diff --git "a/checkpoint-1130/trainer_state.json" "b/checkpoint-1130/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1130/trainer_state.json"
@@ -0,0 +1,7995 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9991154356479433,
+  "eval_steps": 142,
+  "global_step": 1130,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.349672317504883,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.4697,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 3.547457695007324,
+      "eval_runtime": 14.4843,
+      "eval_samples_per_second": 32.863,
+      "eval_steps_per_second": 8.216,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.466770648956299,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.4361,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.768375873565674,
+      "learning_rate": 3e-06,
+      "loss": 3.5871,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.6878485679626465,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.4894,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.205628871917725,
+      "learning_rate": 5e-06,
+      "loss": 3.501,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.880322456359863,
+      "learning_rate": 6e-06,
+      "loss": 3.5771,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.782011032104492,
+      "learning_rate": 7e-06,
+      "loss": 3.5119,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.285853385925293,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 3.4089,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.527816295623779,
+      "learning_rate": 9e-06,
+      "loss": 3.4341,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.505781650543213,
+      "learning_rate": 1e-05,
+      "loss": 3.4175,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.26746940612793,
+      "learning_rate": 9.999991309598975e-06,
+      "loss": 3.3163,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.086071968078613,
+      "learning_rate": 9.999965238426104e-06,
+      "loss": 3.274,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.271662712097168,
+      "learning_rate": 9.999921786572015e-06,
+      "loss": 3.2964,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.7532830238342285,
+      "learning_rate": 9.999860954187756e-06,
+      "loss": 3.1292,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.79344367980957,
+      "learning_rate": 9.99978274148479e-06,
+      "loss": 3.0481,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.707954406738281,
+      "learning_rate": 9.999687148734996e-06,
+      "loss": 2.9757,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.709782600402832,
+      "learning_rate": 9.999574176270667e-06,
+      "loss": 2.8673,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.751430988311768,
+      "learning_rate": 9.999443824484519e-06,
+      "loss": 2.8341,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.603825092315674,
+      "learning_rate": 9.999296093829672e-06,
+      "loss": 2.6421,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.514200210571289,
+      "learning_rate": 9.999130984819662e-06,
+      "loss": 2.5699,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.427622318267822,
+      "learning_rate": 9.998948498028435e-06,
+      "loss": 2.434,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.507719993591309,
+      "learning_rate": 9.998748634090344e-06,
+      "loss": 2.3301,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.355949878692627,
+      "learning_rate": 9.998531393700149e-06,
+      "loss": 2.0848,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.219519138336182,
+      "learning_rate": 9.99829677761301e-06,
+      "loss": 1.9115,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.065847873687744,
+      "learning_rate": 9.998044786644492e-06,
+      "loss": 1.7682,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7316126823425293,
+      "learning_rate": 9.997775421670558e-06,
+      "loss": 1.6426,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5228142738342285,
+      "learning_rate": 9.997488683627558e-06,
+      "loss": 1.5855,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2372498512268066,
+      "learning_rate": 9.997184573512245e-06,
+      "loss": 1.4059,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.072031259536743,
+      "learning_rate": 9.996863092381753e-06,
+      "loss": 1.2633,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.941805124282837,
+      "learning_rate": 9.9965242413536e-06,
+      "loss": 1.1709,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8302178382873535,
+      "learning_rate": 9.99616802160569e-06,
+      "loss": 1.0615,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7408287525177,
+      "learning_rate": 9.995794434376297e-06,
+      "loss": 1.0031,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6635422706604004,
+      "learning_rate": 9.995403480964072e-06,
+      "loss": 0.9273,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.538907766342163,
+      "learning_rate": 9.994995162728029e-06,
+      "loss": 0.8141,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.457651138305664,
+      "learning_rate": 9.994569481087552e-06,
+      "loss": 0.7208,
+      "step": 35
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.383510112762451,
+      "learning_rate": 9.994126437522376e-06,
+      "loss": 0.6763,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.170029401779175,
+      "learning_rate": 9.99366603357259e-06,
+      "loss": 0.6163,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0166823863983154,
+      "learning_rate": 9.993188270838636e-06,
+      "loss": 0.5146,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9549278020858765,
+      "learning_rate": 9.992693150981293e-06,
+      "loss": 0.4851,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7454789876937866,
+      "learning_rate": 9.992180675721671e-06,
+      "loss": 0.4015,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5932520627975464,
+      "learning_rate": 9.991650846841226e-06,
+      "loss": 0.3704,
+      "step": 41
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4394928216934204,
+      "learning_rate": 9.991103666181721e-06,
+      "loss": 0.3194,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2608178853988647,
+      "learning_rate": 9.990539135645246e-06,
+      "loss": 0.2621,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0825426578521729,
+      "learning_rate": 9.989957257194199e-06,
+      "loss": 0.2489,
+      "step": 44
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0383331775665283,
+      "learning_rate": 9.989358032851283e-06,
+      "loss": 0.249,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7754441499710083,
+      "learning_rate": 9.9887414646995e-06,
+      "loss": 0.162,
+      "step": 46
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7348763942718506,
+      "learning_rate": 9.988107554882138e-06,
+      "loss": 0.1713,
+      "step": 47
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5076927542686462,
+      "learning_rate": 9.987456305602769e-06,
+      "loss": 0.1438,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5017581582069397,
+      "learning_rate": 9.986787719125241e-06,
+      "loss": 0.1386,
+      "step": 49
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.42143169045448303,
+      "learning_rate": 9.986101797773667e-06,
+      "loss": 0.138,
+      "step": 50
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2827802896499634,
+      "learning_rate": 9.985398543932421e-06,
+      "loss": 0.1165,
+      "step": 51
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.33519247174263,
+      "learning_rate": 9.984677960046123e-06,
+      "loss": 0.119,
+      "step": 52
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2341219186782837,
+      "learning_rate": 9.983940048619641e-06,
+      "loss": 0.0748,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.20707747340202332,
+      "learning_rate": 9.983184812218071e-06,
+      "loss": 0.0949,
+      "step": 54
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3519020974636078,
+      "learning_rate": 9.98241225346674e-06,
+      "loss": 0.0964,
+      "step": 55
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2268177717924118,
+      "learning_rate": 9.981622375051183e-06,
+      "loss": 0.0771,
+      "step": 56
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.16427360475063324,
+      "learning_rate": 9.980815179717144e-06,
+      "loss": 0.08,
+      "step": 57
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3412397503852844,
+      "learning_rate": 9.979990670270565e-06,
+      "loss": 0.085,
+      "step": 58
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.13405166566371918,
+      "learning_rate": 9.979148849577574e-06,
+      "loss": 0.0852,
+      "step": 59
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.23308596014976501,
+      "learning_rate": 9.978289720564471e-06,
+      "loss": 0.0895,
+      "step": 60
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.20687691867351532,
+      "learning_rate": 9.97741328621773e-06,
+      "loss": 0.0875,
+      "step": 61
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.14851821959018707,
+      "learning_rate": 9.976519549583974e-06,
+      "loss": 0.0898,
+      "step": 62
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.31722521781921387,
+      "learning_rate": 9.975608513769977e-06,
+      "loss": 0.0902,
+      "step": 63
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.13900260627269745,
+      "learning_rate": 9.974680181942645e-06,
+      "loss": 0.0987,
+      "step": 64
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.21426311135292053,
+      "learning_rate": 9.97373455732901e-06,
+      "loss": 0.0765,
+      "step": 65
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.18942435085773468,
+      "learning_rate": 9.972771643216213e-06,
+      "loss": 0.092,
+      "step": 66
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.15527579188346863,
+      "learning_rate": 9.971791442951498e-06,
+      "loss": 0.0667,
+      "step": 67
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2748473286628723,
+      "learning_rate": 9.970793959942197e-06,
+      "loss": 0.0905,
+      "step": 68
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2710763216018677,
+      "learning_rate": 9.969779197655726e-06,
+      "loss": 0.0767,
+      "step": 69
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.19998699426651,
+      "learning_rate": 9.968747159619556e-06,
+      "loss": 0.0836,
+      "step": 70
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5467928051948547,
+      "learning_rate": 9.96769784942122e-06,
+      "loss": 0.1069,
+      "step": 71
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1530974954366684,
+      "learning_rate": 9.966631270708288e-06,
+      "loss": 0.0867,
+      "step": 72
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2446594089269638,
+      "learning_rate": 9.965547427188358e-06,
+      "loss": 0.0771,
+      "step": 73
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.14469081163406372,
+      "learning_rate": 9.964446322629044e-06,
+      "loss": 0.0892,
+      "step": 74
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2065022885799408,
+      "learning_rate": 9.963327960857962e-06,
+      "loss": 0.0729,
+      "step": 75
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.27265021204948425,
+      "learning_rate": 9.962192345762717e-06,
+      "loss": 0.0684,
+      "step": 76
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3528543710708618,
+      "learning_rate": 9.961039481290888e-06,
+      "loss": 0.0656,
+      "step": 77
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.14887100458145142,
+      "learning_rate": 9.959869371450022e-06,
+      "loss": 0.0794,
+      "step": 78
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.24467024207115173,
+      "learning_rate": 9.958682020307602e-06,
+      "loss": 0.0749,
+      "step": 79
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4109407961368561,
+      "learning_rate": 9.957477431991053e-06,
+      "loss": 0.1062,
+      "step": 80
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.20120254158973694,
+      "learning_rate": 9.95625561068772e-06,
+      "loss": 0.101,
+      "step": 81
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.30039364099502563,
+      "learning_rate": 9.955016560644847e-06,
+      "loss": 0.1015,
+      "step": 82
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3657380938529968,
+      "learning_rate": 9.953760286169571e-06,
+      "loss": 0.1124,
+      "step": 83
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3613692820072174,
+      "learning_rate": 9.952486791628905e-06,
+      "loss": 0.0836,
+      "step": 84
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3250538110733032,
+      "learning_rate": 9.95119608144972e-06,
+      "loss": 0.0753,
+      "step": 85
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.21736566722393036,
+      "learning_rate": 9.94988816011873e-06,
+      "loss": 0.075,
+      "step": 86
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8842391967773438,
+      "learning_rate": 9.948563032182482e-06,
+      "loss": 0.1067,
+      "step": 87
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2460835725069046,
+      "learning_rate": 9.947220702247329e-06,
+      "loss": 0.0832,
+      "step": 88
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5178576707839966,
+      "learning_rate": 9.94586117497943e-06,
+      "loss": 0.1007,
+      "step": 89
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.19302992522716522,
+      "learning_rate": 9.944484455104716e-06,
+      "loss": 0.0705,
+      "step": 90
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5060548186302185,
+      "learning_rate": 9.943090547408888e-06,
+      "loss": 0.1216,
+      "step": 91
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3466648757457733,
+      "learning_rate": 9.941679456737395e-06,
+      "loss": 0.0938,
+      "step": 92
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.22704587876796722,
+      "learning_rate": 9.940251187995412e-06,
+      "loss": 0.0803,
+      "step": 93
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.25966060161590576,
+      "learning_rate": 9.938805746147827e-06,
+      "loss": 0.0966,
+      "step": 94
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.20877382159233093,
+      "learning_rate": 9.937343136219234e-06,
+      "loss": 0.0667,
+      "step": 95
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1972026228904724,
+      "learning_rate": 9.935863363293896e-06,
+      "loss": 0.0642,
+      "step": 96
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.21666616201400757,
+      "learning_rate": 9.934366432515741e-06,
+      "loss": 0.0943,
+      "step": 97
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3361506760120392,
+      "learning_rate": 9.932852349088342e-06,
+      "loss": 0.0797,
+      "step": 98
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3094469904899597,
+      "learning_rate": 9.931321118274897e-06,
+      "loss": 0.0762,
+      "step": 99
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.33961156010627747,
+      "learning_rate": 9.929772745398207e-06,
+      "loss": 0.0744,
+      "step": 100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3448229730129242,
+      "learning_rate": 9.928207235840664e-06,
+      "loss": 0.0562,
+      "step": 101
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2657065689563751,
+      "learning_rate": 9.926624595044235e-06,
+      "loss": 0.0922,
+      "step": 102
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2084828019142151,
+      "learning_rate": 9.925024828510429e-06,
+      "loss": 0.0616,
+      "step": 103
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.342433899641037,
+      "learning_rate": 9.92340794180029e-06,
+      "loss": 0.0827,
+      "step": 104
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.21574071049690247,
+      "learning_rate": 9.921773940534382e-06,
+      "loss": 0.0593,
+      "step": 105
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.15846671164035797,
+      "learning_rate": 9.920122830392748e-06,
+      "loss": 0.0732,
+      "step": 106
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2687283456325531,
+      "learning_rate": 9.91845461711492e-06,
+      "loss": 0.0699,
+      "step": 107
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.18184144794940948,
+      "learning_rate": 9.916769306499866e-06,
+      "loss": 0.0632,
+      "step": 108
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.21744874119758606,
+      "learning_rate": 9.915066904406e-06,
+      "loss": 0.0805,
+      "step": 109
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3575034737586975,
+      "learning_rate": 9.913347416751148e-06,
+      "loss": 0.0819,
+      "step": 110
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.23368307948112488,
+      "learning_rate": 9.91161084951252e-06,
+      "loss": 0.0576,
+      "step": 111
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4599943161010742,
+      "learning_rate": 9.909857208726705e-06,
+      "loss": 0.0867,
+      "step": 112
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.26656806468963623,
+      "learning_rate": 9.908086500489638e-06,
+      "loss": 0.0586,
+      "step": 113
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5311251282691956,
+      "learning_rate": 9.906298730956585e-06,
+      "loss": 0.102,
+      "step": 114
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3186182677745819,
+      "learning_rate": 9.904493906342124e-06,
+      "loss": 0.0743,
+      "step": 115
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.20787174999713898,
+      "learning_rate": 9.902672032920106e-06,
+      "loss": 0.0536,
+      "step": 116
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.18734194338321686,
+      "learning_rate": 9.900833117023665e-06,
+      "loss": 0.06,
+      "step": 117
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.46386289596557617,
+      "learning_rate": 9.898977165045161e-06,
+      "loss": 0.0861,
+      "step": 118
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2560313940048218,
+      "learning_rate": 9.897104183436184e-06,
+      "loss": 0.0574,
+      "step": 119
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.22062335908412933,
+      "learning_rate": 9.895214178707516e-06,
+      "loss": 0.0631,
+      "step": 120
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2438971847295761,
+      "learning_rate": 9.89330715742912e-06,
+      "loss": 0.0906,
+      "step": 121
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.17582635581493378,
+      "learning_rate": 9.891383126230105e-06,
+      "loss": 0.0507,
+      "step": 122
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2775309979915619,
+      "learning_rate": 9.889442091798712e-06,
+      "loss": 0.0741,
+      "step": 123
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.18693320453166962,
+      "learning_rate": 9.887484060882292e-06,
+      "loss": 0.0624,
+      "step": 124
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5480987429618835,
+      "learning_rate": 9.885509040287267e-06,
+      "loss": 0.1104,
+      "step": 125
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.25776028633117676,
+      "learning_rate": 9.883517036879133e-06,
+      "loss": 0.0876,
+      "step": 126
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6148080825805664,
+      "learning_rate": 9.881508057582411e-06,
+      "loss": 0.0678,
+      "step": 127
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8079890608787537,
+      "learning_rate": 9.879482109380634e-06,
+      "loss": 0.0801,
+      "step": 128
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.22970955073833466,
+      "learning_rate": 9.877439199316324e-06,
+      "loss": 0.0662,
+      "step": 129
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4607698917388916,
+      "learning_rate": 9.875379334490962e-06,
+      "loss": 0.0863,
+      "step": 130
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4692334532737732,
+      "learning_rate": 9.873302522064972e-06,
+      "loss": 0.0968,
+      "step": 131
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8688197135925293,
+      "learning_rate": 9.871208769257686e-06,
+      "loss": 0.0963,
+      "step": 132
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2747116982936859,
+      "learning_rate": 9.869098083347323e-06,
+      "loss": 0.0801,
+      "step": 133
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5444130897521973,
+      "learning_rate": 9.866970471670968e-06,
+      "loss": 0.0965,
+      "step": 134
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.303763210773468,
+      "learning_rate": 9.864825941624538e-06,
+      "loss": 0.0727,
+      "step": 135
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.16060137748718262,
+      "learning_rate": 9.862664500662763e-06,
+      "loss": 0.0468,
+      "step": 136
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.22011107206344604,
+      "learning_rate": 9.860486156299164e-06,
+      "loss": 0.0832,
+      "step": 137
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.38508448004722595,
+      "learning_rate": 9.85829091610601e-06,
+      "loss": 0.0947,
+      "step": 138
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2385941594839096,
+      "learning_rate": 9.856078787714309e-06,
+      "loss": 0.0676,
+      "step": 139
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.15575379133224487,
+      "learning_rate": 9.853849778813777e-06,
+      "loss": 0.0833,
+      "step": 140
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.23725704848766327,
+      "learning_rate": 9.851603897152804e-06,
+      "loss": 0.0807,
+      "step": 141
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.20930880308151245,
+      "learning_rate": 9.849341150538434e-06,
+      "loss": 0.0881,
+      "step": 142
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.08187390118837357,
+      "eval_runtime": 14.7383,
+      "eval_samples_per_second": 32.297,
+      "eval_steps_per_second": 8.074,
+      "step": 142
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2966826260089874,
+      "learning_rate": 9.84706154683634e-06,
+      "loss": 0.0567,
+      "step": 143
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3383941948413849,
+      "learning_rate": 9.844765093970787e-06,
+      "loss": 0.0597,
+      "step": 144
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.20434536039829254,
+      "learning_rate": 9.842451799924616e-06,
+      "loss": 0.0873,
+      "step": 145
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.27841946482658386,
