diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,6143 +1,199 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 4.458762886597938,
-  "eval_steps": 97,
-  "global_step": 865,
+  "epoch": 0.03235198964736331,
+  "eval_steps": 386,
+  "global_step": 25,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.005154639175257732,
-      "grad_norm": 14.970044136047363,
+      "epoch": 0.0012940795858945326,
+      "grad_norm": 1.7405146360397339,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 9.1577,
+      "loss": 2.4269,
       "step": 1
     },
     {
-      "epoch": 0.005154639175257732,
-      "eval_loss": 9.00115966796875,
-      "eval_runtime": 22.9959,
-      "eval_samples_per_second": 7.132,
-      "eval_steps_per_second": 1.783,
+      "epoch": 0.0012940795858945326,
+      "eval_loss": 2.247628688812256,
+      "eval_runtime": 189.8853,
+      "eval_samples_per_second": 3.428,
+      "eval_steps_per_second": 0.858,
       "step": 1
     },
     {
-      "epoch": 0.010309278350515464,
-      "grad_norm": 15.33629035949707,
+      "epoch": 0.002588159171789065,
+      "grad_norm": 1.6643542051315308,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 9.2777,
+      "loss": 2.2583,
       "step": 2
     },
     {
-      "epoch": 0.015463917525773196,
-      "grad_norm": 16.92699432373047,
+      "epoch": 0.0038822387576835974,
+      "grad_norm": 1.8690767288208008,
       "learning_rate": 6e-06,
-      "loss": 8.744,
+      "loss": 2.2696,
       "step": 3
     },
     {
-      "epoch": 0.020618556701030927,
-      "grad_norm": 19.3622989654541,
+      "epoch": 0.00517631834357813,
+      "grad_norm": 1.828118085861206,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 8.545,
+      "loss": 2.3646,
       "step": 4
     },
     {
-      "epoch": 0.02577319587628866,
-      "grad_norm": 20.165861129760742,
+      "epoch": 0.006470397929472662,
+      "grad_norm": 1.9319926500320435,
       "learning_rate": 1e-05,
-      "loss": 9.7134,
+      "loss": 2.4196,
       "step": 5
     },
     {
-      "epoch": 0.030927835051546393,
-      "grad_norm": 18.959148406982422,
+      "epoch": 0.007764477515367195,
+      "grad_norm": 1.7723782062530518,
       "learning_rate": 1.2e-05,
-      "loss": 9.7526,
+      "loss": 2.4177,
       "step": 6
     },
     {
-      "epoch": 0.03608247422680412,
-      "grad_norm": 16.62849235534668,
+      "epoch": 0.009058557101261728,
+      "grad_norm": 1.9500815868377686,
       "learning_rate": 1.4000000000000001e-05,
-      "loss": 8.9248,
+      "loss": 2.3497,
       "step": 7
     },
     {
-      "epoch": 0.041237113402061855,
-      "grad_norm": 19.083742141723633,
+      "epoch": 0.01035263668715626,
+      "grad_norm": 2.3909075260162354,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 9.0543,
+      "loss": 2.405,
       "step": 8
     },
     {
-      "epoch": 0.04639175257731959,
-      "grad_norm": 16.922298431396484,
+      "epoch": 0.011646716273050793,
+      "grad_norm": 2.0620856285095215,
       "learning_rate": 1.8e-05,
-      "loss": 8.7968,
+      "loss": 2.4098,
       "step": 9
     },
     {
-      "epoch": 0.05154639175257732,
-      "grad_norm": 15.013986587524414,
+      "epoch": 0.012940795858945324,
+      "grad_norm": 1.8054910898208618,
       "learning_rate": 2e-05,
-      "loss": 8.8618,
+      "loss": 2.1233,
       "step": 10
     },
     {
-      "epoch": 0.05670103092783505,
-      "grad_norm": 16.929433822631836,
+      "epoch": 0.014234875444839857,
+      "grad_norm": 2.190964937210083,
       "learning_rate": 2.2000000000000003e-05,
-      "loss": 8.3198,
+      "loss": 2.3985,
       "step": 11
     },
     {
-      "epoch": 0.061855670103092786,
-      "grad_norm": 14.889703750610352,
+      "epoch": 0.01552895503073439,
+      "grad_norm": 1.9412921667099,
       "learning_rate": 2.4e-05,
-      "loss": 8.1047,
+      "loss": 2.462,
       "step": 12
     },
     {
-      "epoch": 0.06701030927835051,
-      "grad_norm": 15.865534782409668,
+      "epoch": 0.016823034616628922,
+      "grad_norm": 1.9161555767059326,
       "learning_rate": 2.6000000000000002e-05,
-      "loss": 7.6863,
+      "loss": 2.2118,
       "step": 13
     },
     {
-      "epoch": 0.07216494845360824,
-      "grad_norm": 15.19003963470459,
+      "epoch": 0.018117114202523456,
+      "grad_norm": 1.7161599397659302,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 7.779,
+      "loss": 2.2175,
       "step": 14
     },
     {
-      "epoch": 0.07731958762886598,
-      "grad_norm": 17.154193878173828,
+      "epoch": 0.019411193788417987,
+      "grad_norm": 2.173877000808716,
       "learning_rate": 3e-05,
-      "loss": 7.7273,
+      "loss": 2.2521,
       "step": 15
     },
     {
-      "epoch": 0.08247422680412371,
-      "grad_norm": 14.37552547454834,
+      "epoch": 0.02070527337431252,
+      "grad_norm": 2.0000555515289307,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 7.1202,
+      "loss": 2.1615,
       "step": 16
     },
     {
-      "epoch": 0.08762886597938144,
-      "grad_norm": 13.732095718383789,
+      "epoch": 0.021999352960207053,
+      "grad_norm": 1.5915080308914185,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 6.9298,
+      "loss": 1.9522,
       "step": 17
     },
     {
-      "epoch": 0.09278350515463918,
-      "grad_norm": 13.2391357421875,
+      "epoch": 0.023293432546101587,
+      "grad_norm": 1.6972448825836182,
       "learning_rate": 3.6e-05,
-      "loss": 6.308,
+      "loss": 1.7224,
       "step": 18
     },
     {
-      "epoch": 0.0979381443298969,
-      "grad_norm": 11.839639663696289,
+      "epoch": 0.024587512131996118,
+      "grad_norm": 1.7509772777557373,
       "learning_rate": 3.8e-05,
-      "loss": 6.1335,
+      "loss": 2.0414,
       "step": 19
     },
     {
-      "epoch": 0.10309278350515463,
-      "grad_norm": 11.819024085998535,
+      "epoch": 0.02588159171789065,
+      "grad_norm": 1.697340488433838,
       "learning_rate": 4e-05,
-      "loss": 5.7396,
+      "loss": 2.0427,
       "step": 20
     },
     {
-      "epoch": 0.10824742268041238,
-      "grad_norm": 11.764089584350586,
+      "epoch": 0.027175671303785183,
+      "grad_norm": 1.8733758926391602,
       "learning_rate": 4.2e-05,
-      "loss": 5.1633,
+      "loss": 1.6772,
       "step": 21
     },
     {
-      "epoch": 0.1134020618556701,
-      "grad_norm": 12.526201248168945,
+      "epoch": 0.028469750889679714,
+      "grad_norm": 1.6085255146026611,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 4.7586,
+      "loss": 1.6527,
       "step": 22
     },
     {
-      "epoch": 0.11855670103092783,
-      "grad_norm": 10.944100379943848,
+      "epoch": 0.029763830475574248,
+      "grad_norm": 1.5792337656021118,
       "learning_rate": 4.600000000000001e-05,
-      "loss": 4.47,
+      "loss": 1.6567,
       "step": 23
     },
     {
-      "epoch": 0.12371134020618557,
-      "grad_norm": 14.008569717407227,
+      "epoch": 0.03105791006146878,
+      "grad_norm": 1.4392567873001099,
       "learning_rate": 4.8e-05,
-      "loss": 4.2032,
+      "loss": 1.508,
       "step": 24
     },
     {
-      "epoch": 0.12886597938144329,
-      "grad_norm": 14.705824851989746,
+      "epoch": 0.03235198964736331,
+      "grad_norm": 1.5222433805465698,
       "learning_rate": 5e-05,
-      "loss": 3.2385,
+      "loss": 1.4606,
       "step": 25
-    },
-    {
-      "epoch": 0.13402061855670103,
-      "grad_norm": 16.897640228271484,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 3.0181,
-      "step": 26
-    },
-    {
-      "epoch": 0.13917525773195877,
-      "grad_norm": 15.038681983947754,
-      "learning_rate": 5.4000000000000005e-05,
-      "loss": 2.171,
-      "step": 27
-    },
-    {
-      "epoch": 0.14432989690721648,
-      "grad_norm": 15.12243938446045,
-      "learning_rate": 5.6000000000000006e-05,
-      "loss": 2.1848,
-      "step": 28
-    },
-    {
-      "epoch": 0.14948453608247422,
-      "grad_norm": 12.384464263916016,
-      "learning_rate": 5.8e-05,
-      "loss": 1.6113,
-      "step": 29
-    },
-    {
-      "epoch": 0.15463917525773196,
-      "grad_norm": 10.626302719116211,
-      "learning_rate": 6e-05,
-      "loss": 0.8058,
-      "step": 30
-    },
-    {
-      "epoch": 0.15979381443298968,
-      "grad_norm": 10.231700897216797,
-      "learning_rate": 6.2e-05,
-      "loss": 1.1714,
-      "step": 31
-    },
-    {
-      "epoch": 0.16494845360824742,
-      "grad_norm": 11.243683815002441,
-      "learning_rate": 6.400000000000001e-05,
-      "loss": 0.8837,
-      "step": 32
-    },
-    {
-      "epoch": 0.17010309278350516,
-      "grad_norm": 13.52505874633789,
-      "learning_rate": 6.6e-05,
-      "loss": 0.9935,
-      "step": 33
-    },
-    {
-      "epoch": 0.17525773195876287,
-      "grad_norm": 11.277730941772461,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 0.7997,
-      "step": 34
-    },
-    {
-      "epoch": 0.18041237113402062,
-      "grad_norm": 8.148871421813965,
-      "learning_rate": 7e-05,
-      "loss": 0.5211,
-      "step": 35
-    },
-    {
-      "epoch": 0.18556701030927836,
-      "grad_norm": 7.315363883972168,
-      "learning_rate": 7.2e-05,
-      "loss": 0.5289,
-      "step": 36
-    },
-    {
-      "epoch": 0.19072164948453607,
-      "grad_norm": 11.134007453918457,
-      "learning_rate": 7.4e-05,
-      "loss": 0.4685,
-      "step": 37
-    },
-    {
-      "epoch": 0.1958762886597938,
-      "grad_norm": 12.332499504089355,
-      "learning_rate": 7.6e-05,
-      "loss": 0.6791,
-      "step": 38
-    },
-    {
-      "epoch": 0.20103092783505155,
-      "grad_norm": 9.159335136413574,
-      "learning_rate": 7.800000000000001e-05,
-      "loss": 0.6563,
-      "step": 39
-    },
-    {
-      "epoch": 0.20618556701030927,
-      "grad_norm": 5.620604038238525,
-      "learning_rate": 8e-05,
-      "loss": 0.3402,
-      "step": 40
-    },
-    {
-      "epoch": 0.211340206185567,
-      "grad_norm": 10.507390975952148,
-      "learning_rate": 8.2e-05,
-      "loss": 0.6456,
-      "step": 41
-    },
-    {
-      "epoch": 0.21649484536082475,
-      "grad_norm": 6.238654613494873,
-      "learning_rate": 8.4e-05,
-      "loss": 0.3945,
-      "step": 42
-    },
-    {
-      "epoch": 0.22164948453608246,
-      "grad_norm": 5.950511455535889,
-      "learning_rate": 8.6e-05,
-      "loss": 0.5033,
-      "step": 43
-    },
-    {
-      "epoch": 0.2268041237113402,
-      "grad_norm": 3.9956507682800293,
-      "learning_rate": 8.800000000000001e-05,
-      "loss": 0.4067,
-      "step": 44
-    },
-    {
-      "epoch": 0.23195876288659795,
-      "grad_norm": 6.797879695892334,
-      "learning_rate": 9e-05,
-      "loss": 0.4202,
-      "step": 45
-    },
-    {
-      "epoch": 0.23711340206185566,
-      "grad_norm": 5.048930644989014,
-      "learning_rate": 9.200000000000001e-05,
-      "loss": 0.3595,
-      "step": 46
-    },
-    {
-      "epoch": 0.2422680412371134,
-      "grad_norm": 4.503322124481201,
-      "learning_rate": 9.4e-05,
-      "loss": 0.3251,
-      "step": 47
-    },
-    {
-      "epoch": 0.24742268041237114,
-      "grad_norm": 6.15740966796875,
-      "learning_rate": 9.6e-05,
-      "loss": 0.5591,
-      "step": 48
-    },
-    {
-      "epoch": 0.25257731958762886,
-      "grad_norm": 4.286350727081299,
-      "learning_rate": 9.8e-05,
-      "loss": 0.3105,
-      "step": 49
-    },
-    {
-      "epoch": 0.25773195876288657,
-      "grad_norm": 5.595468044281006,
-      "learning_rate": 0.0001,
-      "loss": 0.2741,
-      "step": 50
-    },
-    {
-      "epoch": 0.26288659793814434,
-      "grad_norm": 3.9799349308013916,
-      "learning_rate": 9.999970848314005e-05,
-      "loss": 0.194,
-      "step": 51
-    },
-    {
-      "epoch": 0.26804123711340205,
-      "grad_norm": 5.833397388458252,
-      "learning_rate": 9.999883393595947e-05,
-      "loss": 0.316,
-      "step": 52
-    },
-    {
-      "epoch": 0.27319587628865977,
-      "grad_norm": 6.637378215789795,
-      "learning_rate": 9.999737636865609e-05,
-      "loss": 0.3404,
-      "step": 53
-    },
-    {
-      "epoch": 0.27835051546391754,
-      "grad_norm": 8.41789722442627,
-      "learning_rate": 9.99953357982261e-05,
-      "loss": 0.5317,
-      "step": 54
-    },
-    {
-      "epoch": 0.28350515463917525,
-      "grad_norm": 6.7356367111206055,
-      "learning_rate": 9.999271224846396e-05,
-      "loss": 0.3503,
-      "step": 55
-    },
-    {
-      "epoch": 0.28865979381443296,
-      "grad_norm": 3.6454246044158936,
-      "learning_rate": 9.998950574996199e-05,
-      "loss": 0.13,
-      "step": 56
-    },
-    {
-      "epoch": 0.29381443298969073,
-      "grad_norm": 4.760809898376465,
-      "learning_rate": 9.998571634011015e-05,
-      "loss": 0.3055,
-      "step": 57
-    },
-    {
-      "epoch": 0.29896907216494845,
-      "grad_norm": 4.675559043884277,
-      "learning_rate": 9.998134406309554e-05,
-      "loss": 0.2142,
-      "step": 58
-    },
-    {
-      "epoch": 0.30412371134020616,
-      "grad_norm": 5.181208610534668,
-      "learning_rate": 9.99763889699018e-05,
-      "loss": 0.332,
-      "step": 59
-    },
-    {
-      "epoch": 0.30927835051546393,
-      "grad_norm": 6.778941631317139,
-      "learning_rate": 9.99708511183087e-05,
-      "loss": 0.2931,
-      "step": 60
-    },
-    {
-      "epoch": 0.31443298969072164,
-      "grad_norm": 3.8104896545410156,
-      "learning_rate": 9.996473057289132e-05,
-      "loss": 0.1599,
-      "step": 61
-    },
-    {
-      "epoch": 0.31958762886597936,
-      "grad_norm": 4.8184123039245605,
-      "learning_rate": 9.995802740501933e-05,
-      "loss": 0.3225,
-      "step": 62
-    },
-    {
-      "epoch": 0.3247422680412371,
-      "grad_norm": 3.0463945865631104,
-      "learning_rate": 9.99507416928562e-05,
-      "loss": 0.2123,
-      "step": 63
-    },
-    {
-      "epoch": 0.32989690721649484,
-      "grad_norm": 3.5274605751037598,
-      "learning_rate": 9.994287352135825e-05,
-      "loss": 0.2277,
-      "step": 64
-    },
-    {
-      "epoch": 0.33505154639175255,
-      "grad_norm": 3.9188365936279297,
-      "learning_rate": 9.993442298227365e-05,
-      "loss": 0.2356,
-      "step": 65
-    },
-    {
-      "epoch": 0.3402061855670103,
-      "grad_norm": 3.182507038116455,
-      "learning_rate": 9.99253901741414e-05,
-      "loss": 0.1928,
-      "step": 66
-    },
-    {
-      "epoch": 0.34536082474226804,
-      "grad_norm": 5.660282135009766,
-      "learning_rate": 9.991577520229014e-05,
-      "loss": 0.2753,
-      "step": 67
-    },
-    {
-      "epoch": 0.35051546391752575,
-      "grad_norm": 4.6711201667785645,
-      "learning_rate": 9.99055781788369e-05,
-      "loss": 0.1546,
-      "step": 68
-    },
-    {
-      "epoch": 0.3556701030927835,
-      "grad_norm": 6.8249287605285645,
-      "learning_rate": 9.989479922268588e-05,
-      "loss": 0.3186,
-      "step": 69
-    },
-    {
-      "epoch": 0.36082474226804123,
-      "grad_norm": 4.09270715713501,
-      "learning_rate": 9.988343845952697e-05,
-      "loss": 0.289,
-      "step": 70
-    },
-    {
-      "epoch": 0.36597938144329895,
-      "grad_norm": 3.7769954204559326,
-      "learning_rate": 9.98714960218343e-05,
-      "loss": 0.2404,
-      "step": 71
-    },
-    {
-      "epoch": 0.3711340206185567,
-      "grad_norm": 5.468443870544434,
-      "learning_rate": 9.985897204886481e-05,
-      "loss": 0.3089,
-      "step": 72
-    },
-    {
-      "epoch": 0.37628865979381443,
-      "grad_norm": 3.5569119453430176,
-      "learning_rate": 9.98458666866564e-05,
-      "loss": 0.2668,
-      "step": 73
-    },
-    {
-      "epoch": 0.38144329896907214,
-      "grad_norm": 3.872802495956421,
-      "learning_rate": 9.983218008802648e-05,
-      "loss": 0.2248,
-      "step": 74
-    },
-    {
-      "epoch": 0.3865979381443299,
-      "grad_norm": 1.6926909685134888,
-      "learning_rate": 9.981791241257e-05,
-      "loss": 0.0595,
-      "step": 75
-    },
-    {
-      "epoch": 0.3917525773195876,
-      "grad_norm": 2.683844804763794,
-      "learning_rate": 9.98030638266577e-05,
-      "loss": 0.101,
-      "step": 76
-    },
-    {
-      "epoch": 0.39690721649484534,
-      "grad_norm": 2.6626157760620117,
-      "learning_rate": 9.978763450343407e-05,
-      "loss": 0.2987,
-      "step": 77
-    },
-    {
-      "epoch": 0.4020618556701031,
-      "grad_norm": 3.233513116836548,
-      "learning_rate": 9.977162462281544e-05,
-      "loss": 0.1434,
-      "step": 78
-    },
-    {
-      "epoch": 0.4072164948453608,
-      "grad_norm": 3.793926954269409,
-      "learning_rate": 9.975503437148783e-05,
-      "loss": 0.1916,
-      "step": 79
-    },
-    {
-      "epoch": 0.41237113402061853,
-      "grad_norm": 5.386580467224121,
-      "learning_rate": 9.973786394290474e-05,
-      "loss": 0.5362,
-      "step": 80
-    },
-    {
-      "epoch": 0.4175257731958763,
-      "grad_norm": 2.760202407836914,
-      "learning_rate": 9.972011353728496e-05,
-      "loss": 0.1676,
-      "step": 81
-    },
-    {
-      "epoch": 0.422680412371134,
-      "grad_norm": 2.2399468421936035,
-      "learning_rate": 9.970178336161018e-05,
-      "loss": 0.1063,
-      "step": 82
-    },
-    {
-      "epoch": 0.42783505154639173,
-      "grad_norm": 4.6277360916137695,
-      "learning_rate": 9.968287362962264e-05,
-      "loss": 0.3259,
-      "step": 83
-    },
-    {
-      "epoch": 0.4329896907216495,
-      "grad_norm": 2.2781248092651367,
-      "learning_rate": 9.96633845618225e-05,
-      "loss": 0.0777,
-      "step": 84
-    },
-    {
-      "epoch": 0.4381443298969072,
-      "grad_norm": 4.080474376678467,
-      "learning_rate": 9.96433163854655e-05,
-      "loss": 0.1406,
-      "step": 85
-    },
-    {
-      "epoch": 0.44329896907216493,
-      "grad_norm": 2.5774247646331787,
-      "learning_rate": 9.962266933456008e-05,
-      "loss": 0.2172,
-      "step": 86
-    },
-    {
-      "epoch": 0.4484536082474227,
-      "grad_norm": 2.8727529048919678,
-      "learning_rate": 9.96014436498648e-05,
-      "loss": 0.1189,
-      "step": 87
-    },
-    {
-      "epoch": 0.4536082474226804,
-      "grad_norm": 4.630618572235107,
-      "learning_rate": 9.957963957888542e-05,
-      "loss": 0.3777,
-      "step": 88
-    },
-    {
-      "epoch": 0.4587628865979381,
-      "grad_norm": 3.9078369140625,
-      "learning_rate": 9.955725737587214e-05,
-      "loss": 0.393,
-      "step": 89
-    },
-    {
-      "epoch": 0.4639175257731959,
-      "grad_norm": 4.288754940032959,
-      "learning_rate": 9.953429730181653e-05,
-      "loss": 0.3116,
-      "step": 90
-    },
-    {
-      "epoch": 0.4690721649484536,
-      "grad_norm": 4.296566486358643,
-      "learning_rate": 9.951075962444856e-05,
-      "loss": 0.2739,
-      "step": 91
-    },
-    {
-      "epoch": 0.4742268041237113,
-      "grad_norm": 3.088003635406494,
-      "learning_rate": 9.94866446182334e-05,
-      "loss": 0.2998,
-      "step": 92
-    },
-    {
-      "epoch": 0.4793814432989691,
-      "grad_norm": 2.5414445400238037,
-      "learning_rate": 9.94619525643683e-05,
-      "loss": 0.1035,
-      "step": 93
-    },
-    {
-      "epoch": 0.4845360824742268,
-      "grad_norm": 3.514005184173584,
-      "learning_rate": 9.943668375077925e-05,
-      "loss": 0.4353,
-      "step": 94
-    },
-    {
-      "epoch": 0.4896907216494845,
-      "grad_norm": 3.296506404876709,
-      "learning_rate": 9.941083847211765e-05,
-      "loss": 0.1903,
-      "step": 95
-    },
-    {
-      "epoch": 0.4948453608247423,
-      "grad_norm": 3.438098669052124,
-      "learning_rate": 9.938441702975689e-05,
-      "loss": 0.286,
-      "step": 96
-    },
-    {
-      "epoch": 0.5,
-      "grad_norm": 2.7748489379882812,
-      "learning_rate": 9.93574197317888e-05,
-      "loss": 0.1028,
-      "step": 97
-    },
