PEFT
Safetensors
domain0529 / trainer_state.json
chen
Upload folder using huggingface_hub
2b6380d verified
Raw
History Blame Contribute Delete
135 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.0,
"eval_steps": 500,
"global_step": 780,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.005148005148005148,
"grad_norm": 28.892786026000977,
"learning_rate": 0.0,
"loss": 0.6744,
"step": 1
},
{
"epoch": 0.010296010296010296,
"grad_norm": 36.62438201904297,
"learning_rate": 2.061855670103093e-06,
"loss": 1.1938,
"step": 2
},
{
"epoch": 0.015444015444015444,
"grad_norm": 30.183523178100586,
"learning_rate": 4.123711340206186e-06,
"loss": 0.9132,
"step": 3
},
{
"epoch": 0.02059202059202059,
"grad_norm": 30.808048248291016,
"learning_rate": 6.185567010309279e-06,
"loss": 1.0766,
"step": 4
},
{
"epoch": 0.02574002574002574,
"grad_norm": 31.998327255249023,
"learning_rate": 8.247422680412371e-06,
"loss": 1.3448,
"step": 5
},
{
"epoch": 0.03088803088803089,
"grad_norm": 36.63566207885742,
"learning_rate": 1.0309278350515464e-05,
"loss": 1.2847,
"step": 6
},
{
"epoch": 0.036036036036036036,
"grad_norm": 42.04302978515625,
"learning_rate": 1.2371134020618558e-05,
"loss": 0.5938,
"step": 7
},
{
"epoch": 0.04118404118404118,
"grad_norm": 14.437341690063477,
"learning_rate": 1.4432989690721649e-05,
"loss": 0.4745,
"step": 8
},
{
"epoch": 0.04633204633204633,
"grad_norm": 17.014678955078125,
"learning_rate": 1.6494845360824743e-05,
"loss": 0.5739,
"step": 9
},
{
"epoch": 0.05148005148005148,
"grad_norm": 16.1011905670166,
"learning_rate": 1.8556701030927837e-05,
"loss": 0.2984,
"step": 10
},
{
"epoch": 0.05662805662805663,
"grad_norm": 12.759788513183594,
"learning_rate": 2.0618556701030927e-05,
"loss": 0.3813,
"step": 11
},
{
"epoch": 0.06177606177606178,
"grad_norm": 8.758112907409668,
"learning_rate": 2.268041237113402e-05,
"loss": 0.2739,
"step": 12
},
{
"epoch": 0.06692406692406692,
"grad_norm": 12.049139976501465,
"learning_rate": 2.4742268041237116e-05,
"loss": 0.3238,
"step": 13
},
{
"epoch": 0.07207207207207207,
"grad_norm": 20.903139114379883,
"learning_rate": 2.6804123711340206e-05,
"loss": 0.4468,
"step": 14
},
{
"epoch": 0.07722007722007722,
"grad_norm": 24.865018844604492,
"learning_rate": 2.8865979381443297e-05,
"loss": 0.3775,
"step": 15
},
{
"epoch": 0.08236808236808237,
"grad_norm": 10.060315132141113,
"learning_rate": 3.0927835051546395e-05,
"loss": 0.3289,
"step": 16
},
{
"epoch": 0.08751608751608751,
"grad_norm": 5.874386787414551,
"learning_rate": 3.2989690721649485e-05,
"loss": 0.261,
"step": 17
},
{
"epoch": 0.09266409266409266,
"grad_norm": 11.170652389526367,
"learning_rate": 3.5051546391752576e-05,
"loss": 0.4735,
"step": 18
},
{
"epoch": 0.0978120978120978,
"grad_norm": 8.786273002624512,
"learning_rate": 3.7113402061855674e-05,
"loss": 0.3048,
"step": 19
},
{
"epoch": 0.10296010296010295,
"grad_norm": 5.152289867401123,
"learning_rate": 3.9175257731958764e-05,
"loss": 0.1589,
"step": 20
},
{
"epoch": 0.10810810810810811,
"grad_norm": 28.42683219909668,
"learning_rate": 4.1237113402061855e-05,
"loss": 0.2435,
"step": 21
},
{
"epoch": 0.11325611325611326,
"grad_norm": 11.394163131713867,
"learning_rate": 4.329896907216495e-05,
"loss": 0.5999,
"step": 22
},
{
"epoch": 0.11840411840411841,
"grad_norm": 5.172152519226074,
"learning_rate": 4.536082474226804e-05,
"loss": 0.1704,
"step": 23
},
{
"epoch": 0.12355212355212356,
"grad_norm": 3.7361834049224854,
"learning_rate": 4.7422680412371134e-05,
"loss": 0.071,
"step": 24
},
{
"epoch": 0.1287001287001287,
"grad_norm": 10.789998054504395,