+      "learning_rate": 9.840121672739208e-06,
+      "loss": 0.0746,
+      "step": 146
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.18767426908016205,
+      "learning_rate": 9.837774720514456e-06,
+      "loss": 0.0928,
+      "step": 147
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4455524981021881,
+      "learning_rate": 9.835410951408748e-06,
+      "loss": 0.0692,
+      "step": 148
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.24763479828834534,
+      "learning_rate": 9.83303037363892e-06,
+      "loss": 0.0643,
+      "step": 149
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4498063325881958,
+      "learning_rate": 9.830632995480243e-06,
+      "loss": 0.0736,
+      "step": 150
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2298639714717865,
+      "learning_rate": 9.828218825266389e-06,
+      "loss": 0.0678,
+      "step": 151
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.46498408913612366,
+      "learning_rate": 9.8257878713894e-06,
+      "loss": 0.0775,
+      "step": 152
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3503478169441223,
+      "learning_rate": 9.823340142299662e-06,
+      "loss": 0.0749,
+      "step": 153
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3784450590610504,
+      "learning_rate": 9.820875646505874e-06,
+      "loss": 0.0806,
+      "step": 154
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2675660252571106,
+      "learning_rate": 9.818394392575018e-06,
+      "loss": 0.1054,
+      "step": 155
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.18610945343971252,
+      "learning_rate": 9.815896389132333e-06,
+      "loss": 0.0793,
+      "step": 156
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.25484928488731384,
+      "learning_rate": 9.813381644861276e-06,
+      "loss": 0.1004,
+      "step": 157
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2842435836791992,
+      "learning_rate": 9.810850168503506e-06,
+      "loss": 0.0413,
+      "step": 158
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2782611548900604,
+      "learning_rate": 9.808301968858838e-06,
+      "loss": 0.1083,
+      "step": 159
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1917373687028885,
+      "learning_rate": 9.805737054785223e-06,
+      "loss": 0.0727,
+      "step": 160
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2308584451675415,
+      "learning_rate": 9.803155435198713e-06,
+      "loss": 0.0629,
+      "step": 161
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.20095452666282654,
+      "learning_rate": 9.800557119073433e-06,
+      "loss": 0.0857,
+      "step": 162
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.28956976532936096,
+      "learning_rate": 9.797942115441546e-06,
+      "loss": 0.053,
+      "step": 163
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.24081195890903473,
+      "learning_rate": 9.795310433393227e-06,
+      "loss": 0.0611,
+      "step": 164
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4568108022212982,
+      "learning_rate": 9.792662082076618e-06,
+      "loss": 0.1011,
+      "step": 165
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.16725283861160278,
+      "learning_rate": 9.789997070697821e-06,
+      "loss": 0.0525,
+      "step": 166
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.27492183446884155,
+      "learning_rate": 9.787315408520839e-06,
+      "loss": 0.0581,
+      "step": 167
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.21994265913963318,
+      "learning_rate": 9.78461710486756e-06,
+      "loss": 0.068,
+      "step": 168
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2217988818883896,
+      "learning_rate": 9.78190216911772e-06,
+      "loss": 0.0663,
+      "step": 169
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43498608469963074,
+      "learning_rate": 9.779170610708872e-06,
+      "loss": 0.1003,
+      "step": 170
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5494738817214966,
+      "learning_rate": 9.776422439136351e-06,
+      "loss": 0.0901,
+      "step": 171
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.30544596910476685,
+      "learning_rate": 9.773657663953244e-06,
+      "loss": 0.1049,
+      "step": 172
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5173572301864624,
+      "learning_rate": 9.77087629477035e-06,
+      "loss": 0.084,
+      "step": 173
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.1275845617055893,
+      "learning_rate": 9.768078341256156e-06,
+      "loss": 0.0668,
+      "step": 174
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.28769540786743164,
+      "learning_rate": 9.765263813136796e-06,
+      "loss": 0.0743,
+      "step": 175
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7122567296028137,
+      "learning_rate": 9.762432720196024e-06,
+      "loss": 0.129,
+      "step": 176
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.28987741470336914,
+      "learning_rate": 9.759585072275171e-06,
+      "loss": 0.077,
+      "step": 177
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.24338261783123016,
+      "learning_rate": 9.756720879273117e-06,
+      "loss": 0.0763,
+      "step": 178
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3709527254104614,
+      "learning_rate": 9.753840151146259e-06,
+      "loss": 0.0639,
+      "step": 179
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1488633006811142,
+      "learning_rate": 9.750942897908468e-06,
+      "loss": 0.0825,
+      "step": 180
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13455910980701447,
+      "learning_rate": 9.748029129631062e-06,
+      "loss": 0.0594,
+      "step": 181
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2483726590871811,
+      "learning_rate": 9.745098856442769e-06,
+      "loss": 0.0621,
+      "step": 182
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.34576350450515747,
+      "learning_rate": 9.742152088529683e-06,
+      "loss": 0.083,
+      "step": 183
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.28745996952056885,
+      "learning_rate": 9.739188836135247e-06,
+      "loss": 0.0517,
+      "step": 184
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.18788190186023712,
+      "learning_rate": 9.736209109560201e-06,
+      "loss": 0.0831,
+      "step": 185
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1737545132637024,
+      "learning_rate": 9.733212919162551e-06,
+      "loss": 0.0597,
+      "step": 186
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.31250202655792236,
+      "learning_rate": 9.730200275357535e-06,
+      "loss": 0.0591,
+      "step": 187
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.20151746273040771,
+      "learning_rate": 9.727171188617588e-06,
+      "loss": 0.0687,
+      "step": 188
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.19415058195590973,
+      "learning_rate": 9.7241256694723e-06,
+      "loss": 0.0463,
+      "step": 189
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1791050285100937,
+      "learning_rate": 9.721063728508384e-06,
+      "loss": 0.0727,
+      "step": 190
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5859791040420532,
+      "learning_rate": 9.71798537636964e-06,
+      "loss": 0.0996,
+      "step": 191
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4107828438282013,
+      "learning_rate": 9.714890623756912e-06,
+      "loss": 0.0675,
+      "step": 192
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.339235782623291,
+      "learning_rate": 9.711779481428057e-06,
+      "loss": 0.0916,
+      "step": 193
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3153787851333618,
+      "learning_rate": 9.708651960197904e-06,
+      "loss": 0.0838,
+      "step": 194
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.38205188512802124,
+      "learning_rate": 9.705508070938219e-06,
+      "loss": 0.0773,
+      "step": 195
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.48745977878570557,
+      "learning_rate": 9.702347824577667e-06,
+      "loss": 0.0571,
+      "step": 196
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3050192594528198,
+      "learning_rate": 9.699171232101769e-06,
+      "loss": 0.0473,
+      "step": 197
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.22496673464775085,
+      "learning_rate": 9.695978304552871e-06,
+      "loss": 0.0589,
+      "step": 198
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3087387979030609,
+      "learning_rate": 9.6927690530301e-06,
+      "loss": 0.0814,
+      "step": 199
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.1646536886692047,
+      "learning_rate": 9.689543488689332e-06,
+      "loss": 0.0694,
+      "step": 200
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3601110279560089,
+      "learning_rate": 9.686301622743144e-06,
+      "loss": 0.0919,
+      "step": 201
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.20077168941497803,
+      "learning_rate": 9.683043466460783e-06,
+      "loss": 0.0754,
+      "step": 202
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2863204777240753,
+      "learning_rate": 9.67976903116812e-06,
+      "loss": 0.0814,
+      "step": 203
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.30959010124206543,
+      "learning_rate": 9.676478328247623e-06,
+      "loss": 0.1012,
+      "step": 204
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2969251871109009,
+      "learning_rate": 9.673171369138297e-06,
+      "loss": 0.0983,
+      "step": 205
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.19835108518600464,
+      "learning_rate": 9.669848165335668e-06,
+      "loss": 0.0814,
+      "step": 206
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.30629757046699524,
+      "learning_rate": 9.666508728391719e-06,
+      "loss": 0.0985,
+      "step": 207
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.17222163081169128,
+      "learning_rate": 9.663153069914874e-06,
+      "loss": 0.0789,
+      "step": 208
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4108015298843384,
+      "learning_rate": 9.65978120156994e-06,
+      "loss": 0.0797,
+      "step": 209
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2489665299654007,
+      "learning_rate": 9.656393135078067e-06,
+      "loss": 0.0927,
+      "step": 210
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.27541467547416687,
+      "learning_rate": 9.652988882216725e-06,
+      "loss": 0.0496,
+      "step": 211
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.17665977776050568,
+      "learning_rate": 9.649568454819637e-06,
+      "loss": 0.0666,
+      "step": 212
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.20858129858970642,
+      "learning_rate": 9.646131864776762e-06,
+      "loss": 0.0708,
+      "step": 213
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.18341350555419922,
+      "learning_rate": 9.642679124034234e-06,
+      "loss": 0.0805,
+      "step": 214
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.26587384939193726,
+      "learning_rate": 9.639210244594335e-06,
+      "loss": 0.0744,
+      "step": 215
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.20824970304965973,
+      "learning_rate": 9.635725238515447e-06,
+      "loss": 0.0821,
+      "step": 216
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.20785243809223175,
+      "learning_rate": 9.63222411791201e-06,
+      "loss": 0.0435,
+      "step": 217
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.45304587483406067,
+      "learning_rate": 9.628706894954481e-06,
+      "loss": 0.0791,
+      "step": 218
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.34389665722846985,
+      "learning_rate": 9.62517358186929e-06,
+      "loss": 0.0829,
+      "step": 219
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.24149852991104126,
+      "learning_rate": 9.621624190938802e-06,
+      "loss": 0.0555,
+      "step": 220
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.24253778159618378,
+      "learning_rate": 9.61805873450127e-06,
+      "loss": 0.0948,
+      "step": 221
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.24377629160881042,
+      "learning_rate": 9.614477224950788e-06,
+      "loss": 0.0758,
+      "step": 222
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1714078187942505,
+      "learning_rate": 9.610879674737263e-06,
+      "loss": 0.0773,
+      "step": 223
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.15262386202812195,
+      "learning_rate": 9.607266096366353e-06,
+      "loss": 0.0523,
+      "step": 224
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.34207943081855774,
+      "learning_rate": 9.603636502399436e-06,
+      "loss": 0.0981,
+      "step": 225
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2515898048877716,
+      "learning_rate": 9.599990905453567e-06,
+      "loss": 0.0576,
+      "step": 226
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.509267270565033,
+      "learning_rate": 9.59632931820142e-06,
+      "loss": 0.0819,
+      "step": 227
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.21576029062271118,
+      "learning_rate": 9.592651753371264e-06,
+      "loss": 0.0758,
+      "step": 228
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.44684383273124695,
+      "learning_rate": 9.588958223746903e-06,
+      "loss": 0.0543,
+      "step": 229
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4631918668746948,
+      "learning_rate": 9.585248742167638e-06,
+      "loss": 0.0795,
+      "step": 230
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3433883488178253,
+      "learning_rate": 9.581523321528224e-06,
+      "loss": 0.0505,
+      "step": 231
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23749905824661255,
+      "learning_rate": 9.577781974778817e-06,
+      "loss": 0.0463,
+      "step": 232
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.26895391941070557,
+      "learning_rate": 9.574024714924941e-06,
+      "loss": 0.0501,
+      "step": 233
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.45005208253860474,
+      "learning_rate": 9.570251555027432e-06,
+      "loss": 0.1112,
+      "step": 234
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.21089734137058258,
+      "learning_rate": 9.566462508202403e-06,
+      "loss": 0.0912,
+      "step": 235
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.22349245846271515,
+      "learning_rate": 9.562657587621186e-06,
+      "loss": 0.0671,
+      "step": 236
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20209869742393494,
+      "learning_rate": 9.558836806510292e-06,
+      "loss": 0.0507,
+      "step": 237
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.29422205686569214,
+      "learning_rate": 9.555000178151375e-06,
+      "loss": 0.0744,
+      "step": 238
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2201872318983078,
+      "learning_rate": 9.551147715881167e-06,
+      "loss": 0.0784,
+      "step": 239
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2708396017551422,
+      "learning_rate": 9.547279433091446e-06,
+      "loss": 0.0574,
+      "step": 240
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1722852736711502,
+      "learning_rate": 9.543395343228984e-06,
+      "loss": 0.0788,
+      "step": 241
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5377947688102722,
+      "learning_rate": 9.5394954597955e-06,
+      "loss": 0.0908,
+      "step": 242
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2269710898399353,
+      "learning_rate": 9.535579796347612e-06,
+      "loss": 0.078,
+      "step": 243
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2239355891942978,
+      "learning_rate": 9.531648366496799e-06,
+      "loss": 0.0501,
+      "step": 244
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.20963357388973236,
+      "learning_rate": 9.527701183909336e-06,
+      "loss": 0.0611,
+      "step": 245
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2847237288951874,
+      "learning_rate": 9.52373826230627e-06,
+      "loss": 0.0761,
+      "step": 246
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.20428815484046936,
+      "learning_rate": 9.519759615463346e-06,
+      "loss": 0.0684,
+      "step": 247
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.18516795337200165,
+      "learning_rate": 9.51576525721098e-06,
+      "loss": 0.0667,
+      "step": 248
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4787046015262604,
+      "learning_rate": 9.511755201434206e-06,
+      "loss": 0.1339,
+      "step": 249
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5016088485717773,
+      "learning_rate": 9.507729462072615e-06,
+      "loss": 0.0723,
+      "step": 250
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.24285417795181274,
+      "learning_rate": 9.503688053120327e-06,
+      "loss": 0.0553,
+      "step": 251
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.44292014837265015,
+      "learning_rate": 9.499630988625926e-06,
+      "loss": 0.071,
+      "step": 252
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7392637133598328,
+      "learning_rate": 9.495558282692421e-06,
+      "loss": 0.1074,
+      "step": 253
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.24650004506111145,
+      "learning_rate": 9.491469949477189e-06,
+      "loss": 0.0505,
+      "step": 254
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.22753530740737915,
+      "learning_rate": 9.48736600319193e-06,
+      "loss": 0.065,
+      "step": 255
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3731183111667633,
+      "learning_rate": 9.483246458102626e-06,
+      "loss": 0.0809,
+      "step": 256
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.16581465303897858,
+      "learning_rate": 9.479111328529473e-06,
+      "loss": 0.0575,
+      "step": 257
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2750982344150543,
+      "learning_rate": 9.474960628846844e-06,
+      "loss": 0.0967,
+      "step": 258
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3518216609954834,
+      "learning_rate": 9.470794373483236e-06,
+      "loss": 0.09,
+      "step": 259
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3871642053127289,
+      "learning_rate": 9.466612576921223e-06,
+      "loss": 0.0741,
+      "step": 260
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23743928968906403,
+      "learning_rate": 9.462415253697402e-06,
+      "loss": 0.0764,
+      "step": 261
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4438982903957367,
+      "learning_rate": 9.458202418402339e-06,
+      "loss": 0.089,
+      "step": 262
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1571800857782364,
+      "learning_rate": 9.453974085680527e-06,
+      "loss": 0.0481,
+      "step": 263
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.243282288312912,
+      "learning_rate": 9.449730270230326e-06,
+      "loss": 0.0843,
+      "step": 264
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.21889477968215942,
+      "learning_rate": 9.445470986803922e-06,
+      "loss": 0.0456,
+      "step": 265
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.14643190801143646,
+      "learning_rate": 9.441196250207267e-06,
+      "loss": 0.0555,
+      "step": 266
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6666358709335327,
+      "learning_rate": 9.436906075300032e-06,
+      "loss": 0.0775,
+      "step": 267
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.16846437752246857,
+      "learning_rate": 9.432600476995552e-06,
+      "loss": 0.0354,
+      "step": 268
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.23625656962394714,
+      "learning_rate": 9.428279470260776e-06,
+      "loss": 0.0837,
+      "step": 269
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.25802698731422424,
+      "learning_rate": 9.423943070116219e-06,
+      "loss": 0.0685,
+      "step": 270
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1842992752790451,
+      "learning_rate": 9.419591291635901e-06,
+      "loss": 0.0418,
+      "step": 271
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3693784773349762,
+      "learning_rate": 9.415224149947307e-06,
+      "loss": 0.0619,
+      "step": 272
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36359575390815735,
+      "learning_rate": 9.410841660231315e-06,
+      "loss": 0.0675,
+      "step": 273
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.25318172574043274,
+      "learning_rate": 9.406443837722168e-06,
+      "loss": 0.0521,
+      "step": 274
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.32068488001823425,
+      "learning_rate": 9.402030697707398e-06,
+      "loss": 0.0744,
+      "step": 275
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2709636390209198,
+      "learning_rate": 9.397602255527792e-06,
+      "loss": 0.0446,
+      "step": 276
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4686291813850403,
+      "learning_rate": 9.393158526577322e-06,
+      "loss": 0.0683,
+      "step": 277
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.32740774750709534,
+      "learning_rate": 9.388699526303106e-06,
+      "loss": 0.061,
+      "step": 278
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5102832913398743,
+      "learning_rate": 9.38422527020534e-06,
+      "loss": 0.0904,
+      "step": 279
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3581833243370056,
+      "learning_rate": 9.37973577383726e-06,
+      "loss": 0.0691,
+      "step": 280
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.22778642177581787,
+      "learning_rate": 9.375231052805074e-06,
+      "loss": 0.0507,
+      "step": 281
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3017262816429138,
+      "learning_rate": 9.370711122767912e-06,
+      "loss": 0.0909,
+      "step": 282
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.24568532407283783,
+      "learning_rate": 9.36617599943778e-06,
+      "loss": 0.0622,
+      "step": 283
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3963547348976135,
+      "learning_rate": 9.361625698579493e-06,
+      "loss": 0.1131,
+      "step": 284
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.07626692205667496,
+      "eval_runtime": 14.7121,
+      "eval_samples_per_second": 32.354,
+      "eval_steps_per_second": 8.089,
+      "step": 284
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6276136636734009,
+      "learning_rate": 9.357060236010626e-06,
+      "loss": 0.1067,
+      "step": 285
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.30459901690483093,
+      "learning_rate": 9.35247962760146e-06,
+      "loss": 0.0863,
+      "step": 286
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5768241286277771,