-    {
-      "epoch": 0.5,
-      "eval_loss": 0.14588667452335358,
-      "eval_runtime": 23.0705,
-      "eval_samples_per_second": 7.109,
-      "eval_steps_per_second": 1.777,
-      "step": 97
-    },
-    {
-      "epoch": 0.5051546391752577,
-      "grad_norm": 2.874598264694214,
-      "learning_rate": 9.93298468930201e-05,
-      "loss": 0.212,
-      "step": 98
-    },
-    {
-      "epoch": 0.5103092783505154,
-      "grad_norm": 6.394598007202148,
-      "learning_rate": 9.930169883496867e-05,
-      "loss": 0.3593,
-      "step": 99
-    },
-    {
-      "epoch": 0.5154639175257731,
-      "grad_norm": 3.2982473373413086,
-      "learning_rate": 9.927297588585984e-05,
-      "loss": 0.2588,
-      "step": 100
-    },
-    {
-      "epoch": 0.520618556701031,
-      "grad_norm": 1.4498558044433594,
-      "learning_rate": 9.924367838062259e-05,
-      "loss": 0.0973,
-      "step": 101
-    },
-    {
-      "epoch": 0.5257731958762887,
-      "grad_norm": 4.289880275726318,
-      "learning_rate": 9.921380666088558e-05,
-      "loss": 0.3103,
-      "step": 102
-    },
-    {
-      "epoch": 0.5309278350515464,
-      "grad_norm": 2.9333765506744385,
-      "learning_rate": 9.91833610749732e-05,
-      "loss": 0.2417,
-      "step": 103
-    },
-    {
-      "epoch": 0.5360824742268041,
-      "grad_norm": 2.2084240913391113,
-      "learning_rate": 9.915234197790152e-05,
-      "loss": 0.0934,
-      "step": 104
-    },
-    {
-      "epoch": 0.5412371134020618,
-      "grad_norm": 4.149613857269287,
-      "learning_rate": 9.912074973137412e-05,
-      "loss": 0.4745,
-      "step": 105
-    },
-    {
-      "epoch": 0.5463917525773195,
-      "grad_norm": 1.4988398551940918,
-      "learning_rate": 9.908858470377793e-05,
-      "loss": 0.0984,
-      "step": 106
-    },
-    {
-      "epoch": 0.5515463917525774,
-      "grad_norm": 2.332045316696167,
-      "learning_rate": 9.905584727017884e-05,
-      "loss": 0.1449,
-      "step": 107
-    },
-    {
-      "epoch": 0.5567010309278351,
-      "grad_norm": 2.528547525405884,
-      "learning_rate": 9.90225378123174e-05,
-      "loss": 0.196,
-      "step": 108
-    },
-    {
-      "epoch": 0.5618556701030928,
-      "grad_norm": 1.529741883277893,
-      "learning_rate": 9.898865671860438e-05,
-      "loss": 0.1164,
-      "step": 109
-    },
-    {
-      "epoch": 0.5670103092783505,
-      "grad_norm": 2.2940495014190674,
-      "learning_rate": 9.895420438411616e-05,
-      "loss": 0.0998,
-      "step": 110
-    },
-    {
-      "epoch": 0.5721649484536082,
-      "grad_norm": 4.618880271911621,
-      "learning_rate": 9.891918121059019e-05,
-      "loss": 0.2512,
-      "step": 111
-    },
-    {
-      "epoch": 0.5773195876288659,
-      "grad_norm": 2.642099142074585,
-      "learning_rate": 9.888358760642029e-05,
-      "loss": 0.1523,
-      "step": 112
-    },
-    {
-      "epoch": 0.5824742268041238,
-      "grad_norm": 1.374038577079773,
-      "learning_rate": 9.884742398665191e-05,
-      "loss": 0.0801,
-      "step": 113
-    },
-    {
-      "epoch": 0.5876288659793815,
-      "grad_norm": 4.316773891448975,
-      "learning_rate": 9.881069077297723e-05,
-      "loss": 0.3471,
-      "step": 114
-    },
-    {
-      "epoch": 0.5927835051546392,
-      "grad_norm": 2.5112979412078857,
-      "learning_rate": 9.877338839373032e-05,
-      "loss": 0.121,
-      "step": 115
-    },
-    {
-      "epoch": 0.5979381443298969,
-      "grad_norm": 2.17130184173584,
-      "learning_rate": 9.873551728388203e-05,
-      "loss": 0.1461,
-      "step": 116
-    },
-    {
-      "epoch": 0.6030927835051546,
-      "grad_norm": 3.4084630012512207,
-      "learning_rate": 9.869707788503508e-05,
-      "loss": 0.1443,
-      "step": 117
-    },
-    {
-      "epoch": 0.6082474226804123,
-      "grad_norm": 1.7009655237197876,
-      "learning_rate": 9.865807064541877e-05,
-      "loss": 0.0999,
-      "step": 118
-    },
-    {
-      "epoch": 0.6134020618556701,
-      "grad_norm": 2.5199472904205322,
-      "learning_rate": 9.861849601988383e-05,
-      "loss": 0.1068,
-      "step": 119
-    },
-    {
-      "epoch": 0.6185567010309279,
-      "grad_norm": 2.2809436321258545,
-      "learning_rate": 9.857835446989707e-05,
-      "loss": 0.1214,
-      "step": 120
-    },
-    {
-      "epoch": 0.6237113402061856,
-      "grad_norm": 0.899928092956543,
-      "learning_rate": 9.853764646353605e-05,
-      "loss": 0.0325,
-      "step": 121
-    },
-    {
-      "epoch": 0.6288659793814433,
-      "grad_norm": 2.0530076026916504,
-      "learning_rate": 9.849637247548356e-05,
-      "loss": 0.1237,
-      "step": 122
-    },
-    {
-      "epoch": 0.634020618556701,
-      "grad_norm": 4.350875377655029,
-      "learning_rate": 9.845453298702216e-05,
-      "loss": 0.3754,
-      "step": 123
-    },
-    {
-      "epoch": 0.6391752577319587,
-      "grad_norm": 5.752893447875977,
-      "learning_rate": 9.841212848602846e-05,
-      "loss": 0.3497,
-      "step": 124
-    },
-    {
-      "epoch": 0.6443298969072165,
-      "grad_norm": 4.320384979248047,
-      "learning_rate": 9.836915946696759e-05,
-      "loss": 0.2156,
-      "step": 125
-    },
-    {
-      "epoch": 0.6494845360824743,
-      "grad_norm": 3.15738582611084,
-      "learning_rate": 9.832562643088724e-05,
-      "loss": 0.2713,
-      "step": 126
-    },
-    {
-      "epoch": 0.654639175257732,
-      "grad_norm": 1.3500375747680664,
-      "learning_rate": 9.828152988541201e-05,
-      "loss": 0.0746,
-      "step": 127
-    },
-    {
-      "epoch": 0.6597938144329897,
-      "grad_norm": 2.566340923309326,
-      "learning_rate": 9.823687034473735e-05,
-      "loss": 0.0502,
-      "step": 128
-    },
-    {
-      "epoch": 0.6649484536082474,
-      "grad_norm": 2.275930404663086,
-      "learning_rate": 9.81916483296236e-05,
-      "loss": 0.0944,
-      "step": 129
-    },
-    {
-      "epoch": 0.6701030927835051,
-      "grad_norm": 3.953136444091797,
-      "learning_rate": 9.814586436738998e-05,
-      "loss": 0.3553,
-      "step": 130
-    },
-    {
-      "epoch": 0.6752577319587629,
-      "grad_norm": 6.270533084869385,
-      "learning_rate": 9.809951899190835e-05,
-      "loss": 0.3139,
-      "step": 131
-    },
-    {
-      "epoch": 0.6804123711340206,
-      "grad_norm": 3.558518409729004,
-      "learning_rate": 9.805261274359705e-05,
-      "loss": 0.1273,
-      "step": 132
-    },
-    {
-      "epoch": 0.6855670103092784,
-      "grad_norm": 2.322179079055786,
-      "learning_rate": 9.800514616941457e-05,
-      "loss": 0.0881,
-      "step": 133
-    },
-    {
-      "epoch": 0.6907216494845361,
-      "grad_norm": 2.145798683166504,
-      "learning_rate": 9.795711982285316e-05,
-      "loss": 0.099,
-      "step": 134
-    },
-    {
-      "epoch": 0.6958762886597938,
-      "grad_norm": 2.0605945587158203,
-      "learning_rate": 9.790853426393245e-05,
-      "loss": 0.1114,
-      "step": 135
-    },
-    {
-      "epoch": 0.7010309278350515,
-      "grad_norm": 2.2030012607574463,
-      "learning_rate": 9.785939005919278e-05,
-      "loss": 0.0786,
-      "step": 136
-    },
-    {
-      "epoch": 0.7061855670103093,
-      "grad_norm": 2.1371002197265625,
-      "learning_rate": 9.780968778168874e-05,
-      "loss": 0.0472,
-      "step": 137
-    },
-    {
-      "epoch": 0.711340206185567,
-      "grad_norm": 2.5156478881835938,
-      "learning_rate": 9.77594280109824e-05,
-      "loss": 0.2225,
-      "step": 138
-    },
-    {
-      "epoch": 0.7164948453608248,
-      "grad_norm": 3.4769577980041504,
-      "learning_rate": 9.77086113331366e-05,
-      "loss": 0.1936,
-      "step": 139
-    },
-    {
-      "epoch": 0.7216494845360825,
-      "grad_norm": 2.1080641746520996,
-      "learning_rate": 9.765723834070804e-05,
-      "loss": 0.0703,
-      "step": 140
-    },
-    {
-      "epoch": 0.7268041237113402,
-      "grad_norm": 2.564608573913574,
-      "learning_rate": 9.760530963274048e-05,
-      "loss": 0.1402,
-      "step": 141
-    },
-    {
-      "epoch": 0.7319587628865979,
-      "grad_norm": 3.0748331546783447,
-      "learning_rate": 9.755282581475769e-05,
-      "loss": 0.1316,
-      "step": 142
-    },
-    {
-      "epoch": 0.7371134020618557,
-      "grad_norm": 2.478696346282959,
-      "learning_rate": 9.749978749875635e-05,
-      "loss": 0.1484,
-      "step": 143
-    },
-    {
-      "epoch": 0.7422680412371134,
-      "grad_norm": 2.250427722930908,
-      "learning_rate": 9.744619530319899e-05,
-      "loss": 0.1175,
-      "step": 144
-    },
-    {
-      "epoch": 0.7474226804123711,
-      "grad_norm": 4.462061405181885,
-      "learning_rate": 9.739204985300679e-05,
-      "loss": 0.1377,
-      "step": 145
-    },
-    {
-      "epoch": 0.7525773195876289,
-      "grad_norm": 3.592965841293335,
-      "learning_rate": 9.733735177955219e-05,
-      "loss": 0.158,
-      "step": 146
-    },
-    {
-      "epoch": 0.7577319587628866,
-      "grad_norm": 3.2920875549316406,
-      "learning_rate": 9.728210172065162e-05,
-      "loss": 0.1549,
-      "step": 147
-    },
-    {
-      "epoch": 0.7628865979381443,
-      "grad_norm": 3.6123318672180176,
-      "learning_rate": 9.722630032055803e-05,
-      "loss": 0.2841,
-      "step": 148
-    },
-    {
-      "epoch": 0.7680412371134021,
-      "grad_norm": 1.8174854516983032,
-      "learning_rate": 9.716994822995338e-05,
-      "loss": 0.0614,
-      "step": 149
-    },
-    {
-      "epoch": 0.7731958762886598,
-      "grad_norm": 3.1688687801361084,
-      "learning_rate": 9.711304610594104e-05,
-      "loss": 0.1094,
-      "step": 150
-    },
-    {
-      "epoch": 0.7783505154639175,
-      "grad_norm": 3.0467987060546875,
-      "learning_rate": 9.705559461203815e-05,
-      "loss": 0.1497,
-      "step": 151
-    },
-    {
-      "epoch": 0.7835051546391752,
-      "grad_norm": 3.366333484649658,
-      "learning_rate": 9.699759441816787e-05,
-      "loss": 0.2093,
-      "step": 152
-    },
-    {
-      "epoch": 0.788659793814433,
-      "grad_norm": 1.9911260604858398,
-      "learning_rate": 9.69390462006516e-05,
-      "loss": 0.0747,
-      "step": 153
-    },
-    {
-      "epoch": 0.7938144329896907,
-      "grad_norm": 2.2760016918182373,
-      "learning_rate": 9.687995064220102e-05,
-      "loss": 0.1229,
-      "step": 154
-    },
-    {
-      "epoch": 0.7989690721649485,
-      "grad_norm": 2.0865273475646973,
-      "learning_rate": 9.682030843191022e-05,
-      "loss": 0.106,
-      "step": 155
-    },
-    {
-      "epoch": 0.8041237113402062,
-      "grad_norm": 2.1471030712127686,
-      "learning_rate": 9.676012026524755e-05,
-      "loss": 0.1264,
-      "step": 156
-    },
-    {
-      "epoch": 0.8092783505154639,
-      "grad_norm": 1.8155637979507446,
-      "learning_rate": 9.669938684404766e-05,
-      "loss": 0.096,
-      "step": 157
-    },
-    {
-      "epoch": 0.8144329896907216,
-      "grad_norm": 1.922451138496399,
-      "learning_rate": 9.663810887650318e-05,
-      "loss": 0.1148,
-      "step": 158
-    },
-    {
-      "epoch": 0.8195876288659794,
-      "grad_norm": 2.0736405849456787,
-      "learning_rate": 9.657628707715655e-05,
-      "loss": 0.0584,
-      "step": 159
-    },
-    {
-      "epoch": 0.8247422680412371,
-      "grad_norm": 2.7208549976348877,
-      "learning_rate": 9.651392216689165e-05,
-      "loss": 0.2158,
-      "step": 160
-    },
-    {
-      "epoch": 0.8298969072164949,
-      "grad_norm": 1.785387396812439,
-      "learning_rate": 9.645101487292539e-05,
-      "loss": 0.0931,
-      "step": 161
-    },
-    {
-      "epoch": 0.8350515463917526,
-      "grad_norm": 3.9423131942749023,
-      "learning_rate": 9.638756592879922e-05,
-      "loss": 0.2373,
-      "step": 162
-    },
-    {
-      "epoch": 0.8402061855670103,
-      "grad_norm": 2.107962131500244,
-      "learning_rate": 9.632357607437065e-05,
-      "loss": 0.0856,
-      "step": 163
-    },
-    {
-      "epoch": 0.845360824742268,
-      "grad_norm": 3.6012613773345947,
-      "learning_rate": 9.625904605580452e-05,
-      "loss": 0.2367,
-      "step": 164
-    },
-    {
-      "epoch": 0.8505154639175257,
-      "grad_norm": 2.260842800140381,
-      "learning_rate": 9.619397662556435e-05,
-      "loss": 0.1051,
-      "step": 165
-    },
-    {
-      "epoch": 0.8556701030927835,
-      "grad_norm": 2.0315983295440674,
-      "learning_rate": 9.612836854240358e-05,
-      "loss": 0.0865,
-      "step": 166
-    },
-    {
-      "epoch": 0.8608247422680413,
-      "grad_norm": 1.805436372756958,
-      "learning_rate": 9.606222257135675e-05,
-      "loss": 0.0548,
-      "step": 167
-    },
-    {
-      "epoch": 0.865979381443299,
-      "grad_norm": 3.063310146331787,
-      "learning_rate": 9.599553948373045e-05,
-      "loss": 0.1935,
-      "step": 168
-    },
-    {
-      "epoch": 0.8711340206185567,
-      "grad_norm": 3.7688794136047363,
-      "learning_rate": 9.592832005709448e-05,
-      "loss": 0.3533,
-      "step": 169
-    },
-    {
-      "epoch": 0.8762886597938144,
-      "grad_norm": 3.773287534713745,
-      "learning_rate": 9.586056507527266e-05,
-      "loss": 0.1988,
-      "step": 170
-    },
-    {
-      "epoch": 0.8814432989690721,
-      "grad_norm": 2.5585780143737793,
-      "learning_rate": 9.579227532833377e-05,
-      "loss": 0.0867,
-      "step": 171
-    },
-    {
-      "epoch": 0.8865979381443299,
-      "grad_norm": 3.5680785179138184,
-      "learning_rate": 9.572345161258235e-05,
-      "loss": 0.2216,
-      "step": 172
-    },
-    {
-      "epoch": 0.8917525773195877,
-      "grad_norm": 2.894637107849121,
-      "learning_rate": 9.565409473054932e-05,
-      "loss": 0.2954,
-      "step": 173
-    },
-    {
-      "epoch": 0.8969072164948454,
-      "grad_norm": 2.3045082092285156,
-      "learning_rate": 9.558420549098268e-05,
-      "loss": 0.1069,
-      "step": 174
-    },
-    {
-      "epoch": 0.9020618556701031,
-      "grad_norm": 1.5784564018249512,
-      "learning_rate": 9.551378470883812e-05,
-      "loss": 0.0812,
-      "step": 175
-    },
-    {
-      "epoch": 0.9072164948453608,
-      "grad_norm": 0.7501576542854309,
-      "learning_rate": 9.544283320526943e-05,
-      "loss": 0.0496,
-      "step": 176
-    },
-    {
-      "epoch": 0.9123711340206185,
-      "grad_norm": 2.6611945629119873,
-      "learning_rate": 9.537135180761903e-05,
-      "loss": 0.1711,
-      "step": 177
-    },
-    {
-      "epoch": 0.9175257731958762,
-      "grad_norm": 2.171032190322876,
-      "learning_rate": 9.52993413494082e-05,
-      "loss": 0.1465,
-      "step": 178
-    },
-    {
-      "epoch": 0.9226804123711341,
-      "grad_norm": 1.442818522453308,
-      "learning_rate": 9.522680267032742e-05,
-      "loss": 0.0545,
-      "step": 179
-    },
-    {
-      "epoch": 0.9278350515463918,
-      "grad_norm": 1.5512781143188477,
-      "learning_rate": 9.515373661622664e-05,
-      "loss": 0.0637,
-      "step": 180
-    },
-    {
-      "epoch": 0.9329896907216495,
-      "grad_norm": 2.2158899307250977,
-      "learning_rate": 9.508014403910533e-05,
-      "loss": 0.142,
-      "step": 181
-    },
-    {
-      "epoch": 0.9381443298969072,
-      "grad_norm": 0.619306743144989,
-      "learning_rate": 9.500602579710256e-05,
-      "loss": 0.0265,
-      "step": 182
-    },
-    {
-      "epoch": 0.9432989690721649,
-      "grad_norm": 0.3452989459037781,
-      "learning_rate": 9.4931382754487e-05,
-      "loss": 0.0122,
-      "step": 183
-    },
-    {
-      "epoch": 0.9484536082474226,
-      "grad_norm": 2.468613386154175,
-      "learning_rate": 9.485621578164689e-05,
-      "loss": 0.1154,
-      "step": 184
-    },
-    {
-      "epoch": 0.9536082474226805,
-      "grad_norm": 1.288886547088623,
-      "learning_rate": 9.478052575507982e-05,
-      "loss": 0.0564,
-      "step": 185
-    },
-    {
-      "epoch": 0.9587628865979382,
-      "grad_norm": 1.4700998067855835,
-      "learning_rate": 9.470431355738257e-05,
-      "loss": 0.0598,
-      "step": 186
-    },
-    {
-      "epoch": 0.9639175257731959,
-      "grad_norm": 1.8623106479644775,
-      "learning_rate": 9.46275800772407e-05,
-      "loss": 0.0958,
-      "step": 187
-    },
-    {
-      "epoch": 0.9690721649484536,
-      "grad_norm": 3.3809943199157715,
-      "learning_rate": 9.45503262094184e-05,
-      "loss": 0.1816,
-      "step": 188
-    },
-    {
-      "epoch": 0.9742268041237113,
-      "grad_norm": 1.7910330295562744,
-      "learning_rate": 9.447255285474783e-05,
-      "loss": 0.0803,
-      "step": 189
-    },
-    {
-      "epoch": 0.979381443298969,
-      "grad_norm": 0.992155134677887,
-      "learning_rate": 9.439426092011875e-05,
-      "loss": 0.0589,
-      "step": 190
-    },
-    {
-      "epoch": 0.9845360824742269,
-      "grad_norm": 2.7251176834106445,
-      "learning_rate": 9.431545131846797e-05,
-      "loss": 0.1786,
-      "step": 191
-    },
-    {
-      "epoch": 0.9896907216494846,
-      "grad_norm": 1.8005549907684326,
-      "learning_rate": 9.423612496876855e-05,
-      "loss": 0.1158,
-      "step": 192
-    },
-    {
-      "epoch": 0.9948453608247423,
-      "grad_norm": 3.290626287460327,
-      "learning_rate": 9.415628279601923e-05,
-      "loss": 0.2714,
-      "step": 193
-    },
-    {
-      "epoch": 1.0,
-      "grad_norm": 1.748504400253296,
-      "learning_rate": 9.407592573123358e-05,
-      "loss": 0.1168,
-      "step": 194
-    },
-    {
-      "epoch": 1.0,
-      "eval_loss": 0.08979101479053497,
-      "eval_runtime": 23.0231,
-      "eval_samples_per_second": 7.123,
-      "eval_steps_per_second": 1.781,
-      "step": 194
-    },
-    {
-      "epoch": 1.0051546391752577,
-      "grad_norm": 2.3739848136901855,
-      "learning_rate": 9.39950547114292e-05,
-      "loss": 0.1064,
-      "step": 195
-    },
-    {
-      "epoch": 1.0103092783505154,
-      "grad_norm": 1.4507896900177002,
-      "learning_rate": 9.39136706796167e-05,
-      "loss": 0.0388,
-      "step": 196
-    },
-    {
-      "epoch": 1.0154639175257731,
-      "grad_norm": 0.6086592674255371,
-      "learning_rate": 9.383177458478878e-05,
-      "loss": 0.0182,
-      "step": 197
-    },
-    {
-      "epoch": 1.0206185567010309,
-      "grad_norm": 1.4689748287200928,
-      "learning_rate": 9.374936738190914e-05,
-      "loss": 0.0843,
-      "step": 198
-    },
-    {
-      "epoch": 1.0257731958762886,
-      "grad_norm": 4.061261177062988,
-      "learning_rate": 9.366645003190132e-05,
-      "loss": 0.219,
-      "step": 199
-    },
-    {
-      "epoch": 1.0309278350515463,
-      "grad_norm": 2.736973285675049,
-      "learning_rate": 9.358302350163757e-05,
-      "loss": 0.2438,
-      "step": 200
-    },
-    {
-      "epoch": 1.0360824742268042,
-      "grad_norm": 3.9617350101470947,
-      "learning_rate": 9.349908876392748e-05,
-      "loss": 0.2772,
-      "step": 201
-    },
-    {
-      "epoch": 1.041237113402062,
-      "grad_norm": 2.085519790649414,
-      "learning_rate": 9.341464679750669e-05,
-      "loss": 0.1464,
-      "step": 202
-    },
-    {
-      "epoch": 1.0463917525773196,
-      "grad_norm": 4.2934794425964355,
-      "learning_rate": 9.33296985870255e-05,
-      "loss": 0.1695,
-      "step": 203
-    },
-    {
-      "epoch": 1.0515463917525774,
-      "grad_norm": 1.9166585206985474,
-      "learning_rate": 9.32442451230373e-05,
-      "loss": 0.0841,
-      "step": 204
-    },
-    {
-      "epoch": 1.056701030927835,
-      "grad_norm": 3.341161012649536,
-      "learning_rate": 9.315828740198714e-05,
-      "loss": 0.3112,
-      "step": 205
-    },
-    {
-      "epoch": 1.0618556701030928,
-      "grad_norm": 0.9464735388755798,
-      "learning_rate": 9.30718264262e-05,
-      "loss": 0.0329,
-      "step": 206
-    },
-    {
-      "epoch": 1.0670103092783505,
-      "grad_norm": 1.8928874731063843,