"learning_rate": 4.948453608247423e-05,
"loss": 0.1626,
"step": 25
},
{
"epoch": 0.13384813384813385,
"grad_norm": 29.464887619018555,
"learning_rate": 5.1546391752577315e-05,
"loss": 0.3338,
"step": 26
},
{
"epoch": 0.138996138996139,
"grad_norm": 4.975821495056152,
"learning_rate": 5.360824742268041e-05,
"loss": 0.1128,
"step": 27
},
{
"epoch": 0.14414414414414414,
"grad_norm": 18.57765007019043,
"learning_rate": 5.567010309278351e-05,
"loss": 0.4199,
"step": 28
},
{
"epoch": 0.1492921492921493,
"grad_norm": 7.429575443267822,
"learning_rate": 5.7731958762886594e-05,
"loss": 0.3278,
"step": 29
},
{
"epoch": 0.15444015444015444,
"grad_norm": 3.0536420345306396,
"learning_rate": 5.979381443298969e-05,
"loss": 0.1363,
"step": 30
},
{
"epoch": 0.15958815958815958,
"grad_norm": 5.408520698547363,
"learning_rate": 6.185567010309279e-05,
"loss": 0.2471,
"step": 31
},
{
"epoch": 0.16473616473616473,
"grad_norm": 6.894387245178223,
"learning_rate": 6.391752577319587e-05,
"loss": 0.1888,
"step": 32
},
{
"epoch": 0.16988416988416988,
"grad_norm": 3.299340009689331,
"learning_rate": 6.597938144329897e-05,
"loss": 0.1156,
"step": 33
},
{
"epoch": 0.17503217503217502,
"grad_norm": 20.644685745239258,
"learning_rate": 6.804123711340207e-05,
"loss": 0.4801,
"step": 34
},
{
"epoch": 0.18018018018018017,
"grad_norm": 12.999178886413574,
"learning_rate": 7.010309278350515e-05,
"loss": 0.5774,
"step": 35
},
{
"epoch": 0.18532818532818532,
"grad_norm": 5.521422386169434,
"learning_rate": 7.216494845360825e-05,
"loss": 0.1196,
"step": 36
},
{
"epoch": 0.19047619047619047,
"grad_norm": 8.625732421875,
"learning_rate": 7.422680412371135e-05,
"loss": 0.4604,
"step": 37
},
{
"epoch": 0.1956241956241956,
"grad_norm": 5.741725921630859,
"learning_rate": 7.628865979381443e-05,
"loss": 0.2876,
"step": 38
},
{
"epoch": 0.20077220077220076,
"grad_norm": 6.012145519256592,
"learning_rate": 7.835051546391753e-05,
"loss": 0.2902,
"step": 39
},
{
"epoch": 0.2059202059202059,
"grad_norm": 13.354425430297852,
"learning_rate": 8.041237113402063e-05,
"loss": 0.3557,
"step": 40
},
{
"epoch": 0.21106821106821108,
"grad_norm": 4.402190208435059,
"learning_rate": 8.247422680412371e-05,
"loss": 0.2048,
"step": 41
},
{
"epoch": 0.21621621621621623,
"grad_norm": 11.984685897827148,
"learning_rate": 8.453608247422681e-05,
"loss": 0.6509,
"step": 42
},
{
"epoch": 0.22136422136422138,
"grad_norm": 11.039689064025879,
"learning_rate": 8.65979381443299e-05,
"loss": 0.4794,
"step": 43
},
{
"epoch": 0.22651222651222652,
"grad_norm": 5.657270431518555,
"learning_rate": 8.865979381443299e-05,
"loss": 0.3023,
"step": 44
},
{
"epoch": 0.23166023166023167,
"grad_norm": 6.657912731170654,
"learning_rate": 9.072164948453609e-05,
"loss": 0.2527,
"step": 45
},
{
"epoch": 0.23680823680823682,
"grad_norm": 3.1037261486053467,
"learning_rate": 9.278350515463918e-05,
"loss": 0.1864,
"step": 46
},
{
"epoch": 0.24195624195624196,
"grad_norm": 2.0633435249328613,
"learning_rate": 9.484536082474227e-05,
"loss": 0.0518,
"step": 47
},
{
"epoch": 0.2471042471042471,
"grad_norm": 11.29289722442627,
"learning_rate": 9.690721649484537e-05,
"loss": 0.2116,
"step": 48
},
{
"epoch": 0.25225225225225223,
"grad_norm": 8.282283782958984,
"learning_rate": 9.896907216494846e-05,
"loss": 0.1405,
"step": 49
},
{
"epoch": 0.2574002574002574,
"grad_norm": 2.075028657913208,
"learning_rate": 0.00010103092783505155,
"loss": 0.0998,
"step": 50
},
{
"epoch": 0.2625482625482625,
"grad_norm": 12.903587341308594,
"learning_rate": 0.00010309278350515463,
"loss": 0.3451,
"step": 51
},
{
"epoch": 0.2676962676962677,
