step": 602
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.375,
+      "learning_rate": 4.531448006562194e-06,
+      "loss": 0.9035,
+      "step": 603
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.529917930801427e-06,
+      "loss": 0.9121,
+      "step": 604
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.52838562005576e-06,
+      "loss": 0.9022,
+      "step": 605
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.5268510760123e-06,
+      "loss": 0.8698,
+      "step": 606
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.525314300360617e-06,
+      "loss": 0.9267,
+      "step": 607
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5,
+      "learning_rate": 4.523775294792732e-06,
+      "loss": 0.916,
+      "step": 608
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.5222340610031275e-06,
+      "loss": 0.8951,
+      "step": 609
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.520690600688734e-06,
+      "loss": 0.9054,
+      "step": 610
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.519144915548937e-06,
+      "loss": 0.8976,
+      "step": 611
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.517597007285568e-06,
+      "loss": 0.9012,
+      "step": 612
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.5160468776029114e-06,
+      "loss": 0.9214,
+      "step": 613
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.5144945282076935e-06,
+      "loss": 0.907,
+      "step": 614
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.512939960809085e-06,
+      "loss": 0.9226,
+      "step": 615
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.5113831771187e-06,
+      "loss": 0.9008,
+      "step": 616
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.75,
+      "learning_rate": 4.509824178850591e-06,
+      "loss": 0.8953,
+      "step": 617
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.508262967721252e-06,
+      "loss": 0.8876,
+      "step": 618
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.506699545449608e-06,
+      "loss": 0.9004,
+      "step": 619
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.505133913757026e-06,
+      "loss": 0.9125,
+      "step": 620
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.503566074367298e-06,
+      "loss": 0.9274,
+      "step": 621
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.501996029006651e-06,
+      "loss": 0.911,
+      "step": 622
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.50042377940374e-06,
+      "loss": 0.9026,
+      "step": 623
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.498849327289647e-06,
+      "loss": 0.9238,
+      "step": 624
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.497272674397879e-06,
+      "loss": 0.8924,
+      "step": 625
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.495693822464366e-06,
+      "loss": 0.9082,
+      "step": 626
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.49411277322746e-06,
+      "loss": 0.9023,
+      "step": 627
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.492529528427929e-06,
+      "loss": 0.9072,
+      "step": 628
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.490944089808964e-06,
+      "loss": 0.8998,
+      "step": 629
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.489356459116165e-06,
+      "loss": 0.9152,
+      "step": 630
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.487766638097551e-06,
+      "loss": 0.8784,
+      "step": 631
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.48617462850355e-06,
+      "loss": 0.8893,
+      "step": 632
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.484580432086999e-06,
+      "loss": 0.8815,
+      "step": 633
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5,
+      "learning_rate": 4.4829840506031455e-06,
+      "loss": 0.8862,
+      "step": 634
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.48138548580964e-06,
+      "loss": 0.9293,
+      "step": 635
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.479784739466539e-06,
+      "loss": 0.9128,
+      "step": 636
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.478181813336297e-06,
+      "loss": 0.8722,
+      "step": 637
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.375,
+      "learning_rate": 4.476576709183775e-06,
+      "loss": 0.9293,
+      "step": 638
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.474969428776228e-06,
+      "loss": 0.9013,
+      "step": 639
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.473359973883305e-06,
+      "loss": 0.929,
+      "step": 640
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.471748346277054e-06,
+      "loss": 0.8905,
+      "step": 641
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.470134547731912e-06,
+      "loss": 0.8828,
+      "step": 642
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.468518580024708e-06,
+      "loss": 0.9458,
+      "step": 643
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.466900444934657e-06,
+      "loss": 0.9242,
+      "step": 644
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.465280144243363e-06,
+      "loss": 0.8903,
+      "step": 645
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.463657679734813e-06,
+      "loss": 0.8833,
+      "step": 646
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.462033053195376e-06,
+      "loss": 0.908,
+      "step": 647
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.460406266413803e-06,
+      "loss": 0.8984,
+      "step": 648
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.458777321181222e-06,
+      "loss": 0.9178,
+      "step": 649
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.457146219291137e-06,
+      "loss": 0.8878,
+      "step": 650
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.45551296253943e-06,
+      "loss": 0.9174,
+      "step": 651
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.453877552724352e-06,
+      "loss": 0.9021,
+      "step": 652
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.452239991646524e-06,