+      "learning_rate": 9.347883889274922e-06,
+      "loss": 0.0966,
+      "step": 287
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.14496010541915894,
+      "learning_rate": 9.34327303700654e-06,
+      "loss": 0.0242,
+      "step": 288
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.25412389636039734,
+      "learning_rate": 9.338647086824373e-06,
+      "loss": 0.071,
+      "step": 289
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23926912248134613,
+      "learning_rate": 9.334006054808966e-06,
+      "loss": 0.0378,
+      "step": 290
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7999410629272461,
+      "learning_rate": 9.329349957093293e-06,
+      "loss": 0.0912,
+      "step": 291
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.18663926422595978,
+      "learning_rate": 9.324678809862696e-06,
+      "loss": 0.0658,
+      "step": 292
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.27844029664993286,
+      "learning_rate": 9.319992629354828e-06,
+      "loss": 0.0657,
+      "step": 293
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.295076847076416,
+      "learning_rate": 9.31529143185961e-06,
+      "loss": 0.0497,
+      "step": 294
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.25130167603492737,
+      "learning_rate": 9.310575233719155e-06,
+      "loss": 0.0771,
+      "step": 295
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3607313632965088,
+      "learning_rate": 9.305844051327725e-06,
+      "loss": 0.0852,
+      "step": 296
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2247592657804489,
+      "learning_rate": 9.301097901131671e-06,
+      "loss": 0.0793,
+      "step": 297
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.34789037704467773,
+      "learning_rate": 9.296336799629368e-06,
+      "loss": 0.0602,
+      "step": 298
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.27349668741226196,
+      "learning_rate": 9.291560763371173e-06,
+      "loss": 0.0546,
+      "step": 299
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.15801368653774261,
+      "learning_rate": 9.28676980895935e-06,
+      "loss": 0.0545,
+      "step": 300
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.22296564280986786,
+      "learning_rate": 9.28196395304803e-06,
+      "loss": 0.0512,
+      "step": 301
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.18935079872608185,
+      "learning_rate": 9.277143212343134e-06,
+      "loss": 0.0382,
+      "step": 302
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.41481491923332214,
+      "learning_rate": 9.272307603602334e-06,
+      "loss": 0.0924,
+      "step": 303
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4681742489337921,
+      "learning_rate": 9.26745714363498e-06,
+      "loss": 0.0644,
+      "step": 304
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2106870412826538,
+      "learning_rate": 9.262591849302049e-06,
+      "loss": 0.0562,
+      "step": 305
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.45636868476867676,
+      "learning_rate": 9.257711737516083e-06,
+      "loss": 0.0751,
+      "step": 306
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.21162806451320648,
+      "learning_rate": 9.252816825241135e-06,
+      "loss": 0.0356,
+      "step": 307
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.31407129764556885,
+      "learning_rate": 9.247907129492707e-06,
+      "loss": 0.0713,
+      "step": 308
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.15091249346733093,
+      "learning_rate": 9.242982667337686e-06,
+      "loss": 0.066,
+      "step": 309
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.22152362763881683,
+      "learning_rate": 9.238043455894294e-06,
+      "loss": 0.0732,
+      "step": 310
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.21332816779613495,
+      "learning_rate": 9.233089512332021e-06,
+      "loss": 0.0744,
+      "step": 311
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3001808524131775,
+      "learning_rate": 9.228120853871571e-06,
+      "loss": 0.0337,
+      "step": 312
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.45393407344818115,
+      "learning_rate": 9.223137497784798e-06,
+      "loss": 0.0704,
+      "step": 313
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.36986440420150757,
+      "learning_rate": 9.218139461394644e-06,
+      "loss": 0.0751,
+      "step": 314
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.25236037373542786,
+      "learning_rate": 9.213126762075088e-06,
+      "loss": 0.0782,
+      "step": 315
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.19866494834423065,
+      "learning_rate": 9.208099417251077e-06,
+      "loss": 0.0404,
+      "step": 316
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18358244001865387,
+      "learning_rate": 9.203057444398469e-06,
+      "loss": 0.0362,
+      "step": 317
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38209205865859985,
+      "learning_rate": 9.198000861043967e-06,
+      "loss": 0.0531,
+      "step": 318
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2181481420993805,
+      "learning_rate": 9.19292968476507e-06,
+      "loss": 0.0607,
+      "step": 319
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.19979895651340485,
+      "learning_rate": 9.187843933189994e-06,
+      "loss": 0.0654,
+      "step": 320
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.46315455436706543,
+      "learning_rate": 9.182743623997634e-06,
+      "loss": 0.0654,
+      "step": 321
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.31687167286872864,
+      "learning_rate": 9.17762877491748e-06,
+      "loss": 0.0628,
+      "step": 322
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3118394613265991,
+      "learning_rate": 9.172499403729567e-06,
+      "loss": 0.0808,
+      "step": 323
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8999722599983215,
+      "learning_rate": 9.167355528264415e-06,
+      "loss": 0.1028,
+      "step": 324
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.41446566581726074,
+      "learning_rate": 9.162197166402957e-06,
+      "loss": 0.0896,
+      "step": 325
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.23004719614982605,
+      "learning_rate": 9.157024336076488e-06,
+      "loss": 0.067,
+      "step": 326
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.42118194699287415,
+      "learning_rate": 9.151837055266595e-06,
+      "loss": 0.0391,
+      "step": 327
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.21042917668819427,
+      "learning_rate": 9.1466353420051e-06,
+      "loss": 0.0677,
+      "step": 328
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.22170236706733704,
+      "learning_rate": 9.14141921437399e-06,
+      "loss": 0.0551,
+      "step": 329
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.24892501533031464,
+      "learning_rate": 9.136188690505363e-06,
+      "loss": 0.0527,
+      "step": 330
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.319352924823761,
+      "learning_rate": 9.130943788581359e-06,
+      "loss": 0.0843,
+      "step": 331
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.36097249388694763,
+      "learning_rate": 9.1256845268341e-06,
+      "loss": 0.108,
+      "step": 332
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.39498621225357056,
+      "learning_rate": 9.120410923545619e-06,
+      "loss": 0.053,
+      "step": 333
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5976508855819702,
+      "learning_rate": 9.115122997047812e-06,
+      "loss": 0.093,
+      "step": 334
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3573974072933197,
+      "learning_rate": 9.109820765722357e-06,
+      "loss": 0.0988,
+      "step": 335
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3447941839694977,
+      "learning_rate": 9.10450424800066e-06,
+      "loss": 0.1083,
+      "step": 336
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.36789995431900024,
+      "learning_rate": 9.099173462363794e-06,
+      "loss": 0.048,
+      "step": 337
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.23102298378944397,
+      "learning_rate": 9.093828427342419e-06,
+      "loss": 0.0615,
+      "step": 338
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2154015451669693,
+      "learning_rate": 9.088469161516735e-06,
+      "loss": 0.0775,
+      "step": 339
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2312682718038559,
+      "learning_rate": 9.083095683516414e-06,
+      "loss": 0.0708,
+      "step": 340
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1561180055141449,
+      "learning_rate": 9.077708012020525e-06,
+      "loss": 0.0628,
+      "step": 341
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2561321556568146,
+      "learning_rate": 9.072306165757476e-06,
+      "loss": 0.0913,
+      "step": 342
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4026300013065338,
+      "learning_rate": 9.066890163504956e-06,
+      "loss": 0.0757,
+      "step": 343
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2731042802333832,
+      "learning_rate": 9.061460024089853e-06,
+      "loss": 0.1009,
+      "step": 344
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.32153868675231934,
+      "learning_rate": 9.056015766388205e-06,
+      "loss": 0.0841,
+      "step": 345
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2845655381679535,
+      "learning_rate": 9.050557409325126e-06,
+      "loss": 0.0573,
+      "step": 346
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2339751124382019,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.083,
+      "step": 347
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1685064285993576,
+      "learning_rate": 9.039598473060114e-06,
+      "loss": 0.0598,
+      "step": 348
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2221630960702896,
+      "learning_rate": 9.0340979319532e-06,
+      "loss": 0.0639,
+      "step": 349
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.24915754795074463,
+      "learning_rate": 9.028583367674767e-06,
+      "loss": 0.1017,
+      "step": 350
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.21576786041259766,
+      "learning_rate": 9.023054799394316e-06,
+      "loss": 0.0598,
+      "step": 351
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.46793073415756226,
+      "learning_rate": 9.017512246330043e-06,
+      "loss": 0.0845,
+      "step": 352
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.23132359981536865,
+      "learning_rate": 9.01195572774875e-06,
+      "loss": 0.0865,
+      "step": 353
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1407833993434906,
+      "learning_rate": 9.006385262965786e-06,
+      "loss": 0.0771,
+      "step": 354
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.34604090452194214,
+      "learning_rate": 9.00080087134498e-06,
+      "loss": 0.1058,
+      "step": 355
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.31735068559646606,
+      "learning_rate": 8.995202572298575e-06,
+      "loss": 0.0833,
+      "step": 356
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2618115246295929,
+      "learning_rate": 8.989590385287156e-06,
+      "loss": 0.0688,
+      "step": 357
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.22336295247077942,
+      "learning_rate": 8.983964329819584e-06,
+      "loss": 0.0963,
+      "step": 358
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.27905017137527466,
+      "learning_rate": 8.97832442545293e-06,
+      "loss": 0.068,
+      "step": 359
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.16449055075645447,
+      "learning_rate": 8.972670691792409e-06,
+      "loss": 0.0951,
+      "step": 360
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2156919538974762,
+      "learning_rate": 8.967003148491305e-06,
+      "loss": 0.0495,
+      "step": 361
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.17048591375350952,
+      "learning_rate": 8.961321815250905e-06,
+      "loss": 0.0739,
+      "step": 362
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.21207985281944275,
+      "learning_rate": 8.955626711820438e-06,
+      "loss": 0.0817,
+      "step": 363
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.17985820770263672,
+      "learning_rate": 8.949917857996996e-06,
+      "loss": 0.0798,
+      "step": 364
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.18554697930812836,
+      "learning_rate": 8.944195273625472e-06,
+      "loss": 0.0511,
+      "step": 365
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.41460761427879333,
+      "learning_rate": 8.938458978598483e-06,
+      "loss": 0.104,
+      "step": 366
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1609378457069397,
+      "learning_rate": 8.932708992856315e-06,
+      "loss": 0.078,
+      "step": 367
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5837603211402893,
+      "learning_rate": 8.926945336386838e-06,
+      "loss": 0.0916,
+      "step": 368
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.20917074382305145,
+      "learning_rate": 8.921168029225448e-06,
+      "loss": 0.0439,
+      "step": 369
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.16059182584285736,
+      "learning_rate": 8.915377091454992e-06,
+      "loss": 0.0622,
+      "step": 370
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.42982298135757446,
+      "learning_rate": 8.909572543205697e-06,
+      "loss": 0.0688,
+      "step": 371
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.204355388879776,
+      "learning_rate": 8.903754404655107e-06,
+      "loss": 0.0355,
+      "step": 372
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1616220325231552,
+      "learning_rate": 8.897922696027998e-06,
+      "loss": 0.0751,
+      "step": 373
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.22931227087974548,
+      "learning_rate": 8.892077437596333e-06,
+      "loss": 0.064,
+      "step": 374
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24884217977523804,
+      "learning_rate": 8.886218649679162e-06,
+      "loss": 0.092,
+      "step": 375
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.23468102514743805,
+      "learning_rate": 8.880346352642575e-06,
+      "loss": 0.074,
+      "step": 376
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40000632405281067,
+      "learning_rate": 8.874460566899616e-06,
+      "loss": 0.0553,
+      "step": 377
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.24387776851654053,
+      "learning_rate": 8.868561312910222e-06,
+      "loss": 0.0469,
+      "step": 378
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.27059826254844666,
+      "learning_rate": 8.862648611181145e-06,
+      "loss": 0.0561,
+      "step": 379
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2047024518251419,
+      "learning_rate": 8.856722482265886e-06,
+      "loss": 0.0425,
+      "step": 380
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.19120073318481445,
+      "learning_rate": 8.850782946764618e-06,
+      "loss": 0.0683,
+      "step": 381
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.25501665472984314,
+      "learning_rate": 8.844830025324123e-06,
+      "loss": 0.0625,
+      "step": 382
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5250377058982849,
+      "learning_rate": 8.838863738637707e-06,
+      "loss": 0.0729,
+      "step": 383
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.18302053213119507,
+      "learning_rate": 8.83288410744514e-06,
+      "loss": 0.0509,
+      "step": 384
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3579452633857727,
+      "learning_rate": 8.826891152532579e-06,
+      "loss": 0.0815,
+      "step": 385
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4426077902317047,
+      "learning_rate": 8.820884894732498e-06,
+      "loss": 0.0868,
+      "step": 386
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.18046371638774872,
+      "learning_rate": 8.814865354923614e-06,
+      "loss": 0.0545,
+      "step": 387
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.17153623700141907,
+      "learning_rate": 8.808832554030809e-06,
+      "loss": 0.0407,
+      "step": 388
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.29191163182258606,
+      "learning_rate": 8.802786513025069e-06,
+      "loss": 0.0551,
+      "step": 389
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3224787414073944,
+      "learning_rate": 8.796727252923403e-06,
+      "loss": 0.0435,
+      "step": 390
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5621923208236694,
+      "learning_rate": 8.79065479478877e-06,
+      "loss": 0.0522,
+      "step": 391
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3209080398082733,
+      "learning_rate": 8.784569159730008e-06,
+      "loss": 0.0546,
+      "step": 392
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6716540455818176,
+      "learning_rate": 8.778470368901761e-06,
+      "loss": 0.0754,
+      "step": 393
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.19775713980197906,
+      "learning_rate": 8.772358443504406e-06,
+      "loss": 0.013,
+      "step": 394
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.25971564650535583,
+      "learning_rate": 8.766233404783975e-06,
+      "loss": 0.0344,
+      "step": 395
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5102748870849609,
+      "learning_rate": 8.760095274032083e-06,
+      "loss": 0.0646,
+      "step": 396
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.30724096298217773,
+      "learning_rate": 8.75394407258586e-06,
+      "loss": 0.0444,
+      "step": 397
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2920438051223755,
+      "learning_rate": 8.747779821827868e-06,
+      "loss": 0.0714,
+      "step": 398
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5672811269760132,
+      "learning_rate": 8.741602543186032e-06,
+      "loss": 0.144,
+      "step": 399
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.29323235154151917,
+      "learning_rate": 8.735412258133562e-06,
+      "loss": 0.0467,
+      "step": 400
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7952798008918762,
+      "learning_rate": 8.729208988188882e-06,
+      "loss": 0.1006,
+      "step": 401
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1980268657207489,
+      "learning_rate": 8.722992754915555e-06,
+      "loss": 0.0282,
+      "step": 402
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3340957462787628,
+      "learning_rate": 8.716763579922203e-06,
+      "loss": 0.0587,
+      "step": 403
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3219829499721527,
+      "learning_rate": 8.71052148486244e-06,
+      "loss": 0.0484,
+      "step": 404
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4473950266838074,
+      "learning_rate": 8.704266491434787e-06,
+      "loss": 0.0613,
+      "step": 405
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4419131875038147,
+      "learning_rate": 8.697998621382608e-06,
+      "loss": 0.0569,
+      "step": 406
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3174911141395569,
+      "learning_rate": 8.69171789649402e-06,
+      "loss": 0.0381,
+      "step": 407
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.445470929145813,
+      "learning_rate": 8.685424338601833e-06,
+      "loss": 0.0818,
+      "step": 408
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.30242660641670227,
+      "learning_rate": 8.679117969583464e-06,
+      "loss": 0.0733,
+      "step": 409
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.30104532837867737,
+      "learning_rate": 8.672798811360863e-06,
+      "loss": 0.0635,
+      "step": 410
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4942661225795746,
+      "learning_rate": 8.66646688590044e-06,
+      "loss": 0.0652,
+      "step": 411
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.48222100734710693,
+      "learning_rate": 8.660122215212976e-06,
+      "loss": 0.0519,
+      "step": 412
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.18950317800045013,
+      "learning_rate": 8.653764821353575e-06,
+      "loss": 0.0372,
+      "step": 413
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.0479586124420166,
+      "learning_rate": 8.647394726421547e-06,
+      "loss": 0.1133,
+      "step": 414
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9393060803413391,
+      "learning_rate": 8.641011952560372e-06,
+      "loss": 0.1114,
+      "step": 415
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6437638998031616,
+      "learning_rate": 8.63461652195759e-06,
+      "loss": 0.0834,
+      "step": 416
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5016691088676453,
+      "learning_rate": 8.628208456844749e-06,
+      "loss": 0.1051,
+      "step": 417
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2746430039405823,
+      "learning_rate": 8.621787779497307e-06,
+      "loss": 0.0756,
+      "step": 418
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.1799677610397339,
+      "learning_rate": 8.615354512234569e-06,
+      "loss": 0.0748,
+      "step": 419
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.32306981086730957,
+      "learning_rate": 8.608908677419606e-06,
+      "loss": 0.09,
+      "step": 420
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.40375712513923645,
+      "learning_rate": 8.602450297459173e-06,
+      "loss": 0.1321,
+      "step": 421
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1651470959186554,
+      "learning_rate": 8.595979394803633e-06,
+      "loss": 0.0613,
+      "step": 422
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.17097824811935425,
+      "learning_rate": 8.589495991946885e-06,
+      "loss": 0.0701,
+      "step": 423
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5410835146903992,
+      "learning_rate": 8.583000111426277e-06,
+      "loss": 0.0655,
+      "step": 424
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2845303416252136,
+      "learning_rate": 8.576491775822527e-06,
+      "loss": 0.0859,
+      "step": 425
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2452799528837204,
+      "learning_rate": 8.569971007759657e-06,
+      "loss": 0.0538,
+      "step": 426
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.07323230057954788,
+      "eval_runtime": 14.7111,
+      "eval_samples_per_second": 32.356,
+      "eval_steps_per_second": 8.089,
+      "step": 426
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6456173062324524,
+      "learning_rate": 8.563437829904904e-06,
+      "loss": 0.1131,
+      "step": 427
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3903690278530121,