-      "learning_rate": 9.298486320386919e-05,
-      "loss": 0.1272,
-      "step": 207
-    },
-    {
-      "epoch": 1.0721649484536082,
-      "grad_norm": 1.175891399383545,
-      "learning_rate": 9.289739874904449e-05,
-      "loss": 0.023,
-      "step": 208
-    },
-    {
-      "epoch": 1.077319587628866,
-      "grad_norm": 1.9952442646026611,
-      "learning_rate": 9.280943408162046e-05,
-      "loss": 0.1426,
-      "step": 209
-    },
-    {
-      "epoch": 1.0824742268041236,
-      "grad_norm": 0.945824384689331,
-      "learning_rate": 9.272097022732443e-05,
-      "loss": 0.0435,
-      "step": 210
-    },
-    {
-      "epoch": 1.0876288659793814,
-      "grad_norm": 3.451200485229492,
-      "learning_rate": 9.263200821770461e-05,
-      "loss": 0.2378,
-      "step": 211
-    },
-    {
-      "epoch": 1.0927835051546393,
-      "grad_norm": 1.6046141386032104,
-      "learning_rate": 9.254254909011804e-05,
-      "loss": 0.0563,
-      "step": 212
-    },
-    {
-      "epoch": 1.097938144329897,
-      "grad_norm": 3.0407392978668213,
-      "learning_rate": 9.245259388771845e-05,
-      "loss": 0.2115,
-      "step": 213
-    },
-    {
-      "epoch": 1.1030927835051547,
-      "grad_norm": 2.295111894607544,
-      "learning_rate": 9.236214365944418e-05,
-      "loss": 0.1761,
-      "step": 214
-    },
-    {
-      "epoch": 1.1082474226804124,
-      "grad_norm": 3.0166642665863037,
-      "learning_rate": 9.22711994600059e-05,
-      "loss": 0.1294,
-      "step": 215
-    },
-    {
-      "epoch": 1.1134020618556701,
-      "grad_norm": 0.8205422759056091,
-      "learning_rate": 9.217976234987428e-05,
-      "loss": 0.0279,
-      "step": 216
-    },
-    {
-      "epoch": 1.1185567010309279,
-      "grad_norm": 1.37704336643219,
-      "learning_rate": 9.208783339526773e-05,
-      "loss": 0.0329,
-      "step": 217
-    },
-    {
-      "epoch": 1.1237113402061856,
-      "grad_norm": 1.7456235885620117,
-      "learning_rate": 9.199541366813982e-05,
-      "loss": 0.0357,
-      "step": 218
-    },
-    {
-      "epoch": 1.1288659793814433,
-      "grad_norm": 0.9686729311943054,
-      "learning_rate": 9.190250424616693e-05,
-      "loss": 0.026,
-      "step": 219
-    },
-    {
-      "epoch": 1.134020618556701,
-      "grad_norm": 1.0743412971496582,
-      "learning_rate": 9.180910621273555e-05,
-      "loss": 0.0618,
-      "step": 220
-    },
-    {
-      "epoch": 1.1391752577319587,
-      "grad_norm": 1.515604853630066,
-      "learning_rate": 9.171522065692975e-05,
-      "loss": 0.0923,
-      "step": 221
-    },
-    {
-      "epoch": 1.1443298969072164,
-      "grad_norm": 1.8872064352035522,
-      "learning_rate": 9.162084867351842e-05,
-      "loss": 0.0943,
-      "step": 222
-    },
-    {
-      "epoch": 1.1494845360824741,
-      "grad_norm": 0.26518696546554565,
-      "learning_rate": 9.152599136294253e-05,
-      "loss": 0.0062,
-      "step": 223
-    },
-    {
-      "epoch": 1.1546391752577319,
-      "grad_norm": 2.0155885219573975,
-      "learning_rate": 9.14306498313023e-05,
-      "loss": 0.1104,
-      "step": 224
-    },
-    {
-      "epoch": 1.1597938144329896,
-      "grad_norm": 1.6406800746917725,
-      "learning_rate": 9.133482519034428e-05,
-      "loss": 0.0518,
-      "step": 225
-    },
-    {
-      "epoch": 1.1649484536082475,
-      "grad_norm": 2.0402371883392334,
-      "learning_rate": 9.123851855744843e-05,
-      "loss": 0.1368,
-      "step": 226
-    },
-    {
-      "epoch": 1.1701030927835052,
-      "grad_norm": 1.0710994005203247,
-      "learning_rate": 9.114173105561501e-05,
-      "loss": 0.0246,
-      "step": 227
-    },
-    {
-      "epoch": 1.175257731958763,
-      "grad_norm": 0.7198122143745422,
-      "learning_rate": 9.104446381345159e-05,
-      "loss": 0.0255,
-      "step": 228
-    },
-    {
-      "epoch": 1.1804123711340206,
-      "grad_norm": 1.8444178104400635,
-      "learning_rate": 9.094671796515978e-05,
-      "loss": 0.0691,
-      "step": 229
-    },
-    {
-      "epoch": 1.1855670103092784,
-      "grad_norm": 0.8436192274093628,
-      "learning_rate": 9.08484946505221e-05,
-      "loss": 0.0135,
-      "step": 230
-    },
-    {
-      "epoch": 1.190721649484536,
-      "grad_norm": 2.076083183288574,
-      "learning_rate": 9.074979501488867e-05,
-      "loss": 0.0766,
-      "step": 231
-    },
-    {
-      "epoch": 1.1958762886597938,
-      "grad_norm": 0.5280607342720032,
-      "learning_rate": 9.065062020916377e-05,
-      "loss": 0.0135,
-      "step": 232
-    },
-    {
-      "epoch": 1.2010309278350515,
-      "grad_norm": 5.431139945983887,
-      "learning_rate": 9.055097138979252e-05,
-      "loss": 0.3477,
-      "step": 233
-    },
-    {
-      "epoch": 1.2061855670103092,
-      "grad_norm": 2.9912867546081543,
-      "learning_rate": 9.045084971874738e-05,
-      "loss": 0.1651,
-      "step": 234
-    },
-    {
-      "epoch": 1.211340206185567,
-      "grad_norm": 3.116377830505371,
-      "learning_rate": 9.035025636351452e-05,
-      "loss": 0.0958,
-      "step": 235
-    },
-    {
-      "epoch": 1.2164948453608249,
-      "grad_norm": 2.6776368618011475,
-      "learning_rate": 9.024919249708035e-05,
-      "loss": 0.0679,
-      "step": 236
-    },
-    {
-      "epoch": 1.2216494845360826,
-      "grad_norm": 1.5956416130065918,
-      "learning_rate": 9.014765929791768e-05,
-      "loss": 0.0292,
-      "step": 237
-    },
-    {
-      "epoch": 1.2268041237113403,
-      "grad_norm": 2.6305689811706543,
-      "learning_rate": 9.004565794997209e-05,
-      "loss": 0.0979,
-      "step": 238
-    },
-    {
-      "epoch": 1.231958762886598,
-      "grad_norm": 0.8152199387550354,
-      "learning_rate": 8.994318964264809e-05,
-      "loss": 0.017,
-      "step": 239
-    },
-    {
-      "epoch": 1.2371134020618557,
-      "grad_norm": 2.0557334423065186,
-      "learning_rate": 8.984025557079523e-05,
-      "loss": 0.0471,
-      "step": 240
-    },
-    {
-      "epoch": 1.2422680412371134,
-      "grad_norm": 1.7707017660140991,
-      "learning_rate": 8.973685693469423e-05,
-      "loss": 0.0545,
-      "step": 241
-    },
-    {
-      "epoch": 1.2474226804123711,
-      "grad_norm": 1.4157006740570068,
-      "learning_rate": 8.963299494004291e-05,
-      "loss": 0.0454,
-      "step": 242
-    },
-    {
-      "epoch": 1.2525773195876289,
-      "grad_norm": 2.159468650817871,
-      "learning_rate": 8.952867079794218e-05,
-      "loss": 0.0411,
-      "step": 243
-    },
-    {
-      "epoch": 1.2577319587628866,
-      "grad_norm": 3.038072109222412,
-      "learning_rate": 8.942388572488187e-05,
-      "loss": 0.1618,
-      "step": 244
-    },
-    {
-      "epoch": 1.2628865979381443,
-      "grad_norm": 2.6952202320098877,
-      "learning_rate": 8.931864094272663e-05,
-      "loss": 0.1251,
-      "step": 245
-    },
-    {
-      "epoch": 1.268041237113402,
-      "grad_norm": 1.2491196393966675,
-      "learning_rate": 8.921293767870157e-05,
-      "loss": 0.0339,
-      "step": 246
-    },
-    {
-      "epoch": 1.2731958762886597,
-      "grad_norm": 1.2598530054092407,
-      "learning_rate": 8.910677716537806e-05,
-      "loss": 0.0333,
-      "step": 247
-    },
-    {
-      "epoch": 1.2783505154639174,
-      "grad_norm": 2.025956869125366,
-      "learning_rate": 8.900016064065923e-05,
-      "loss": 0.0672,
-      "step": 248
-    },
-    {
-      "epoch": 1.2835051546391751,
-      "grad_norm": 2.79536771774292,
-      "learning_rate": 8.889308934776572e-05,
-      "loss": 0.1278,
-      "step": 249
-    },
-    {
-      "epoch": 1.2886597938144329,
-      "grad_norm": 1.9153764247894287,
-      "learning_rate": 8.8785564535221e-05,
-      "loss": 0.0495,
-      "step": 250
-    },
-    {
-      "epoch": 1.2938144329896908,
-      "grad_norm": 2.235368013381958,
-      "learning_rate": 8.867758745683687e-05,
-      "loss": 0.0573,
-      "step": 251
-    },
-    {
-      "epoch": 1.2989690721649485,
-      "grad_norm": 0.9632558226585388,
-      "learning_rate": 8.85691593716989e-05,
-      "loss": 0.03,
-      "step": 252
-    },
-    {
-      "epoch": 1.3041237113402062,
-      "grad_norm": 0.7992522716522217,
-      "learning_rate": 8.84602815441517e-05,
-      "loss": 0.0194,
-      "step": 253
-    },
-    {
-      "epoch": 1.309278350515464,
-      "grad_norm": 1.7946758270263672,
-      "learning_rate": 8.835095524378414e-05,
-      "loss": 0.0344,
-      "step": 254
-    },
-    {
-      "epoch": 1.3144329896907216,
-      "grad_norm": 1.6255040168762207,
-      "learning_rate": 8.824118174541464e-05,
-      "loss": 0.0815,
-      "step": 255
-    },
-    {
-      "epoch": 1.3195876288659794,
-      "grad_norm": 3.2107818126678467,
-      "learning_rate": 8.81309623290762e-05,
-      "loss": 0.1047,
-      "step": 256
-    },
-    {
-      "epoch": 1.324742268041237,
-      "grad_norm": 3.3725228309631348,
-      "learning_rate": 8.802029828000156e-05,
-      "loss": 0.132,
-      "step": 257
-    },
-    {
-      "epoch": 1.3298969072164948,
-      "grad_norm": 3.7058792114257812,
-      "learning_rate": 8.790919088860814e-05,
-      "loss": 0.127,
-      "step": 258
-    },
-    {
-      "epoch": 1.3350515463917525,
-      "grad_norm": 2.2644128799438477,
-      "learning_rate": 8.779764145048308e-05,
-      "loss": 0.0506,
-      "step": 259
-    },
-    {
-      "epoch": 1.3402061855670104,
-      "grad_norm": 1.035138726234436,
-      "learning_rate": 8.768565126636806e-05,
-      "loss": 0.0165,
-      "step": 260
-    },
-    {
-      "epoch": 1.3453608247422681,
-      "grad_norm": 2.1964375972747803,
-      "learning_rate": 8.757322164214413e-05,
-      "loss": 0.0643,
-      "step": 261
-    },
-    {
-      "epoch": 1.3505154639175259,
-      "grad_norm": 1.7320009469985962,
-      "learning_rate": 8.746035388881655e-05,
-      "loss": 0.0418,
-      "step": 262
-    },
-    {
-      "epoch": 1.3556701030927836,
-      "grad_norm": 0.9327526092529297,
-      "learning_rate": 8.734704932249944e-05,
-      "loss": 0.0145,
-      "step": 263
-    },
-    {
-      "epoch": 1.3608247422680413,
-      "grad_norm": 3.7060790061950684,
-      "learning_rate": 8.723330926440045e-05,
-      "loss": 0.2553,
-      "step": 264
-    },
-    {
-      "epoch": 1.365979381443299,
-      "grad_norm": 2.5351924896240234,
-      "learning_rate": 8.711913504080534e-05,
-      "loss": 0.0692,
-      "step": 265
-    },
-    {
-      "epoch": 1.3711340206185567,
-      "grad_norm": 0.6212343573570251,
-      "learning_rate": 8.70045279830626e-05,
-      "loss": 0.0092,
-      "step": 266
-    },
-    {
-      "epoch": 1.3762886597938144,
-      "grad_norm": 5.307930946350098,
-      "learning_rate": 8.688948942756778e-05,
-      "loss": 0.2139,
-      "step": 267
-    },
-    {
-      "epoch": 1.3814432989690721,
-      "grad_norm": 2.6784942150115967,
-      "learning_rate": 8.677402071574805e-05,
-      "loss": 0.1175,
-      "step": 268
-    },
-    {
-      "epoch": 1.3865979381443299,
-      "grad_norm": 3.658653736114502,
-      "learning_rate": 8.665812319404643e-05,
-      "loss": 0.1758,
-      "step": 269
-    },
-    {
-      "epoch": 1.3917525773195876,
-      "grad_norm": 0.7189630270004272,
-      "learning_rate": 8.654179821390621e-05,
-      "loss": 0.0173,
-      "step": 270
-    },
-    {
-      "epoch": 1.3969072164948453,
-      "grad_norm": 1.0501234531402588,
-      "learning_rate": 8.642504713175508e-05,
-      "loss": 0.0255,
-      "step": 271
-    },
-    {
-      "epoch": 1.402061855670103,
-      "grad_norm": 1.4844290018081665,
-      "learning_rate": 8.630787130898943e-05,
-      "loss": 0.0522,
-      "step": 272
-    },
-    {
-      "epoch": 1.4072164948453607,
-      "grad_norm": 2.0968682765960693,
-      "learning_rate": 8.619027211195836e-05,
-      "loss": 0.044,
-      "step": 273
-    },
-    {
-      "epoch": 1.4123711340206184,
-      "grad_norm": 3.405850887298584,
-      "learning_rate": 8.607225091194779e-05,
-      "loss": 0.2777,
-      "step": 274
-    },
-    {
-      "epoch": 1.4175257731958764,
-      "grad_norm": 0.47108590602874756,
-      "learning_rate": 8.595380908516454e-05,
-      "loss": 0.0079,
-      "step": 275
-    },
-    {
-      "epoch": 1.422680412371134,
-      "grad_norm": 5.354682445526123,
-      "learning_rate": 8.583494801272018e-05,
-      "loss": 0.2747,
-      "step": 276
-    },
-    {
-      "epoch": 1.4278350515463918,
-      "grad_norm": 2.489128589630127,
-      "learning_rate": 8.571566908061497e-05,
-      "loss": 0.0863,
-      "step": 277
-    },
-    {
-      "epoch": 1.4329896907216495,
-      "grad_norm": 3.1641483306884766,
-      "learning_rate": 8.559597367972168e-05,
-      "loss": 0.1414,
-      "step": 278
-    },
-    {
-      "epoch": 1.4381443298969072,
-      "grad_norm": 2.6844987869262695,
-      "learning_rate": 8.547586320576945e-05,
-      "loss": 0.1925,
-      "step": 279
-    },
-    {
-      "epoch": 1.443298969072165,
-      "grad_norm": 0.3662746250629425,
-      "learning_rate": 8.535533905932738e-05,
-      "loss": 0.0079,
-      "step": 280
-    },
-    {
-      "epoch": 1.4484536082474226,
-      "grad_norm": 2.521512269973755,
-      "learning_rate": 8.52344026457883e-05,
-      "loss": 0.2552,
-      "step": 281
-    },
-    {
-      "epoch": 1.4536082474226804,
-      "grad_norm": 1.6905772686004639,
-      "learning_rate": 8.511305537535237e-05,
-      "loss": 0.0426,
-      "step": 282
-    },
-    {
-      "epoch": 1.458762886597938,
-      "grad_norm": 2.184577226638794,
-      "learning_rate": 8.499129866301057e-05,
-      "loss": 0.0719,
-      "step": 283
-    },
-    {
-      "epoch": 1.463917525773196,
-      "grad_norm": 2.447618007659912,
-      "learning_rate": 8.48691339285283e-05,
-      "loss": 0.1018,
-      "step": 284
-    },
-    {
-      "epoch": 1.4690721649484537,
-      "grad_norm": 3.0394506454467773,
-      "learning_rate": 8.474656259642873e-05,
-      "loss": 0.1984,
-      "step": 285
-    },
-    {
-      "epoch": 1.4742268041237114,
-      "grad_norm": 0.8660880923271179,
-      "learning_rate": 8.46235860959763e-05,
-      "loss": 0.0198,
-      "step": 286
-    },
-    {
-      "epoch": 1.4793814432989691,
-      "grad_norm": 2.118943929672241,
-      "learning_rate": 8.450020586115987e-05,
-      "loss": 0.1439,
-      "step": 287
-    },
-    {
-      "epoch": 1.4845360824742269,
-      "grad_norm": 1.2160487174987793,
-      "learning_rate": 8.437642333067625e-05,
-      "loss": 0.0511,
-      "step": 288
-    },
-    {
-      "epoch": 1.4896907216494846,
-      "grad_norm": 2.628675699234009,
-      "learning_rate": 8.42522399479132e-05,
-      "loss": 0.1341,
-      "step": 289
-    },
-    {
-      "epoch": 1.4948453608247423,
-      "grad_norm": 1.4727402925491333,
-      "learning_rate": 8.412765716093272e-05,
-      "loss": 0.0472,
-      "step": 290
-    },
-    {
-      "epoch": 1.5,
-      "grad_norm": 1.7666587829589844,
-      "learning_rate": 8.40026764224541e-05,
-      "loss": 0.0689,
-      "step": 291
-    },
-    {
-      "epoch": 1.5,
-      "eval_loss": 0.10398600995540619,
-      "eval_runtime": 22.9785,
-      "eval_samples_per_second": 7.137,
-      "eval_steps_per_second": 1.784,
-      "step": 291
-    },
-    {
-      "epoch": 1.5051546391752577,
-      "grad_norm": 2.2388651371002197,
-      "learning_rate": 8.387729918983706e-05,
-      "loss": 0.0711,
-      "step": 292
-    },
-    {
-      "epoch": 1.5103092783505154,
-      "grad_norm": 1.5305780172348022,
-      "learning_rate": 8.375152692506468e-05,
-      "loss": 0.0564,
-      "step": 293
-    },
-    {
-      "epoch": 1.5154639175257731,
-      "grad_norm": 3.6230311393737793,
-      "learning_rate": 8.362536109472636e-05,
-      "loss": 0.313,
-      "step": 294
-    },
-    {
-      "epoch": 1.5206185567010309,
-      "grad_norm": 1.3446002006530762,
-      "learning_rate": 8.349880317000082e-05,
-      "loss": 0.0531,
-      "step": 295
-    },
-    {
-      "epoch": 1.5257731958762886,
-      "grad_norm": 1.283772349357605,
-      "learning_rate": 8.337185462663878e-05,
-      "loss": 0.086,
-      "step": 296
-    },
-    {
-      "epoch": 1.5309278350515463,
-      "grad_norm": 1.9061647653579712,
-      "learning_rate": 8.32445169449459e-05,
-      "loss": 0.0897,
-      "step": 297
-    },
-    {
-      "epoch": 1.536082474226804,
-      "grad_norm": 0.9783570766448975,
-      "learning_rate": 8.311679160976539e-05,
-      "loss": 0.0401,
-      "step": 298
-    },
-    {
-      "epoch": 1.5412371134020617,
-      "grad_norm": 1.3108831644058228,
-      "learning_rate": 8.29886801104608e-05,
-      "loss": 0.0583,
-      "step": 299
-    },
-    {
-      "epoch": 1.5463917525773194,
-      "grad_norm": 2.761857032775879,
-      "learning_rate": 8.286018394089863e-05,
-      "loss": 0.1497,
-      "step": 300
-    },
-    {
-      "epoch": 1.5515463917525774,
-      "grad_norm": 1.9834277629852295,
-      "learning_rate": 8.273130459943086e-05,
-      "loss": 0.0695,
-      "step": 301
-    },
-    {
-      "epoch": 1.556701030927835,
-      "grad_norm": 1.5544965267181396,
-      "learning_rate": 8.260204358887754e-05,
-      "loss": 0.1192,
-      "step": 302
-    },
-    {
-      "epoch": 1.5618556701030928,
-      "grad_norm": 0.6277759671211243,
-      "learning_rate": 8.247240241650918e-05,
-      "loss": 0.0158,
-      "step": 303
-    },
-    {
-      "epoch": 1.5670103092783505,
-      "grad_norm": 0.8964148163795471,
-      "learning_rate": 8.234238259402935e-05,
-      "loss": 0.0364,
-      "step": 304
-    },
-    {
-      "epoch": 1.5721649484536082,
-      "grad_norm": 1.8108261823654175,
-      "learning_rate": 8.221198563755682e-05,
-      "loss": 0.1198,
-      "step": 305
-    },
-    {
-      "epoch": 1.577319587628866,
-      "grad_norm": 1.8986718654632568,
-      "learning_rate": 8.208121306760805e-05,
-      "loss": 0.0912,
-      "step": 306
-    },
-    {
-      "epoch": 1.5824742268041239,
-      "grad_norm": 0.25971317291259766,
-      "learning_rate": 8.195006640907942e-05,
-      "loss": 0.0047,
-      "step": 307
-    },
-    {
-      "epoch": 1.5876288659793816,
-      "grad_norm": 0.2667933702468872,
-      "learning_rate": 8.181854719122939e-05,
-      "loss": 0.0069,
-      "step": 308
-    },
-    {
-      "epoch": 1.5927835051546393,
-      "grad_norm": 1.9358141422271729,
-      "learning_rate": 8.168665694766073e-05,
-      "loss": 0.0873,
-      "step": 309
-    },
-    {
-      "epoch": 1.597938144329897,
-      "grad_norm": 1.8080211877822876,
-      "learning_rate": 8.155439721630264e-05,
-      "loss": 0.0706,
-      "step": 310
-    },
-    {
-      "epoch": 1.6030927835051547,
-      "grad_norm": 0.8074958324432373,
-      "learning_rate": 8.142176953939279e-05,
-      "loss": 0.0401,
-      "step": 311
-    },
-    {
-      "epoch": 1.6082474226804124,
-      "grad_norm": 0.34044063091278076,
-      "learning_rate": 8.128877546345933e-05,
-      "loss": 0.0088,
-      "step": 312
-    },
-    {
-      "epoch": 1.6134020618556701,
-      "grad_norm": 2.4520204067230225,
-      "learning_rate": 8.115541653930286e-05,
-      "loss": 0.0972,
-      "step": 313
-    },
-    {
-      "epoch": 1.6185567010309279,
-      "grad_norm": 2.7366955280303955,
-      "learning_rate": 8.102169432197842e-05,
-      "loss": 0.1658,
-      "step": 314
-    },
-    {
-      "epoch": 1.6237113402061856,
-      "grad_norm": 1.3256347179412842,
-      "learning_rate": 8.088761037077718e-05,
-      "loss": 0.0465,
-      "step": 315
-    },
-    {
-      "epoch": 1.6288659793814433,
-      "grad_norm": 0.23313826322555542,
-      "learning_rate": 8.075316624920848e-05,
-      "loss": 0.0077,
-      "step": 316
-    },
-    {
-      "epoch": 1.634020618556701,
-      "grad_norm": 1.9687016010284424,
-      "learning_rate": 8.061836352498145e-05,
-      "loss": 0.0768,
-      "step": 317