"grad_norm": 7.385539531707764,
"learning_rate": 0.00010515463917525774,
"loss": 0.2367,
"step": 52
},
{
"epoch": 0.2728442728442728,
"grad_norm": 14.151900291442871,
"learning_rate": 0.00010721649484536083,
"loss": 0.0793,
"step": 53
},
{
"epoch": 0.277992277992278,
"grad_norm": 55.025394439697266,
"learning_rate": 0.00010927835051546391,
"loss": 0.2893,
"step": 54
},
{
"epoch": 0.28314028314028317,
"grad_norm": 6.179469585418701,
"learning_rate": 0.00011134020618556702,
"loss": 0.1815,
"step": 55
},
{
"epoch": 0.2882882882882883,
"grad_norm": 8.201168060302734,
"learning_rate": 0.0001134020618556701,
"loss": 0.2443,
"step": 56
},
{
"epoch": 0.29343629343629346,
"grad_norm": 11.909571647644043,
"learning_rate": 0.00011546391752577319,
"loss": 0.2542,
"step": 57
},
{
"epoch": 0.2985842985842986,
"grad_norm": 11.174131393432617,
"learning_rate": 0.0001175257731958763,
"loss": 0.0862,
"step": 58
},
{
"epoch": 0.30373230373230375,
"grad_norm": 4.6562819480896,
"learning_rate": 0.00011958762886597938,
"loss": 0.1845,
"step": 59
},
{
"epoch": 0.3088803088803089,
"grad_norm": 1.4798322916030884,
"learning_rate": 0.00012164948453608247,
"loss": 0.046,
"step": 60
},
{
"epoch": 0.31402831402831405,
"grad_norm": 5.802827835083008,
"learning_rate": 0.00012371134020618558,
"loss": 0.1478,
"step": 61
},
{
"epoch": 0.31917631917631917,
"grad_norm": 8.945999145507812,
"learning_rate": 0.00012577319587628866,
"loss": 0.5062,
"step": 62
},
{
"epoch": 0.32432432432432434,
"grad_norm": 3.4113495349884033,
"learning_rate": 0.00012783505154639175,
"loss": 0.1137,
"step": 63
},
{
"epoch": 0.32947232947232946,
"grad_norm": 7.847507476806641,
"learning_rate": 0.00012989690721649486,
"loss": 0.1599,
"step": 64
},
{
"epoch": 0.33462033462033464,
"grad_norm": 6.716371059417725,
"learning_rate": 0.00013195876288659794,
"loss": 0.2323,
"step": 65
},
{
"epoch": 0.33976833976833976,
"grad_norm": 8.012821197509766,
"learning_rate": 0.00013402061855670103,
"loss": 0.3496,
"step": 66
},
{
"epoch": 0.34491634491634493,
"grad_norm": 10.279366493225098,
"learning_rate": 0.00013608247422680414,
"loss": 0.5191,
"step": 67
},
{
"epoch": 0.35006435006435005,
"grad_norm": 6.560108661651611,
"learning_rate": 0.00013814432989690722,
"loss": 0.3342,
"step": 68
},
{
"epoch": 0.3552123552123552,
"grad_norm": 9.097430229187012,
"learning_rate": 0.0001402061855670103,
"loss": 0.355,
"step": 69
},
{
"epoch": 0.36036036036036034,
"grad_norm": 6.984493732452393,
"learning_rate": 0.00014226804123711342,
"loss": 0.3015,
"step": 70
},
{
"epoch": 0.3655083655083655,
"grad_norm": 4.377732276916504,
"learning_rate": 0.0001443298969072165,
"loss": 0.1736,
"step": 71
},
{
"epoch": 0.37065637065637064,
"grad_norm": 3.6228346824645996,
"learning_rate": 0.00014639175257731958,
"loss": 0.1391,
"step": 72
},
{
"epoch": 0.3758043758043758,
"grad_norm": 2.835405111312866,
"learning_rate": 0.0001484536082474227,
"loss": 0.085,
"step": 73
},
{
"epoch": 0.38095238095238093,
"grad_norm": 2.9045298099517822,
"learning_rate": 0.00015051546391752578,
"loss": 0.0973,
"step": 74
},
{
"epoch": 0.3861003861003861,
"grad_norm": 15.416662216186523,
"learning_rate": 0.00015257731958762886,
"loss": 0.7117,
"step": 75
},
{
"epoch": 0.3912483912483912,
"grad_norm": 12.579750061035156,
"learning_rate": 0.00015463917525773197,
"loss": 0.6393,
"step": 76
},
{
"epoch": 0.3963963963963964,
"grad_norm": 3.5863471031188965,
"learning_rate": 0.00015670103092783506,
"loss": 0.0516,
"step": 77
},
{
"epoch": 0.4015444015444015,
"grad_norm": 18.148759841918945,
"learning_rate": 0.00015876288659793814,