+      "learning_rate": 8.556892264968639e-06,
+      "loss": 0.0997,
+      "step": 428
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.33470049500465393,
+      "learning_rate": 8.550334335704298e-06,
+      "loss": 0.0987,
+      "step": 429
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1498459130525589,
+      "learning_rate": 8.543764064908295e-06,
+      "loss": 0.0524,
+      "step": 430
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.21974940598011017,
+      "learning_rate": 8.537181475419944e-06,
+      "loss": 0.0789,
+      "step": 431
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.16694916784763336,
+      "learning_rate": 8.530586590121384e-06,
+      "loss": 0.0731,
+      "step": 432
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.12150876969099045,
+      "learning_rate": 8.523979431937493e-06,
+      "loss": 0.0694,
+      "step": 433
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.17073531448841095,
+      "learning_rate": 8.51736002383581e-06,
+      "loss": 0.0698,
+      "step": 434
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2708394527435303,
+      "learning_rate": 8.510728388826464e-06,
+      "loss": 0.0739,
+      "step": 435
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1602393388748169,
+      "learning_rate": 8.504084549962079e-06,
+      "loss": 0.0709,
+      "step": 436
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2071549892425537,
+      "learning_rate": 8.497428530337707e-06,
+      "loss": 0.0757,
+      "step": 437
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1717323213815689,
+      "learning_rate": 8.490760353090738e-06,
+      "loss": 0.0802,
+      "step": 438
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2539728879928589,
+      "learning_rate": 8.484080041400827e-06,
+      "loss": 0.0852,
+      "step": 439
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.15313653647899628,
+      "learning_rate": 8.477387618489808e-06,
+      "loss": 0.0788,
+      "step": 440
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.20286281406879425,
+      "learning_rate": 8.470683107621616e-06,
+      "loss": 0.0423,
+      "step": 441
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21172399818897247,
+      "learning_rate": 8.463966532102207e-06,
+      "loss": 0.0575,
+      "step": 442
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.23021037876605988,
+      "learning_rate": 8.457237915279477e-06,
+      "loss": 0.0774,
+      "step": 443
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.14592242240905762,
+      "learning_rate": 8.450497280543174e-06,
+      "loss": 0.0699,
+      "step": 444
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2122061848640442,
+      "learning_rate": 8.443744651324828e-06,
+      "loss": 0.0801,
+      "step": 445
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.21682047843933105,
+      "learning_rate": 8.43698005109766e-06,
+      "loss": 0.0478,
+      "step": 446
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.19426396489143372,
+      "learning_rate": 8.430203503376506e-06,
+      "loss": 0.0508,
+      "step": 447
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.14614954590797424,
+      "learning_rate": 8.423415031717734e-06,
+      "loss": 0.0712,
+      "step": 448
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1340399831533432,
+      "learning_rate": 8.416614659719158e-06,
+      "loss": 0.0723,
+      "step": 449
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.16521310806274414,
+      "learning_rate": 8.409802411019962e-06,
+      "loss": 0.0486,
+      "step": 450
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3560049533843994,
+      "learning_rate": 8.40297830930062e-06,
+      "loss": 0.0868,
+      "step": 451
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1964522898197174,
+      "learning_rate": 8.396142378282799e-06,
+      "loss": 0.046,
+      "step": 452
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.175230011343956,
+      "learning_rate": 8.389294641729293e-06,
+      "loss": 0.0659,
+      "step": 453
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.22042769193649292,
+      "learning_rate": 8.382435123443934e-06,
+      "loss": 0.0746,
+      "step": 454
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.19862250983715057,
+      "learning_rate": 8.375563847271506e-06,
+      "loss": 0.055,
+      "step": 455
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.24993905425071716,
+      "learning_rate": 8.36868083709767e-06,
+      "loss": 0.0858,
+      "step": 456
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.14945238828659058,
+      "learning_rate": 8.361786116848871e-06,
+      "loss": 0.0573,
+      "step": 457
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.43160539865493774,
+      "learning_rate": 8.354879710492264e-06,
+      "loss": 0.094,
+      "step": 458
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5086230635643005,
+      "learning_rate": 8.347961642035624e-06,
+      "loss": 0.0956,
+      "step": 459
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1694391518831253,
+      "learning_rate": 8.341031935527267e-06,
+      "loss": 0.0539,
+      "step": 460
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2055732011795044,
+      "learning_rate": 8.334090615055966e-06,
+      "loss": 0.0564,
+      "step": 461
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.14689050614833832,
+      "learning_rate": 8.327137704750863e-06,
+      "loss": 0.0554,
+      "step": 462
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3445436656475067,
+      "learning_rate": 8.32017322878139e-06,
+      "loss": 0.1029,
+      "step": 463
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1985045075416565,
+      "learning_rate": 8.31319721135718e-06,
+      "loss": 0.0703,
+      "step": 464
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.24056944251060486,
+      "learning_rate": 8.306209676727994e-06,
+      "loss": 0.0909,
+      "step": 465
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.20218642055988312,
+      "learning_rate": 8.29921064918362e-06,
+      "loss": 0.0658,
+      "step": 466
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2052248865365982,
+      "learning_rate": 8.2922001530538e-06,
+      "loss": 0.0522,
+      "step": 467
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2361009120941162,
+      "learning_rate": 8.285178212708143e-06,
+      "loss": 0.0674,
+      "step": 468
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.30261772871017456,
+      "learning_rate": 8.278144852556042e-06,
+      "loss": 0.0672,
+      "step": 469
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3114418685436249,
+      "learning_rate": 8.271100097046585e-06,
+      "loss": 0.0762,
+      "step": 470
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4094521701335907,
+      "learning_rate": 8.26404397066847e-06,
+      "loss": 0.0725,
+      "step": 471
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.23902684450149536,
+      "learning_rate": 8.256976497949924e-06,
+      "loss": 0.0408,
+      "step": 472
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2393728494644165,
+      "learning_rate": 8.249897703458619e-06,
+      "loss": 0.0608,
+      "step": 473
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.24134708940982819,
+      "learning_rate": 8.242807611801578e-06,
+      "loss": 0.0369,
+      "step": 474
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14594919979572296,
+      "learning_rate": 8.235706247625098e-06,
+      "loss": 0.0261,
+      "step": 475
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.32239043712615967,
+      "learning_rate": 8.228593635614659e-06,
+      "loss": 0.1011,
+      "step": 476
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2879891097545624,
+      "learning_rate": 8.22146980049484e-06,
+      "loss": 0.0742,
+      "step": 477
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5778201818466187,
+      "learning_rate": 8.21433476702924e-06,
+      "loss": 0.1146,
+      "step": 478
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.214900940656662,
+      "learning_rate": 8.207188560020374e-06,
+      "loss": 0.037,
+      "step": 479
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.22797343134880066,
+      "learning_rate": 8.200031204309604e-06,
+      "loss": 0.0595,
+      "step": 480
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.19148996472358704,
+      "learning_rate": 8.192862724777052e-06,
+      "loss": 0.0339,
+      "step": 481
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15512730181217194,
+      "learning_rate": 8.185683146341496e-06,
+      "loss": 0.051,
+      "step": 482
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.20564667880535126,
+      "learning_rate": 8.178492493960309e-06,
+      "loss": 0.0453,
+      "step": 483
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3061021566390991,
+      "learning_rate": 8.171290792629348e-06,
+      "loss": 0.033,
+      "step": 484
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.23765882849693298,
+      "learning_rate": 8.16407806738288e-06,
+      "loss": 0.0614,
+      "step": 485
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3384253978729248,
+      "learning_rate": 8.156854343293501e-06,
+      "loss": 0.0662,
+      "step": 486
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1890600174665451,
+      "learning_rate": 8.149619645472031e-06,
+      "loss": 0.0495,
+      "step": 487
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.30018725991249084,
+      "learning_rate": 8.14237399906744e-06,
+      "loss": 0.0722,
+      "step": 488
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.31263217329978943,
+      "learning_rate": 8.135117429266756e-06,
+      "loss": 0.0693,
+      "step": 489
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.18034443259239197,
+      "learning_rate": 8.127849961294984e-06,
+      "loss": 0.0409,
+      "step": 490
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.23813746869564056,
+      "learning_rate": 8.120571620415007e-06,
+      "loss": 0.0433,
+      "step": 491
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.23636382818222046,
+      "learning_rate": 8.113282431927502e-06,
+      "loss": 0.0544,
+      "step": 492
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5150755047798157,
+      "learning_rate": 8.10598242117086e-06,
+      "loss": 0.1072,
+      "step": 493
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3001669645309448,
+      "learning_rate": 8.09867161352109e-06,
+      "loss": 0.0414,
+      "step": 494
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.228012353181839,
+      "learning_rate": 8.091350034391732e-06,
+      "loss": 0.0701,
+      "step": 495
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3816164433956146,
+      "learning_rate": 8.084017709233767e-06,
+      "loss": 0.0723,
+      "step": 496
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.32659652829170227,
+      "learning_rate": 8.076674663535537e-06,
+      "loss": 0.0697,
+      "step": 497
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.48343512415885925,
+      "learning_rate": 8.069320922822644e-06,
+      "loss": 0.1034,
+      "step": 498
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.29286321997642517,
+      "learning_rate": 8.061956512657872e-06,
+      "loss": 0.075,
+      "step": 499
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4552519917488098,
+      "learning_rate": 8.05458145864109e-06,
+      "loss": 0.0568,
+      "step": 500
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3469892740249634,
+      "learning_rate": 8.047195786409172e-06,
+      "loss": 0.0661,
+      "step": 501
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1452968716621399,
+      "learning_rate": 8.039799521635896e-06,
+      "loss": 0.0226,
+      "step": 502
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.25091221928596497,
+      "learning_rate": 8.032392690031868e-06,
+      "loss": 0.0486,
+      "step": 503
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1830379068851471,
+      "learning_rate": 8.024975317344421e-06,
+      "loss": 0.0448,
+      "step": 504
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.32811227440834045,
+      "learning_rate": 8.017547429357532e-06,
+      "loss": 0.055,
+      "step": 505
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.24694731831550598,
+      "learning_rate": 8.010109051891731e-06,
+      "loss": 0.0614,
+      "step": 506
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3512776792049408,
+      "learning_rate": 8.002660210804011e-06,
+      "loss": 0.0699,
+      "step": 507
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2562338709831238,
+      "learning_rate": 7.995200931987744e-06,
+      "loss": 0.0726,
+      "step": 508
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38721486926078796,
+      "learning_rate": 7.987731241372572e-06,
+      "loss": 0.0798,
+      "step": 509
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.24697037041187286,
+      "learning_rate": 7.980251164924342e-06,
+      "loss": 0.0657,
+      "step": 510
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5245312452316284,
+      "learning_rate": 7.972760728644995e-06,
+      "loss": 0.0575,
+      "step": 511
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.29805997014045715,
+      "learning_rate": 7.965259958572495e-06,
+      "loss": 0.052,
+      "step": 512
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.21135053038597107,
+      "learning_rate": 7.957748880780722e-06,
+      "loss": 0.0378,
+      "step": 513
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.23149773478507996,
+      "learning_rate": 7.950227521379382e-06,
+      "loss": 0.0726,
+      "step": 514
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.19880171120166779,
+      "learning_rate": 7.94269590651393e-06,
+      "loss": 0.0714,
+      "step": 515
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.28021228313446045,
+      "learning_rate": 7.935154062365468e-06,
+      "loss": 0.0634,
+      "step": 516
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.21841171383857727,
+      "learning_rate": 7.927602015150655e-06,
+      "loss": 0.0625,
+      "step": 517
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1774914562702179,
+      "learning_rate": 7.920039791121617e-06,
+      "loss": 0.0321,
+      "step": 518
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38757917284965515,
+      "learning_rate": 7.91246741656586e-06,
+      "loss": 0.1069,
+      "step": 519
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19643576443195343,
+      "learning_rate": 7.904884917806174e-06,
+      "loss": 0.0409,
+      "step": 520
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2885828912258148,
+      "learning_rate": 7.897292321200538e-06,
+      "loss": 0.0514,
+      "step": 521
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4019085168838501,
+      "learning_rate": 7.889689653142037e-06,
+      "loss": 0.089,
+      "step": 522
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.350379079580307,
+      "learning_rate": 7.882076940058764e-06,
+      "loss": 0.0505,
+      "step": 523
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3123965859413147,
+      "learning_rate": 7.87445420841373e-06,
+      "loss": 0.0555,
+      "step": 524
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.21883922815322876,
+      "learning_rate": 7.866821484704777e-06,
+      "loss": 0.0439,
+      "step": 525
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2795931100845337,
+      "learning_rate": 7.859178795464473e-06,
+      "loss": 0.0757,
+      "step": 526
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3848627507686615,
+      "learning_rate": 7.851526167260034e-06,
+      "loss": 0.0995,
+      "step": 527
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5650475025177002,
+      "learning_rate": 7.843863626693221e-06,
+      "loss": 0.0956,
+      "step": 528
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.20236073434352875,
+      "learning_rate": 7.836191200400257e-06,
+      "loss": 0.0629,
+      "step": 529
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19835765659809113,
+      "learning_rate": 7.828508915051724e-06,
+      "loss": 0.0661,
+      "step": 530
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2083461880683899,
+      "learning_rate": 7.82081679735248e-06,
+      "loss": 0.051,
+      "step": 531
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4042919874191284,
+      "learning_rate": 7.813114874041558e-06,
+      "loss": 0.0736,
+      "step": 532
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.20774157345294952,
+      "learning_rate": 7.80540317189208e-06,
+      "loss": 0.0578,
+      "step": 533
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.20473290979862213,
+      "learning_rate": 7.797681717711162e-06,
+      "loss": 0.0471,
+      "step": 534
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2514810562133789,
+      "learning_rate": 7.789950538339813e-06,
+      "loss": 0.0665,
+      "step": 535
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.43802833557128906,
+      "learning_rate": 7.782209660652855e-06,
+      "loss": 0.087,
+      "step": 536
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2503105103969574,
+      "learning_rate": 7.774459111558821e-06,
+      "loss": 0.0819,
+      "step": 537
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.16841083765029907,
+      "learning_rate": 7.766698917999862e-06,
+      "loss": 0.0505,
+      "step": 538
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3313782513141632,
+      "learning_rate": 7.758929106951656e-06,
+      "loss": 0.0713,
+      "step": 539
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.16334685683250427,
+      "learning_rate": 7.751149705423313e-06,
+      "loss": 0.044,
+      "step": 540
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4028722941875458,
+      "learning_rate": 7.743360740457278e-06,
+      "loss": 0.0847,
+      "step": 541
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.24219095706939697,
+      "learning_rate": 7.735562239129248e-06,
+      "loss": 0.0848,
+      "step": 542
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.19837279617786407,
+      "learning_rate": 7.72775422854806e-06,
+      "loss": 0.0331,
+      "step": 543
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.21855801343917847,
+      "learning_rate": 7.719936735855611e-06,
+      "loss": 0.0473,
+      "step": 544
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1909436583518982,
+      "learning_rate": 7.712109788226763e-06,
+      "loss": 0.0694,
+      "step": 545
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.37223267555236816,
+      "learning_rate": 7.704273412869239e-06,
+      "loss": 0.077,
+      "step": 546
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.33873534202575684,
+      "learning_rate": 7.696427637023537e-06,
+      "loss": 0.0391,
+      "step": 547
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.22075894474983215,
+      "learning_rate": 7.688572487962836e-06,
+      "loss": 0.0363,
+      "step": 548
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3139945864677429,
+      "learning_rate": 7.680707992992889e-06,
+      "loss": 0.0676,
+      "step": 549
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.24971792101860046,
+      "learning_rate": 7.672834179451943e-06,
+      "loss": 0.0561,
+      "step": 550
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2864936292171478,
+      "learning_rate": 7.664951074710638e-06,
+      "loss": 0.0885,
+      "step": 551
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.32206228375434875,
+      "learning_rate": 7.657058706171912e-06,
+      "loss": 0.065,
+      "step": 552
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28918203711509705,
+      "learning_rate": 7.649157101270904e-06,
+      "loss": 0.108,
+      "step": 553
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2585963308811188,
+      "learning_rate": 7.641246287474856e-06,
+      "loss": 0.0536,
+      "step": 554
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4212508797645569,
+      "learning_rate": 7.633326292283028e-06,
+      "loss": 0.0857,
+      "step": 555
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1916050761938095,
+      "learning_rate": 7.625397143226596e-06,
+      "loss": 0.0597,
+      "step": 556
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.678855299949646,
+      "learning_rate": 7.617458867868554e-06,
+      "loss": 0.0759,
+      "step": 557
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3328985273838043,
+      "learning_rate": 7.609511493803616e-06,
+      "loss": 0.0545,
+      "step": 558
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2722048759460449,
+      "learning_rate": 7.601555048658133e-06,
+      "loss": 0.0633,
+      "step": 559
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2508256733417511,
+      "learning_rate": 7.593589560089984e-06,
+      "loss": 0.0571,
+      "step": 560
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.49203789234161377,
+      "learning_rate": 7.585615055788484e-06,
+      "loss": 0.0834,
+      "step": 561
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2127446085214615,
+      "learning_rate": 7.577631563474291e-06,
+      "loss": 0.0658,
+      "step": 562
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6191400289535522,
+      "learning_rate": 7.569639110899303e-06,
+      "loss": 0.0848,
+      "step": 563
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.22974306344985962,
+      "learning_rate": 7.561637725846568e-06,
+      "loss": 0.0493,
+      "step": 564
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6743472814559937,
+      "learning_rate": 7.553627436130183e-06,
+      "loss": 0.139,
+      "step": 565
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2306680679321289,
+      "learning_rate": 7.545608269595201e-06,
+      "loss": 0.0575,
+      "step": 566
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.25629836320877075,
+      "learning_rate": 7.537580254117531e-06,
+      "loss": 0.0596,
+      "step": 567
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.27573418617248535,
+      "learning_rate": 7.529543417603844e-06,
+      "loss": 0.0425,
+      "step": 568
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.06561362743377686,
+      "eval_runtime": 14.6894,
+      "eval_samples_per_second": 32.404,
+      "eval_steps_per_second": 8.101,
+      "step": 568
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.34876716136932373,