-    },
-    {
-      "epoch": 1.6391752577319587,
-      "grad_norm": 1.2009949684143066,
-      "learning_rate": 8.048320376998673e-05,
-      "loss": 0.128,
-      "step": 318
-    },
-    {
-      "epoch": 1.6443298969072164,
-      "grad_norm": 1.9169154167175293,
-      "learning_rate": 8.034768856027826e-05,
-      "loss": 0.0903,
-      "step": 319
-    },
-    {
-      "epoch": 1.6494845360824741,
-      "grad_norm": 2.0573318004608154,
-      "learning_rate": 8.021181947605473e-05,
-      "loss": 0.1958,
-      "step": 320
-    },
-    {
-      "epoch": 1.6546391752577319,
-      "grad_norm": 1.6862727403640747,
-      "learning_rate": 8.007559810164133e-05,
-      "loss": 0.1499,
-      "step": 321
-    },
-    {
-      "epoch": 1.6597938144329896,
-      "grad_norm": 2.078341007232666,
-      "learning_rate": 7.993902602547113e-05,
-      "loss": 0.0639,
-      "step": 322
-    },
-    {
-      "epoch": 1.6649484536082473,
-      "grad_norm": 1.3753212690353394,
-      "learning_rate": 7.980210484006666e-05,
-      "loss": 0.0471,
-      "step": 323
-    },
-    {
-      "epoch": 1.670103092783505,
-      "grad_norm": 2.7627837657928467,
-      "learning_rate": 7.966483614202128e-05,
-      "loss": 0.1081,
-      "step": 324
-    },
-    {
-      "epoch": 1.675257731958763,
-      "grad_norm": 2.5081069469451904,
-      "learning_rate": 7.952722153198054e-05,
-      "loss": 0.1212,
-      "step": 325
-    },
-    {
-      "epoch": 1.6804123711340206,
-      "grad_norm": 2.0369913578033447,
-      "learning_rate": 7.938926261462366e-05,
-      "loss": 0.0373,
-      "step": 326
-    },
-    {
-      "epoch": 1.6855670103092784,
-      "grad_norm": 2.289896249771118,
-      "learning_rate": 7.925096099864464e-05,
-      "loss": 0.0828,
-      "step": 327
-    },
-    {
-      "epoch": 1.690721649484536,
-      "grad_norm": 2.909376621246338,
-      "learning_rate": 7.911231829673356e-05,
-      "loss": 0.1396,
-      "step": 328
-    },
-    {
-      "epoch": 1.6958762886597938,
-      "grad_norm": 2.371479034423828,
-      "learning_rate": 7.897333612555785e-05,
-      "loss": 0.0491,
-      "step": 329
-    },
-    {
-      "epoch": 1.7010309278350515,
-      "grad_norm": 1.4082870483398438,
-      "learning_rate": 7.883401610574336e-05,
-      "loss": 0.0393,
-      "step": 330
-    },
-    {
-      "epoch": 1.7061855670103094,
-      "grad_norm": 3.49005126953125,
-      "learning_rate": 7.869435986185547e-05,
-      "loss": 0.1891,
-      "step": 331
-    },
-    {
-      "epoch": 1.7113402061855671,
-      "grad_norm": 2.8195505142211914,
-      "learning_rate": 7.855436902238017e-05,
-      "loss": 0.1249,
-      "step": 332
-    },
-    {
-      "epoch": 1.7164948453608249,
-      "grad_norm": 1.555109977722168,
-      "learning_rate": 7.841404521970505e-05,
-      "loss": 0.1454,
-      "step": 333
-    },
-    {
-      "epoch": 1.7216494845360826,
-      "grad_norm": 2.4946799278259277,
-      "learning_rate": 7.82733900901003e-05,
-      "loss": 0.0818,
-      "step": 334
-    },
-    {
-      "epoch": 1.7268041237113403,
-      "grad_norm": 0.8252114653587341,
-      "learning_rate": 7.813240527369959e-05,
-      "loss": 0.0296,
-      "step": 335
-    },
-    {
-      "epoch": 1.731958762886598,
-      "grad_norm": 1.7780154943466187,
-      "learning_rate": 7.799109241448091e-05,
-      "loss": 0.0818,
-      "step": 336
-    },
-    {
-      "epoch": 1.7371134020618557,
-      "grad_norm": 0.4910796284675598,
-      "learning_rate": 7.784945316024756e-05,
-      "loss": 0.0242,
-      "step": 337
-    },
-    {
-      "epoch": 1.7422680412371134,
-      "grad_norm": 2.092083215713501,
-      "learning_rate": 7.770748916260875e-05,
-      "loss": 0.1391,
-      "step": 338
-    },
-    {
-      "epoch": 1.7474226804123711,
-      "grad_norm": 1.151160478591919,
-      "learning_rate": 7.756520207696041e-05,
-      "loss": 0.0362,
-      "step": 339
-    },
-    {
-      "epoch": 1.7525773195876289,
-      "grad_norm": 1.400128960609436,
-      "learning_rate": 7.742259356246593e-05,
-      "loss": 0.0566,
-      "step": 340
-    },
-    {
-      "epoch": 1.7577319587628866,
-      "grad_norm": 1.9376780986785889,
-      "learning_rate": 7.727966528203678e-05,
-      "loss": 0.1098,
-      "step": 341
-    },
-    {
-      "epoch": 1.7628865979381443,
-      "grad_norm": 0.3112329840660095,
-      "learning_rate": 7.71364189023131e-05,
-      "loss": 0.011,
-      "step": 342
-    },
-    {
-      "epoch": 1.768041237113402,
-      "grad_norm": 1.3194454908370972,
-      "learning_rate": 7.699285609364424e-05,
-      "loss": 0.061,
-      "step": 343
-    },
-    {
-      "epoch": 1.7731958762886597,
-      "grad_norm": 1.4510351419448853,
-      "learning_rate": 7.68489785300694e-05,
-      "loss": 0.0577,
-      "step": 344
-    },
-    {
-      "epoch": 1.7783505154639174,
-      "grad_norm": 0.23605522513389587,
-      "learning_rate": 7.670478788929802e-05,
-      "loss": 0.0058,
-      "step": 345
-    },
-    {
-      "epoch": 1.7835051546391751,
-      "grad_norm": 2.1307742595672607,
-      "learning_rate": 7.656028585269018e-05,
-      "loss": 0.1553,
-      "step": 346
-    },
-    {
-      "epoch": 1.7886597938144329,
-      "grad_norm": 1.0913602113723755,
-      "learning_rate": 7.641547410523709e-05,
-      "loss": 0.0582,
-      "step": 347
-    },
-    {
-      "epoch": 1.7938144329896906,
-      "grad_norm": 0.9072116017341614,
-      "learning_rate": 7.627035433554138e-05,
-      "loss": 0.0298,
-      "step": 348
-    },
-    {
-      "epoch": 1.7989690721649485,
-      "grad_norm": 0.515763521194458,
-      "learning_rate": 7.612492823579745e-05,
-      "loss": 0.01,
-      "step": 349
-    },
-    {
-      "epoch": 1.8041237113402062,
-      "grad_norm": 0.8141824007034302,
-      "learning_rate": 7.597919750177168e-05,
-      "loss": 0.0182,
-      "step": 350
-    },
-    {
-      "epoch": 1.809278350515464,
-      "grad_norm": 2.7248380184173584,
-      "learning_rate": 7.583316383278273e-05,
-      "loss": 0.1133,
-      "step": 351
-    },
-    {
-      "epoch": 1.8144329896907216,
-      "grad_norm": 2.5838863849639893,
-      "learning_rate": 7.568682893168164e-05,
-      "loss": 0.0996,
-      "step": 352
-    },
-    {
-      "epoch": 1.8195876288659794,
-      "grad_norm": 1.152638554573059,
-      "learning_rate": 7.554019450483208e-05,
-      "loss": 0.015,
-      "step": 353
-    },
-    {
-      "epoch": 1.824742268041237,
-      "grad_norm": 1.3408974409103394,
-      "learning_rate": 7.539326226209031e-05,
-      "loss": 0.0599,
-      "step": 354
-    },
-    {
-      "epoch": 1.829896907216495,
-      "grad_norm": 2.786916494369507,
-      "learning_rate": 7.524603391678541e-05,
-      "loss": 0.1024,
-      "step": 355
-    },
-    {
-      "epoch": 1.8350515463917527,
-      "grad_norm": 0.8519769906997681,
-      "learning_rate": 7.509851118569915e-05,
-      "loss": 0.0269,
-      "step": 356
-    },
-    {
-      "epoch": 1.8402061855670104,
-      "grad_norm": 0.3977208435535431,
-      "learning_rate": 7.495069578904608e-05,
-      "loss": 0.009,
-      "step": 357
-    },
-    {
-      "epoch": 1.8453608247422681,
-      "grad_norm": 0.6682721376419067,
-      "learning_rate": 7.48025894504534e-05,
-      "loss": 0.014,
-      "step": 358
-    },
-    {
-      "epoch": 1.8505154639175259,
-      "grad_norm": 1.844580888748169,
-      "learning_rate": 7.465419389694092e-05,
-      "loss": 0.0935,
-      "step": 359
-    },
-    {
-      "epoch": 1.8556701030927836,
-      "grad_norm": 2.259485960006714,
-      "learning_rate": 7.450551085890087e-05,
-      "loss": 0.0708,
-      "step": 360
-    },
-    {
-      "epoch": 1.8608247422680413,
-      "grad_norm": 0.8421556949615479,
-      "learning_rate": 7.435654207007773e-05,
-      "loss": 0.0279,
-      "step": 361
-    },
-    {
-      "epoch": 1.865979381443299,
-      "grad_norm": 0.994603157043457,
-      "learning_rate": 7.420728926754803e-05,
-      "loss": 0.0273,
-      "step": 362
-    },
-    {
-      "epoch": 1.8711340206185567,
-      "grad_norm": 2.5991320610046387,
-      "learning_rate": 7.405775419170014e-05,
-      "loss": 0.1439,
-      "step": 363
-    },
-    {
-      "epoch": 1.8762886597938144,
-      "grad_norm": 3.4226911067962646,
-      "learning_rate": 7.390793858621386e-05,
-      "loss": 0.0539,
-      "step": 364
-    },
-    {
-      "epoch": 1.8814432989690721,
-      "grad_norm": 1.0767531394958496,
-      "learning_rate": 7.375784419804019e-05,
-      "loss": 0.0284,
-      "step": 365
-    },
-    {
-      "epoch": 1.8865979381443299,
-      "grad_norm": 0.8647091388702393,
-      "learning_rate": 7.360747277738094e-05,
-      "loss": 0.0178,
-      "step": 366
-    },
-    {
-      "epoch": 1.8917525773195876,
-      "grad_norm": 0.6758021116256714,
-      "learning_rate": 7.345682607766826e-05,
-      "loss": 0.0092,
-      "step": 367
-    },
-    {
-      "epoch": 1.8969072164948453,
-      "grad_norm": 0.6764877438545227,
-      "learning_rate": 7.330590585554428e-05,
-      "loss": 0.0114,
-      "step": 368
-    },
-    {
-      "epoch": 1.902061855670103,
-      "grad_norm": 1.526297926902771,
-      "learning_rate": 7.315471387084056e-05,
-      "loss": 0.0368,
-      "step": 369
-    },
-    {
-      "epoch": 1.9072164948453607,
-      "grad_norm": 2.972353219985962,
-      "learning_rate": 7.300325188655761e-05,
-      "loss": 0.18,
-      "step": 370
-    },
-    {
-      "epoch": 1.9123711340206184,
-      "grad_norm": 0.385691374540329,
-      "learning_rate": 7.285152166884432e-05,
-      "loss": 0.0104,
-      "step": 371
-    },
-    {
-      "epoch": 1.9175257731958761,
-      "grad_norm": 4.219260215759277,
-      "learning_rate": 7.269952498697734e-05,
-      "loss": 0.1956,
-      "step": 372
-    },
-    {
-      "epoch": 1.922680412371134,
-      "grad_norm": 0.18770945072174072,
-      "learning_rate": 7.25472636133405e-05,
-      "loss": 0.0041,
-      "step": 373
-    },
-    {
-      "epoch": 1.9278350515463918,
-      "grad_norm": 1.2908581495285034,
-      "learning_rate": 7.23947393234041e-05,
-      "loss": 0.0237,
-      "step": 374
-    },
-    {
-      "epoch": 1.9329896907216495,
-      "grad_norm": 2.868788242340088,
-      "learning_rate": 7.224195389570422e-05,
-      "loss": 0.1838,
-      "step": 375
-    },
-    {
-      "epoch": 1.9381443298969072,
-      "grad_norm": 2.256253719329834,
-      "learning_rate": 7.208890911182197e-05,
-      "loss": 0.0469,
-      "step": 376
-    },
-    {
-      "epoch": 1.943298969072165,
-      "grad_norm": 1.5368409156799316,
-      "learning_rate": 7.193560675636277e-05,
-      "loss": 0.0516,
-      "step": 377
-    },
-    {
-      "epoch": 1.9484536082474226,
-      "grad_norm": 0.9154139757156372,
-      "learning_rate": 7.178204861693545e-05,
-      "loss": 0.0178,
-      "step": 378
-    },
-    {
-      "epoch": 1.9536082474226806,
-      "grad_norm": 0.3536202311515808,
-      "learning_rate": 7.162823648413151e-05,
-      "loss": 0.0065,
-      "step": 379
-    },
-    {
-      "epoch": 1.9587628865979383,
-      "grad_norm": 3.183803081512451,
-      "learning_rate": 7.14741721515041e-05,
-      "loss": 0.1542,
-      "step": 380
-    },
-    {
-      "epoch": 1.963917525773196,
-      "grad_norm": 2.011648416519165,
-      "learning_rate": 7.131985741554728e-05,
-      "loss": 0.0602,
-      "step": 381
-    },
-    {
-      "epoch": 1.9690721649484537,
-      "grad_norm": 2.658155679702759,
-      "learning_rate": 7.116529407567489e-05,
-      "loss": 0.1196,
-      "step": 382
-    },
-    {
-      "epoch": 1.9742268041237114,
-      "grad_norm": 1.5329421758651733,
-      "learning_rate": 7.101048393419977e-05,
-      "loss": 0.0444,
-      "step": 383
-    },
-    {
-      "epoch": 1.9793814432989691,
-      "grad_norm": 0.7631077766418457,
-      "learning_rate": 7.085542879631253e-05,
-      "loss": 0.0241,
-      "step": 384
-    },
-    {
-      "epoch": 1.9845360824742269,
-      "grad_norm": 2.961472749710083,
-      "learning_rate": 7.070013047006068e-05,
-      "loss": 0.1172,
-      "step": 385
-    },
-    {
-      "epoch": 1.9896907216494846,
-      "grad_norm": 1.429437279701233,
-      "learning_rate": 7.054459076632743e-05,
-      "loss": 0.0483,
-      "step": 386
-    },
-    {
-      "epoch": 1.9948453608247423,
-      "grad_norm": 2.7817625999450684,
-      "learning_rate": 7.038881149881058e-05,
-      "loss": 0.1143,
-      "step": 387
-    },
-    {
-      "epoch": 2.0,
-      "grad_norm": 0.6764756441116333,
-      "learning_rate": 7.02327944840015e-05,
-      "loss": 0.0185,
-      "step": 388
-    },
-    {
-      "epoch": 2.0,
-      "eval_loss": 0.07160064578056335,
-      "eval_runtime": 23.0429,
-      "eval_samples_per_second": 7.117,
-      "eval_steps_per_second": 1.779,
-      "step": 388
-    },
-    {
-      "epoch": 2.0051546391752577,
-      "grad_norm": 0.07525697350502014,
-      "learning_rate": 7.007654154116377e-05,
-      "loss": 0.0027,
-      "step": 389
-    },
-    {
-      "epoch": 2.0103092783505154,
-      "grad_norm": 1.171702265739441,
-      "learning_rate": 6.992005449231208e-05,
-      "loss": 0.0673,
-      "step": 390
-    },
-    {
-      "epoch": 2.015463917525773,
-      "grad_norm": 0.2757341265678406,
-      "learning_rate": 6.976333516219096e-05,
-      "loss": 0.008,
-      "step": 391
-    },
-    {
-      "epoch": 2.020618556701031,
-      "grad_norm": 0.18395589292049408,
-      "learning_rate": 6.960638537825352e-05,
-      "loss": 0.0037,
-      "step": 392
-    },
-    {
-      "epoch": 2.0257731958762886,
-      "grad_norm": 0.2340579330921173,
-      "learning_rate": 6.944920697064004e-05,
-      "loss": 0.0073,
-      "step": 393
-    },
-    {
-      "epoch": 2.0309278350515463,
-      "grad_norm": 0.7616062164306641,
-      "learning_rate": 6.929180177215678e-05,
-      "loss": 0.0352,
-      "step": 394
-    },
-    {
-      "epoch": 2.036082474226804,
-      "grad_norm": 0.4245890974998474,
-      "learning_rate": 6.91341716182545e-05,
-      "loss": 0.0126,
-      "step": 395
-    },
-    {
-      "epoch": 2.0412371134020617,
-      "grad_norm": 2.350367546081543,
-      "learning_rate": 6.897631834700709e-05,
-      "loss": 0.0362,
-      "step": 396
-    },
-    {
-      "epoch": 2.0463917525773194,
-      "grad_norm": 1.0493096113204956,
-      "learning_rate": 6.881824379909017e-05,
-      "loss": 0.0292,
-      "step": 397
-    },
-    {
-      "epoch": 2.051546391752577,
-      "grad_norm": 1.125183343887329,
-      "learning_rate": 6.865994981775957e-05,
-      "loss": 0.0508,
-      "step": 398
-    },
-    {
-      "epoch": 2.056701030927835,
-      "grad_norm": 0.7406489253044128,
-      "learning_rate": 6.850143824882986e-05,
-      "loss": 0.0189,
-      "step": 399
-    },
-    {
-      "epoch": 2.0618556701030926,
-      "grad_norm": 1.1769431829452515,
-      "learning_rate": 6.834271094065283e-05,
-      "loss": 0.0347,
-      "step": 400
-    },
-    {
-      "epoch": 2.0670103092783507,
-      "grad_norm": 0.06829158961772919,
-      "learning_rate": 6.818376974409593e-05,
-      "loss": 0.0024,
-      "step": 401
-    },
-    {
-      "epoch": 2.0721649484536084,
-      "grad_norm": 0.5625203847885132,
-      "learning_rate": 6.802461651252073e-05,
-      "loss": 0.0107,
-      "step": 402
-    },
-    {
-      "epoch": 2.077319587628866,
-      "grad_norm": 0.4858710765838623,
-      "learning_rate": 6.786525310176123e-05,
-      "loss": 0.0231,
-      "step": 403
-    },
-    {
-      "epoch": 2.082474226804124,
-      "grad_norm": 0.5900213718414307,
-      "learning_rate": 6.770568137010226e-05,
-      "loss": 0.017,
-      "step": 404
-    },
-    {
-      "epoch": 2.0876288659793816,
-      "grad_norm": 2.3651366233825684,
-      "learning_rate": 6.754590317825785e-05,
-      "loss": 0.0628,
-      "step": 405
-    },
-    {
-      "epoch": 2.0927835051546393,
-      "grad_norm": 1.993862271308899,
-      "learning_rate": 6.738592038934946e-05,
-      "loss": 0.0581,
-      "step": 406
-    },
-    {
-      "epoch": 2.097938144329897,
-      "grad_norm": 1.991409182548523,
-      "learning_rate": 6.722573486888427e-05,
-      "loss": 0.055,
-      "step": 407
-    },
-    {
-      "epoch": 2.1030927835051547,
-      "grad_norm": 0.10092249512672424,
-      "learning_rate": 6.706534848473352e-05,
-      "loss": 0.0021,
-      "step": 408
-    },
-    {
-      "epoch": 2.1082474226804124,
-      "grad_norm": 0.5169625282287598,
-      "learning_rate": 6.69047631071106e-05,
-      "loss": 0.0116,
-      "step": 409
-    },
-    {
-      "epoch": 2.11340206185567,
-      "grad_norm": 0.5807163119316101,
-      "learning_rate": 6.674398060854931e-05,
-      "loss": 0.0111,
-      "step": 410
-    },
-    {
-      "epoch": 2.118556701030928,
-      "grad_norm": 2.279827833175659,
-      "learning_rate": 6.658300286388203e-05,
-      "loss": 0.059,
-      "step": 411
-    },
-    {
-      "epoch": 2.1237113402061856,
-      "grad_norm": 1.3535722494125366,
-      "learning_rate": 6.642183175021779e-05,
-      "loss": 0.0326,
-      "step": 412
-    },
-    {
-      "epoch": 2.1288659793814433,
-      "grad_norm": 0.06376868486404419,
-      "learning_rate": 6.62604691469205e-05,
-      "loss": 0.0021,
-      "step": 413
-    },
-    {
-      "epoch": 2.134020618556701,
-      "grad_norm": 0.6728596091270447,
-      "learning_rate": 6.609891693558692e-05,
-      "loss": 0.0139,
-      "step": 414
-    },
-    {
-      "epoch": 2.1391752577319587,
-      "grad_norm": 0.8516107797622681,
-      "learning_rate": 6.59371770000248e-05,
-      "loss": 0.0214,
-      "step": 415
-    },
-    {
-      "epoch": 2.1443298969072164,
-      "grad_norm": 1.3128242492675781,
-      "learning_rate": 6.577525122623084e-05,
-      "loss": 0.029,
-      "step": 416
-    },
-    {
-      "epoch": 2.149484536082474,
-      "grad_norm": 0.3968404233455658,
-      "learning_rate": 6.561314150236882e-05,
-      "loss": 0.0063,
-      "step": 417
-    },
-    {
-      "epoch": 2.154639175257732,
-      "grad_norm": 2.2075860500335693,
-      "learning_rate": 6.545084971874738e-05,
-      "loss": 0.04,
-      "step": 418
-    },
-    {
-      "epoch": 2.1597938144329896,
-      "grad_norm": 2.252422332763672,
-      "learning_rate": 6.528837776779819e-05,
-      "loss": 0.0291,
-      "step": 419
-    },
-    {
-      "epoch": 2.1649484536082473,
-      "grad_norm": 0.9669781923294067,
-      "learning_rate": 6.51257275440538e-05,
-      "loss": 0.0189,
-      "step": 420
-    },
-    {
-      "epoch": 2.170103092783505,
-      "grad_norm": 0.15717247128486633,
-      "learning_rate": 6.496290094412546e-05,
-      "loss": 0.0035,
-      "step": 421
-    },
-    {
-      "epoch": 2.1752577319587627,
-      "grad_norm": 1.4802827835083008,
-      "learning_rate": 6.479989986668118e-05,
-      "loss": 0.0245,
-      "step": 422
-    },
-    {
-      "epoch": 2.1804123711340204,
-      "grad_norm": 1.0676665306091309,
-      "learning_rate": 6.463672621242342e-05,
-      "loss": 0.0145,
-      "step": 423
-    },
-    {
-      "epoch": 2.1855670103092786,
-      "grad_norm": 3.3371965885162354,
-      "learning_rate": 6.447338188406704e-05,
-      "loss": 0.0616,
-      "step": 424
-    },
-    {
-      "epoch": 2.1907216494845363,
-      "grad_norm": 0.09870132803916931,
-      "learning_rate": 6.430986878631707e-05,
-      "loss": 0.0018,
-      "step": 425
-    },
-    {
-      "epoch": 2.195876288659794,
-      "grad_norm": 2.374746799468994,
-      "learning_rate": 6.41461888258465e-05,
-      "loss": 0.0527,
-      "step": 426
-    },
-    {
-      "epoch": 2.2010309278350517,
-      "grad_norm": 0.7650613188743591,
-      "learning_rate": 6.398234391127406e-05,
-      "loss": 0.0067,
-      "step": 427
-    },
-    {
-      "epoch": 2.2061855670103094,
-      "grad_norm": 0.8953922390937805,