+      "loss": 0.8848,
+      "step": 653
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.450600281108939e-06,
+      "loss": 0.883,
+      "step": 654
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5,
+      "learning_rate": 4.448958422916955e-06,
+      "loss": 0.9285,
+      "step": 655
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.447314418878294e-06,
+      "loss": 0.9376,
+      "step": 656
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.4456682708030405e-06,
+      "loss": 0.9148,
+      "step": 657
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.444019980503641e-06,
+      "loss": 0.9486,
+      "step": 658
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.4423695497949e-06,
+      "loss": 0.8893,
+      "step": 659
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.440716980493978e-06,
+      "loss": 0.9185,
+      "step": 660
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.439062274420391e-06,
+      "loss": 0.9164,
+      "step": 661
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.437405433396006e-06,
+      "loss": 0.9074,
+      "step": 662
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.435746459245044e-06,
+      "loss": 0.9056,
+      "step": 663
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5,
+      "learning_rate": 4.4340853537940715e-06,
+      "loss": 0.919,
+      "step": 664
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5,
+      "learning_rate": 4.432422118872003e-06,
+      "loss": 0.9421,
+      "step": 665
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.430756756310096e-06,
+      "loss": 0.8764,
+      "step": 666
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.429089267941953e-06,
+      "loss": 0.8854,
+      "step": 667
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.427419655603515e-06,
+      "loss": 0.9039,
+      "step": 668
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.425747921133064e-06,
+      "loss": 0.8807,
+      "step": 669
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.424074066371216e-06,
+      "loss": 0.8806,
+      "step": 670
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.422398093160921e-06,
+      "loss": 0.896,
+      "step": 671
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.420720003347465e-06,
+      "loss": 0.9087,
+      "step": 672
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.419039798778463e-06,
+      "loss": 0.8918,
+      "step": 673
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.417357481303854e-06,
+      "loss": 0.8981,
+      "step": 674
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5,
+      "learning_rate": 4.41567305277591e-06,
+      "loss": 0.89,
+      "step": 675
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.4139865150492235e-06,
+      "loss": 0.9133,
+      "step": 676
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.412297869980711e-06,
+      "loss": 0.905,
+      "step": 677
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.4106071194296065e-06,
+      "loss": 0.9136,
+      "step": 678
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.408914265257466e-06,
+      "loss": 0.9163,
+      "step": 679
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.407219309328159e-06,
+      "loss": 0.8835,
+      "step": 680
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.405522253507869e-06,
+      "loss": 0.883,
+      "step": 681
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.403823099665093e-06,
+      "loss": 0.9026,
+      "step": 682
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.402121849670637e-06,
+      "loss": 0.8959,
+      "step": 683
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.400418505397616e-06,
+      "loss": 0.8907,
+      "step": 684
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.398713068721449e-06,
+      "loss": 0.9037,
+      "step": 685
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.39700554151986e-06,
+      "loss": 0.8921,
+      "step": 686
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.375,
+      "learning_rate": 4.395295925672874e-06,
+      "loss": 0.9063,
+      "step": 687
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.393584223062819e-06,
+      "loss": 0.8844,
+      "step": 688
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.3918704355743144e-06,
+      "loss": 0.8801,
+      "step": 689
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.375,
+      "learning_rate": 4.390154565094279e-06,
+      "loss": 0.8886,
+      "step": 690
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.388436613511926e-06,
+      "loss": 0.8975,
+      "step": 691
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.3867165827187575e-06,
+      "loss": 0.899,
+      "step": 692
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.384994474608565e-06,
+      "loss": 0.8694,
+      "step": 693
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.375,
+      "learning_rate": 4.38327029107743e-06,
+      "loss": 0.8856,
+      "step": 694
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.381544034023713e-06,
+      "loss": 0.8712,
+      "step": 695
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.3798157053480675e-06,
+      "loss": 0.9077,
+      "step": 696
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.378085306953417e-06,
+      "loss": 0.8955,
+      "step": 697
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.37635284074497e-06,
+      "loss": 0.9213,
+      "step": 698
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.374618308630211e-06,
+      "loss": 0.8937,
+      "step": 699
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.372881712518898e-06,
+      "loss": 0.8883,
+      "step": 700
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.371143054323063e-06,
+      "loss": 0.8791,
+      "step": 701
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.369402335957006e-06,
+      "loss": 0.9023,
+      "step": 702
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.367659559337299e-06,
+      "loss": 0.9068,
+      "step": 703