+      "learning_rate": 7.521497787991472e-06,
+      "loss": 0.1116,
+      "step": 569
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.17311374843120575,
+      "learning_rate": 7.513443393248312e-06,
+      "loss": 0.0543,
+      "step": 570
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.38628146052360535,
+      "learning_rate": 7.505380261372735e-06,
+      "loss": 0.0912,
+      "step": 571
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3152383863925934,
+      "learning_rate": 7.497308420393478e-06,
+      "loss": 0.0964,
+      "step": 572
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.2237791121006012,
+      "learning_rate": 7.489227898369558e-06,
+      "loss": 0.0776,
+      "step": 573
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3081691563129425,
+      "learning_rate": 7.481138723390165e-06,
+      "loss": 0.091,
+      "step": 574
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.23158672451972961,
+      "learning_rate": 7.473040923574567e-06,
+      "loss": 0.042,
+      "step": 575
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.20039266347885132,
+      "learning_rate": 7.464934527072016e-06,
+      "loss": 0.0467,
+      "step": 576
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.15118126571178436,
+      "learning_rate": 7.456819562061649e-06,
+      "loss": 0.0573,
+      "step": 577
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.25211474299430847,
+      "learning_rate": 7.448696056752383e-06,
+      "loss": 0.0767,
+      "step": 578
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.1856004297733307,
+      "learning_rate": 7.440564039382827e-06,
+      "loss": 0.0446,
+      "step": 579
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.19141128659248352,
+      "learning_rate": 7.432423538221179e-06,
+      "loss": 0.0481,
+      "step": 580
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.3133082091808319,
+      "learning_rate": 7.424274581565123e-06,
+      "loss": 0.0444,
+      "step": 581
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.15396596491336823,
+      "learning_rate": 7.416117197741742e-06,
+      "loss": 0.052,
+      "step": 582
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.21921683847904205,
+      "learning_rate": 7.407951415107413e-06,
+      "loss": 0.0577,
+      "step": 583
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.2533078193664551,
+      "learning_rate": 7.3997772620477e-06,
+      "loss": 0.0769,
+      "step": 584
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.16214172542095184,
+      "learning_rate": 7.391594766977277e-06,
+      "loss": 0.0415,
+      "step": 585
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.1633971482515335,
+      "learning_rate": 7.383403958339806e-06,
+      "loss": 0.0589,
+      "step": 586
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.223504438996315,
+      "learning_rate": 7.375204864607852e-06,
+      "loss": 0.0644,
+      "step": 587
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.3524682819843292,
+      "learning_rate": 7.366997514282782e-06,
+      "loss": 0.0793,
+      "step": 588
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.2214106321334839,
+      "learning_rate": 7.358781935894659e-06,
+      "loss": 0.059,
+      "step": 589
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.23937495052814484,
+      "learning_rate": 7.350558158002154e-06,
+      "loss": 0.0286,
+      "step": 590
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2844971716403961,
+      "learning_rate": 7.342326209192435e-06,
+      "loss": 0.0516,
+      "step": 591
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2881384491920471,
+      "learning_rate": 7.334086118081081e-06,
+      "loss": 0.0672,
+      "step": 592
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.20290230214595795,
+      "learning_rate": 7.3258379133119665e-06,
+      "loss": 0.0411,
+      "step": 593
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.26997798681259155,
+      "learning_rate": 7.317581623557177e-06,
+      "loss": 0.048,
+      "step": 594
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.296772301197052,
+      "learning_rate": 7.3093172775169e-06,
+      "loss": 0.0662,
+      "step": 595
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.3392322063446045,
+      "learning_rate": 7.3010449039193255e-06,
+      "loss": 0.0745,
+      "step": 596
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.3285764455795288,
+      "learning_rate": 7.2927645315205535e-06,
+      "loss": 0.0848,
+      "step": 597
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.20147007703781128,
+      "learning_rate": 7.284476189104486e-06,
+      "loss": 0.0544,
+      "step": 598
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.2233363389968872,
+      "learning_rate": 7.27617990548273e-06,
+      "loss": 0.048,
+      "step": 599
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.3063652217388153,
+      "learning_rate": 7.2678757094945e-06,
+      "loss": 0.0879,
+      "step": 600
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.34670838713645935,
+      "learning_rate": 7.259563630006512e-06,
+      "loss": 0.0753,
+      "step": 601
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2491820603609085,
+      "learning_rate": 7.251243695912886e-06,
+      "loss": 0.0492,
+      "step": 602
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2033078521490097,
+      "learning_rate": 7.242915936135052e-06,
+      "loss": 0.0468,
+      "step": 603
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.47694021463394165,
+      "learning_rate": 7.234580379621636e-06,
+      "loss": 0.1053,
+      "step": 604
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.25937044620513916,
+      "learning_rate": 7.226237055348369e-06,
+      "loss": 0.0632,
+      "step": 605
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.45980414748191833,
+      "learning_rate": 7.2178859923179855e-06,
+      "loss": 0.0611,
+      "step": 606
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2813967168331146,
+      "learning_rate": 7.20952721956012e-06,
+      "loss": 0.0826,
+      "step": 607
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.35964691638946533,
+      "learning_rate": 7.201160766131208e-06,
+      "loss": 0.0573,
+      "step": 608
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.4538634717464447,
+      "learning_rate": 7.192786661114384e-06,
+      "loss": 0.0615,
+      "step": 609
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.43191802501678467,
+      "learning_rate": 7.184404933619377e-06,
+      "loss": 0.071,
+      "step": 610
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.24433721601963043,
+      "learning_rate": 7.176015612782421e-06,
+      "loss": 0.032,
+      "step": 611
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.3911650478839874,
+      "learning_rate": 7.167618727766138e-06,
+      "loss": 0.0901,
+      "step": 612
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.25476253032684326,
+      "learning_rate": 7.1592143077594475e-06,
+      "loss": 0.0386,
+      "step": 613
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.2659028172492981,
+      "learning_rate": 7.1508023819774644e-06,
+      "loss": 0.0617,
+      "step": 614
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.26452651619911194,
+      "learning_rate": 7.142382979661386e-06,
+      "loss": 0.0489,
+      "step": 615
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.2677484154701233,
+      "learning_rate": 7.133956130078412e-06,
+      "loss": 0.0444,
+      "step": 616
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.2430453896522522,
+      "learning_rate": 7.12552186252162e-06,
+      "loss": 0.0632,
+      "step": 617
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.2421427220106125,
+      "learning_rate": 7.117080206309879e-06,
+      "loss": 0.0597,
+      "step": 618
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.45028990507125854,
+      "learning_rate": 7.1086311907877346e-06,
+      "loss": 0.0779,
+      "step": 619
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4141206443309784,
+      "learning_rate": 7.100174845325327e-06,
+      "loss": 0.0991,
+      "step": 620
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.23673787713050842,
+      "learning_rate": 7.091711199318265e-06,
+      "loss": 0.0728,
+      "step": 621
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.3629487454891205,
+      "learning_rate": 7.083240282187542e-06,
+      "loss": 0.0547,
+      "step": 622
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.23040874302387238,
+      "learning_rate": 7.074762123379424e-06,
+      "loss": 0.0398,
+      "step": 623
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.31620001792907715,
+      "learning_rate": 7.0662767523653515e-06,
+      "loss": 0.0509,
+      "step": 624
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.2546658217906952,
+      "learning_rate": 7.057784198641835e-06,
+      "loss": 0.0577,
+      "step": 625
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.3494987487792969,
+      "learning_rate": 7.0492844917303535e-06,
+      "loss": 0.088,
+      "step": 626
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.4441275894641876,
+      "learning_rate": 7.040777661177251e-06,
+      "loss": 0.06,
+      "step": 627
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.21604092419147491,
+      "learning_rate": 7.032263736553635e-06,
+      "loss": 0.0424,
+      "step": 628
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.2695225179195404,
+      "learning_rate": 7.0237427474552755e-06,
+      "loss": 0.0622,
+      "step": 629
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.2582091689109802,
+      "learning_rate": 7.015214723502496e-06,
+      "loss": 0.0777,
+      "step": 630
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.348019540309906,
+      "learning_rate": 7.006679694340074e-06,
+      "loss": 0.0666,
+      "step": 631
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.21345749497413635,
+      "learning_rate": 6.998137689637142e-06,
+      "loss": 0.0245,
+      "step": 632
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.19356510043144226,
+      "learning_rate": 6.989588739087079e-06,
+      "loss": 0.0445,
+      "step": 633
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.33389925956726074,
+      "learning_rate": 6.981032872407406e-06,
+      "loss": 0.0793,
+      "step": 634
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.21888035535812378,
+      "learning_rate": 6.972470119339692e-06,
+      "loss": 0.0583,
+      "step": 635
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.4273500442504883,
+      "learning_rate": 6.963900509649435e-06,
+      "loss": 0.0739,
+      "step": 636
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.2740592956542969,
+      "learning_rate": 6.955324073125979e-06,
+      "loss": 0.065,
+      "step": 637
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.4859815537929535,
+      "learning_rate": 6.946740839582388e-06,
+      "loss": 0.0498,
+      "step": 638
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.1959264874458313,
+      "learning_rate": 6.93815083885536e-06,
+      "loss": 0.0375,
+      "step": 639
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.28902924060821533,
+      "learning_rate": 6.929554100805118e-06,
+      "loss": 0.0493,
+      "step": 640
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.18830162286758423,
+      "learning_rate": 6.920950655315298e-06,
+      "loss": 0.0271,
+      "step": 641
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.31182587146759033,
+      "learning_rate": 6.912340532292861e-06,
+      "loss": 0.086,
+      "step": 642
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.4169754683971405,
+      "learning_rate": 6.903723761667973e-06,
+      "loss": 0.0639,
+      "step": 643
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.48527055978775024,
+      "learning_rate": 6.8951003733939125e-06,
+      "loss": 0.0852,
+      "step": 644
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.21746531128883362,
+      "learning_rate": 6.886470397446958e-06,
+      "loss": 0.06,
+      "step": 645
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.18846824765205383,
+      "learning_rate": 6.877833863826295e-06,
+      "loss": 0.0486,
+      "step": 646
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.24251620471477509,
+      "learning_rate": 6.869190802553895e-06,
+      "loss": 0.0663,
+      "step": 647
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.175017848610878,
+      "learning_rate": 6.860541243674427e-06,
+      "loss": 0.022,
+      "step": 648
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.33023568987846375,
+      "learning_rate": 6.8518852172551454e-06,
+      "loss": 0.0799,
+      "step": 649
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.4299830198287964,
+      "learning_rate": 6.843222753385785e-06,
+      "loss": 0.0401,
+      "step": 650
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.24229450523853302,
+      "learning_rate": 6.834553882178464e-06,
+      "loss": 0.0481,
+      "step": 651
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.21505345404148102,
+      "learning_rate": 6.825878633767564e-06,
+      "loss": 0.05,
+      "step": 652
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29249686002731323,
+      "learning_rate": 6.817197038309644e-06,
+      "loss": 0.0344,
+      "step": 653
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.3737218677997589,
+      "learning_rate": 6.808509125983321e-06,
+      "loss": 0.068,
+      "step": 654
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.2989976108074188,
+      "learning_rate": 6.799814926989171e-06,
+      "loss": 0.0583,
+      "step": 655
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.2579036056995392,
+      "learning_rate": 6.791114471549627e-06,
+      "loss": 0.0347,
+      "step": 656
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.19712986052036285,
+      "learning_rate": 6.782407789908864e-06,
+      "loss": 0.056,
+      "step": 657
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.3034021854400635,
+      "learning_rate": 6.773694912332706e-06,
+      "loss": 0.0694,
+      "step": 658
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.2495863288640976,
+      "learning_rate": 6.764975869108514e-06,
+      "loss": 0.0562,
+      "step": 659
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.22996436059474945,
+      "learning_rate": 6.756250690545079e-06,
+      "loss": 0.0519,
+      "step": 660
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.42392170429229736,
+      "learning_rate": 6.747519406972525e-06,
+      "loss": 0.0487,
+      "step": 661
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.502571702003479,
+      "learning_rate": 6.738782048742187e-06,
+      "loss": 0.0956,
+      "step": 662
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.4921998083591461,
+      "learning_rate": 6.730038646226532e-06,
+      "loss": 0.037,
+      "step": 663
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5764889717102051,
+      "learning_rate": 6.721289229819024e-06,
+      "loss": 0.0761,
+      "step": 664
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6852768659591675,
+      "learning_rate": 6.712533829934042e-06,
+      "loss": 0.0516,
+      "step": 665
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.32368728518486023,
+      "learning_rate": 6.703772477006758e-06,
+      "loss": 0.0279,
+      "step": 666
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3027501702308655,
+      "learning_rate": 6.6950052014930375e-06,
+      "loss": 0.0823,
+      "step": 667
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.21375176310539246,
+      "learning_rate": 6.686232033869343e-06,
+      "loss": 0.0565,
+      "step": 668
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.25710931420326233,
+      "learning_rate": 6.677453004632608e-06,
+      "loss": 0.0462,
+      "step": 669
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.3554043471813202,
+      "learning_rate": 6.6686681443001485e-06,
+      "loss": 0.0639,
+      "step": 670
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.44096800684928894,
+      "learning_rate": 6.659877483409545e-06,
+      "loss": 0.0585,
+      "step": 671
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.33739519119262695,
+      "learning_rate": 6.65108105251855e-06,
+      "loss": 0.0794,
+      "step": 672
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.3041416108608246,
+      "learning_rate": 6.6422788822049635e-06,
+      "loss": 0.0762,
+      "step": 673
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.45933637022972107,
+      "learning_rate": 6.633471003066543e-06,
+      "loss": 0.0652,
+      "step": 674
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.30925846099853516,
+      "learning_rate": 6.62465744572089e-06,
+      "loss": 0.0746,
+      "step": 675
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9105139970779419,
+      "learning_rate": 6.615838240805344e-06,
+      "loss": 0.1138,
+      "step": 676
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.33594557642936707,
+      "learning_rate": 6.607013418976874e-06,
+      "loss": 0.0375,
+      "step": 677
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.22061163187026978,
+      "learning_rate": 6.598183010911978e-06,
+      "loss": 0.0472,
+      "step": 678
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.3593815863132477,
+      "learning_rate": 6.5893470473065716e-06,
+      "loss": 0.0636,
+      "step": 679
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.29337939620018005,
+      "learning_rate": 6.580505558875878e-06,
+      "loss": 0.0529,
+      "step": 680
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.16969534754753113,
+      "learning_rate": 6.571658576354334e-06,
+      "loss": 0.0741,
+      "step": 681
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.23921915888786316,
+      "learning_rate": 6.562806130495467e-06,
+      "loss": 0.0423,
+      "step": 682
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.22711874544620514,
+      "learning_rate": 6.5539482520718e-06,
+      "loss": 0.0615,
+      "step": 683
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.17673003673553467,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.0521,
+      "step": 684
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.22866903245449066,
+      "learning_rate": 6.536216320714466e-06,
+      "loss": 0.0452,
+      "step": 685
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5617073178291321,
+      "learning_rate": 6.527342329419837e-06,
+      "loss": 0.0838,
+      "step": 686
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.45371171832084656,
+      "learning_rate": 6.518463028838271e-06,
+      "loss": 0.0485,
+      "step": 687
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.27730685472488403,
+      "learning_rate": 6.5095784498356365e-06,
+      "loss": 0.0348,
+      "step": 688
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.35711053013801575,
+      "learning_rate": 6.5006886232961585e-06,
+      "loss": 0.0639,
+      "step": 689
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5427911281585693,
+      "learning_rate": 6.491793580122301e-06,
+      "loss": 0.0729,
+      "step": 690
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.4389806091785431,
+      "learning_rate": 6.482893351234658e-06,
+      "loss": 0.0547,
+      "step": 691
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.2546471357345581,
+      "learning_rate": 6.473987967571855e-06,
+      "loss": 0.0419,
+      "step": 692
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.28974685072898865,
+      "learning_rate": 6.465077460090431e-06,
+      "loss": 0.035,
+      "step": 693
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.4222591519355774,
+      "learning_rate": 6.4561618597647445e-06,
+      "loss": 0.047,
+      "step": 694
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.45247283577919006,
+      "learning_rate": 6.447241197586847e-06,
+      "loss": 0.1075,
+      "step": 695
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5222399830818176,
+      "learning_rate": 6.438315504566397e-06,
+      "loss": 0.1131,
+      "step": 696
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.36488956212997437,
+      "learning_rate": 6.429384811730528e-06,
+      "loss": 0.0498,
+      "step": 697
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.39750203490257263,
+      "learning_rate": 6.420449150123768e-06,
+      "loss": 0.0623,
+      "step": 698
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.27273041009902954,
+      "learning_rate": 6.411508550807905e-06,
+      "loss": 0.058,
+      "step": 699
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.4950941205024719,
+      "learning_rate": 6.402563044861899e-06,
+      "loss": 0.0491,
+      "step": 700
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.30463680624961853,
+      "learning_rate": 6.393612663381763e-06,
+      "loss": 0.0677,
+      "step": 701
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.2015472948551178,
+      "learning_rate": 6.384657437480458e-06,
+      "loss": 0.0454,
+      "step": 702
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.2115863561630249,
+      "learning_rate": 6.375697398287788e-06,
+      "loss": 0.0437,
+      "step": 703
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.1840757578611374,
+      "learning_rate": 6.3667325769502845e-06,
+      "loss": 0.0712,
+      "step": 704
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3236899971961975,
+      "learning_rate": 6.357763004631104e-06,
+      "loss": 0.057,
+      "step": 705
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.2529219686985016,
+      "learning_rate": 6.34878871250992e-06,
+      "loss": 0.0453,
+      "step": 706
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.20748858153820038,
+      "learning_rate": 6.3398097317828114e-06,
+      "loss": 0.0433,
+      "step": 707
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3028950095176697,
+      "learning_rate": 6.330826093662157e-06,
+      "loss": 0.0545,
+      "step": 708
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.2200806140899658,
+      "learning_rate": 6.321837829376521e-06,
+      "loss": 0.1283,
+      "step": 709
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.44462844729423523,
+      "learning_rate": 6.312844970170551e-06,
+      "loss": 0.0866,
+      "step": 710
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 0.05823693051934242,