-      "learning_rate": 6.381833595314195e-05,
-      "loss": 0.0279,
-      "step": 428
-    },
-    {
-      "epoch": 2.211340206185567,
-      "grad_norm": 0.2915783226490021,
-      "learning_rate": 6.365416686389358e-05,
-      "loss": 0.0035,
-      "step": 429
-    },
-    {
-      "epoch": 2.216494845360825,
-      "grad_norm": 2.34714412689209,
-      "learning_rate": 6.348983855785121e-05,
-      "loss": 0.1444,
-      "step": 430
-    },
-    {
-      "epoch": 2.2216494845360826,
-      "grad_norm": 0.20125645399093628,
-      "learning_rate": 6.332535295119377e-05,
-      "loss": 0.0024,
-      "step": 431
-    },
-    {
-      "epoch": 2.2268041237113403,
-      "grad_norm": 0.358877032995224,
-      "learning_rate": 6.31607119619343e-05,
-      "loss": 0.007,
-      "step": 432
-    },
-    {
-      "epoch": 2.231958762886598,
-      "grad_norm": 0.021968163549900055,
-      "learning_rate": 6.299591750989779e-05,
-      "loss": 0.0006,
-      "step": 433
-    },
-    {
-      "epoch": 2.2371134020618557,
-      "grad_norm": 0.09418534487485886,
-      "learning_rate": 6.283097151669869e-05,
-      "loss": 0.0018,
-      "step": 434
-    },
-    {
-      "epoch": 2.2422680412371134,
-      "grad_norm": 0.0898284986615181,
-      "learning_rate": 6.266587590571852e-05,
-      "loss": 0.0013,
-      "step": 435
-    },
-    {
-      "epoch": 2.247422680412371,
-      "grad_norm": 0.12911048531532288,
-      "learning_rate": 6.250063260208346e-05,
-      "loss": 0.0034,
-      "step": 436
-    },
-    {
-      "epoch": 2.252577319587629,
-      "grad_norm": 2.0727717876434326,
-      "learning_rate": 6.233524353264187e-05,
-      "loss": 0.0277,
-      "step": 437
-    },
-    {
-      "epoch": 2.2577319587628866,
-      "grad_norm": 0.8507752418518066,
-      "learning_rate": 6.216971062594179e-05,
-      "loss": 0.0093,
-      "step": 438
-    },
-    {
-      "epoch": 2.2628865979381443,
-      "grad_norm": 0.17573043704032898,
-      "learning_rate": 6.200403581220861e-05,
-      "loss": 0.0023,
-      "step": 439
-    },
-    {
-      "epoch": 2.268041237113402,
-      "grad_norm": 1.0571852922439575,
-      "learning_rate": 6.183822102332234e-05,
-      "loss": 0.018,
-      "step": 440
-    },
-    {
-      "epoch": 2.2731958762886597,
-      "grad_norm": 0.14709679782390594,
-      "learning_rate": 6.167226819279528e-05,
-      "loss": 0.0022,
-      "step": 441
-    },
-    {
-      "epoch": 2.2783505154639174,
-      "grad_norm": 0.7632313370704651,
-      "learning_rate": 6.150617925574933e-05,
-      "loss": 0.0123,
-      "step": 442
-    },
-    {
-      "epoch": 2.283505154639175,
-      "grad_norm": 0.09419295191764832,
-      "learning_rate": 6.13399561488935e-05,
-      "loss": 0.0016,
-      "step": 443
-    },
-    {
-      "epoch": 2.288659793814433,
-      "grad_norm": 0.7709088325500488,
-      "learning_rate": 6.117360081050136e-05,
-      "loss": 0.0133,
-      "step": 444
-    },
-    {
-      "epoch": 2.2938144329896906,
-      "grad_norm": 0.6842679977416992,
-      "learning_rate": 6.1007115180388285e-05,
-      "loss": 0.0116,
-      "step": 445
-    },
-    {
-      "epoch": 2.2989690721649483,
-      "grad_norm": 0.0223514623939991,
-      "learning_rate": 6.0840501199889046e-05,
-      "loss": 0.0005,
-      "step": 446
-    },
-    {
-      "epoch": 2.304123711340206,
-      "grad_norm": 1.7335599660873413,
-      "learning_rate": 6.067376081183499e-05,
-      "loss": 0.0198,
-      "step": 447
-    },
-    {
-      "epoch": 2.3092783505154637,
-      "grad_norm": 0.3301581144332886,
-      "learning_rate": 6.050689596053151e-05,
-      "loss": 0.0036,
-      "step": 448
-    },
-    {
-      "epoch": 2.3144329896907214,
-      "grad_norm": 4.583668231964111,
-      "learning_rate": 6.0339908591735296e-05,
-      "loss": 0.186,
-      "step": 449
-    },
-    {
-      "epoch": 2.319587628865979,
-      "grad_norm": 2.8353846073150635,
-      "learning_rate": 6.01728006526317e-05,
-      "loss": 0.0844,
-      "step": 450
-    },
-    {
-      "epoch": 2.3247422680412373,
-      "grad_norm": 1.3765830993652344,
-      "learning_rate": 6.0005574091811964e-05,
-      "loss": 0.027,
-      "step": 451
-    },
-    {
-      "epoch": 2.329896907216495,
-      "grad_norm": 3.2432947158813477,
-      "learning_rate": 5.9838230859250586e-05,
-      "loss": 0.2141,
-      "step": 452
-    },
-    {
-      "epoch": 2.3350515463917527,
-      "grad_norm": 0.536233127117157,
-      "learning_rate": 5.967077290628249e-05,
-      "loss": 0.007,
-      "step": 453
-    },
-    {
-      "epoch": 2.3402061855670104,
-      "grad_norm": 1.7645577192306519,
-      "learning_rate": 5.950320218558037e-05,
-      "loss": 0.1288,
-      "step": 454
-    },
-    {
-      "epoch": 2.345360824742268,
-      "grad_norm": 1.1916286945343018,
-      "learning_rate": 5.9335520651131814e-05,
-      "loss": 0.0184,
-      "step": 455
-    },
-    {
-      "epoch": 2.350515463917526,
-      "grad_norm": 2.7826900482177734,
-      "learning_rate": 5.9167730258216627e-05,
-      "loss": 0.1094,
-      "step": 456
-    },
-    {
-      "epoch": 2.3556701030927836,
-      "grad_norm": 0.09810366481542587,
-      "learning_rate": 5.899983296338392e-05,
-      "loss": 0.0017,
-      "step": 457
-    },
-    {
-      "epoch": 2.3608247422680413,
-      "grad_norm": 2.2794928550720215,
-      "learning_rate": 5.8831830724429384e-05,
-      "loss": 0.0475,
-      "step": 458
-    },
-    {
-      "epoch": 2.365979381443299,
-      "grad_norm": 0.5080482363700867,
-      "learning_rate": 5.866372550037242e-05,
-      "loss": 0.0049,
-      "step": 459
-    },
-    {
-      "epoch": 2.3711340206185567,
-      "grad_norm": 0.8963924050331116,
-      "learning_rate": 5.849551925143334e-05,
-      "loss": 0.0171,
-      "step": 460
-    },
-    {
-      "epoch": 2.3762886597938144,
-      "grad_norm": 1.555064082145691,
-      "learning_rate": 5.8327213939010414e-05,
-      "loss": 0.0299,
-      "step": 461
-    },
-    {
-      "epoch": 2.381443298969072,
-      "grad_norm": 2.659799337387085,
-      "learning_rate": 5.815881152565712e-05,
-      "loss": 0.0445,
-      "step": 462
-    },
-    {
-      "epoch": 2.38659793814433,
-      "grad_norm": 0.690310001373291,
-      "learning_rate": 5.799031397505913e-05,
-      "loss": 0.0074,
-      "step": 463
-    },
-    {
-      "epoch": 2.3917525773195876,
-      "grad_norm": 0.796109139919281,
-      "learning_rate": 5.782172325201155e-05,
-      "loss": 0.0095,
-      "step": 464
-    },
-    {
-      "epoch": 2.3969072164948453,
-      "grad_norm": 0.16811159253120422,
-      "learning_rate": 5.7653041322395895e-05,
-      "loss": 0.0038,
-      "step": 465
-    },
-    {
-      "epoch": 2.402061855670103,
-      "grad_norm": 1.949715495109558,
-      "learning_rate": 5.748427015315722e-05,
-      "loss": 0.0131,
-      "step": 466
-    },
-    {
-      "epoch": 2.4072164948453607,
-      "grad_norm": 1.220193862915039,
-      "learning_rate": 5.7315411712281186e-05,
-      "loss": 0.0177,
-      "step": 467
-    },
-    {
-      "epoch": 2.4123711340206184,
-      "grad_norm": 2.5171456336975098,
-      "learning_rate": 5.714646796877108e-05,
-      "loss": 0.0518,
-      "step": 468
-    },
-    {
-      "epoch": 2.417525773195876,
-      "grad_norm": 0.662219226360321,
-      "learning_rate": 5.697744089262491e-05,
-      "loss": 0.011,
-      "step": 469
-    },
-    {
-      "epoch": 2.422680412371134,
-      "grad_norm": 0.6332148313522339,
-      "learning_rate": 5.680833245481234e-05,
-      "loss": 0.0096,
-      "step": 470
-    },
-    {
-      "epoch": 2.4278350515463916,
-      "grad_norm": 2.559650182723999,
-      "learning_rate": 5.6639144627251816e-05,
-      "loss": 0.0927,
-      "step": 471
-    },
-    {
-      "epoch": 2.4329896907216497,
-      "grad_norm": 4.00107479095459,
-      "learning_rate": 5.646987938278753e-05,
-      "loss": 0.1422,
-      "step": 472
-    },
-    {
-      "epoch": 2.4381443298969074,
-      "grad_norm": 2.220475673675537,
-      "learning_rate": 5.630053869516635e-05,
-      "loss": 0.0452,
-      "step": 473
-    },
-    {
-      "epoch": 2.443298969072165,
-      "grad_norm": 0.17029526829719543,
-      "learning_rate": 5.6131124539014926e-05,
-      "loss": 0.0029,
-      "step": 474
-    },
-    {
-      "epoch": 2.448453608247423,
-      "grad_norm": 1.107358455657959,
-      "learning_rate": 5.596163888981656e-05,
-      "loss": 0.0108,
-      "step": 475
-    },
-    {
-      "epoch": 2.4536082474226806,
-      "grad_norm": 0.07260440289974213,
-      "learning_rate": 5.5792083723888225e-05,
-      "loss": 0.0019,
-      "step": 476
-    },
-    {
-      "epoch": 2.4587628865979383,
-      "grad_norm": 1.1532561779022217,
-      "learning_rate": 5.5622461018357486e-05,
-      "loss": 0.0205,
-      "step": 477
-    },
-    {
-      "epoch": 2.463917525773196,
-      "grad_norm": 1.5596643686294556,
-      "learning_rate": 5.5452772751139496e-05,
-      "loss": 0.0361,
-      "step": 478
-    },
-    {
-      "epoch": 2.4690721649484537,
-      "grad_norm": 0.033296748995780945,
-      "learning_rate": 5.5283020900913886e-05,
-      "loss": 0.0008,
-      "step": 479
-    },
-    {
-      "epoch": 2.4742268041237114,
-      "grad_norm": 1.283778190612793,
-      "learning_rate": 5.511320744710171e-05,
-      "loss": 0.02,
-      "step": 480
-    },
-    {
-      "epoch": 2.479381443298969,
-      "grad_norm": 1.2095892429351807,
-      "learning_rate": 5.494333436984238e-05,
-      "loss": 0.0115,
-      "step": 481
-    },
-    {
-      "epoch": 2.484536082474227,
-      "grad_norm": 2.116044282913208,
-      "learning_rate": 5.477340364997051e-05,
-      "loss": 0.0457,
-      "step": 482
-    },
-    {
-      "epoch": 2.4896907216494846,
-      "grad_norm": 1.7883367538452148,
-      "learning_rate": 5.460341726899291e-05,
-      "loss": 0.0319,
-      "step": 483
-    },
-    {
-      "epoch": 2.4948453608247423,
-      "grad_norm": 0.627947211265564,
-      "learning_rate": 5.4433377209065414e-05,
-      "loss": 0.0065,
-      "step": 484
-    },
-    {
-      "epoch": 2.5,
-      "grad_norm": 0.03853435814380646,
-      "learning_rate": 5.4263285452969806e-05,
-      "loss": 0.0007,
-      "step": 485
-    },
-    {
-      "epoch": 2.5,
-      "eval_loss": 0.09605833142995834,
-      "eval_runtime": 22.9335,
-      "eval_samples_per_second": 7.151,
-      "eval_steps_per_second": 1.788,
-      "step": 485
-    },
-    {
-      "epoch": 2.5051546391752577,
-      "grad_norm": 0.8205093145370483,
-      "learning_rate": 5.409314398409067e-05,
-      "loss": 0.0129,
-      "step": 486
-    },
-    {
-      "epoch": 2.5103092783505154,
-      "grad_norm": 0.027210909873247147,
-      "learning_rate": 5.392295478639225e-05,
-      "loss": 0.0005,
-      "step": 487
-    },
-    {
-      "epoch": 2.515463917525773,
-      "grad_norm": 0.6312366724014282,
-      "learning_rate": 5.3752719844395405e-05,
-      "loss": 0.0075,
-      "step": 488
-    },
-    {
-      "epoch": 2.520618556701031,
-      "grad_norm": 0.26939788460731506,
-      "learning_rate": 5.358244114315434e-05,
-      "loss": 0.0037,
-      "step": 489
-    },
-    {
-      "epoch": 2.5257731958762886,
-      "grad_norm": 0.18861384689807892,
-      "learning_rate": 5.341212066823355e-05,
-      "loss": 0.0028,
-      "step": 490
-    },
-    {
-      "epoch": 2.5309278350515463,
-      "grad_norm": 0.18018989264965057,
-      "learning_rate": 5.324176040568465e-05,
-      "loss": 0.0023,
-      "step": 491
-    },
-    {
-      "epoch": 2.536082474226804,
-      "grad_norm": 0.2840721011161804,
-      "learning_rate": 5.307136234202318e-05,
-      "loss": 0.0036,
-      "step": 492
-    },
-    {
-      "epoch": 2.5412371134020617,
-      "grad_norm": 0.11515682190656662,
-      "learning_rate": 5.290092846420548e-05,
-      "loss": 0.0021,
-      "step": 493
-    },
-    {
-      "epoch": 2.5463917525773194,
-      "grad_norm": 3.078644037246704,
-      "learning_rate": 5.27304607596055e-05,
-      "loss": 0.0914,
-      "step": 494
-    },
-    {
-      "epoch": 2.551546391752577,
-      "grad_norm": 2.2690036296844482,
-      "learning_rate": 5.255996121599167e-05,
-      "loss": 0.0396,
-      "step": 495
-    },
-    {
-      "epoch": 2.556701030927835,
-      "grad_norm": 2.1380650997161865,
-      "learning_rate": 5.2389431821503606e-05,
-      "loss": 0.0346,
-      "step": 496
-    },
-    {
-      "epoch": 2.5618556701030926,
-      "grad_norm": 0.0314573273062706,
-      "learning_rate": 5.221887456462907e-05,
-      "loss": 0.0005,
-      "step": 497
-    },
-    {
-      "epoch": 2.5670103092783503,
-      "grad_norm": 0.434132844209671,
-      "learning_rate": 5.2048291434180716e-05,
-      "loss": 0.0044,
-      "step": 498
-    },
-    {
-      "epoch": 2.572164948453608,
-      "grad_norm": 0.48740851879119873,
-      "learning_rate": 5.1877684419272875e-05,
-      "loss": 0.0067,
-      "step": 499
-    },
-    {
-      "epoch": 2.5773195876288657,
-      "grad_norm": 0.2123924344778061,
-      "learning_rate": 5.1707055509298396e-05,
-      "loss": 0.0027,
-      "step": 500
-    },
-    {
-      "epoch": 2.582474226804124,
-      "grad_norm": 2.360799789428711,
-      "learning_rate": 5.153640669390546e-05,
-      "loss": 0.0923,
-      "step": 501
-    },
-    {
-      "epoch": 2.5876288659793816,
-      "grad_norm": 2.2972326278686523,
-      "learning_rate": 5.1365739962974304e-05,
-      "loss": 0.0257,
-      "step": 502
-    },
-    {
-      "epoch": 2.5927835051546393,
-      "grad_norm": 0.734491229057312,
-      "learning_rate": 5.119505730659413e-05,
-      "loss": 0.0069,
-      "step": 503
-    },
-    {
-      "epoch": 2.597938144329897,
-      "grad_norm": 0.681890606880188,
-      "learning_rate": 5.102436071503982e-05,
-      "loss": 0.0054,
-      "step": 504
-    },
-    {
-      "epoch": 2.6030927835051547,
-      "grad_norm": 0.29352545738220215,
-      "learning_rate": 5.0853652178748746e-05,
-      "loss": 0.0037,
-      "step": 505
-    },
-    {
-      "epoch": 2.6082474226804124,
-      "grad_norm": 0.1541396975517273,
-      "learning_rate": 5.068293368829755e-05,
-      "loss": 0.0014,
-      "step": 506
-    },
-    {
-      "epoch": 2.61340206185567,
-      "grad_norm": 3.912865161895752,
-      "learning_rate": 5.0512207234379004e-05,
-      "loss": 0.06,
-      "step": 507
-    },
-    {
-      "epoch": 2.618556701030928,
-      "grad_norm": 2.996270179748535,
-      "learning_rate": 5.0341474807778663e-05,
-      "loss": 0.0982,
-      "step": 508
-    },
-    {
-      "epoch": 2.6237113402061856,
-      "grad_norm": 0.02659071795642376,
-      "learning_rate": 5.017073839935178e-05,
-      "loss": 0.0006,
-      "step": 509
-    },
-    {
-      "epoch": 2.6288659793814433,
-      "grad_norm": 0.17042477428913116,
-      "learning_rate": 5e-05,
-      "loss": 0.0018,
-      "step": 510
-    },
-    {
-      "epoch": 2.634020618556701,
-      "grad_norm": 0.02339012175798416,
-      "learning_rate": 4.982926160064823e-05,
-      "loss": 0.0004,
-      "step": 511
-    },
-    {
-      "epoch": 2.6391752577319587,
-      "grad_norm": 2.022866725921631,
-      "learning_rate": 4.965852519222134e-05,
-      "loss": 0.0534,
-      "step": 512
-    },
-    {
-      "epoch": 2.6443298969072164,
-      "grad_norm": 0.9663375020027161,
-      "learning_rate": 4.948779276562101e-05,
-      "loss": 0.0172,
-      "step": 513
-    },
-    {
-      "epoch": 2.649484536082474,
-      "grad_norm": 1.1391932964324951,
-      "learning_rate": 4.9317066311702456e-05,
-      "loss": 0.0105,
-      "step": 514
-    },
-    {
-      "epoch": 2.654639175257732,
-      "grad_norm": 0.11315444111824036,
-      "learning_rate": 4.9146347821251266e-05,
-      "loss": 0.0011,
-      "step": 515
-    },
-    {
-      "epoch": 2.6597938144329896,
-      "grad_norm": 2.212524652481079,
-      "learning_rate": 4.89756392849602e-05,
-      "loss": 0.0243,
-      "step": 516
-    },
-    {
-      "epoch": 2.6649484536082473,
-      "grad_norm": 5.0199294090271,
-      "learning_rate": 4.880494269340588e-05,
-      "loss": 0.1882,
-      "step": 517
-    },
-    {
-      "epoch": 2.670103092783505,
-      "grad_norm": 0.7238196134567261,
-      "learning_rate": 4.863426003702572e-05,
-      "loss": 0.0059,
-      "step": 518
-    },
-    {
-      "epoch": 2.675257731958763,
-      "grad_norm": 3.742496967315674,
-      "learning_rate": 4.8463593306094555e-05,
-      "loss": 0.019,
-      "step": 519
-    },
-    {
-      "epoch": 2.680412371134021,
-      "grad_norm": 0.03769111633300781,
-      "learning_rate": 4.829294449070161e-05,
-      "loss": 0.0008,
-      "step": 520
-    },
-    {
-      "epoch": 2.6855670103092786,
-      "grad_norm": 0.20755574107170105,
-      "learning_rate": 4.8122315580727136e-05,
-      "loss": 0.0023,
-      "step": 521
-    },
-    {
-      "epoch": 2.6907216494845363,
-      "grad_norm": 1.4307230710983276,
-      "learning_rate": 4.795170856581929e-05,
-      "loss": 0.0152,
-      "step": 522
-    },
-    {
-      "epoch": 2.695876288659794,
-      "grad_norm": 0.22543154656887054,
-      "learning_rate": 4.778112543537094e-05,
-      "loss": 0.0028,
-      "step": 523
-    },
-    {
-      "epoch": 2.7010309278350517,
-      "grad_norm": 0.31340834498405457,
-      "learning_rate": 4.7610568178496405e-05,
-      "loss": 0.0036,
-      "step": 524
-    },
-    {
-      "epoch": 2.7061855670103094,
-      "grad_norm": 2.5833778381347656,
-      "learning_rate": 4.744003878400835e-05,
-      "loss": 0.0361,
-      "step": 525
-    },
-    {
-      "epoch": 2.711340206185567,
-      "grad_norm": 3.0378775596618652,
-      "learning_rate": 4.726953924039451e-05,
-      "loss": 0.0751,
-      "step": 526
-    },
-    {
-      "epoch": 2.716494845360825,
-      "grad_norm": 0.24335841834545135,
-      "learning_rate": 4.709907153579454e-05,
-      "loss": 0.002,
-      "step": 527
-    },
-    {
-      "epoch": 2.7216494845360826,
-      "grad_norm": 0.7776563763618469,
-      "learning_rate": 4.692863765797683e-05,
-      "loss": 0.0082,
-      "step": 528
-    },
-    {
-      "epoch": 2.7268041237113403,
-      "grad_norm": 0.052026282995939255,
-      "learning_rate": 4.675823959431535e-05,
-      "loss": 0.0009,
-      "step": 529
-    },
-    {
-      "epoch": 2.731958762886598,
-      "grad_norm": 2.127627372741699,
-      "learning_rate": 4.658787933176646e-05,
-      "loss": 0.0297,
-      "step": 530
-    },
-    {
-      "epoch": 2.7371134020618557,
-      "grad_norm": 0.5500102043151855,
-      "learning_rate": 4.641755885684566e-05,
-      "loss": 0.0051,
-      "step": 531
-    },
-    {
-      "epoch": 2.7422680412371134,
-      "grad_norm": 0.05251020938158035,
-      "learning_rate": 4.624728015560461e-05,
-      "loss": 0.0013,
-      "step": 532
-    },
-    {
-      "epoch": 2.747422680412371,
-      "grad_norm": 0.04040883108973503,
-      "learning_rate": 4.607704521360776e-05,
-      "loss": 0.0008,
-      "step": 533
-    },
-    {
-      "epoch": 2.752577319587629,
-      "grad_norm": 0.2735699415206909,
-      "learning_rate": 4.590685601590936e-05,
-      "loss": 0.003,
-      "step": 534
-    },
-    {
-      "epoch": 2.7577319587628866,
-      "grad_norm": 0.2323101907968521,
-      "learning_rate": 4.57367145470302e-05,
-      "loss": 0.003,
-      "step": 535
-    },
-    {
-      "epoch": 2.7628865979381443,
-      "grad_norm": 0.025867881253361702,
-      "learning_rate": 4.5566622790934604e-05,
-      "loss": 0.0006,
-      "step": 536
-    },
-    {
-      "epoch": 2.768041237113402,
-      "grad_norm": 2.5169122219085693,
-      "learning_rate": 4.5396582731007095e-05,
-      "loss": 0.0334,
-      "step": 537
-    },
-    {
-      "epoch": 2.7731958762886597,
-      "grad_norm": 0.39224404096603394,
-      "learning_rate": 4.52265963500295e-05,
-      "loss": 0.0049,