"loss": 0.3573,
"step": 78
},
{
"epoch": 0.4066924066924067,
"grad_norm": 5.694168567657471,
"learning_rate": 0.00016082474226804125,
"loss": 0.0653,
"step": 79
},
{
"epoch": 0.4118404118404118,
"grad_norm": 11.648760795593262,
"learning_rate": 0.00016288659793814434,
"loss": 0.6276,
"step": 80
},
{
"epoch": 0.416988416988417,
"grad_norm": 6.861691474914551,
"learning_rate": 0.00016494845360824742,
"loss": 0.3568,
"step": 81
},
{
"epoch": 0.42213642213642216,
"grad_norm": 10.421208381652832,
"learning_rate": 0.00016701030927835053,
"loss": 0.3878,
"step": 82
},
{
"epoch": 0.4272844272844273,
"grad_norm": 8.286598205566406,
"learning_rate": 0.00016907216494845361,
"loss": 0.2377,
"step": 83
},
{
"epoch": 0.43243243243243246,
"grad_norm": 10.079511642456055,
"learning_rate": 0.0001711340206185567,
"loss": 0.3389,
"step": 84
},
{
"epoch": 0.4375804375804376,
"grad_norm": 4.2360053062438965,
"learning_rate": 0.0001731958762886598,
"loss": 0.1506,
"step": 85
},
{
"epoch": 0.44272844272844275,
"grad_norm": 6.217354774475098,
"learning_rate": 0.0001752577319587629,
"loss": 0.2424,
"step": 86
},
{
"epoch": 0.44787644787644787,
"grad_norm": 19.61492347717285,
"learning_rate": 0.00017731958762886598,
"loss": 0.3983,
"step": 87
},
{
"epoch": 0.45302445302445304,
"grad_norm": 4.8819403648376465,
"learning_rate": 0.0001793814432989691,
"loss": 0.0863,
"step": 88
},
{
"epoch": 0.45817245817245816,
"grad_norm": 1.8875865936279297,
"learning_rate": 0.00018144329896907217,
"loss": 0.0825,
"step": 89
},
{
"epoch": 0.46332046332046334,
"grad_norm": 10.996556282043457,
"learning_rate": 0.00018350515463917526,
"loss": 0.1688,
"step": 90
},
{
"epoch": 0.46846846846846846,
"grad_norm": 12.937451362609863,
"learning_rate": 0.00018556701030927837,
"loss": 0.3039,
"step": 91
},
{
"epoch": 0.47361647361647363,
"grad_norm": 7.5320940017700195,
"learning_rate": 0.00018762886597938145,
"loss": 0.1127,
"step": 92
},
{
"epoch": 0.47876447876447875,
"grad_norm": 8.6849946975708,
"learning_rate": 0.00018969072164948454,
"loss": 0.6294,
"step": 93
},
{
"epoch": 0.4839124839124839,
"grad_norm": 9.145796775817871,
"learning_rate": 0.00019175257731958765,
"loss": 0.1106,
"step": 94
},
{
"epoch": 0.48906048906048905,
"grad_norm": 19.894432067871094,
"learning_rate": 0.00019381443298969073,
"loss": 0.5281,
"step": 95
},
{
"epoch": 0.4942084942084942,
"grad_norm": 4.2731194496154785,
"learning_rate": 0.00019587628865979381,
"loss": 0.0629,
"step": 96
},
{
"epoch": 0.49935649935649934,
"grad_norm": 1.811614751815796,
"learning_rate": 0.00019793814432989693,
"loss": 0.012,
"step": 97
},
{
"epoch": 0.5045045045045045,
"grad_norm": 14.820219993591309,
"learning_rate": 0.0002,
"loss": 0.5524,
"step": 98
},
{
"epoch": 0.5096525096525096,
"grad_norm": 10.390414237976074,
"learning_rate": 0.00019999935249847796,
"loss": 0.4835,
"step": 99
},
{
"epoch": 0.5148005148005148,
"grad_norm": 20.594295501708984,
"learning_rate": 0.00019999741000229694,
"loss": 0.4877,
"step": 100
},
{
"epoch": 0.51994851994852,
"grad_norm": 8.217048645019531,
"learning_rate": 0.00019999417253661235,
"loss": 0.3074,
"step": 101
},
{
"epoch": 0.525096525096525,
"grad_norm": 4.669956207275391,
"learning_rate": 0.00019998964014334946,
"loss": 0.178,
"step": 102
},
{
"epoch": 0.5302445302445302,
"grad_norm": 4.6300129890441895,
"learning_rate": 0.00019998381288120295,
"loss": 0.457,
"step": 103
},
{
"epoch": 0.5353925353925354,
"grad_norm": 8.305853843688965,
"learning_rate": 0.00019997669082563597,
"loss": 0.6123,
"step": 104
},
{
"epoch": 0.5405405405405406,
"grad_norm": 6.06637716293335,