+      "eval_runtime": 14.7341,
+      "eval_samples_per_second": 32.306,
+      "eval_steps_per_second": 8.076,
+      "step": 710
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.23898513615131378,
+      "learning_rate": 6.303847547304872e-06,
+      "loss": 0.0644,
+      "step": 711
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.2102598398923874,
+      "learning_rate": 6.294845592055967e-06,
+      "loss": 0.065,
+      "step": 712
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.13285306096076965,
+      "learning_rate": 6.2858391357160785e-06,
+      "loss": 0.0537,
+      "step": 713
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5347241163253784,
+      "learning_rate": 6.27682820959309e-06,
+      "loss": 0.0458,
+      "step": 714
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.4614933431148529,
+      "learning_rate": 6.267812845010431e-06,
+      "loss": 0.07,
+      "step": 715
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.2043062001466751,
+      "learning_rate": 6.258793073306949e-06,
+      "loss": 0.0506,
+      "step": 716
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.2690112292766571,
+      "learning_rate": 6.2497689258368225e-06,
+      "loss": 0.0924,
+      "step": 717
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.3479662537574768,
+      "learning_rate": 6.2407404339694324e-06,
+      "loss": 0.0802,
+      "step": 718
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.20581212639808655,
+      "learning_rate": 6.231707629089263e-06,
+      "loss": 0.0677,
+      "step": 719
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.22926701605319977,
+      "learning_rate": 6.2226705425958e-06,
+      "loss": 0.0446,
+      "step": 720
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.39380067586898804,
+      "learning_rate": 6.2136292059034e-06,
+      "loss": 0.0418,
+      "step": 721
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.22926107048988342,
+      "learning_rate": 6.204583650441201e-06,
+      "loss": 0.0553,
+      "step": 722
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.3458244502544403,
+      "learning_rate": 6.1955339076530045e-06,
+      "loss": 0.0834,
+      "step": 723
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5059794783592224,
+      "learning_rate": 6.18648000899717e-06,
+      "loss": 0.0524,
+      "step": 724
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.19327516853809357,
+      "learning_rate": 6.177421985946499e-06,
+      "loss": 0.0324,
+      "step": 725
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.27781131863594055,
+      "learning_rate": 6.168359869988134e-06,
+      "loss": 0.0677,
+      "step": 726
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6419954299926758,
+      "learning_rate": 6.159293692623443e-06,
+      "loss": 0.1251,
+      "step": 727
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.1929435431957245,
+      "learning_rate": 6.150223485367914e-06,
+      "loss": 0.0547,
+      "step": 728
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.39410218596458435,
+      "learning_rate": 6.141149279751043e-06,
+      "loss": 0.0516,
+      "step": 729
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.2835167348384857,
+      "learning_rate": 6.132071107316221e-06,
+      "loss": 0.0719,
+      "step": 730
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.4182420074939728,
+      "learning_rate": 6.122988999620634e-06,
+      "loss": 0.0346,
+      "step": 731
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.3160066604614258,
+      "learning_rate": 6.113902988235145e-06,
+      "loss": 0.08,
+      "step": 732
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.24593767523765564,
+      "learning_rate": 6.1048131047441876e-06,
+      "loss": 0.052,
+      "step": 733
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.2121860831975937,
+      "learning_rate": 6.095719380745654e-06,
+      "loss": 0.0437,
+      "step": 734
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.605929970741272,
+      "learning_rate": 6.0866218478507875e-06,
+      "loss": 0.0886,
+      "step": 735
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5460467338562012,
+      "learning_rate": 6.0775205376840715e-06,
+      "loss": 0.0617,
+      "step": 736
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.2423592209815979,
+      "learning_rate": 6.068415481883122e-06,
+      "loss": 0.0599,
+      "step": 737
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.38910943269729614,
+      "learning_rate": 6.059306712098571e-06,
+      "loss": 0.0756,
+      "step": 738
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.16872254014015198,
+      "learning_rate": 6.050194259993967e-06,
+      "loss": 0.0342,
+      "step": 739
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.44383174180984497,
+      "learning_rate": 6.041078157245649e-06,
+      "loss": 0.1051,
+      "step": 740
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5203137397766113,
+      "learning_rate": 6.031958435542659e-06,
+      "loss": 0.1132,
+      "step": 741
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.614219605922699,
+      "learning_rate": 6.022835126586609e-06,
+      "loss": 0.0534,
+      "step": 742
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.1791837066411972,
+      "learning_rate": 6.0137082620915865e-06,
+      "loss": 0.0508,
+      "step": 743
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.28818053007125854,
+      "learning_rate": 6.004577873784035e-06,
+      "loss": 0.0623,
+      "step": 744
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.43602147698402405,
+      "learning_rate": 5.995443993402647e-06,
+      "loss": 0.0659,
+      "step": 745
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.2554510235786438,
+      "learning_rate": 5.986306652698261e-06,
+      "loss": 0.0762,
+      "step": 746
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.22898496687412262,
+      "learning_rate": 5.977165883433734e-06,
+      "loss": 0.0363,
+      "step": 747
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5783344507217407,
+      "learning_rate": 5.968021717383849e-06,
+      "loss": 0.0927,
+      "step": 748
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.25347182154655457,
+      "learning_rate": 5.958874186335193e-06,
+      "loss": 0.0391,
+      "step": 749
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.2889784872531891,
+      "learning_rate": 5.949723322086053e-06,
+      "loss": 0.0589,
+      "step": 750
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.3338702321052551,
+      "learning_rate": 5.940569156446299e-06,
+      "loss": 0.0779,
+      "step": 751
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.23281054198741913,
+      "learning_rate": 5.931411721237279e-06,
+      "loss": 0.0398,
+      "step": 752
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.24785956740379333,
+      "learning_rate": 5.9222510482917075e-06,
+      "loss": 0.0445,
+      "step": 753
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.42631295323371887,
+      "learning_rate": 5.9130871694535545e-06,
+      "loss": 0.0522,
+      "step": 754
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.4015826880931854,
+      "learning_rate": 5.9039201165779315e-06,
+      "loss": 0.0434,
+      "step": 755
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.2290300726890564,
+      "learning_rate": 5.8947499215309836e-06,
+      "loss": 0.0587,
+      "step": 756
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.328885942697525,
+      "learning_rate": 5.885576616189781e-06,
+      "loss": 0.049,
+      "step": 757
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5737412571907043,
+      "learning_rate": 5.876400232442206e-06,
+      "loss": 0.0765,
+      "step": 758
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3481307327747345,
+      "learning_rate": 5.867220802186837e-06,
+      "loss": 0.0432,
+      "step": 759
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.20464670658111572,
+      "learning_rate": 5.858038357332851e-06,
+      "loss": 0.0589,
+      "step": 760
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.645138144493103,
+      "learning_rate": 5.8488529297998946e-06,
+      "loss": 0.0763,
+      "step": 761
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.22710151970386505,
+      "learning_rate": 5.839664551517989e-06,
+      "loss": 0.0452,
+      "step": 762
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.2501658499240875,
+      "learning_rate": 5.83047325442741e-06,
+      "loss": 0.0392,
+      "step": 763
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.40542712807655334,
+      "learning_rate": 5.821279070478582e-06,
+      "loss": 0.0449,
+      "step": 764
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.49357643723487854,
+      "learning_rate": 5.812082031631966e-06,
+      "loss": 0.0848,
+      "step": 765
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.33942776918411255,
+      "learning_rate": 5.8028821698579385e-06,
+      "loss": 0.0674,
+      "step": 766
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.33811476826667786,
+      "learning_rate": 5.7936795171367e-06,
+      "loss": 0.0854,
+      "step": 767
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.35965660214424133,
+      "learning_rate": 5.784474105458143e-06,
+      "loss": 0.0532,
+      "step": 768
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.27447643876075745,
+      "learning_rate": 5.77526596682176e-06,
+      "loss": 0.0496,
+      "step": 769
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.3869398534297943,
+      "learning_rate": 5.766055133236513e-06,
+      "loss": 0.09,
+      "step": 770
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.3004852831363678,
+      "learning_rate": 5.75684163672074e-06,
+      "loss": 0.0729,
+      "step": 771
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.30032235383987427,
+      "learning_rate": 5.747625509302032e-06,
+      "loss": 0.0611,
+      "step": 772
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3252098858356476,
+      "learning_rate": 5.7384067830171276e-06,
+      "loss": 0.074,
+      "step": 773
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.43575194478034973,
+      "learning_rate": 5.729185489911797e-06,
+      "loss": 0.065,
+      "step": 774
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.2744060158729553,
+      "learning_rate": 5.7199616620407325e-06,
+      "loss": 0.0684,
+      "step": 775
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.47385725378990173,
+      "learning_rate": 5.710735331467444e-06,
+      "loss": 0.1052,
+      "step": 776
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.23851534724235535,
+      "learning_rate": 5.701506530264133e-06,
+      "loss": 0.0372,
+      "step": 777
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.44428861141204834,
+      "learning_rate": 5.692275290511592e-06,
+      "loss": 0.0685,
+      "step": 778
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.23831063508987427,
+      "learning_rate": 5.683041644299094e-06,
+      "loss": 0.055,
+      "step": 779
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.3996879458427429,
+      "learning_rate": 5.673805623724272e-06,
+      "loss": 0.0709,
+      "step": 780
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.29323792457580566,
+      "learning_rate": 5.664567260893019e-06,
+      "loss": 0.0638,
+      "step": 781
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.2788830101490021,
+      "learning_rate": 5.655326587919361e-06,
+      "loss": 0.0652,
+      "step": 782
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.2126566469669342,
+      "learning_rate": 5.646083636925363e-06,
+      "loss": 0.0463,
+      "step": 783
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.3818773925304413,
+      "learning_rate": 5.636838440041004e-06,
+      "loss": 0.058,
+      "step": 784
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.20874321460723877,
+      "learning_rate": 5.627591029404072e-06,
+      "loss": 0.069,
+      "step": 785
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.25264501571655273,
+      "learning_rate": 5.61834143716005e-06,
+      "loss": 0.0669,
+      "step": 786
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.22084520757198334,
+      "learning_rate": 5.609089695462002e-06,
+      "loss": 0.0435,
+      "step": 787
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.22601231932640076,
+      "learning_rate": 5.599835836470469e-06,
+      "loss": 0.0571,
+      "step": 788
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.19402346014976501,
+      "learning_rate": 5.590579892353348e-06,
+      "loss": 0.0418,
+      "step": 789
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.40445828437805176,
+      "learning_rate": 5.581321895285787e-06,
+      "loss": 0.0777,
+      "step": 790
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.25165826082229614,
+      "learning_rate": 5.572061877450068e-06,
+      "loss": 0.0469,
+      "step": 791
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.22140294313430786,
+      "learning_rate": 5.562799871035496e-06,
+      "loss": 0.059,
+      "step": 792
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.3147905170917511,
+      "learning_rate": 5.553535908238295e-06,
+      "loss": 0.0815,
+      "step": 793
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.2372606098651886,
+      "learning_rate": 5.544270021261483e-06,
+      "loss": 0.0699,
+      "step": 794
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5909821391105652,
+      "learning_rate": 5.535002242314772e-06,
+      "loss": 0.0957,
+      "step": 795
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.3787746727466583,
+      "learning_rate": 5.5257326036144446e-06,
+      "loss": 0.0716,
+      "step": 796
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.20804928243160248,
+      "learning_rate": 5.516461137383254e-06,
+      "loss": 0.0392,
+      "step": 797
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.20450638234615326,
+      "learning_rate": 5.507187875850305e-06,
+      "loss": 0.0365,
+      "step": 798
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.30582720041275024,
+      "learning_rate": 5.497912851250941e-06,
+      "loss": 0.0455,
+      "step": 799
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.3252048194408417,
+      "learning_rate": 5.488636095826636e-06,
+      "loss": 0.0542,
+      "step": 800
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.44586294889450073,
+      "learning_rate": 5.4793576418248775e-06,
+      "loss": 0.0769,
+      "step": 801
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.21377098560333252,
+      "learning_rate": 5.470077521499063e-06,
+      "loss": 0.0517,
+      "step": 802
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.39969006180763245,
+      "learning_rate": 5.460795767108379e-06,
+      "loss": 0.0743,
+      "step": 803
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.2593373656272888,
+      "learning_rate": 5.451512410917691e-06,
+      "loss": 0.0364,
+      "step": 804
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.2312588095664978,
+      "learning_rate": 5.4422274851974356e-06,
+      "loss": 0.0743,
+      "step": 805
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3301038444042206,
+      "learning_rate": 5.432941022223503e-06,
+      "loss": 0.0278,
+      "step": 806
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.1854841262102127,
+      "learning_rate": 5.42365305427713e-06,
+      "loss": 0.0617,
+      "step": 807
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.2691740095615387,
+      "learning_rate": 5.414363613644782e-06,
+      "loss": 0.0476,
+      "step": 808
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3060298562049866,
+      "learning_rate": 5.4050727326180426e-06,
+      "loss": 0.053,
+      "step": 809
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.34074532985687256,
+      "learning_rate": 5.395780443493508e-06,
+      "loss": 0.0706,
+      "step": 810
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7154234051704407,
+      "learning_rate": 5.386486778572665e-06,
+      "loss": 0.0592,
+      "step": 811
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2346647083759308,
+      "learning_rate": 5.377191770161783e-06,
+      "loss": 0.0525,
+      "step": 812
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.46330153942108154,
+      "learning_rate": 5.3678954505718016e-06,
+      "loss": 0.0729,
+      "step": 813
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.3810093402862549,
+      "learning_rate": 5.358597852118219e-06,
+      "loss": 0.0993,
+      "step": 814
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.22972767055034637,
+      "learning_rate": 5.34929900712098e-06,
+      "loss": 0.0517,
+      "step": 815
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2932167053222656,
+      "learning_rate": 5.339998947904362e-06,
+      "loss": 0.0583,
+      "step": 816
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.43843695521354675,
+      "learning_rate": 5.330697706796861e-06,
+      "loss": 0.0528,
+      "step": 817
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.29898717999458313,
+      "learning_rate": 5.3213953161310825e-06,
+      "loss": 0.0722,
+      "step": 818
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.23077327013015747,
+      "learning_rate": 5.3120918082436315e-06,
+      "loss": 0.045,
+      "step": 819
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.31443125009536743,
+      "learning_rate": 5.302787215474992e-06,
+      "loss": 0.0686,
+      "step": 820
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.30536675453186035,
+      "learning_rate": 5.293481570169421e-06,
+      "loss": 0.056,
+      "step": 821
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.8416706919670105,
+      "learning_rate": 5.284174904674835e-06,
+      "loss": 0.1033,
+      "step": 822
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.35498178005218506,
+      "learning_rate": 5.274867251342694e-06,
+      "loss": 0.0428,
+      "step": 823
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.3787842094898224,
+      "learning_rate": 5.265558642527897e-06,
+      "loss": 0.0896,
+      "step": 824
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.3150663673877716,
+      "learning_rate": 5.256249110588659e-06,
+      "loss": 0.0612,
+      "step": 825
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.4420692026615143,
+      "learning_rate": 5.246938687886409e-06,
+      "loss": 0.102,
+      "step": 826
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5929310321807861,
+      "learning_rate": 5.237627406785667e-06,
+      "loss": 0.0862,
+      "step": 827
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.0055999755859375,
+      "learning_rate": 5.228315299653942e-06,
+      "loss": 0.1209,
+      "step": 828
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.37658628821372986,
+      "learning_rate": 5.219002398861611e-06,
+      "loss": 0.0911,
+      "step": 829
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.26989948749542236,
+      "learning_rate": 5.209688736781811e-06,
+      "loss": 0.0696,
+      "step": 830
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.24650172889232635,
+      "learning_rate": 5.200374345790326e-06,
+      "loss": 0.0712,
+      "step": 831
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.2733075022697449,
+      "learning_rate": 5.1910592582654715e-06,
+      "loss": 0.0739,
+      "step": 832
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.20648646354675293,
+      "learning_rate": 5.18174350658799e-06,
+      "loss": 0.0509,
+      "step": 833
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.5087945461273193,
+      "learning_rate": 5.172427123140923e-06,
+      "loss": 0.1139,
+      "step": 834
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.27857109904289246,
+      "learning_rate": 5.163110140309518e-06,
+      "loss": 0.0565,
+      "step": 835
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.2023928463459015,
+      "learning_rate": 5.1537925904811004e-06,
+      "loss": 0.0718,
+      "step": 836
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.2805935740470886,
+      "learning_rate": 5.144474506044968e-06,
+      "loss": 0.0558,
+      "step": 837
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.22690273821353912,
+      "learning_rate": 5.13515591939228e-06,
+      "loss": 0.0534,
+      "step": 838
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.2374301701784134,
+      "learning_rate": 5.125836862915934e-06,
+      "loss": 0.0447,
+      "step": 839
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.2171458601951599,
+      "learning_rate": 5.116517369010467e-06,
+      "loss": 0.0429,
+      "step": 840
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.22108136117458344,
+      "learning_rate": 5.107197470071933e-06,
+      "loss": 0.0556,
+      "step": 841
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.3796743154525757,
+      "learning_rate": 5.0978771984978e-06,
+      "loss": 0.0854,
+      "step": 842
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.2067977786064148,
+      "learning_rate": 5.088556586686822e-06,
+      "loss": 0.0368,
+      "step": 843
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.3032953441143036,
+      "learning_rate": 5.079235667038944e-06,
+      "loss": 0.0785,
+      "step": 844
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.46437495946884155,
+      "learning_rate": 5.069914471955179e-06,
+      "loss": 0.0834,
+      "step": 845
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.25711363554000854,
+      "learning_rate": 5.060593033837493e-06,
+      "loss": 0.0309,
+      "step": 846
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.4127390384674072,
+      "learning_rate": 5.051271385088702e-06,
+      "loss": 0.0653,
+      "step": 847
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.2039380520582199,
+      "learning_rate": 5.041949558112351e-06,
+      "loss": 0.0469,
+      "step": 848
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.1909148395061493,
+      "learning_rate": 5.032627585312608e-06,
+      "loss": 0.0569,
+      "step": 849
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3275524973869324,
+      "learning_rate": 5.023305499094145e-06,
+      "loss": 0.075,
+      "step": 850
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.2979578971862793,
+      "learning_rate": 5.013983331862027e-06,
+      "loss": 0.0748,
+      "step": 851