-      "step": 538
-    },
-    {
-      "epoch": 2.7783505154639174,
-      "grad_norm": 0.15631736814975739,
-      "learning_rate": 4.505666563015763e-05,
-      "loss": 0.0025,
-      "step": 539
-    },
-    {
-      "epoch": 2.783505154639175,
-      "grad_norm": 1.025899052619934,
-      "learning_rate": 4.4886792552898286e-05,
-      "loss": 0.0211,
-      "step": 540
-    },
-    {
-      "epoch": 2.788659793814433,
-      "grad_norm": 0.10308719426393509,
-      "learning_rate": 4.471697909908613e-05,
-      "loss": 0.0013,
-      "step": 541
-    },
-    {
-      "epoch": 2.7938144329896906,
-      "grad_norm": 0.9129118919372559,
-      "learning_rate": 4.454722724886051e-05,
-      "loss": 0.0106,
-      "step": 542
-    },
-    {
-      "epoch": 2.7989690721649483,
-      "grad_norm": 0.37369224429130554,
-      "learning_rate": 4.437753898164254e-05,
-      "loss": 0.0043,
-      "step": 543
-    },
-    {
-      "epoch": 2.804123711340206,
-      "grad_norm": 0.12186601012945175,
-      "learning_rate": 4.420791627611179e-05,
-      "loss": 0.0017,
-      "step": 544
-    },
-    {
-      "epoch": 2.8092783505154637,
-      "grad_norm": 0.03575112298130989,
-      "learning_rate": 4.403836111018346e-05,
-      "loss": 0.0005,
-      "step": 545
-    },
-    {
-      "epoch": 2.8144329896907214,
-      "grad_norm": 0.0675966814160347,
-      "learning_rate": 4.3868875460985085e-05,
-      "loss": 0.0008,
-      "step": 546
-    },
-    {
-      "epoch": 2.819587628865979,
-      "grad_norm": 0.6237895488739014,
-      "learning_rate": 4.369946130483364e-05,
-      "loss": 0.0081,
-      "step": 547
-    },
-    {
-      "epoch": 2.824742268041237,
-      "grad_norm": 0.6258450746536255,
-      "learning_rate": 4.353012061721249e-05,
-      "loss": 0.009,
-      "step": 548
-    },
-    {
-      "epoch": 2.829896907216495,
-      "grad_norm": 1.2090260982513428,
-      "learning_rate": 4.336085537274818e-05,
-      "loss": 0.0127,
-      "step": 549
-    },
-    {
-      "epoch": 2.8350515463917527,
-      "grad_norm": 0.41710153222084045,
-      "learning_rate": 4.319166754518768e-05,
-      "loss": 0.0047,
-      "step": 550
-    },
-    {
-      "epoch": 2.8402061855670104,
-      "grad_norm": 0.38828399777412415,
-      "learning_rate": 4.3022559107375106e-05,
-      "loss": 0.0037,
-      "step": 551
-    },
-    {
-      "epoch": 2.845360824742268,
-      "grad_norm": 0.008606897667050362,
-      "learning_rate": 4.285353203122893e-05,
-      "loss": 0.0003,
-      "step": 552
-    },
-    {
-      "epoch": 2.850515463917526,
-      "grad_norm": 0.21119487285614014,
-      "learning_rate": 4.268458828771883e-05,
-      "loss": 0.002,
-      "step": 553
-    },
-    {
-      "epoch": 2.8556701030927836,
-      "grad_norm": 0.38432881236076355,
-      "learning_rate": 4.251572984684281e-05,
-      "loss": 0.0056,
-      "step": 554
-    },
-    {
-      "epoch": 2.8608247422680413,
-      "grad_norm": 1.544846773147583,
-      "learning_rate": 4.234695867760412e-05,
-      "loss": 0.0234,
-      "step": 555
-    },
-    {
-      "epoch": 2.865979381443299,
-      "grad_norm": 0.02039588801562786,
-      "learning_rate": 4.2178276747988446e-05,
-      "loss": 0.0005,
-      "step": 556
-    },
-    {
-      "epoch": 2.8711340206185567,
-      "grad_norm": 0.2800588309764862,
-      "learning_rate": 4.200968602494087e-05,
-      "loss": 0.0035,
-      "step": 557
-    },
-    {
-      "epoch": 2.8762886597938144,
-      "grad_norm": 2.2519681453704834,
-      "learning_rate": 4.18411884743429e-05,
-      "loss": 0.0577,
-      "step": 558
-    },
-    {
-      "epoch": 2.881443298969072,
-      "grad_norm": 0.7249019742012024,
-      "learning_rate": 4.16727860609896e-05,
-      "loss": 0.0064,
-      "step": 559
-    },
-    {
-      "epoch": 2.88659793814433,
-      "grad_norm": 0.8100419640541077,
-      "learning_rate": 4.150448074856667e-05,
-      "loss": 0.0066,
-      "step": 560
-    },
-    {
-      "epoch": 2.8917525773195876,
-      "grad_norm": 1.3413070440292358,
-      "learning_rate": 4.1336274499627596e-05,
-      "loss": 0.0219,
-      "step": 561
-    },
-    {
-      "epoch": 2.8969072164948453,
-      "grad_norm": 3.6059770584106445,
-      "learning_rate": 4.1168169275570635e-05,
-      "loss": 0.0575,
-      "step": 562
-    },
-    {
-      "epoch": 2.902061855670103,
-      "grad_norm": 4.4368181228637695,
-      "learning_rate": 4.1000167036616113e-05,
-      "loss": 0.1659,
-      "step": 563
-    },
-    {
-      "epoch": 2.9072164948453607,
-      "grad_norm": 0.06619425863027573,
-      "learning_rate": 4.083226974178339e-05,
-      "loss": 0.0012,
-      "step": 564
-    },
-    {
-      "epoch": 2.9123711340206184,
-      "grad_norm": 0.5384992957115173,
-      "learning_rate": 4.066447934886819e-05,
-      "loss": 0.007,
-      "step": 565
-    },
-    {
-      "epoch": 2.917525773195876,
-      "grad_norm": 5.088779926300049,
-      "learning_rate": 4.049679781441965e-05,
-      "loss": 0.1876,
-      "step": 566
-    },
-    {
-      "epoch": 2.9226804123711343,
-      "grad_norm": 2.96576189994812,
-      "learning_rate": 4.0329227093717515e-05,
-      "loss": 0.0468,
-      "step": 567
-    },
-    {
-      "epoch": 2.927835051546392,
-      "grad_norm": 0.0372987799346447,
-      "learning_rate": 4.016176914074944e-05,
-      "loss": 0.0007,
-      "step": 568
-    },
-    {
-      "epoch": 2.9329896907216497,
-      "grad_norm": 1.4958171844482422,
-      "learning_rate": 3.999442590818804e-05,
-      "loss": 0.037,
-      "step": 569
-    },
-    {
-      "epoch": 2.9381443298969074,
-      "grad_norm": 0.24370728433132172,
-      "learning_rate": 3.982719934736832e-05,
-      "loss": 0.0029,
-      "step": 570
-    },
-    {
-      "epoch": 2.943298969072165,
-      "grad_norm": 1.0667810440063477,
-      "learning_rate": 3.9660091408264716e-05,
-      "loss": 0.0227,
-      "step": 571
-    },
-    {
-      "epoch": 2.948453608247423,
-      "grad_norm": 0.06778667867183685,
-      "learning_rate": 3.949310403946849e-05,
-      "loss": 0.0012,
-      "step": 572
-    },
-    {
-      "epoch": 2.9536082474226806,
-      "grad_norm": 1.0466609001159668,
-      "learning_rate": 3.9326239188165025e-05,
-      "loss": 0.0082,
-      "step": 573
-    },
-    {
-      "epoch": 2.9587628865979383,
-      "grad_norm": 0.33205440640449524,
-      "learning_rate": 3.915949880011096e-05,
-      "loss": 0.0037,
-      "step": 574
-    },
-    {
-      "epoch": 2.963917525773196,
-      "grad_norm": 0.016779989004135132,
-      "learning_rate": 3.899288481961173e-05,
-      "loss": 0.0004,
-      "step": 575
-    },
-    {
-      "epoch": 2.9690721649484537,
-      "grad_norm": 0.9853585958480835,
-      "learning_rate": 3.8826399189498654e-05,
-      "loss": 0.0098,
-      "step": 576
-    },
-    {
-      "epoch": 2.9742268041237114,
-      "grad_norm": 0.5929998159408569,
-      "learning_rate": 3.86600438511065e-05,
-      "loss": 0.0043,
-      "step": 577
-    },
-    {
-      "epoch": 2.979381443298969,
-      "grad_norm": 3.938533306121826,
-      "learning_rate": 3.8493820744250685e-05,
-      "loss": 0.131,
-      "step": 578
-    },
-    {
-      "epoch": 2.984536082474227,
-      "grad_norm": 2.7528624534606934,
-      "learning_rate": 3.832773180720475e-05,
-      "loss": 0.0992,
-      "step": 579
-    },
-    {
-      "epoch": 2.9896907216494846,
-      "grad_norm": 1.0223017930984497,
-      "learning_rate": 3.8161778976677666e-05,
-      "loss": 0.0069,
-      "step": 580
-    },
-    {
-      "epoch": 2.9948453608247423,
-      "grad_norm": 0.7486586570739746,
-      "learning_rate": 3.79959641877914e-05,
-      "loss": 0.0071,
-      "step": 581
-    },
-    {
-      "epoch": 3.0,
-      "grad_norm": 0.2129761278629303,
-      "learning_rate": 3.783028937405821e-05,
-      "loss": 0.0026,
-      "step": 582
-    },
-    {
-      "epoch": 3.0,
-      "eval_loss": 0.1013411208987236,
-      "eval_runtime": 22.9543,
-      "eval_samples_per_second": 7.145,
-      "eval_steps_per_second": 1.786,
-      "step": 582
-    },
-    {
-      "epoch": 3.0051546391752577,
-      "grad_norm": 0.47546765208244324,
-      "learning_rate": 3.766475646735815e-05,
-      "loss": 0.0029,
-      "step": 583
-    },
-    {
-      "epoch": 3.0103092783505154,
-      "grad_norm": 0.024188758805394173,
-      "learning_rate": 3.7499367397916555e-05,
-      "loss": 0.0006,
-      "step": 584
-    },
-    {
-      "epoch": 3.015463917525773,
-      "grad_norm": 0.03356657549738884,
-      "learning_rate": 3.733412409428148e-05,
-      "loss": 0.0006,
-      "step": 585
-    },
-    {
-      "epoch": 3.020618556701031,
-      "grad_norm": 0.19245874881744385,
-      "learning_rate": 3.716902848330133e-05,
-      "loss": 0.0025,
-      "step": 586
-    },
-    {
-      "epoch": 3.0257731958762886,
-      "grad_norm": 0.992292046546936,
-      "learning_rate": 3.7004082490102226e-05,
-      "loss": 0.0109,
-      "step": 587
-    },
-    {
-      "epoch": 3.0309278350515463,
-      "grad_norm": 0.0995234027504921,
-      "learning_rate": 3.6839288038065734e-05,
-      "loss": 0.0012,
-      "step": 588
-    },
-    {
-      "epoch": 3.036082474226804,
-      "grad_norm": 0.07061074674129486,
-      "learning_rate": 3.667464704880625e-05,
-      "loss": 0.001,
-      "step": 589
-    },
-    {
-      "epoch": 3.0412371134020617,
-      "grad_norm": 0.23088815808296204,
-      "learning_rate": 3.651016144214878e-05,
-      "loss": 0.003,
-      "step": 590
-    },
-    {
-      "epoch": 3.0463917525773194,
-      "grad_norm": 0.18078461289405823,
-      "learning_rate": 3.634583313610644e-05,
-      "loss": 0.0015,
-      "step": 591
-    },
-    {
-      "epoch": 3.051546391752577,
-      "grad_norm": 0.22349312901496887,
-      "learning_rate": 3.618166404685805e-05,
-      "loss": 0.0019,
-      "step": 592
-    },
-    {
-      "epoch": 3.056701030927835,
-      "grad_norm": 0.5327549576759338,
-      "learning_rate": 3.601765608872595e-05,
-      "loss": 0.0048,
-      "step": 593
-    },
-    {
-      "epoch": 3.0618556701030926,
-      "grad_norm": 0.10766857862472534,
-      "learning_rate": 3.585381117415349e-05,
-      "loss": 0.0015,
-      "step": 594
-    },
-    {
-      "epoch": 3.0670103092783507,
-      "grad_norm": 0.08378089964389801,
-      "learning_rate": 3.5690131213682944e-05,
-      "loss": 0.0014,
-      "step": 595
-    },
-    {
-      "epoch": 3.0721649484536084,
-      "grad_norm": 0.12249798327684402,
-      "learning_rate": 3.5526618115932975e-05,
-      "loss": 0.0014,
-      "step": 596
-    },
-    {
-      "epoch": 3.077319587628866,
-      "grad_norm": 0.09342166781425476,
-      "learning_rate": 3.53632737875766e-05,
-      "loss": 0.001,
-      "step": 597
-    },
-    {
-      "epoch": 3.082474226804124,
-      "grad_norm": 0.030285261571407318,
-      "learning_rate": 3.5200100133318834e-05,
-      "loss": 0.0005,
-      "step": 598
-    },
-    {
-      "epoch": 3.0876288659793816,
-      "grad_norm": 0.6078401803970337,
-      "learning_rate": 3.5037099055874536e-05,
-      "loss": 0.0053,
-      "step": 599
-    },
-    {
-      "epoch": 3.0927835051546393,
-      "grad_norm": 0.031891413033008575,
-      "learning_rate": 3.487427245594622e-05,
-      "loss": 0.0006,
-      "step": 600
-    },
-    {
-      "epoch": 3.097938144329897,
-      "grad_norm": 0.22419054806232452,
-      "learning_rate": 3.47116222322018e-05,
-      "loss": 0.0026,
-      "step": 601
-    },
-    {
-      "epoch": 3.1030927835051547,
-      "grad_norm": 3.222254753112793,
-      "learning_rate": 3.4549150281252636e-05,
-      "loss": 0.0673,
-      "step": 602
-    },
-    {
-      "epoch": 3.1082474226804124,
-      "grad_norm": 0.07035762071609497,
-      "learning_rate": 3.4386858497631205e-05,
-      "loss": 0.0012,
-      "step": 603
-    },
-    {
-      "epoch": 3.11340206185567,
-      "grad_norm": 0.0403384231030941,
-      "learning_rate": 3.422474877376917e-05,
-      "loss": 0.0006,
-      "step": 604
-    },
-    {
-      "epoch": 3.118556701030928,
-      "grad_norm": 0.8612932562828064,
-      "learning_rate": 3.406282299997521e-05,
-      "loss": 0.0059,
-      "step": 605
-    },
-    {
-      "epoch": 3.1237113402061856,
-      "grad_norm": 0.04132150113582611,
-      "learning_rate": 3.3901083064413095e-05,
-      "loss": 0.0005,
-      "step": 606
-    },
-    {
-      "epoch": 3.1288659793814433,
-      "grad_norm": 0.06787485629320145,
-      "learning_rate": 3.3739530853079516e-05,
-      "loss": 0.0009,
-      "step": 607
-    },
-    {
-      "epoch": 3.134020618556701,
-      "grad_norm": 0.1399659663438797,
-      "learning_rate": 3.357816824978222e-05,
-      "loss": 0.0009,
-      "step": 608
-    },
-    {
-      "epoch": 3.1391752577319587,
-      "grad_norm": 0.4071262776851654,
-      "learning_rate": 3.341699713611799e-05,
-      "loss": 0.003,
-      "step": 609
-    },
-    {
-      "epoch": 3.1443298969072164,
-      "grad_norm": 0.0502166785299778,
-      "learning_rate": 3.325601939145069e-05,
-      "loss": 0.0005,
-      "step": 610
-    },
-    {
-      "epoch": 3.149484536082474,
-      "grad_norm": 0.044071536511182785,
-      "learning_rate": 3.309523689288941e-05,
-      "loss": 0.0008,
-      "step": 611
-    },
-    {
-      "epoch": 3.154639175257732,
-      "grad_norm": 0.00721757160499692,
-      "learning_rate": 3.293465151526649e-05,
-      "loss": 0.0002,
-      "step": 612
-    },
-    {
-      "epoch": 3.1597938144329896,
-      "grad_norm": 0.02341277524828911,
-      "learning_rate": 3.277426513111575e-05,
-      "loss": 0.0005,
-      "step": 613
-    },
-    {
-      "epoch": 3.1649484536082473,
-      "grad_norm": 0.06053034961223602,
-      "learning_rate": 3.261407961065056e-05,
-      "loss": 0.001,
-      "step": 614
-    },
-    {
-      "epoch": 3.170103092783505,
-      "grad_norm": 0.27043405175209045,
-      "learning_rate": 3.245409682174217e-05,
-      "loss": 0.0031,
-      "step": 615
-    },
-    {
-      "epoch": 3.1752577319587627,
-      "grad_norm": 0.03400631621479988,
-      "learning_rate": 3.229431862989775e-05,
-      "loss": 0.0006,
-      "step": 616
-    },
-    {
-      "epoch": 3.1804123711340204,
-      "grad_norm": 1.3103770017623901,
-      "learning_rate": 3.2134746898238774e-05,
-      "loss": 0.0111,
-      "step": 617
-    },
-    {
-      "epoch": 3.1855670103092786,
-      "grad_norm": 0.20999719202518463,
-      "learning_rate": 3.197538348747927e-05,
-      "loss": 0.003,
-      "step": 618
-    },
-    {
-      "epoch": 3.1907216494845363,
-      "grad_norm": 0.04834004119038582,
-      "learning_rate": 3.181623025590405e-05,
-      "loss": 0.0005,
-      "step": 619
-    },
-    {
-      "epoch": 3.195876288659794,
-      "grad_norm": 0.060436930507421494,
-      "learning_rate": 3.165728905934718e-05,
-      "loss": 0.0007,
-      "step": 620
-    },
-    {
-      "epoch": 3.2010309278350517,
-      "grad_norm": 0.12254571914672852,
-      "learning_rate": 3.149856175117014e-05,
-      "loss": 0.0014,
-      "step": 621
-    },
-    {
-      "epoch": 3.2061855670103094,
-      "grad_norm": 0.17747434973716736,
-      "learning_rate": 3.134005018224044e-05,
-      "loss": 0.0014,
-      "step": 622
-    },
-    {
-      "epoch": 3.211340206185567,
-      "grad_norm": 0.12086270749568939,
-      "learning_rate": 3.118175620090983e-05,
-      "loss": 0.0013,
-      "step": 623
-    },
-    {
-      "epoch": 3.216494845360825,
-      "grad_norm": 0.01552607398480177,
-      "learning_rate": 3.1023681652992926e-05,
-      "loss": 0.0003,
-      "step": 624
-    },
-    {
-      "epoch": 3.2216494845360826,
-      "grad_norm": 0.04116373881697655,
-      "learning_rate": 3.086582838174551e-05,
-      "loss": 0.0007,
-      "step": 625
-    },
-    {
-      "epoch": 3.2268041237113403,
-      "grad_norm": 0.018168117851018906,
-      "learning_rate": 3.070819822784323e-05,
-      "loss": 0.0003,
-      "step": 626
-    },
-    {
-      "epoch": 3.231958762886598,
-      "grad_norm": 0.07606551796197891,
-      "learning_rate": 3.055079302935997e-05,
-      "loss": 0.0009,
-      "step": 627
-    },
-    {
-      "epoch": 3.2371134020618557,
-      "grad_norm": 0.0784323439002037,
-      "learning_rate": 3.0393614621746498e-05,
-      "loss": 0.0007,
-      "step": 628
-    },
-    {
-      "epoch": 3.2422680412371134,
-      "grad_norm": 0.07449536770582199,
-      "learning_rate": 3.023666483780905e-05,
-      "loss": 0.0009,
-      "step": 629
-    },
-    {
-      "epoch": 3.247422680412371,
-      "grad_norm": 0.049209631979465485,
-      "learning_rate": 3.007994550768793e-05,
-      "loss": 0.0008,
-      "step": 630
-    },
-    {
-      "epoch": 3.252577319587629,
-      "grad_norm": 0.006166890263557434,
-      "learning_rate": 2.9923458458836258e-05,
-      "loss": 0.0002,
-      "step": 631
-    },
-    {
-      "epoch": 3.2577319587628866,
-      "grad_norm": 0.2379947453737259,
-      "learning_rate": 2.9767205515998518e-05,
-      "loss": 0.0016,
-      "step": 632
-    },
-    {
-      "epoch": 3.2628865979381443,
-      "grad_norm": 0.00971717108041048,
-      "learning_rate": 2.9611188501189435e-05,
-      "loss": 0.0002,
-      "step": 633
-    },
-    {
-      "epoch": 3.268041237113402,
-      "grad_norm": 0.003669175785034895,
-      "learning_rate": 2.9455409233672592e-05,
-      "loss": 0.0001,
-      "step": 634
-    },
-    {
-      "epoch": 3.2731958762886597,
-      "grad_norm": 0.006943593733012676,
-      "learning_rate": 2.929986952993933e-05,
-      "loss": 0.0002,
-      "step": 635
-    },
-    {
-      "epoch": 3.2783505154639174,
-      "grad_norm": 0.030062507838010788,
-      "learning_rate": 2.9144571203687476e-05,
-      "loss": 0.0005,
-      "step": 636
-    },
-    {
-      "epoch": 3.283505154639175,
-      "grad_norm": 0.009839212521910667,
-      "learning_rate": 2.8989516065800238e-05,
-      "loss": 0.0002,
-      "step": 637
-    },
-    {
-      "epoch": 3.288659793814433,
-      "grad_norm": 0.0119356419891119,
-      "learning_rate": 2.8834705924325118e-05,
-      "loss": 0.0002,
-      "step": 638
-    },
-    {
-      "epoch": 3.2938144329896906,
-      "grad_norm": 0.4045789837837219,
-      "learning_rate": 2.8680142584452742e-05,
-      "loss": 0.0032,
-      "step": 639
-    },
-    {
-      "epoch": 3.2989690721649483,
-      "grad_norm": 0.12698271870613098,
-      "learning_rate": 2.8525827848495913e-05,
-      "loss": 0.0016,
-      "step": 640
-    },
-    {
-      "epoch": 3.304123711340206,
-      "grad_norm": 0.45483699440956116,
-      "learning_rate": 2.83717635158685e-05,
-      "loss": 0.0057,
-      "step": 641
-    },
-    {
-      "epoch": 3.3092783505154637,
-      "grad_norm": 0.061637796461582184,
-      "learning_rate": 2.8217951383064544e-05,
-      "loss": 0.0012,
-      "step": 642
-    },
-    {
-      "epoch": 3.3144329896907214,
-      "grad_norm": 0.18873506784439087,
-      "learning_rate": 2.8064393243637222e-05,
-      "loss": 0.0018,
-      "step": 643
-    },
-    {
-      "epoch": 3.319587628865979,
-      "grad_norm": 0.013112235814332962,
-      "learning_rate": 2.791109088817803e-05,
-      "loss": 0.0003,
-      "step": 644
-    },
-    {
-      "epoch": 3.3247422680412373,
-      "grad_norm": 0.006930911913514137,
-      "learning_rate": 2.7758046104295797e-05,
-      "loss": 0.0002,
-      "step": 645
-    },
-    {
-      "epoch": 3.329896907216495,
-      "grad_norm": 0.017246872186660767,
-      "learning_rate": 2.760526067659591e-05,
-      "loss": 0.0003,
-      "step": 646
-    },
-    {
-      "epoch": 3.3350515463917527,
-      "grad_norm": 0.5544341206550598,
-      "learning_rate": 2.7452736386659516e-05,
-      "loss": 0.0042,
-      "step": 647
-    },
-    {
-      "epoch": 3.3402061855670104,
-      "grad_norm": 0.01653260365128517,
-      "learning_rate": 2.7300475013022663e-05,