"learning_rate": 0.0001999682740688794,
"loss": 0.3565,
"step": 105
},
{
"epoch": 0.5456885456885456,
"grad_norm": 2.230668306350708,
"learning_rate": 0.0001999585627199305,
"loss": 0.1088,
"step": 106
},
{
"epoch": 0.5508365508365508,
"grad_norm": 2.4483532905578613,
"learning_rate": 0.00019994755690455152,
"loss": 0.2637,
"step": 107
},
{
"epoch": 0.555984555984556,
"grad_norm": 7.294671058654785,
"learning_rate": 0.00019993525676526805,
"loss": 0.3268,
"step": 108
},
{
"epoch": 0.5611325611325612,
"grad_norm": 4.615152359008789,
"learning_rate": 0.00019992166246136738,
"loss": 0.2324,
"step": 109
},
{
"epoch": 0.5662805662805663,
"grad_norm": 6.605710983276367,
"learning_rate": 0.00019990677416889608,
"loss": 0.3835,
"step": 110
},
{
"epoch": 0.5714285714285714,
"grad_norm": 7.689223289489746,
"learning_rate": 0.000199890592080658,
"loss": 0.2309,
"step": 111
},
{
"epoch": 0.5765765765765766,
"grad_norm": 7.037056922912598,
"learning_rate": 0.00019987311640621173,
"loss": 0.2134,
"step": 112
},
{
"epoch": 0.5817245817245817,
"grad_norm": 8.55246639251709,
"learning_rate": 0.0001998543473718677,
"loss": 0.3593,
"step": 113
},
{
"epoch": 0.5868725868725869,
"grad_norm": 5.665684223175049,
"learning_rate": 0.00019983428522068556,
"loss": 0.2502,
"step": 114
},
{
"epoch": 0.592020592020592,
"grad_norm": 6.286493301391602,
"learning_rate": 0.0001998129302124707,
"loss": 0.2534,
"step": 115
},
{
"epoch": 0.5971685971685972,
"grad_norm": 11.630946159362793,
"learning_rate": 0.00019979028262377118,
"loss": 0.2662,
"step": 116
},
{
"epoch": 0.6023166023166023,
"grad_norm": 6.08538293838501,
"learning_rate": 0.00019976634274787393,
"loss": 0.322,
"step": 117
},
{
"epoch": 0.6074646074646075,
"grad_norm": 2.3434488773345947,
"learning_rate": 0.00019974111089480112,
"loss": 0.049,
"step": 118
},
{
"epoch": 0.6126126126126126,
"grad_norm": 4.001118183135986,
"learning_rate": 0.00019971458739130598,
"loss": 0.1217,
"step": 119
},
{
"epoch": 0.6177606177606177,
"grad_norm": 4.836251258850098,
"learning_rate": 0.00019968677258086866,
"loss": 0.128,
"step": 120
},
{
"epoch": 0.6229086229086229,
"grad_norm": 11.091276168823242,
"learning_rate": 0.00019965766682369186,
"loss": 0.0122,
"step": 121
},
{
"epoch": 0.6280566280566281,
"grad_norm": 6.09678840637207,
"learning_rate": 0.000199627270496696,
"loss": 0.3839,
"step": 122
},
{
"epoch": 0.6332046332046332,
"grad_norm": 4.328238487243652,
"learning_rate": 0.00019959558399351444,
"loss": 0.0516,
"step": 123
},
{
"epoch": 0.6383526383526383,
"grad_norm": 34.09149932861328,
"learning_rate": 0.00019956260772448833,
"loss": 0.4919,
"step": 124
},
{
"epoch": 0.6435006435006435,
"grad_norm": 6.8349504470825195,
"learning_rate": 0.0001995283421166614,
"loss": 0.0291,
"step": 125
},
{
"epoch": 0.6486486486486487,
"grad_norm": 8.841316223144531,
"learning_rate": 0.0001994927876137743,
"loss": 0.2145,
"step": 126
},
{
"epoch": 0.6537966537966537,
"grad_norm": 13.123899459838867,
"learning_rate": 0.00019945594467625895,
"loss": 0.5017,
"step": 127
},
{
"epoch": 0.6589446589446589,
"grad_norm": 5.323240280151367,
"learning_rate": 0.00019941781378123244,
"loss": 0.0709,
"step": 128
},
{
"epoch": 0.6640926640926641,
"grad_norm": 3.08840274810791,
"learning_rate": 0.00019937839542249108,
"loss": 0.0475,
"step": 129
},
{
"epoch": 0.6692406692406693,
"grad_norm": 0.6424345970153809,
"learning_rate": 0.0001993376901105038,
"loss": 0.0119,
"step": 130
},
{
"epoch": 0.6743886743886743,
"grad_norm": 6.012360095977783,
"learning_rate": 0.00019929569837240564,
"loss": 0.1478,
"step": 131
},
{