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.36494141817092896,
+      "learning_rate": 5.004661116021605e-06,
+      "loss": 0.0705,
+      "step": 852
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 0.05927066504955292,
+      "eval_runtime": 14.7269,
+      "eval_samples_per_second": 32.322,
+      "eval_steps_per_second": 8.08,
+      "step": 852
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.2674633860588074,
+      "learning_rate": 4.995338883978396e-06,
+      "loss": 0.0612,
+      "step": 853
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.22576095163822174,
+      "learning_rate": 4.986016668137975e-06,
+      "loss": 0.0552,
+      "step": 854
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.3576613664627075,
+      "learning_rate": 4.976694500905858e-06,
+      "loss": 0.0536,
+      "step": 855
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.4389832615852356,
+      "learning_rate": 4.967372414687394e-06,
+      "loss": 0.0794,
+      "step": 856
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.40479084849357605,
+      "learning_rate": 4.958050441887651e-06,
+      "loss": 0.0661,
+      "step": 857
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.2826189398765564,
+      "learning_rate": 4.948728614911299e-06,
+      "loss": 0.0494,
+      "step": 858
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.24314090609550476,
+      "learning_rate": 4.939406966162508e-06,
+      "loss": 0.0708,
+      "step": 859
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.2984601557254791,
+      "learning_rate": 4.930085528044823e-06,
+      "loss": 0.092,
+      "step": 860
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.19549059867858887,
+      "learning_rate": 4.9207643329610565e-06,
+      "loss": 0.0616,
+      "step": 861
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.28568658232688904,
+      "learning_rate": 4.911443413313179e-06,
+      "loss": 0.0416,
+      "step": 862
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.49393466114997864,
+      "learning_rate": 4.902122801502202e-06,
+      "loss": 0.0493,
+      "step": 863
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.21126003563404083,
+      "learning_rate": 4.892802529928067e-06,
+      "loss": 0.0302,
+      "step": 864
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.21899864077568054,
+      "learning_rate": 4.883482630989536e-06,
+      "loss": 0.0364,
+      "step": 865
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.2775663733482361,
+      "learning_rate": 4.874163137084068e-06,
+      "loss": 0.0582,
+      "step": 866
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.472811222076416,
+      "learning_rate": 4.864844080607723e-06,
+      "loss": 0.0564,
+      "step": 867
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.2039356380701065,
+      "learning_rate": 4.855525493955033e-06,
+      "loss": 0.0583,
+      "step": 868
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.2668206989765167,
+      "learning_rate": 4.8462074095188995e-06,
+      "loss": 0.036,
+      "step": 869
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.29921385645866394,
+      "learning_rate": 4.8368898596904834e-06,
+      "loss": 0.047,
+      "step": 870
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.2845795154571533,
+      "learning_rate": 4.827572876859078e-06,
+      "loss": 0.0254,
+      "step": 871
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.361094206571579,
+      "learning_rate": 4.818256493412011e-06,
+      "loss": 0.0297,
+      "step": 872
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.6772926449775696,
+      "learning_rate": 4.80894074173453e-06,
+      "loss": 0.0751,
+      "step": 873
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.14277713000774384,
+      "learning_rate": 4.799625654209675e-06,
+      "loss": 0.0156,
+      "step": 874
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.32318535447120667,
+      "learning_rate": 4.790311263218191e-06,
+      "loss": 0.0722,
+      "step": 875
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6377431750297546,
+      "learning_rate": 4.7809976011383905e-06,
+      "loss": 0.0614,
+      "step": 876
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.26989302039146423,
+      "learning_rate": 4.771684700346059e-06,
+      "loss": 0.071,
+      "step": 877
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.25494495034217834,
+      "learning_rate": 4.762372593214335e-06,
+      "loss": 0.046,
+      "step": 878
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.2811807096004486,
+      "learning_rate": 4.753061312113592e-06,
+      "loss": 0.0369,
+      "step": 879
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.1959286481142044,
+      "learning_rate": 4.743750889411342e-06,
+      "loss": 0.0342,
+      "step": 880
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.7420495748519897,
+      "learning_rate": 4.734441357472105e-06,
+      "loss": 0.0808,
+      "step": 881
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.28202396631240845,
+      "learning_rate": 4.725132748657307e-06,
+      "loss": 0.0277,
+      "step": 882
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.2835392653942108,
+      "learning_rate": 4.715825095325168e-06,
+      "loss": 0.0685,
+      "step": 883
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.2612256109714508,
+      "learning_rate": 4.70651842983058e-06,
+      "loss": 0.0442,
+      "step": 884
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5025742053985596,
+      "learning_rate": 4.697212784525009e-06,
+      "loss": 0.0935,
+      "step": 885
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.46131956577301025,
+      "learning_rate": 4.687908191756369e-06,
+      "loss": 0.0718,
+      "step": 886
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.2350883036851883,
+      "learning_rate": 4.678604683868918e-06,
+      "loss": 0.0404,
+      "step": 887
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.4978535771369934,
+      "learning_rate": 4.6693022932031415e-06,
+      "loss": 0.042,
+      "step": 888
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.1883852779865265,
+      "learning_rate": 4.660001052095639e-06,
+      "loss": 0.0417,
+      "step": 889
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5793114900588989,
+      "learning_rate": 4.65070099287902e-06,
+      "loss": 0.0787,
+      "step": 890
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4304198920726776,
+      "learning_rate": 4.641402147881782e-06,
+      "loss": 0.0608,
+      "step": 891
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.22804735600948334,
+      "learning_rate": 4.6321045494282e-06,
+      "loss": 0.0373,
+      "step": 892
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4823177754878998,
+      "learning_rate": 4.62280822983822e-06,
+      "loss": 0.0896,
+      "step": 893
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.048827052116394,
+      "learning_rate": 4.613513221427337e-06,
+      "loss": 0.0711,
+      "step": 894
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.2886340618133545,
+      "learning_rate": 4.604219556506492e-06,
+      "loss": 0.031,
+      "step": 895
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.5323426723480225,
+      "learning_rate": 4.594927267381958e-06,
+      "loss": 0.1087,
+      "step": 896
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.2986535429954529,
+      "learning_rate": 4.58563638635522e-06,
+      "loss": 0.0636,
+      "step": 897
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.24155835807323456,
+      "learning_rate": 4.57634694572287e-06,
+      "loss": 0.0415,
+      "step": 898
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.23381805419921875,
+      "learning_rate": 4.567058977776498e-06,
+      "loss": 0.0354,
+      "step": 899
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.17703460156917572,
+      "learning_rate": 4.557772514802564e-06,
+      "loss": 0.0408,
+      "step": 900
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.3805024325847626,
+      "learning_rate": 4.548487589082311e-06,
+      "loss": 0.0612,
+      "step": 901
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.341226726770401,
+      "learning_rate": 4.539204232891622e-06,
+      "loss": 0.0843,
+      "step": 902
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.38434794545173645,
+      "learning_rate": 4.529922478500938e-06,
+      "loss": 0.0444,
+      "step": 903
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.2903810739517212,
+      "learning_rate": 4.520642358175125e-06,
+      "loss": 0.0323,
+      "step": 904
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.30602797865867615,
+      "learning_rate": 4.511363904173366e-06,
+      "loss": 0.0592,
+      "step": 905
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5753944516181946,
+      "learning_rate": 4.502087148749061e-06,
+      "loss": 0.0891,
+      "step": 906
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.29496291279792786,
+      "learning_rate": 4.492812124149696e-06,
+      "loss": 0.0684,
+      "step": 907
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.348178505897522,
+      "learning_rate": 4.483538862616747e-06,
+      "loss": 0.0948,
+      "step": 908
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.2786281108856201,
+      "learning_rate": 4.474267396385558e-06,
+      "loss": 0.0505,
+      "step": 909
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.4709155559539795,
+      "learning_rate": 4.46499775768523e-06,
+      "loss": 0.0698,
+      "step": 910
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.23245489597320557,
+      "learning_rate": 4.4557299787385175e-06,
+      "loss": 0.0348,
+      "step": 911
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.22803007066249847,
+      "learning_rate": 4.446464091761706e-06,
+      "loss": 0.0523,
+      "step": 912
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.4116624593734741,
+      "learning_rate": 4.437200128964505e-06,
+      "loss": 0.087,
+      "step": 913
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.26835620403289795,
+      "learning_rate": 4.427938122549935e-06,
+      "loss": 0.0751,
+      "step": 914
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.23349997401237488,
+      "learning_rate": 4.418678104714214e-06,
+      "loss": 0.0575,
+      "step": 915
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.37686672806739807,
+      "learning_rate": 4.409420107646652e-06,
+      "loss": 0.0421,
+      "step": 916
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.17440485954284668,
+      "learning_rate": 4.400164163529532e-06,
+      "loss": 0.0545,
+      "step": 917
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3214772939682007,
+      "learning_rate": 4.390910304537999e-06,
+      "loss": 0.0605,
+      "step": 918
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.5040317177772522,
+      "learning_rate": 4.381658562839954e-06,
+      "loss": 0.0842,
+      "step": 919
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.21060919761657715,
+      "learning_rate": 4.372408970595931e-06,
+      "loss": 0.0416,
+      "step": 920
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.279563307762146,
+      "learning_rate": 4.363161559958996e-06,
+      "loss": 0.0385,
+      "step": 921
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.3074188530445099,
+      "learning_rate": 4.353916363074638e-06,
+      "loss": 0.0521,
+      "step": 922
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.32682177424430847,
+      "learning_rate": 4.34467341208064e-06,
+      "loss": 0.0619,
+      "step": 923
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.2511276602745056,
+      "learning_rate": 4.335432739106983e-06,
+      "loss": 0.0349,
+      "step": 924
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.4092172682285309,
+      "learning_rate": 4.326194376275729e-06,
+      "loss": 0.0671,
+      "step": 925
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5157366394996643,
+      "learning_rate": 4.316958355700906e-06,
+      "loss": 0.082,
+      "step": 926
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.2773836553096771,
+      "learning_rate": 4.307724709488409e-06,
+      "loss": 0.0346,
+      "step": 927
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32867851853370667,
+      "learning_rate": 4.2984934697358695e-06,
+      "loss": 0.0746,
+      "step": 928
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.29042449593544006,
+      "learning_rate": 4.2892646685325575e-06,
+      "loss": 0.06,
+      "step": 929
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.22214840352535248,
+      "learning_rate": 4.280038337959268e-06,
+      "loss": 0.0406,
+      "step": 930
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.3768647611141205,
+      "learning_rate": 4.270814510088203e-06,
+      "loss": 0.0633,
+      "step": 931
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.22813941538333893,
+      "learning_rate": 4.261593216982874e-06,
+      "loss": 0.0676,
+      "step": 932
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.24750201404094696,
+      "learning_rate": 4.2523744906979684e-06,
+      "loss": 0.0681,
+      "step": 933
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.538628101348877,
+      "learning_rate": 4.243158363279261e-06,
+      "loss": 0.0895,
+      "step": 934
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.32601863145828247,
+      "learning_rate": 4.2339448667634885e-06,
+      "loss": 0.0547,
+      "step": 935
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.33620086312294006,
+      "learning_rate": 4.224734033178242e-06,
+      "loss": 0.0581,
+      "step": 936
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.41226306557655334,
+      "learning_rate": 4.215525894541856e-06,
+      "loss": 0.0638,
+      "step": 937
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.42651426792144775,
+      "learning_rate": 4.206320482863301e-06,
+      "loss": 0.0627,
+      "step": 938
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.18828719854354858,
+      "learning_rate": 4.1971178301420615e-06,
+      "loss": 0.0237,
+      "step": 939
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.1825932115316391,
+      "learning_rate": 4.187917968368036e-06,
+      "loss": 0.0409,
+      "step": 940
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.168092742562294,
+      "learning_rate": 4.1787209295214186e-06,
+      "loss": 0.0442,
+      "step": 941
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.31606152653694153,
+      "learning_rate": 4.1695267455725904e-06,
+      "loss": 0.053,
+      "step": 942
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.22704976797103882,
+      "learning_rate": 4.160335448482014e-06,
+      "loss": 0.0483,
+      "step": 943
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.26612523198127747,
+      "learning_rate": 4.151147070200108e-06,
+      "loss": 0.0632,
+      "step": 944
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5621999502182007,
+      "learning_rate": 4.141961642667152e-06,
+      "loss": 0.0786,
+      "step": 945
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.26850804686546326,
+      "learning_rate": 4.132779197813165e-06,
+      "loss": 0.0448,
+      "step": 946
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.6089004278182983,
+      "learning_rate": 4.123599767557795e-06,
+      "loss": 0.0744,
+      "step": 947
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.19964046776294708,
+      "learning_rate": 4.11442338381022e-06,
+      "loss": 0.0297,
+      "step": 948
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.360899955034256,
+      "learning_rate": 4.105250078469018e-06,
+      "loss": 0.0699,
+      "step": 949
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.45353251695632935,
+      "learning_rate": 4.09607988342207e-06,
+      "loss": 0.0543,
+      "step": 950
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.26752933859825134,
+      "learning_rate": 4.086912830546448e-06,
+      "loss": 0.0612,
+      "step": 951
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.57179856300354,
+      "learning_rate": 4.0777489517082925e-06,
+      "loss": 0.0855,
+      "step": 952
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.36693987250328064,
+      "learning_rate": 4.068588278762723e-06,
+      "loss": 0.0508,
+      "step": 953
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.2643117606639862,
+      "learning_rate": 4.059430843553703e-06,
+      "loss": 0.0332,
+      "step": 954
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3400687277317047,
+      "learning_rate": 4.0502766779139485e-06,
+      "loss": 0.0736,
+      "step": 955
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.20526453852653503,
+      "learning_rate": 4.041125813664809e-06,
+      "loss": 0.0319,
+      "step": 956
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.28753894567489624,
+      "learning_rate": 4.0319782826161516e-06,
+      "loss": 0.0782,
+      "step": 957
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.22514447569847107,
+      "learning_rate": 4.022834116566269e-06,
+      "loss": 0.0407,
+      "step": 958
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.27989277243614197,
+      "learning_rate": 4.013693347301741e-06,
+      "loss": 0.0438,
+      "step": 959
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.33890578150749207,
+      "learning_rate": 4.0045560065973535e-06,
+      "loss": 0.0524,
+      "step": 960
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.32162007689476013,
+      "learning_rate": 3.995422126215968e-06,
+      "loss": 0.0756,
+      "step": 961
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.4270891845226288,
+      "learning_rate": 3.986291737908414e-06,
+      "loss": 0.0644,
+      "step": 962
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.7270297408103943,
+      "learning_rate": 3.977164873413391e-06,
+      "loss": 0.075,
+      "step": 963
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.47613292932510376,
+      "learning_rate": 3.968041564457342e-06,
+      "loss": 0.0467,
+      "step": 964
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.17602606117725372,
+      "learning_rate": 3.958921842754351e-06,
+      "loss": 0.0412,
+      "step": 965
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.23278112709522247,
+      "learning_rate": 3.949805740006037e-06,
+      "loss": 0.0569,
+      "step": 966
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.6249252557754517,
+      "learning_rate": 3.94069328790143e-06,
+      "loss": 0.1109,
+      "step": 967
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.3838892877101898,
+      "learning_rate": 3.931584518116878e-06,
+      "loss": 0.0634,
+      "step": 968
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2674984931945801,
+      "learning_rate": 3.922479462315929e-06,
+      "loss": 0.0396,
+      "step": 969
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.7437862753868103,
+      "learning_rate": 3.913378152149214e-06,
+      "loss": 0.1037,
+      "step": 970
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.22365398705005646,
+      "learning_rate": 3.904280619254348e-06,
+      "loss": 0.0472,
+      "step": 971
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.3518374264240265,
+      "learning_rate": 3.895186895255814e-06,
+      "loss": 0.0822,
+      "step": 972
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4127620458602905,
+      "learning_rate": 3.886097011764856e-06,
+      "loss": 0.0435,
+      "step": 973
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.20354340970516205,
+      "learning_rate": 3.877011000379367e-06,
+      "loss": 0.0277,
+      "step": 974
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.23527513444423676,
+      "learning_rate": 3.86792889268378e-06,
+      "loss": 0.0511,
+      "step": 975
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.2532998025417328,
+      "learning_rate": 3.858850720248959e-06,
+      "loss": 0.0565,
+      "step": 976
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.22200430929660797,
+      "learning_rate": 3.8497765146320874e-06,
+      "loss": 0.0542,
+      "step": 977
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.39259976148605347,
+      "learning_rate": 3.8407063073765574e-06,
+      "loss": 0.0631,
+      "step": 978
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.21652863919734955,
+      "learning_rate": 3.831640130011867e-06,
+      "loss": 0.0513,
+      "step": 979
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.21479672193527222,
+      "learning_rate": 3.8225780140535025e-06,
+      "loss": 0.0448,
+      "step": 980
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.21375016868114471,
+      "learning_rate": 3.8135199910028314e-06,
+      "loss": 0.0368,
+      "step": 981
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.37601420283317566,
+      "learning_rate": 3.8044660923469968e-06,
+      "loss": 0.0438,
+      "step": 982
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.4459959864616394,
+      "learning_rate": 3.7954163495588e-06,
+      "loss": 0.0918,
+      "step": 983
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.3222822844982147,
+      "learning_rate": 3.786370794096603e-06,
+      "loss": 0.0356,
+      "step": 984
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.26596394181251526,
+      "learning_rate": 3.777329457404202e-06,
+      "loss": 0.0698,
+      "step": 985
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.4063988924026489,
+      "learning_rate": 3.7682923709107367e-06,
+      "loss": 0.0759,
+      "step": 986
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.34790676832199097,
+      "learning_rate": 3.759259566030571e-06,
+      "loss": 0.0471,
+      "step": 987
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.2744903266429901,
+      "learning_rate": 3.750231074163179e-06,
+      "loss": 0.0402,
+      "step": 988
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.18864701688289642,
+      "learning_rate": 3.741206926693052e-06,
+      "loss": 0.0257,
+      "step": 989
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.3292123079299927,
+      "learning_rate": 3.7321871549895715e-06,
+      "loss": 0.0645,
+      "step": 990
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.2857760488986969,
+      "learning_rate": 3.7231717904069097e-06,
+      "loss": 0.0747,
+      "step": 991
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.4177703857421875,
+      "learning_rate": 3.714160864283923e-06,