-      "loss": 0.0003,
-      "step": 648
-    },
-    {
-      "epoch": 3.345360824742268,
-      "grad_norm": 0.011309400200843811,
-      "learning_rate": 2.7148478331155702e-05,
-      "loss": 0.0003,
-      "step": 649
-    },
-    {
-      "epoch": 3.350515463917526,
-      "grad_norm": 1.8607807159423828,
-      "learning_rate": 2.6996748113442394e-05,
-      "loss": 0.0126,
-      "step": 650
-    },
-    {
-      "epoch": 3.3556701030927836,
-      "grad_norm": 4.186022758483887,
-      "learning_rate": 2.6845286129159464e-05,
-      "loss": 0.0301,
-      "step": 651
-    },
-    {
-      "epoch": 3.3608247422680413,
-      "grad_norm": 1.2969677448272705,
-      "learning_rate": 2.669409414445574e-05,
-      "loss": 0.0136,
-      "step": 652
-    },
-    {
-      "epoch": 3.365979381443299,
-      "grad_norm": 0.030797425657510757,
-      "learning_rate": 2.6543173922331743e-05,
-      "loss": 0.0005,
-      "step": 653
-    },
-    {
-      "epoch": 3.3711340206185567,
-      "grad_norm": 0.16669800877571106,
-      "learning_rate": 2.639252722261908e-05,
-      "loss": 0.0015,
-      "step": 654
-    },
-    {
-      "epoch": 3.3762886597938144,
-      "grad_norm": 0.11004102230072021,
-      "learning_rate": 2.624215580195981e-05,
-      "loss": 0.001,
-      "step": 655
-    },
-    {
-      "epoch": 3.381443298969072,
-      "grad_norm": 1.4919167757034302,
-      "learning_rate": 2.6092061413786156e-05,
-      "loss": 0.0151,
-      "step": 656
-    },
-    {
-      "epoch": 3.38659793814433,
-      "grad_norm": 0.12221274524927139,
-      "learning_rate": 2.5942245808299886e-05,
-      "loss": 0.0011,
-      "step": 657
-    },
-    {
-      "epoch": 3.3917525773195876,
-      "grad_norm": 0.40452706813812256,
-      "learning_rate": 2.5792710732451997e-05,
-      "loss": 0.0028,
-      "step": 658
-    },
-    {
-      "epoch": 3.3969072164948453,
-      "grad_norm": 1.5829023122787476,
-      "learning_rate": 2.56434579299223e-05,
-      "loss": 0.0206,
-      "step": 659
-    },
-    {
-      "epoch": 3.402061855670103,
-      "grad_norm": 0.26589515805244446,
-      "learning_rate": 2.5494489141099153e-05,
-      "loss": 0.0018,
-      "step": 660
-    },
-    {
-      "epoch": 3.4072164948453607,
-      "grad_norm": 0.6049233078956604,
-      "learning_rate": 2.534580610305909e-05,
-      "loss": 0.0049,
-      "step": 661
-    },
-    {
-      "epoch": 3.4123711340206184,
-      "grad_norm": 0.012360837310552597,
-      "learning_rate": 2.5197410549546595e-05,
-      "loss": 0.0002,
-      "step": 662
-    },
-    {
-      "epoch": 3.417525773195876,
-      "grad_norm": 0.03558109700679779,
-      "learning_rate": 2.5049304210953933e-05,
-      "loss": 0.0004,
-      "step": 663
-    },
-    {
-      "epoch": 3.422680412371134,
-      "grad_norm": 0.03851904720067978,
-      "learning_rate": 2.4901488814300856e-05,
-      "loss": 0.0004,
-      "step": 664
-    },
-    {
-      "epoch": 3.4278350515463916,
-      "grad_norm": 0.41700494289398193,
-      "learning_rate": 2.4753966083214615e-05,
-      "loss": 0.0043,
-      "step": 665
-    },
-    {
-      "epoch": 3.4329896907216497,
-      "grad_norm": 0.0076299929060041904,
-      "learning_rate": 2.4606737737909697e-05,
-      "loss": 0.0002,
-      "step": 666
-    },
-    {
-      "epoch": 3.4381443298969074,
-      "grad_norm": 0.029439013451337814,
-      "learning_rate": 2.4459805495167942e-05,
-      "loss": 0.0005,
-      "step": 667
-    },
-    {
-      "epoch": 3.443298969072165,
-      "grad_norm": 0.20996803045272827,
-      "learning_rate": 2.4313171068318357e-05,
-      "loss": 0.0023,
-      "step": 668
-    },
-    {
-      "epoch": 3.448453608247423,
-      "grad_norm": 0.11522570997476578,
-      "learning_rate": 2.4166836167217283e-05,
-      "loss": 0.0013,
-      "step": 669
-    },
-    {
-      "epoch": 3.4536082474226806,
-      "grad_norm": 0.041188135743141174,
-      "learning_rate": 2.4020802498228335e-05,
-      "loss": 0.0004,
-      "step": 670
-    },
-    {
-      "epoch": 3.4587628865979383,
-      "grad_norm": 0.20725612342357635,
-      "learning_rate": 2.3875071764202563e-05,
-      "loss": 0.0017,
-      "step": 671
-    },
-    {
-      "epoch": 3.463917525773196,
-      "grad_norm": 0.8663135170936584,
-      "learning_rate": 2.3729645664458638e-05,
-      "loss": 0.0066,
-      "step": 672
-    },
-    {
-      "epoch": 3.4690721649484537,
-      "grad_norm": 0.36084750294685364,
-      "learning_rate": 2.3584525894762928e-05,
-      "loss": 0.002,
-      "step": 673
-    },
-    {
-      "epoch": 3.4742268041237114,
-      "grad_norm": 0.03471383452415466,
-      "learning_rate": 2.3439714147309845e-05,
-      "loss": 0.0005,
-      "step": 674
-    },
-    {
-      "epoch": 3.479381443298969,
-      "grad_norm": 0.008910657837986946,
-      "learning_rate": 2.329521211070199e-05,
-      "loss": 0.0002,
-      "step": 675
-    },
-    {
-      "epoch": 3.484536082474227,
-      "grad_norm": 0.06672267615795135,
-      "learning_rate": 2.3151021469930613e-05,
-      "loss": 0.0009,
-      "step": 676
-    },
-    {
-      "epoch": 3.4896907216494846,
-      "grad_norm": 0.05928796902298927,
-      "learning_rate": 2.3007143906355767e-05,
-      "loss": 0.0005,
-      "step": 677
-    },
-    {
-      "epoch": 3.4948453608247423,
-      "grad_norm": 0.22010110318660736,
-      "learning_rate": 2.2863581097686925e-05,
-      "loss": 0.0024,
-      "step": 678
-    },
-    {
-      "epoch": 3.5,
-      "grad_norm": 0.012560139410197735,
-      "learning_rate": 2.2720334717963222e-05,
-      "loss": 0.0002,
-      "step": 679
-    },
-    {
-      "epoch": 3.5,
-      "eval_loss": 0.10579764097929001,
-      "eval_runtime": 22.9584,
-      "eval_samples_per_second": 7.143,
-      "eval_steps_per_second": 1.786,
-      "step": 679
-    },
-    {
-      "epoch": 3.5051546391752577,
-      "grad_norm": 1.5844885110855103,
-      "learning_rate": 2.2577406437534054e-05,
-      "loss": 0.0165,
-      "step": 680
-    },
-    {
-      "epoch": 3.5103092783505154,
-      "grad_norm": 0.12943275272846222,
-      "learning_rate": 2.2434797923039598e-05,
-      "loss": 0.0012,
-      "step": 681
-    },
-    {
-      "epoch": 3.515463917525773,
-      "grad_norm": 0.41947558522224426,
-      "learning_rate": 2.2292510837391267e-05,
-      "loss": 0.0028,
-      "step": 682
-    },
-    {
-      "epoch": 3.520618556701031,
-      "grad_norm": 0.279430091381073,
-      "learning_rate": 2.2150546839752438e-05,
-      "loss": 0.0027,
-      "step": 683
-    },
-    {
-      "epoch": 3.5257731958762886,
-      "grad_norm": 0.07363631576299667,
-      "learning_rate": 2.2008907585519095e-05,
-      "loss": 0.0007,
-      "step": 684
-    },
-    {
-      "epoch": 3.5309278350515463,
-      "grad_norm": 0.28761395812034607,
-      "learning_rate": 2.186759472630045e-05,
-      "loss": 0.0024,
-      "step": 685
-    },
-    {
-      "epoch": 3.536082474226804,
-      "grad_norm": 0.013523326255381107,
-      "learning_rate": 2.172660990989971e-05,
-      "loss": 0.0003,
-      "step": 686
-    },
-    {
-      "epoch": 3.5412371134020617,
-      "grad_norm": 0.02186383679509163,
-      "learning_rate": 2.1585954780294947e-05,
-      "loss": 0.0004,
-      "step": 687
-    },
-    {
-      "epoch": 3.5463917525773194,
-      "grad_norm": 0.12467202544212341,
-      "learning_rate": 2.144563097761984e-05,
-      "loss": 0.0008,
-      "step": 688
-    },
-    {
-      "epoch": 3.551546391752577,
-      "grad_norm": 0.0455549992620945,
-      "learning_rate": 2.130564013814453e-05,
-      "loss": 0.0006,
-      "step": 689
-    },
-    {
-      "epoch": 3.556701030927835,
-      "grad_norm": 0.06408347189426422,
-      "learning_rate": 2.1165983894256647e-05,
-      "loss": 0.0009,
-      "step": 690
-    },
-    {
-      "epoch": 3.5618556701030926,
-      "grad_norm": 0.027774274349212646,
-      "learning_rate": 2.102666387444215e-05,
-      "loss": 0.0004,
-      "step": 691
-    },
-    {
-      "epoch": 3.5670103092783503,
-      "grad_norm": 1.0454899072647095,
-      "learning_rate": 2.0887681703266453e-05,
-      "loss": 0.007,
-      "step": 692
-    },
-    {
-      "epoch": 3.572164948453608,
-      "grad_norm": 0.023052293807268143,
-      "learning_rate": 2.0749039001355375e-05,
-      "loss": 0.0003,
-      "step": 693
-    },
-    {
-      "epoch": 3.5773195876288657,
-      "grad_norm": 0.05714479088783264,
-      "learning_rate": 2.061073738537635e-05,
-      "loss": 0.0007,
-      "step": 694
-    },
-    {
-      "epoch": 3.582474226804124,
-      "grad_norm": 0.008403950370848179,
-      "learning_rate": 2.0472778468019454e-05,
-      "loss": 0.0002,
-      "step": 695
-    },
-    {
-      "epoch": 3.5876288659793816,
-      "grad_norm": 0.46188488602638245,
-      "learning_rate": 2.0335163857978744e-05,
-      "loss": 0.0038,
-      "step": 696
-    },
-    {
-      "epoch": 3.5927835051546393,
-      "grad_norm": 0.008672907017171383,
-      "learning_rate": 2.019789515993336e-05,
-      "loss": 0.0002,
-      "step": 697
-    },
-    {
-      "epoch": 3.597938144329897,
-      "grad_norm": 0.6004514098167419,
-      "learning_rate": 2.0060973974528874e-05,
-      "loss": 0.0044,
-      "step": 698
-    },
-    {
-      "epoch": 3.6030927835051547,
-      "grad_norm": 0.049697019159793854,
-      "learning_rate": 1.992440189835869e-05,
-      "loss": 0.0007,
-      "step": 699
-    },
-    {
-      "epoch": 3.6082474226804124,
-      "grad_norm": 0.14393624663352966,
-      "learning_rate": 1.9788180523945277e-05,
-      "loss": 0.0016,
-      "step": 700
-    },
-    {
-      "epoch": 3.61340206185567,
-      "grad_norm": 0.005976496264338493,
-      "learning_rate": 1.9652311439721764e-05,
-      "loss": 0.0001,
-      "step": 701
-    },
-    {
-      "epoch": 3.618556701030928,
-      "grad_norm": 0.45061904191970825,
-      "learning_rate": 1.9516796230013272e-05,
-      "loss": 0.0033,
-      "step": 702
-    },
-    {
-      "epoch": 3.6237113402061856,
-      "grad_norm": 0.004622936248779297,
-      "learning_rate": 1.9381636475018577e-05,
-      "loss": 0.0001,
-      "step": 703
-    },
-    {
-      "epoch": 3.6288659793814433,
-      "grad_norm": 0.05081766098737717,
-      "learning_rate": 1.9246833750791526e-05,
-      "loss": 0.0004,
-      "step": 704
-    },
-    {
-      "epoch": 3.634020618556701,
-      "grad_norm": 0.09289523959159851,
-      "learning_rate": 1.9112389629222823e-05,
-      "loss": 0.0011,
-      "step": 705
-    },
-    {
-      "epoch": 3.6391752577319587,
-      "grad_norm": 0.030106987804174423,
-      "learning_rate": 1.8978305678021595e-05,
-      "loss": 0.0004,
-      "step": 706
-    },
-    {
-      "epoch": 3.6443298969072164,
-      "grad_norm": 0.2676849067211151,
-      "learning_rate": 1.884458346069713e-05,
-      "loss": 0.0024,
-      "step": 707
-    },
-    {
-      "epoch": 3.649484536082474,
-      "grad_norm": 0.011654643341898918,
-      "learning_rate": 1.8711224536540678e-05,
-      "loss": 0.0002,
-      "step": 708
-    },
-    {
-      "epoch": 3.654639175257732,
-      "grad_norm": 3.1462109088897705,
-      "learning_rate": 1.857823046060722e-05,
-      "loss": 0.0191,
-      "step": 709
-    },
-    {
-      "epoch": 3.6597938144329896,
-      "grad_norm": 0.6463388204574585,
-      "learning_rate": 1.8445602783697374e-05,
-      "loss": 0.0083,
-      "step": 710
-    },
-    {
-      "epoch": 3.6649484536082473,
-      "grad_norm": 0.005683832801878452,
-      "learning_rate": 1.831334305233928e-05,
-      "loss": 0.0001,
-      "step": 711
-    },
-    {
-      "epoch": 3.670103092783505,
-      "grad_norm": 0.14142107963562012,
-      "learning_rate": 1.8181452808770637e-05,
-      "loss": 0.0009,
-      "step": 712
-    },
-    {
-      "epoch": 3.675257731958763,
-      "grad_norm": 0.0627388134598732,
-      "learning_rate": 1.804993359092059e-05,
-      "loss": 0.0007,
-      "step": 713
-    },
-    {
-      "epoch": 3.680412371134021,
-      "grad_norm": 0.006492619402706623,
-      "learning_rate": 1.7918786932391944e-05,
-      "loss": 0.0001,
-      "step": 714
-    },
-    {
-      "epoch": 3.6855670103092786,
-      "grad_norm": 0.37058717012405396,
-      "learning_rate": 1.778801436244319e-05,
-      "loss": 0.0039,
-      "step": 715
-    },
-    {
-      "epoch": 3.6907216494845363,
-      "grad_norm": 0.02725241892039776,
-      "learning_rate": 1.765761740597065e-05,
-      "loss": 0.0004,
-      "step": 716
-    },
-    {
-      "epoch": 3.695876288659794,
-      "grad_norm": 0.5024199485778809,
-      "learning_rate": 1.7527597583490822e-05,
-      "loss": 0.0028,
-      "step": 717
-    },
-    {
-      "epoch": 3.7010309278350517,
-      "grad_norm": 0.28971970081329346,
-      "learning_rate": 1.739795641112248e-05,
-      "loss": 0.0029,
-      "step": 718
-    },
-    {
-      "epoch": 3.7061855670103094,
-      "grad_norm": 0.4896758198738098,
-      "learning_rate": 1.726869540056915e-05,
-      "loss": 0.0041,
-      "step": 719
-    },
-    {
-      "epoch": 3.711340206185567,
-      "grad_norm": 0.01545119471848011,
-      "learning_rate": 1.713981605910137e-05,
-      "loss": 0.0003,
-      "step": 720
-    },
-    {
-      "epoch": 3.716494845360825,
-      "grad_norm": 0.8276351094245911,
-      "learning_rate": 1.70113198895392e-05,
-      "loss": 0.0072,
-      "step": 721
-    },
-    {
-      "epoch": 3.7216494845360826,
-      "grad_norm": 0.008877018466591835,
-      "learning_rate": 1.6883208390234628e-05,
-      "loss": 0.0002,
-      "step": 722
-    },
-    {
-      "epoch": 3.7268041237113403,
-      "grad_norm": 0.011914421804249287,
-      "learning_rate": 1.6755483055054105e-05,
-      "loss": 0.0002,
-      "step": 723
-    },
-    {
-      "epoch": 3.731958762886598,
-      "grad_norm": 0.4500105082988739,
-      "learning_rate": 1.662814537336122e-05,
-      "loss": 0.0034,
-      "step": 724
-    },
-    {
-      "epoch": 3.7371134020618557,
-      "grad_norm": 0.022771602496504784,
-      "learning_rate": 1.650119682999918e-05,
-      "loss": 0.0003,
-      "step": 725
-    },
-    {
-      "epoch": 3.7422680412371134,
-      "grad_norm": 1.4144443273544312,
-      "learning_rate": 1.6374638905273643e-05,
-      "loss": 0.0103,
-      "step": 726
-    },
-    {
-      "epoch": 3.747422680412371,
-      "grad_norm": 0.09157131612300873,
-      "learning_rate": 1.624847307493534e-05,
-      "loss": 0.0013,
-      "step": 727
-    },
-    {
-      "epoch": 3.752577319587629,
-      "grad_norm": 0.09352636337280273,
-      "learning_rate": 1.6122700810162966e-05,
-      "loss": 0.0009,
-      "step": 728
-    },
-    {
-      "epoch": 3.7577319587628866,
-      "grad_norm": 0.02057919092476368,
-      "learning_rate": 1.5997323577545915e-05,
-      "loss": 0.0003,
-      "step": 729
-    },
-    {
-      "epoch": 3.7628865979381443,
-      "grad_norm": 0.053213931620121,
-      "learning_rate": 1.5872342839067306e-05,
-      "loss": 0.0004,
-      "step": 730
-    },
-    {
-      "epoch": 3.768041237113402,
-      "grad_norm": 0.21348951756954193,
-      "learning_rate": 1.5747760052086803e-05,
-      "loss": 0.0021,
-      "step": 731
-    },
-    {
-      "epoch": 3.7731958762886597,
-      "grad_norm": 0.03968895971775055,
-      "learning_rate": 1.5623576669323743e-05,
-      "loss": 0.0005,
-      "step": 732
-    },
-    {
-      "epoch": 3.7783505154639174,
-      "grad_norm": 0.00650789774954319,
-      "learning_rate": 1.5499794138840122e-05,
-      "loss": 0.0002,
-      "step": 733
-    },
-    {
-      "epoch": 3.783505154639175,
-      "grad_norm": 1.0022690296173096,
-      "learning_rate": 1.5376413904023722e-05,
-      "loss": 0.0081,
-      "step": 734
-    },
-    {
-      "epoch": 3.788659793814433,
-      "grad_norm": 1.8135733604431152,
-      "learning_rate": 1.525343740357128e-05,
-      "loss": 0.0138,
-      "step": 735
-    },
-    {
-      "epoch": 3.7938144329896906,
-      "grad_norm": 0.0444062165915966,
-      "learning_rate": 1.5130866071471717e-05,
-      "loss": 0.0007,
-      "step": 736
-    },
-    {
-      "epoch": 3.7989690721649483,
-      "grad_norm": 0.06379842758178711,
-      "learning_rate": 1.500870133698945e-05,
-      "loss": 0.0009,
-      "step": 737
-    },
-    {
-      "epoch": 3.804123711340206,
-      "grad_norm": 0.012018664740025997,
-      "learning_rate": 1.4886944624647647e-05,
-      "loss": 0.0002,
-      "step": 738
-    },
-    {
-      "epoch": 3.8092783505154637,
-      "grad_norm": 0.34600210189819336,
-      "learning_rate": 1.4765597354211713e-05,
-      "loss": 0.0027,
-      "step": 739
-    },
-    {
-      "epoch": 3.8144329896907214,
-      "grad_norm": 2.4469614028930664,
-      "learning_rate": 1.4644660940672627e-05,
-      "loss": 0.0204,
-      "step": 740
-    },
-    {
-      "epoch": 3.819587628865979,
-      "grad_norm": 0.017720194533467293,
-      "learning_rate": 1.4524136794230547e-05,
-      "loss": 0.0003,
-      "step": 741
-    },
-    {
-      "epoch": 3.824742268041237,
-      "grad_norm": 0.15242381393909454,
-      "learning_rate": 1.4404026320278318e-05,
-      "loss": 0.0011,
-      "step": 742
-    },
-    {
-      "epoch": 3.829896907216495,
-      "grad_norm": 0.06981607526540756,
-      "learning_rate": 1.4284330919385036e-05,
-      "loss": 0.0007,
-      "step": 743
-    },
-    {
-      "epoch": 3.8350515463917527,
-      "grad_norm": 0.08876276761293411,
-      "learning_rate": 1.4165051987279831e-05,
-      "loss": 0.0006,
-      "step": 744
-    },
-    {
-      "epoch": 3.8402061855670104,
-      "grad_norm": 0.17177937924861908,
-      "learning_rate": 1.404619091483546e-05,
-      "loss": 0.0013,
-      "step": 745
-    },
-    {
-      "epoch": 3.845360824742268,
-      "grad_norm": 0.033125944435596466,
-      "learning_rate": 1.3927749088052217e-05,
-      "loss": 0.0005,
-      "step": 746
-    },
-    {
-      "epoch": 3.850515463917526,
-      "grad_norm": 1.0394357442855835,
-      "learning_rate": 1.3809727888041668e-05,
-      "loss": 0.0086,
-      "step": 747
-    },
-    {
-      "epoch": 3.8556701030927836,
-      "grad_norm": 0.020910169929265976,
-      "learning_rate": 1.3692128691010592e-05,
-      "loss": 0.0003,
-      "step": 748
-    },
-    {
-      "epoch": 3.8608247422680413,
-      "grad_norm": 0.016304798424243927,
-      "learning_rate": 1.3574952868244922e-05,
-      "loss": 0.0003,
-      "step": 749
-    },
-    {
-      "epoch": 3.865979381443299,
-      "grad_norm": 1.6812098026275635,
-      "learning_rate": 1.3458201786093794e-05,
-      "loss": 0.0097,
-      "step": 750
-    },
-    {
-      "epoch": 3.8711340206185567,
-      "grad_norm": 0.09045127779245377,
-      "learning_rate": 1.334187680595358e-05,
-      "loss": 0.0009,
-      "step": 751
-    },
-    {
-      "epoch": 3.8762886597938144,
-      "grad_norm": 0.39485737681388855,
-      "learning_rate": 1.3225979284251954e-05,
-      "loss": 0.0032,
-      "step": 752
-    },
-    {
-      "epoch": 3.881443298969072,
-      "grad_norm": 1.2273677587509155,
-      "learning_rate": 1.3110510572432221e-05,
-      "loss": 0.0108,
-      "step": 753
-    },
-    {
-      "epoch": 3.88659793814433,
-      "grad_norm": 0.03808501362800598,
-      "learning_rate": 1.2995472016937404e-05,
-      "loss": 0.0003,
-      "step": 754
-    },
-    {
-      "epoch": 3.8917525773195876,
-      "grad_norm": 0.02039515972137451,
-      "learning_rate": 1.2880864959194665e-05,
-      "loss": 0.0003,
-      "step": 755
-    },
-    {
-      "epoch": 3.8969072164948453,
-      "grad_norm": 1.4002429246902466,
-      "learning_rate": 1.2766690735599568e-05,
-      "loss": 0.0142,
-      "step": 756
-    },
-    {
-      "epoch": 3.902061855670103,
-      "grad_norm": 0.06780578941106796,
-      "learning_rate": 1.2652950677500574e-05,
-      "loss": 0.0006,
-      "step": 757
-    },
-    {