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.365914726382774e-06,
+      "loss": 0.902,
+      "step": 704
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.364167839014535e-06,
+      "loss": 0.9101,
+      "step": 705
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.362418899155941e-06,
+      "loss": 0.9329,
+      "step": 706
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.360667908732616e-06,
+      "loss": 0.9207,
+      "step": 707
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.3589148696724395e-06,
+      "loss": 0.9197,
+      "step": 708
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5,
+      "learning_rate": 4.357159783905545e-06,
+      "loss": 0.9076,
+      "step": 709
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.355402653364321e-06,
+      "loss": 0.9011,
+      "step": 710
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.353643479983408e-06,
+      "loss": 0.8802,
+      "step": 711
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.351882265699696e-06,
+      "loss": 0.888,
+      "step": 712
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.350119012452321e-06,
+      "loss": 0.8981,
+      "step": 713
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.348353722182662e-06,
+      "loss": 0.8978,
+      "step": 714
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.3465863968343455e-06,
+      "loss": 0.949,
+      "step": 715
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.344817038353235e-06,
+      "loss": 0.9404,
+      "step": 716
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.343045648687434e-06,
+      "loss": 0.8982,
+      "step": 717
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5,
+      "learning_rate": 4.341272229787281e-06,
+      "loss": 0.8961,
+      "step": 718
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.339496783605351e-06,
+      "loss": 0.8729,
+      "step": 719
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.33771931209645e-06,
+      "loss": 0.9091,
+      "step": 720
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.335939817217612e-06,
+      "loss": 0.9266,
+      "step": 721
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.334158300928102e-06,
+      "loss": 0.8649,
+      "step": 722
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.332374765189409e-06,
+      "loss": 0.8755,
+      "step": 723
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.330589211965246e-06,
+      "loss": 0.9158,
+      "step": 724
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.328801643221546e-06,
+      "loss": 0.9189,
+      "step": 725
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.327012060926462e-06,
+      "loss": 0.8881,
+      "step": 726
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.325220467050364e-06,
+      "loss": 0.8659,
+      "step": 727
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.625,
+      "learning_rate": 4.323426863565838e-06,
+      "loss": 0.8801,
+      "step": 728
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.32163125244768e-06,
+      "loss": 0.9503,
+      "step": 729
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.319833635672899e-06,
+      "loss": 0.8773,
+      "step": 730
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.318034015220711e-06,
+      "loss": 0.9003,
+      "step": 731
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.316232393072538e-06,
+      "loss": 0.9126,
+      "step": 732
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.314428771212005e-06,
+      "loss": 0.9077,
+      "step": 733
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.3126231516249416e-06,
+      "loss": 0.9059,
+      "step": 734
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.25,
+      "learning_rate": 4.310815536299374e-06,
+      "loss": 0.9006,
+      "step": 735
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.309005927225528e-06,
+      "loss": 0.9217,
+      "step": 736
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.307194326395823e-06,
+      "loss": 0.9118,
+      "step": 737
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.3053807358048715e-06,
+      "loss": 0.9227,
+      "step": 738
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5,
+      "learning_rate": 4.303565157449478e-06,
+      "loss": 0.8995,
+      "step": 739
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5,
+      "learning_rate": 4.301747593328634e-06,
+      "loss": 0.9246,
+      "step": 740
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.299928045443518e-06,
+      "loss": 0.9216,
+      "step": 741
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.2981065157974955e-06,
+      "loss": 0.892,
+      "step": 742
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.375,
+      "learning_rate": 4.296283006396108e-06,
+      "loss": 0.875,
+      "step": 743
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.421875,
+      "learning_rate": 4.29445751924708e-06,
+      "loss": 0.9035,
+      "step": 744
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.292630056360316e-06,
+      "loss": 0.9139,
+      "step": 745
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.2908006197478914e-06,
+      "loss": 0.9201,
+      "step": 746
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.288969211424057e-06,
+      "loss": 0.9056,
+      "step": 747
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.287135833405235e-06,
+      "loss": 0.8909,
+      "step": 748
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.285300487710014e-06,
+      "loss": 0.8985,
+      "step": 749
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.375,
+      "learning_rate": 4.283463176359153e-06,
+      "loss": 0.8911,
+      "step": 750
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.28162390137557e-06,
+      "loss": 0.8813,
+      "step": 751
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3004,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 751,
+  "total_flos": 8.399260062207443e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}