"epoch": 0.6795366795366795,
"grad_norm": 6.305980682373047,
"learning_rate": 0.0001992524207519909,
"loss": 0.7462,
"step": 132
},
{
"epoch": 0.6846846846846847,
"grad_norm": 3.4723167419433594,
"learning_rate": 0.00019920785780970604,
"loss": 0.0576,
"step": 133
},
{
"epoch": 0.6898326898326899,
"grad_norm": 11.569315910339355,
"learning_rate": 0.00019916201012264254,
"loss": 0.557,
"step": 134
},
{
"epoch": 0.694980694980695,
"grad_norm": 4.63154411315918,
"learning_rate": 0.00019911487828452932,
"loss": 0.0806,
"step": 135
},
{
"epoch": 0.7001287001287001,
"grad_norm": 4.816208839416504,
"learning_rate": 0.00019906646290572514,
"loss": 0.0662,
"step": 136
},
{
"epoch": 0.7052767052767053,
"grad_norm": 1.8962702751159668,
"learning_rate": 0.00019901676461321068,
"loss": 0.0352,
"step": 137
},
{
"epoch": 0.7104247104247104,
"grad_norm": 5.727291584014893,
"learning_rate": 0.00019896578405058027,
"loss": 0.3411,
"step": 138
},
{
"epoch": 0.7155727155727156,
"grad_norm": 5.64797306060791,
"learning_rate": 0.00019891352187803376,
"loss": 0.2624,
"step": 139
},
{
"epoch": 0.7207207207207207,
"grad_norm": 4.403573989868164,
"learning_rate": 0.00019885997877236788,
"loss": 0.1156,
"step": 140
},
{
"epoch": 0.7258687258687259,
"grad_norm": 0.9339085221290588,
"learning_rate": 0.0001988051554269675,
"loss": 0.0262,
"step": 141
},
{
"epoch": 0.731016731016731,
"grad_norm": 5.508331298828125,
"learning_rate": 0.0001987490525517966,
"loss": 0.3531,
"step": 142
},
{
"epoch": 0.7361647361647362,
"grad_norm": 2.545103073120117,
"learning_rate": 0.00019869167087338907,
"loss": 0.0646,
"step": 143
},
{
"epoch": 0.7413127413127413,
"grad_norm": 2.4320385456085205,
"learning_rate": 0.00019863301113483948,
"loss": 0.1268,
"step": 144
},
{
"epoch": 0.7464607464607464,
"grad_norm": 6.993696689605713,
"learning_rate": 0.00019857307409579318,
"loss": 0.2409,
"step": 145
},
{
"epoch": 0.7516087516087516,
"grad_norm": 5.1249213218688965,
"learning_rate": 0.00019851186053243666,
"loss": 0.1423,
"step": 146
},
{
"epoch": 0.7567567567567568,
"grad_norm": 6.732025623321533,
"learning_rate": 0.0001984493712374874,
"loss": 0.2505,
"step": 147
},
{
"epoch": 0.7619047619047619,
"grad_norm": 6.208662986755371,
"learning_rate": 0.00019838560702018374,
"loss": 0.4633,
"step": 148
},
{
"epoch": 0.767052767052767,
"grad_norm": 0.4718399941921234,
"learning_rate": 0.00019832056870627417,
"loss": 0.0152,
"step": 149
},
{
"epoch": 0.7722007722007722,
"grad_norm": 3.0602071285247803,
"learning_rate": 0.00019825425713800688,
"loss": 0.0912,
"step": 150
},
{
"epoch": 0.7773487773487774,
"grad_norm": 11.723753929138184,
"learning_rate": 0.00019818667317411865,
"loss": 0.255,
"step": 151
},
{
"epoch": 0.7824967824967825,
"grad_norm": 2.562699556350708,
"learning_rate": 0.0001981178176898239,
"loss": 0.0631,
"step": 152
},
{
"epoch": 0.7876447876447876,
"grad_norm": 8.836181640625,
"learning_rate": 0.00019804769157680328,
"loss": 0.2614,
"step": 153
},
{
"epoch": 0.7927927927927928,
"grad_norm": 0.6671794056892395,
"learning_rate": 0.000197976295743192,
"loss": 0.0154,
"step": 154
},
{
"epoch": 0.797940797940798,
"grad_norm": 0.9738186001777649,
"learning_rate": 0.00019790363111356837,
"loss": 0.0183,
"step": 155
},
{
"epoch": 0.803088803088803,
"grad_norm": 3.9262330532073975,
"learning_rate": 0.0001978296986289415,
"loss": 0.1642,
"step": 156
},
{
"epoch": 0.8082368082368082,
"grad_norm": 5.610588550567627,
"learning_rate": 0.00019775449924673934,
"loss": 0.1122,
"step": 157
},
{
"epoch": 0.8133848133848134,
"grad_norm": 3.3969929218292236,
"learning_rate": 0.00019767803394079615,