-      "epoch": 3.9072164948453607,
-      "grad_norm": 0.00860658474266529,
-      "learning_rate": 1.253964611118345e-05,
-      "loss": 0.0002,
-      "step": 758
-    },
-    {
-      "epoch": 3.9123711340206184,
-      "grad_norm": 2.367659330368042,
-      "learning_rate": 1.2426778357855873e-05,
-      "loss": 0.024,
-      "step": 759
-    },
-    {
-      "epoch": 3.917525773195876,
-      "grad_norm": 0.3828847110271454,
-      "learning_rate": 1.2314348733631959e-05,
-      "loss": 0.0037,
-      "step": 760
-    },
-    {
-      "epoch": 3.9226804123711343,
-      "grad_norm": 0.0686371847987175,
-      "learning_rate": 1.2202358549516923e-05,
-      "loss": 0.0007,
-      "step": 761
-    },
-    {
-      "epoch": 3.927835051546392,
-      "grad_norm": 0.00528530590236187,
-      "learning_rate": 1.209080911139187e-05,
-      "loss": 0.0002,
-      "step": 762
-    },
-    {
-      "epoch": 3.9329896907216497,
-      "grad_norm": 0.10510650277137756,
-      "learning_rate": 1.1979701719998453e-05,
-      "loss": 0.0013,
-      "step": 763
-    },
-    {
-      "epoch": 3.9381443298969074,
-      "grad_norm": 0.019046945497393608,
-      "learning_rate": 1.1869037670923815e-05,
-      "loss": 0.0002,
-      "step": 764
-    },
-    {
-      "epoch": 3.943298969072165,
-      "grad_norm": 0.012646029703319073,
-      "learning_rate": 1.1758818254585369e-05,
-      "loss": 0.0002,
-      "step": 765
-    },
-    {
-      "epoch": 3.948453608247423,
-      "grad_norm": 2.9753146171569824,
-      "learning_rate": 1.164904475621587e-05,
-      "loss": 0.0265,
-      "step": 766
-    },
-    {
-      "epoch": 3.9536082474226806,
-      "grad_norm": 1.622678279876709,
-      "learning_rate": 1.1539718455848309e-05,
-      "loss": 0.0135,
-      "step": 767
-    },
-    {
-      "epoch": 3.9587628865979383,
-      "grad_norm": 0.0164541844278574,
-      "learning_rate": 1.1430840628301093e-05,
-      "loss": 0.0003,
-      "step": 768
-    },
-    {
-      "epoch": 3.963917525773196,
-      "grad_norm": 0.15376994013786316,
-      "learning_rate": 1.1322412543163135e-05,
-      "loss": 0.0013,
-      "step": 769
-    },
-    {
-      "epoch": 3.9690721649484537,
-      "grad_norm": 0.052942678332328796,
-      "learning_rate": 1.1214435464779006e-05,
-      "loss": 0.0005,
-      "step": 770
-    },
-    {
-      "epoch": 3.9742268041237114,
-      "grad_norm": 0.0726817324757576,
-      "learning_rate": 1.1106910652234276e-05,
-      "loss": 0.0007,
-      "step": 771
-    },
-    {
-      "epoch": 3.979381443298969,
-      "grad_norm": 0.014821125194430351,
-      "learning_rate": 1.099983935934077e-05,
-      "loss": 0.0002,
-      "step": 772
-    },
-    {
-      "epoch": 3.984536082474227,
-      "grad_norm": 0.014423374086618423,
-      "learning_rate": 1.089322283462197e-05,
-      "loss": 0.0003,
-      "step": 773
-    },
-    {
-      "epoch": 3.9896907216494846,
-      "grad_norm": 0.0970507562160492,
-      "learning_rate": 1.0787062321298442e-05,
-      "loss": 0.0009,
-      "step": 774
-    },
-    {
-      "epoch": 3.9948453608247423,
-      "grad_norm": 0.013409281149506569,
-      "learning_rate": 1.0681359057273388e-05,
-      "loss": 0.0003,
-      "step": 775
-    },
-    {
-      "epoch": 4.0,
-      "grad_norm": 0.8209367394447327,
-      "learning_rate": 1.0576114275118131e-05,
-      "loss": 0.0061,
-      "step": 776
-    },
-    {
-      "epoch": 4.0,
-      "eval_loss": 0.10921349376440048,
-      "eval_runtime": 22.9078,
-      "eval_samples_per_second": 7.159,
-      "eval_steps_per_second": 1.79,
-      "step": 776
-    },
-    {
-      "epoch": 4.005154639175258,
-      "grad_norm": 0.08000417053699493,
-      "learning_rate": 1.0471329202057823e-05,
-      "loss": 0.0008,
-      "step": 777
-    },
-    {
-      "epoch": 4.010309278350515,
-      "grad_norm": 0.05137176439166069,
-      "learning_rate": 1.0367005059957096e-05,
-      "loss": 0.0006,
-      "step": 778
-    },
-    {
-      "epoch": 4.015463917525773,
-      "grad_norm": 0.006002445705235004,
-      "learning_rate": 1.0263143065305769e-05,
-      "loss": 0.0001,
-      "step": 779
-    },
-    {
-      "epoch": 4.020618556701031,
-      "grad_norm": 0.00495640654116869,
-      "learning_rate": 1.0159744429204777e-05,
-      "loss": 0.0001,
-      "step": 780
-    },
-    {
-      "epoch": 4.025773195876289,
-      "grad_norm": 0.23104587197303772,
-      "learning_rate": 1.005681035735192e-05,
-      "loss": 0.002,
-      "step": 781
-    },
-    {
-      "epoch": 4.030927835051546,
-      "grad_norm": 0.018206927925348282,
-      "learning_rate": 9.954342050027921e-06,
-      "loss": 0.0002,
-      "step": 782
-    },
-    {
-      "epoch": 4.036082474226804,
-      "grad_norm": 0.03753966838121414,
-      "learning_rate": 9.852340702082318e-06,
-      "loss": 0.0004,
-      "step": 783
-    },
-    {
-      "epoch": 4.041237113402062,
-      "grad_norm": 0.03190065175294876,
-      "learning_rate": 9.750807502919652e-06,
-      "loss": 0.0004,
-      "step": 784
-    },
-    {
-      "epoch": 4.046391752577319,
-      "grad_norm": 0.029177363961935043,
-      "learning_rate": 9.64974363648548e-06,
-      "loss": 0.0003,
-      "step": 785
-    },
-    {
-      "epoch": 4.051546391752577,
-      "grad_norm": 0.013272424228489399,
-      "learning_rate": 9.549150281252633e-06,
-      "loss": 0.0002,
-      "step": 786
-    },
-    {
-      "epoch": 4.056701030927835,
-      "grad_norm": 0.11541124433279037,
-      "learning_rate": 9.449028610207494e-06,
-      "loss": 0.0013,
-      "step": 787
-    },
-    {
-      "epoch": 4.061855670103093,
-      "grad_norm": 0.07466880977153778,
-      "learning_rate": 9.349379790836243e-06,
-      "loss": 0.0006,
-      "step": 788
-    },
-    {
-      "epoch": 4.06701030927835,
-      "grad_norm": 0.01846369542181492,
-      "learning_rate": 9.25020498511135e-06,
-      "loss": 0.0003,
-      "step": 789
-    },
-    {
-      "epoch": 4.072164948453608,
-      "grad_norm": 0.053815506398677826,
-      "learning_rate": 9.151505349477902e-06,
-      "loss": 0.0006,
-      "step": 790
-    },
-    {
-      "epoch": 4.077319587628866,
-      "grad_norm": 0.011579618789255619,
-      "learning_rate": 9.053282034840238e-06,
-      "loss": 0.0002,
-      "step": 791
-    },
-    {
-      "epoch": 4.082474226804123,
-      "grad_norm": 0.007092161104083061,
-      "learning_rate": 8.955536186548425e-06,
-      "loss": 0.0001,
-      "step": 792
-    },
-    {
-      "epoch": 4.087628865979381,
-      "grad_norm": 0.047016046941280365,
-      "learning_rate": 8.858268944384995e-06,
-      "loss": 0.0006,
-      "step": 793
-    },
-    {
-      "epoch": 4.092783505154639,
-      "grad_norm": 0.08547954261302948,
-      "learning_rate": 8.761481442551573e-06,
-      "loss": 0.0008,
-      "step": 794
-    },
-    {
-      "epoch": 4.097938144329897,
-      "grad_norm": 0.008324719965457916,
-      "learning_rate": 8.665174809655708e-06,
-      "loss": 0.0002,
-      "step": 795
-    },
-    {
-      "epoch": 4.103092783505154,
-      "grad_norm": 0.1423901468515396,
-      "learning_rate": 8.569350168697704e-06,
-      "loss": 0.0012,
-      "step": 796
-    },
-    {
-      "epoch": 4.108247422680412,
-      "grad_norm": 0.13605594635009766,
-      "learning_rate": 8.474008637057478e-06,
-      "loss": 0.0008,
-      "step": 797
-    },
-    {
-      "epoch": 4.11340206185567,
-      "grad_norm": 0.01406699325889349,
-      "learning_rate": 8.379151326481587e-06,
-      "loss": 0.0002,
-      "step": 798
-    },
-    {
-      "epoch": 4.118556701030927,
-      "grad_norm": 0.023498743772506714,
-      "learning_rate": 8.284779343070265e-06,
-      "loss": 0.0003,
-      "step": 799
-    },
-    {
-      "epoch": 4.123711340206185,
-      "grad_norm": 0.3762100040912628,
-      "learning_rate": 8.19089378726447e-06,
-      "loss": 0.0023,
-      "step": 800
-    },
-    {
-      "epoch": 4.128865979381443,
-      "grad_norm": 0.007977807894349098,
-      "learning_rate": 8.097495753833078e-06,
-      "loss": 0.0002,
-      "step": 801
-    },
-    {
-      "epoch": 4.134020618556701,
-      "grad_norm": 0.02798437513411045,
-      "learning_rate": 8.004586331860175e-06,
-      "loss": 0.0003,
-      "step": 802
-    },
-    {
-      "epoch": 4.139175257731959,
-      "grad_norm": 0.013621841557323933,
-      "learning_rate": 7.91216660473228e-06,
-      "loss": 0.0003,
-      "step": 803
-    },
-    {
-      "epoch": 4.144329896907217,
-      "grad_norm": 0.010507001541554928,
-      "learning_rate": 7.820237650125712e-06,
-      "loss": 0.0002,
-      "step": 804
-    },
-    {
-      "epoch": 4.149484536082475,
-      "grad_norm": 0.06786844879388809,
-      "learning_rate": 7.728800539994113e-06,
-      "loss": 0.0006,
-      "step": 805
-    },
-    {
-      "epoch": 4.154639175257732,
-      "grad_norm": 0.02216934785246849,
-      "learning_rate": 7.637856340555822e-06,
-      "loss": 0.0003,
-      "step": 806
-    },
-    {
-      "epoch": 4.15979381443299,
-      "grad_norm": 0.021773718297481537,
-      "learning_rate": 7.547406112281557e-06,
-      "loss": 0.0003,
-      "step": 807
-    },
-    {
-      "epoch": 4.164948453608248,
-      "grad_norm": 0.005113635677844286,
-      "learning_rate": 7.457450909881969e-06,
-      "loss": 0.0001,
-      "step": 808
-    },
-    {
-      "epoch": 4.170103092783505,
-      "grad_norm": 0.027230041101574898,
-      "learning_rate": 7.367991782295391e-06,
-      "loss": 0.0002,
-      "step": 809
-    },
-    {
-      "epoch": 4.175257731958763,
-      "grad_norm": 0.12998707592487335,
-      "learning_rate": 7.2790297726755716e-06,
-      "loss": 0.0013,
-      "step": 810
-    },
-    {
-      "epoch": 4.180412371134021,
-      "grad_norm": 0.26570916175842285,
-      "learning_rate": 7.190565918379549e-06,
-      "loss": 0.0018,
-      "step": 811
-    },
-    {
-      "epoch": 4.185567010309279,
-      "grad_norm": 0.014305188320577145,
-      "learning_rate": 7.1026012509555265e-06,
-      "loss": 0.0002,
-      "step": 812
-    },
-    {
-      "epoch": 4.190721649484536,
-      "grad_norm": 0.014115352183580399,
-      "learning_rate": 7.015136796130828e-06,
-      "loss": 0.0003,
-      "step": 813
-    },
-    {
-      "epoch": 4.195876288659794,
-      "grad_norm": 0.06203961372375488,
-      "learning_rate": 6.928173573800006e-06,
-      "loss": 0.0007,
-      "step": 814
-    },
-    {
-      "epoch": 4.201030927835052,
-      "grad_norm": 0.594573438167572,
-      "learning_rate": 6.8417125980128675e-06,
-      "loss": 0.0048,
-      "step": 815
-    },
-    {
-      "epoch": 4.206185567010309,
-      "grad_norm": 0.1779005527496338,
-      "learning_rate": 6.755754876962711e-06,
-      "loss": 0.0013,
-      "step": 816
-    },
-    {
-      "epoch": 4.211340206185567,
-      "grad_norm": 0.005702666938304901,
-      "learning_rate": 6.670301412974511e-06,
-      "loss": 0.0002,
-      "step": 817
-    },
-    {
-      "epoch": 4.216494845360825,
-      "grad_norm": 0.00647127116099,
-      "learning_rate": 6.585353202493322e-06,
-      "loss": 0.0002,
-      "step": 818
-    },
-    {
-      "epoch": 4.221649484536083,
-      "grad_norm": 0.04004105553030968,
-      "learning_rate": 6.500911236072532e-06,
-      "loss": 0.0006,
-      "step": 819
-    },
-    {
-      "epoch": 4.22680412371134,
-      "grad_norm": 0.019848493859171867,
-      "learning_rate": 6.416976498362432e-06,
-      "loss": 0.0002,
-      "step": 820
-    },
-    {
-      "epoch": 4.231958762886598,
-      "grad_norm": 0.009120230562984943,
-      "learning_rate": 6.333549968098684e-06,
-      "loss": 0.0002,
-      "step": 821
-    },
-    {
-      "epoch": 4.237113402061856,
-      "grad_norm": 0.010857068933546543,
-      "learning_rate": 6.250632618090868e-06,
-      "loss": 0.0002,
-      "step": 822
-    },
-    {
-      "epoch": 4.242268041237113,
-      "grad_norm": 0.010758602060377598,
-      "learning_rate": 6.168225415211226e-06,
-      "loss": 0.0002,
-      "step": 823
-    },
-    {
-      "epoch": 4.247422680412371,
-      "grad_norm": 0.033467814326286316,
-      "learning_rate": 6.0863293203833105e-06,
-      "loss": 0.0004,
-      "step": 824
-    },
-    {
-      "epoch": 4.252577319587629,
-      "grad_norm": 0.008194089867174625,
-      "learning_rate": 6.004945288570813e-06,
-      "loss": 0.0002,
-      "step": 825
-    },
-    {
-      "epoch": 4.257731958762887,
-      "grad_norm": 0.04414292797446251,
-      "learning_rate": 5.924074268766422e-06,
-      "loss": 0.0004,
-      "step": 826
-    },
-    {
-      "epoch": 4.262886597938144,
-      "grad_norm": 0.041067712008953094,
-      "learning_rate": 5.843717203980792e-06,
-      "loss": 0.0004,
-      "step": 827
-    },
-    {
-      "epoch": 4.268041237113402,
-      "grad_norm": 0.006359627936035395,
-      "learning_rate": 5.763875031231464e-06,
-      "loss": 0.0001,
-      "step": 828
-    },
-    {
-      "epoch": 4.27319587628866,
-      "grad_norm": 0.016976401209831238,
-      "learning_rate": 5.684548681532032e-06,
-      "loss": 0.0003,
-      "step": 829
-    },
-    {
-      "epoch": 4.278350515463917,
-      "grad_norm": 0.00730324350297451,
-      "learning_rate": 5.605739079881239e-06,
-      "loss": 0.0002,
-      "step": 830
-    },
-    {
-      "epoch": 4.283505154639175,
-      "grad_norm": 0.013788518495857716,
-      "learning_rate": 5.527447145252174e-06,
-      "loss": 0.0003,
-      "step": 831
-    },
-    {
-      "epoch": 4.288659793814433,
-      "grad_norm": 0.1177946925163269,
-      "learning_rate": 5.449673790581611e-06,
-      "loss": 0.0014,
-      "step": 832
-    },
-    {
-      "epoch": 4.293814432989691,
-      "grad_norm": 0.011995314620435238,
-      "learning_rate": 5.372419922759292e-06,
-      "loss": 0.0002,
-      "step": 833
-    },
-    {
-      "epoch": 4.298969072164948,
-      "grad_norm": 0.07039642333984375,
-      "learning_rate": 5.295686442617443e-06,
-      "loss": 0.001,
-      "step": 834
-    },
-    {
-      "epoch": 4.304123711340206,
-      "grad_norm": 0.05830349028110504,
-      "learning_rate": 5.219474244920164e-06,
-      "loss": 0.0005,
-      "step": 835
-    },
-    {
-      "epoch": 4.309278350515464,
-      "grad_norm": 0.05796303227543831,
-      "learning_rate": 5.143784218353103e-06,
-      "loss": 0.0009,
-      "step": 836
-    },
-    {
-      "epoch": 4.314432989690721,
-      "grad_norm": 0.01927541196346283,
-      "learning_rate": 5.068617245513008e-06,
-      "loss": 0.0002,
-      "step": 837
-    },
-    {
-      "epoch": 4.319587628865979,
-      "grad_norm": 0.04202926531434059,
-      "learning_rate": 4.993974202897455e-06,
-      "loss": 0.0004,
-      "step": 838
-    },
-    {
-      "epoch": 4.324742268041237,
-      "grad_norm": 0.6382042169570923,
-      "learning_rate": 4.9198559608946815e-06,
-      "loss": 0.0048,
-      "step": 839
-    },
-    {
-      "epoch": 4.329896907216495,
-      "grad_norm": 0.008230258710682392,
-      "learning_rate": 4.846263383773364e-06,
-      "loss": 0.0002,
-      "step": 840
-    },
-    {
-      "epoch": 4.335051546391752,
-      "grad_norm": 0.0035637454129755497,
-      "learning_rate": 4.773197329672596e-06,
-      "loss": 0.0001,
-      "step": 841
-    },
-    {
-      "epoch": 4.34020618556701,
-      "grad_norm": 0.10410863161087036,
-      "learning_rate": 4.700658650591827e-06,
-      "loss": 0.001,
-      "step": 842
-    },
-    {
-      "epoch": 4.345360824742268,
-      "grad_norm": 0.038275450468063354,
-      "learning_rate": 4.628648192380986e-06,
-      "loss": 0.0005,
-      "step": 843
-    },
-    {
-      "epoch": 4.350515463917525,
-      "grad_norm": 0.006225410848855972,
-      "learning_rate": 4.557166794730572e-06,
-      "loss": 0.0002,
-      "step": 844
-    },
-    {
-      "epoch": 4.355670103092783,
-      "grad_norm": 0.02506340481340885,
-      "learning_rate": 4.4862152911618934e-06,
-      "loss": 0.0003,
-      "step": 845
-    },
-    {
-      "epoch": 4.360824742268041,
-      "grad_norm": 0.03422137349843979,
-      "learning_rate": 4.415794509017329e-06,
-      "loss": 0.0004,
-      "step": 846
-    },
-    {
-      "epoch": 4.365979381443299,
-      "grad_norm": 0.0062943557277321815,
-      "learning_rate": 4.34590526945069e-06,
-      "loss": 0.0001,
-      "step": 847
-    },
-    {
-      "epoch": 4.371134020618557,
-      "grad_norm": 0.006638492923229933,
-      "learning_rate": 4.276548387417656e-06,
-      "loss": 0.0002,
-      "step": 848
-    },
-    {
-      "epoch": 4.376288659793815,
-      "grad_norm": 0.17980645596981049,
-      "learning_rate": 4.20772467166623e-06,
-      "loss": 0.0017,
-      "step": 849
-    },
-    {
-      "epoch": 4.381443298969073,
-      "grad_norm": 0.04007335379719734,
-      "learning_rate": 4.139434924727359e-06,
-      "loss": 0.0006,
-      "step": 850
-    },
-    {
-      "epoch": 4.38659793814433,
-      "grad_norm": 0.06706022471189499,
-      "learning_rate": 4.071679942905532e-06,
-      "loss": 0.0007,
-      "step": 851
-    },
-    {
-      "epoch": 4.391752577319588,
-      "grad_norm": 0.2766573131084442,
-      "learning_rate": 4.004460516269554e-06,
-      "loss": 0.0013,
-      "step": 852
-    },
-    {
-      "epoch": 4.396907216494846,
-      "grad_norm": 0.029405513778328896,
-      "learning_rate": 3.937777428643253e-06,
-      "loss": 0.0004,
-      "step": 853
-    },
-    {
-      "epoch": 4.402061855670103,
-      "grad_norm": 0.031775474548339844,
-      "learning_rate": 3.87163145759642e-06,
-      "loss": 0.0005,
-      "step": 854
-    },
-    {
-      "epoch": 4.407216494845361,
-      "grad_norm": 0.038619689643383026,
-      "learning_rate": 3.8060233744356633e-06,
-      "loss": 0.0004,
-      "step": 855
-    },
-    {
-      "epoch": 4.412371134020619,
-      "grad_norm": 0.01843906380236149,
-      "learning_rate": 3.7409539441954965e-06,
-      "loss": 0.0003,
-      "step": 856
-    },
-    {
-      "epoch": 4.417525773195877,
-      "grad_norm": 0.005279079079627991,
-      "learning_rate": 3.6764239256293577e-06,
-      "loss": 0.0001,
-      "step": 857
-    },
-    {
-      "epoch": 4.422680412371134,
-      "grad_norm": 0.006745612248778343,
-      "learning_rate": 3.612434071200771e-06,
-      "loss": 0.0002,
-      "step": 858
-    },
-    {
-      "epoch": 4.427835051546392,
-      "grad_norm": 0.3258199691772461,
-      "learning_rate": 3.548985127074611e-06,
-      "loss": 0.0014,
-      "step": 859
-    },
-    {
-      "epoch": 4.43298969072165,
-      "grad_norm": 0.017199253663420677,
-      "learning_rate": 3.486077833108342e-06,
-      "loss": 0.0003,
-      "step": 860
-    },
-    {
-      "epoch": 4.438144329896907,
-      "grad_norm": 0.003292699111625552,
-      "learning_rate": 3.4237129228434415e-06,
-      "loss": 0.0001,
-      "step": 861
-    },
-    {
-      "epoch": 4.443298969072165,
-      "grad_norm": 0.13467860221862793,
-      "learning_rate": 3.3618911234968243e-06,
-      "loss": 0.0012,
-      "step": 862
-    },
-    {
-      "epoch": 4.448453608247423,
-      "grad_norm": 0.006807579658925533,
-      "learning_rate": 3.300613155952359e-06,
-      "loss": 0.0001,
-      "step": 863
-    },
-    {
-      "epoch": 4.453608247422681,
-      "grad_norm": 0.04459763318300247,
-      "learning_rate": 3.2398797347524656e-06,
-      "loss": 0.0004,
-      "step": 864
-    },
-    {
-      "epoch": 4.458762886597938,
-      "grad_norm": 0.005752299912273884,
-      "learning_rate": 3.1796915680897988e-06,
-      "loss": 0.0002,
-      "step": 865
     }
   ],
   "logging_steps": 1,
-  "max_steps": 970,
+  "max_steps": 3860,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 5,
   "save_steps": 5,
@@ -6153,7 +209,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3.345416922983301e+17,
+  "total_flos": 2.79484292923392e+16,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null