+      "loss": 0.0628,
+      "step": 992
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.3333231806755066,
+      "learning_rate": 3.705154407944034e-06,
+      "loss": 0.0559,
+      "step": 993
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.45318400859832764,
+      "learning_rate": 3.696152452695128e-06,
+      "loss": 0.0848,
+      "step": 994
+    },
+    {
+      "epoch": 1.76,
+      "eval_loss": 0.056184083223342896,
+      "eval_runtime": 14.7268,
+      "eval_samples_per_second": 32.322,
+      "eval_steps_per_second": 8.081,
+      "step": 994
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.36879318952560425,
+      "learning_rate": 3.68715502982945e-06,
+      "loss": 0.0497,
+      "step": 995
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.2791776657104492,
+      "learning_rate": 3.6781621706234815e-06,
+      "loss": 0.0385,
+      "step": 996
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.7810603380203247,
+      "learning_rate": 3.6691739063378462e-06,
+      "loss": 0.0805,
+      "step": 997
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.4113273620605469,
+      "learning_rate": 3.66019026821719e-06,
+      "loss": 0.0466,
+      "step": 998
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.2155979722738266,
+      "learning_rate": 3.65121128749008e-06,
+      "loss": 0.0199,
+      "step": 999
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.2696681320667267,
+      "learning_rate": 3.6422369953688973e-06,
+      "loss": 0.0518,
+      "step": 1000
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.36340588331222534,
+      "learning_rate": 3.633267423049717e-06,
+      "loss": 0.0644,
+      "step": 1001
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.219515860080719,
+      "learning_rate": 3.624302601712213e-06,
+      "loss": 0.0393,
+      "step": 1002
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.2098739892244339,
+      "learning_rate": 3.6153425625195424e-06,
+      "loss": 0.0492,
+      "step": 1003
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.4837384819984436,
+      "learning_rate": 3.606387336618237e-06,
+      "loss": 0.1312,
+      "step": 1004
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3313275873661041,
+      "learning_rate": 3.5974369551381023e-06,
+      "loss": 0.0638,
+      "step": 1005
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.2670367658138275,
+      "learning_rate": 3.5884914491920963e-06,
+      "loss": 0.056,
+      "step": 1006
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.32359379529953003,
+      "learning_rate": 3.579550849876233e-06,
+      "loss": 0.0736,
+      "step": 1007
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.23541584610939026,
+      "learning_rate": 3.5706151882694727e-06,
+      "loss": 0.0354,
+      "step": 1008
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.3773065507411957,
+      "learning_rate": 3.561684495433605e-06,
+      "loss": 0.0552,
+      "step": 1009
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.4034823179244995,
+      "learning_rate": 3.5527588024131542e-06,
+      "loss": 0.0614,
+      "step": 1010
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.2549152970314026,
+      "learning_rate": 3.543838140235257e-06,
+      "loss": 0.0603,
+      "step": 1011
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.2435804307460785,
+      "learning_rate": 3.5349225399095693e-06,
+      "loss": 0.0422,
+      "step": 1012
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.6781402230262756,
+      "learning_rate": 3.526012032428148e-06,
+      "loss": 0.0816,
+      "step": 1013
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.4394122064113617,
+      "learning_rate": 3.5171066487653427e-06,
+      "loss": 0.0885,
+      "step": 1014
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3086816072463989,
+      "learning_rate": 3.5082064198777e-06,
+      "loss": 0.0616,
+      "step": 1015
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3581552803516388,
+      "learning_rate": 3.4993113767038423e-06,
+      "loss": 0.0557,
+      "step": 1016
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.28149986267089844,
+      "learning_rate": 3.4904215501643647e-06,
+      "loss": 0.0674,
+      "step": 1017
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.4265075623989105,
+      "learning_rate": 3.481536971161732e-06,
+      "loss": 0.0604,
+      "step": 1018
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.28389066457748413,
+      "learning_rate": 3.472657670580164e-06,
+      "loss": 0.032,
+      "step": 1019
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.48513075709342957,
+      "learning_rate": 3.463783679285535e-06,
+      "loss": 0.0518,
+      "step": 1020
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.2950093746185303,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.0516,
+      "step": 1021
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.23902879655361176,
+      "learning_rate": 3.446051747928202e-06,
+      "loss": 0.0493,
+      "step": 1022
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.4256305694580078,
+      "learning_rate": 3.4371938695045347e-06,
+      "loss": 0.0729,
+      "step": 1023
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.22448065876960754,
+      "learning_rate": 3.428341423645668e-06,
+      "loss": 0.0501,
+      "step": 1024
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.19932492077350616,
+      "learning_rate": 3.4194944411241213e-06,
+      "loss": 0.0257,
+      "step": 1025
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5429794788360596,
+      "learning_rate": 3.4106529526934305e-06,
+      "loss": 0.0788,
+      "step": 1026
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.19093726575374603,
+      "learning_rate": 3.4018169890880227e-06,
+      "loss": 0.027,
+      "step": 1027
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.6629717946052551,
+      "learning_rate": 3.3929865810231264e-06,
+      "loss": 0.0627,
+      "step": 1028
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.48709404468536377,
+      "learning_rate": 3.3841617591946584e-06,
+      "loss": 0.041,
+      "step": 1029
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5192286968231201,
+      "learning_rate": 3.3753425542791106e-06,
+      "loss": 0.0729,
+      "step": 1030
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.30288419127464294,
+      "learning_rate": 3.3665289969334587e-06,
+      "loss": 0.0409,
+      "step": 1031
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.3584707975387573,
+      "learning_rate": 3.3577211177950386e-06,
+      "loss": 0.055,
+      "step": 1032
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.43994244933128357,
+      "learning_rate": 3.348918947481452e-06,
+      "loss": 0.0682,
+      "step": 1033
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6027707457542419,
+      "learning_rate": 3.340122516590456e-06,
+      "loss": 0.0659,
+      "step": 1034
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.49970120191574097,
+      "learning_rate": 3.3313318556998523e-06,
+      "loss": 0.0611,
+      "step": 1035
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.21730153262615204,
+      "learning_rate": 3.322546995367394e-06,
+      "loss": 0.048,
+      "step": 1036
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.3189135193824768,
+      "learning_rate": 3.3137679661306578e-06,
+      "loss": 0.0866,
+      "step": 1037
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.798090934753418,
+      "learning_rate": 3.304994798506962e-06,
+      "loss": 0.0869,
+      "step": 1038
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.3200780153274536,
+      "learning_rate": 3.296227522993245e-06,
+      "loss": 0.0613,
+      "step": 1039
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.3401470482349396,
+      "learning_rate": 3.2874661700659586e-06,
+      "loss": 0.0498,
+      "step": 1040
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.2279331088066101,
+      "learning_rate": 3.2787107701809757e-06,
+      "loss": 0.0557,
+      "step": 1041
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.47060710191726685,
+      "learning_rate": 3.2699613537734693e-06,
+      "loss": 0.0604,
+      "step": 1042
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.22766916453838348,
+      "learning_rate": 3.261217951257813e-06,
+      "loss": 0.0394,
+      "step": 1043
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6578858494758606,
+      "learning_rate": 3.252480593027478e-06,
+      "loss": 0.0754,
+      "step": 1044
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6935117840766907,
+      "learning_rate": 3.2437493094549223e-06,
+      "loss": 0.0778,
+      "step": 1045
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.2019956409931183,
+      "learning_rate": 3.2350241308914865e-06,
+      "loss": 0.0369,
+      "step": 1046
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.4244823455810547,
+      "learning_rate": 3.2263050876672954e-06,
+      "loss": 0.0549,
+      "step": 1047
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.24040935933589935,
+      "learning_rate": 3.217592210091137e-06,
+      "loss": 0.0394,
+      "step": 1048
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.8403865098953247,
+      "learning_rate": 3.2088855284503762e-06,
+      "loss": 0.0904,
+      "step": 1049
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.21365146338939667,
+      "learning_rate": 3.200185073010831e-06,
+      "loss": 0.0267,
+      "step": 1050
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.7642082571983337,
+      "learning_rate": 3.1914908740166793e-06,
+      "loss": 0.0951,
+      "step": 1051
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6344515085220337,
+      "learning_rate": 3.182802961690357e-06,
+      "loss": 0.0809,
+      "step": 1052
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5710294246673584,
+      "learning_rate": 3.1741213662324365e-06,
+      "loss": 0.0858,
+      "step": 1053
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.3214423656463623,
+      "learning_rate": 3.165446117821538e-06,
+      "loss": 0.0686,
+      "step": 1054
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.2226896584033966,
+      "learning_rate": 3.1567772466142156e-06,
+      "loss": 0.0177,
+      "step": 1055
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.2788223326206207,
+      "learning_rate": 3.1481147827448554e-06,
+      "loss": 0.0411,
+      "step": 1056
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.22315210103988647,
+      "learning_rate": 3.139458756325576e-06,
+      "loss": 0.0398,
+      "step": 1057
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.2029338926076889,
+      "learning_rate": 3.1308091974461064e-06,
+      "loss": 0.0309,
+      "step": 1058
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.445560097694397,
+      "learning_rate": 3.1221661361737065e-06,
+      "loss": 0.0659,
+      "step": 1059
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.7532091736793518,
+      "learning_rate": 3.1135296025530426e-06,
+      "loss": 0.0883,
+      "step": 1060
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5283082127571106,
+      "learning_rate": 3.1048996266060883e-06,
+      "loss": 0.0812,
+      "step": 1061
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.19848276674747467,
+      "learning_rate": 3.0962762383320288e-06,
+      "loss": 0.0294,
+      "step": 1062
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.3289884030818939,
+      "learning_rate": 3.0876594677071405e-06,
+      "loss": 0.0691,
+      "step": 1063
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.475301057100296,
+      "learning_rate": 3.0790493446847024e-06,
+      "loss": 0.0552,
+      "step": 1064
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.20955568552017212,
+      "learning_rate": 3.070445899194885e-06,
+      "loss": 0.0336,
+      "step": 1065
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.280180424451828,
+      "learning_rate": 3.061849161144641e-06,
+      "loss": 0.0732,
+      "step": 1066
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.23523566126823425,
+      "learning_rate": 3.0532591604176132e-06,
+      "loss": 0.0615,
+      "step": 1067
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4371294379234314,
+      "learning_rate": 3.044675926874023e-06,
+      "loss": 0.0431,
+      "step": 1068
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.34492623805999756,
+      "learning_rate": 3.0360994903505654e-06,
+      "loss": 0.0539,
+      "step": 1069
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.3771841824054718,
+      "learning_rate": 3.0275298806603102e-06,
+      "loss": 0.0707,
+      "step": 1070
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.32946136593818665,
+      "learning_rate": 3.0189671275925954e-06,
+      "loss": 0.0485,
+      "step": 1071
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.34278154373168945,
+      "learning_rate": 3.010411260912922e-06,
+      "loss": 0.072,
+      "step": 1072
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.2122594267129898,
+      "learning_rate": 3.00186231036286e-06,
+      "loss": 0.0386,
+      "step": 1073
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.46752992272377014,
+      "learning_rate": 2.9933203056599277e-06,
+      "loss": 0.1058,
+      "step": 1074
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.4876655638217926,
+      "learning_rate": 2.984785276497507e-06,
+      "loss": 0.0755,
+      "step": 1075
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.4009245038032532,
+      "learning_rate": 2.9762572525447266e-06,
+      "loss": 0.0801,
+      "step": 1076
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.31884658336639404,
+      "learning_rate": 2.9677362634463647e-06,
+      "loss": 0.0491,
+      "step": 1077
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.4274621903896332,
+      "learning_rate": 2.9592223388227505e-06,
+      "loss": 0.0647,
+      "step": 1078
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.2394295334815979,
+      "learning_rate": 2.950715508269648e-06,
+      "loss": 0.0646,
+      "step": 1079
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3970087170600891,
+      "learning_rate": 2.9422158013581658e-06,
+      "loss": 0.0751,
+      "step": 1080
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.33805954456329346,
+      "learning_rate": 2.93372324763465e-06,
+      "loss": 0.064,
+      "step": 1081
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.27474018931388855,
+      "learning_rate": 2.925237876620576e-06,
+      "loss": 0.0646,
+      "step": 1082
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.3348204791545868,
+      "learning_rate": 2.9167597178124584e-06,
+      "loss": 0.0602,
+      "step": 1083
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.22010396420955658,
+      "learning_rate": 2.908288800681737e-06,
+      "loss": 0.0502,
+      "step": 1084
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.46095865964889526,
+      "learning_rate": 2.899825154674674e-06,
+      "loss": 0.0698,
+      "step": 1085
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.3379755914211273,
+      "learning_rate": 2.8913688092122667e-06,
+      "loss": 0.0737,
+      "step": 1086
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.2667890787124634,
+      "learning_rate": 2.882919793690123e-06,
+      "loss": 0.0331,
+      "step": 1087
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.28333139419555664,
+      "learning_rate": 2.8744781374783813e-06,
+      "loss": 0.044,
+      "step": 1088
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5258477330207825,
+      "learning_rate": 2.8660438699215896e-06,
+      "loss": 0.0611,
+      "step": 1089
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.37528035044670105,
+      "learning_rate": 2.8576170203386144e-06,
+      "loss": 0.0593,
+      "step": 1090
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.28744783997535706,
+      "learning_rate": 2.849197618022539e-06,
+      "loss": 0.056,
+      "step": 1091
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.2404329478740692,
+      "learning_rate": 2.840785692240553e-06,
+      "loss": 0.0526,
+      "step": 1092
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.41300275921821594,
+      "learning_rate": 2.832381272233864e-06,
+      "loss": 0.1055,
+      "step": 1093
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.37923821806907654,
+      "learning_rate": 2.8239843872175814e-06,
+      "loss": 0.0611,
+      "step": 1094
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.5538582801818848,
+      "learning_rate": 2.8155950663806234e-06,
+      "loss": 0.0666,
+      "step": 1095
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.44852039217948914,
+      "learning_rate": 2.8072133388856194e-06,
+      "loss": 0.0601,
+      "step": 1096
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.2169903963804245,
+      "learning_rate": 2.7988392338687925e-06,
+      "loss": 0.054,
+      "step": 1097
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3337094187736511,
+      "learning_rate": 2.7904727804398813e-06,
+      "loss": 0.0637,
+      "step": 1098
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.5035146474838257,
+      "learning_rate": 2.782114007682016e-06,
+      "loss": 0.0875,
+      "step": 1099
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.2593541145324707,
+      "learning_rate": 2.7737629446516325e-06,
+      "loss": 0.0629,
+      "step": 1100
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.28386053442955017,
+      "learning_rate": 2.765419620378366e-06,
+      "loss": 0.0544,
+      "step": 1101
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.18949826061725616,
+      "learning_rate": 2.7570840638649487e-06,
+      "loss": 0.0438,
+      "step": 1102
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.26150980591773987,
+      "learning_rate": 2.7487563040871145e-06,
+      "loss": 0.0513,
+      "step": 1103
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.3108008801937103,
+      "learning_rate": 2.740436369993491e-06,
+      "loss": 0.0714,
+      "step": 1104
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.26925361156463623,
+      "learning_rate": 2.732124290505501e-06,
+      "loss": 0.0334,
+      "step": 1105
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.2529263496398926,
+      "learning_rate": 2.72382009451727e-06,
+      "loss": 0.0334,
+      "step": 1106
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5129886865615845,
+      "learning_rate": 2.7155238108955153e-06,
+      "loss": 0.0989,
+      "step": 1107
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5947487354278564,
+      "learning_rate": 2.707235468479449e-06,
+      "loss": 0.0673,
+      "step": 1108
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3080863058567047,
+      "learning_rate": 2.698955096080677e-06,
+      "loss": 0.0565,
+      "step": 1109
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.24549928307533264,
+      "learning_rate": 2.6906827224831024e-06,
+      "loss": 0.0551,
+      "step": 1110
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.24393980205059052,
+      "learning_rate": 2.6824183764428226e-06,
+      "loss": 0.0549,
+      "step": 1111
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5429656505584717,
+      "learning_rate": 2.6741620866880335e-06,
+      "loss": 0.1288,
+      "step": 1112
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.22410114109516144,
+      "learning_rate": 2.665913881918921e-06,
+      "loss": 0.0699,
+      "step": 1113
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6905423998832703,
+      "learning_rate": 2.6576737908075667e-06,
+      "loss": 0.0894,
+      "step": 1114
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.23629222810268402,
+      "learning_rate": 2.6494418419978485e-06,
+      "loss": 0.057,
+      "step": 1115
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.24294273555278778,
+      "learning_rate": 2.641218064105341e-06,
+      "loss": 0.057,
+      "step": 1116
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.3973044753074646,
+      "learning_rate": 2.6330024857172193e-06,
+      "loss": 0.0545,
+      "step": 1117
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.35339024662971497,
+      "learning_rate": 2.6247951353921484e-06,
+      "loss": 0.0443,
+      "step": 1118
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.33749720454216003,
+      "learning_rate": 2.6165960416601944e-06,
+      "loss": 0.0633,
+      "step": 1119
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.19248297810554504,
+      "learning_rate": 2.608405233022724e-06,
+      "loss": 0.0657,
+      "step": 1120
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.30044496059417725,
+      "learning_rate": 2.600222737952299e-06,
+      "loss": 0.0596,
+      "step": 1121
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.18807631731033325,
+      "learning_rate": 2.5920485848925914e-06,
+      "loss": 0.0474,
+      "step": 1122
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.2754729688167572,
+      "learning_rate": 2.5838828022582595e-06,
+      "loss": 0.0382,
+      "step": 1123
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.38777849078178406,
+      "learning_rate": 2.575725418434878e-06,
+      "loss": 0.04,
+      "step": 1124
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.3121282756328583,
+      "learning_rate": 2.5675764617788233e-06,
+      "loss": 0.0592,
+      "step": 1125
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.2995060086250305,
+      "learning_rate": 2.5594359606171728e-06,
+      "loss": 0.0607,
+      "step": 1126
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.22326309978961945,
+      "learning_rate": 2.5513039432476195e-06,
+      "loss": 0.0661,
+      "step": 1127
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.21206530928611755,
+      "learning_rate": 2.543180437938352e-06,
+      "loss": 0.0389,
+      "step": 1128
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.26785749197006226,
+      "learning_rate": 2.5350654729279832e-06,
+      "loss": 0.0393,
+      "step": 1129
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.27896589040756226,
+      "learning_rate": 2.526959076425434e-06,
+      "loss": 0.084,
+      "step": 1130
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1695,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 565,
+  "total_flos": 1.055381254129582e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}