"loss": 0.181,
"step": 158
},
{
"epoch": 0.8185328185328186,
"grad_norm": 7.106573104858398,
"learning_rate": 0.00019760030370134004,
"loss": 0.1641,
"step": 159
},
{
"epoch": 0.8236808236808236,
"grad_norm": 0.9679182767868042,
"learning_rate": 0.0001975213095349799,
"loss": 0.0041,
"step": 160
},
{
"epoch": 0.8288288288288288,
"grad_norm": 6.026078224182129,
"learning_rate": 0.00019744105246469263,
"loss": 0.241,
"step": 161
},
{
"epoch": 0.833976833976834,
"grad_norm": 5.32731819152832,
"learning_rate": 0.0001973595335298097,
"loss": 0.0284,
"step": 162
},
{
"epoch": 0.8391248391248392,
"grad_norm": 9.328866004943848,
"learning_rate": 0.0001972767537860038,
"loss": 0.2987,
"step": 163
},
{
"epoch": 0.8442728442728443,
"grad_norm": 11.435934066772461,
"learning_rate": 0.0001971927143052752,
"loss": 0.2802,
"step": 164
},
{
"epoch": 0.8494208494208494,
"grad_norm": 10.742446899414062,
"learning_rate": 0.00019710741617593767,
"loss": 0.1801,
"step": 165
},
{
"epoch": 0.8545688545688546,
"grad_norm": 4.005350589752197,
"learning_rate": 0.00019702086050260456,
"loss": 0.1937,
"step": 166
},
{
"epoch": 0.8597168597168597,
"grad_norm": 16.507844924926758,
"learning_rate": 0.00019693304840617457,
"loss": 0.0953,
"step": 167
},
{
"epoch": 0.8648648648648649,
"grad_norm": 9.257786750793457,
"learning_rate": 0.00019684398102381694,
"loss": 0.1817,
"step": 168
},
{
"epoch": 0.87001287001287,
"grad_norm": 1.577095627784729,
"learning_rate": 0.000196753659508957,
"loss": 0.0272,
"step": 169
},
{
"epoch": 0.8751608751608752,
"grad_norm": 1.1240413188934326,
"learning_rate": 0.00019666208503126112,
"loss": 0.0201,
"step": 170
},
{
"epoch": 0.8803088803088803,
"grad_norm": 9.716466903686523,
"learning_rate": 0.0001965692587766216,
"loss": 0.442,
"step": 171
},
{
"epoch": 0.8854568854568855,
"grad_norm": 0.4750039279460907,
"learning_rate": 0.00019647518194714124,
"loss": 0.0089,
"step": 172
},
{
"epoch": 0.8906048906048906,
"grad_norm": 8.365714073181152,
"learning_rate": 0.00019637985576111778,
"loss": 0.0668,
"step": 173
},
{
"epoch": 0.8957528957528957,
"grad_norm": 11.201833724975586,
"learning_rate": 0.00019628328145302837,
"loss": 0.1195,
"step": 174
},
{
"epoch": 0.9009009009009009,
"grad_norm": 4.803361892700195,
"learning_rate": 0.00019618546027351316,
"loss": 0.0633,
"step": 175
},
{
"epoch": 0.9060489060489061,
"grad_norm": 6.3179731369018555,
"learning_rate": 0.0001960863934893594,
"loss": 0.0849,
"step": 176
},
{
"epoch": 0.9111969111969112,
"grad_norm": 5.650880813598633,
"learning_rate": 0.00019598608238348493,
"loss": 0.6128,
"step": 177
},
{
"epoch": 0.9163449163449163,
"grad_norm": 3.937826156616211,
"learning_rate": 0.0001958845282549217,
"loss": 0.065,
"step": 178
},
{
"epoch": 0.9214929214929215,
"grad_norm": 9.037993431091309,
"learning_rate": 0.00019578173241879872,
"loss": 0.1074,
"step": 179
},
{
"epoch": 0.9266409266409267,
"grad_norm": 4.078769683837891,
"learning_rate": 0.0001956776962063252,
"loss": 0.0618,
"step": 180
},
{
"epoch": 0.9317889317889317,
"grad_norm": 0.4434206783771515,
"learning_rate": 0.00019557242096477327,
"loss": 0.0082,
"step": 181
},
{
"epoch": 0.9369369369369369,
"grad_norm": 8.550223350524902,
"learning_rate": 0.00019546590805746052,
"loss": 0.2954,
"step": 182
},
{
"epoch": 0.9420849420849421,
"grad_norm": 7.798310279846191,
"learning_rate": 0.00019535815886373233,
"loss": 0.4207,
"step": 183
},
{
"epoch": 0.9472329472329473,
"grad_norm": 3.2412683963775635,
"learning_rate": 0.00019524917477894404,
"loss": 0.039,
"step": 184
},
{
"epoch": 0.9523809523809523,
"grad_norm": 10.730278015136719,
"learning_rate": 0.00019513895721444286,
"loss": 0.6668,
"step": 185
},
{
"epoch": 0.9575289575289575,
"grad_norm": 2.474336624145508,
"learning_rate": 0.00019502750759754962,
"loss": 0.0668,
"step": 186
},
{
"epoch": 0.9626769626769627,
"grad_norm": 6.537830352783203,
"learning_rate": 0.0001949148273715402,
"loss": 0.1131,
"step": 187
},
{
"epoch": 0.9678249678249679,
"grad_norm": 10.1017484664917,
"learning_rate": 0.00019480091799562704,
"loss": 0.242,
"step": 188
},
{
"epoch": 0.972972972972973,
"grad_norm": 40.51047134399414,
"learning_rate": 0.00019468578094493998,
"loss": 0.6634,
"step": 189
},
{
"epoch": 0.9781209781209781,
"grad_norm": 3.2758638858795166,
"learning_rate": 0.00019456941771050734,
"loss": 0.0417,
"step": 190
},
{
"epoch": 0.9832689832689833,
"grad_norm": 3.893338441848755,
"learning_rate": 0.00019445182979923654,
"loss": 0.5034,
"step": 191
},
{
"epoch": 0.9884169884169884,
"grad_norm": 6.016239643096924,
"learning_rate": 0.00019433301873389463,
"loss": 0.1005,
"step": 192
},
{
"epoch": 0.9935649935649936,
"grad_norm": 3.714273452758789,
"learning_rate": 0.00019421298605308847,
"loss": 0.1076,
"step": 193
},
{
"epoch": 0.9987129987129987,
"grad_norm": 7.0243000984191895,
"learning_rate": 0.000194091733311245,
"loss": 0.1508,
"step": 194
},
{
"epoch": 1.0,
"grad_norm": 5.084377765655518,
"learning_rate": 0.00019396926207859084,
"loss": 0.179,
"step": 195
},
{
"epoch": 1.005148005148005,
"grad_norm": 5.120644569396973,
"learning_rate": 0.00019384557394113228,
"loss": 0.3528,
"step": 196
},
{
"epoch": 1.0102960102960103,
"grad_norm": 2.3707666397094727,
"learning_rate": 0.00019372067050063438,
"loss": 0.0649,
"step": 197
},
{
"epoch": 1.0154440154440154,
"grad_norm": 1.5513978004455566,
"learning_rate": 0.00019359455337460054,
"loss": 0.0563,
"step": 198
},
{
"epoch": 1.0205920205920207,
"grad_norm": 3.5720956325531006,
"learning_rate": 0.00019346722419625136,
"loss": 0.0953,
"step": 199
},
{
"epoch": 1.0257400257400258,
"grad_norm": 0.8894923329353333,
"learning_rate": 0.0001933386846145036,
"loss": 0.0207,
"step": 200
},
{
"epoch": 1.0308880308880308,
"grad_norm": 1.2937897443771362,
"learning_rate": 0.00019320893629394873,
"loss": 0.024,
"step": 201
},
{
"epoch": 1.0360360360360361,
"grad_norm": 3.0240986347198486,
"learning_rate": 0.00019307798091483145,
"loss": 0.0398,
"step": 202
},
{
"epoch": 1.0411840411840412,
"grad_norm": 4.447941303253174,
"learning_rate": 0.00019294582017302797,
"loss": 0.067,
"step": 203
},
{
"epoch": 1.0463320463320462,
"grad_norm": 3.8924341201782227,
"learning_rate": 0.0001928124557800238,
"loss": 0.1048,
"step": 204
},
{
"epoch": 1.0514800514800515,
"grad_norm": 21.91023063659668,
"learning_rate": 0.000192677889462892,
"loss": 0.398,
"step": 205
},
{
"epoch": 1.0566280566280566,
"grad_norm": 10.584362983703613,
"learning_rate": 0.00019254212296427044,
"loss": 0.0987,
"step": 206
},
{
"epoch": 1.0617760617760619,
"grad_norm": 4.127809524536133,
"learning_rate": 0.00019240515804233937,
"loss": 0.293,
"step": 207
},
{
"epoch": 1.066924066924067,
"grad_norm": 6.54122257232666,
"learning_rate": 0.00019226699647079874,
"loss": 0.0313,
"step": 208
},
{
"epoch": 1.072072072072072,
"grad_norm": 1.3742715120315552,
"learning_rate": 0.0001921276400388451,
"loss": 0.0026,
"step": 209
},
{
"epoch": 1.0772200772200773,
"grad_norm": 2.312436580657959,
"learning_rate": 0.0001919870905511485,
"loss": 0.0268,
"step": 210
},
{
"epoch": 1.0823680823680824,
"grad_norm": 7.563596725463867,
"learning_rate": 0.00019184534982782904,
"loss": 0.2532,
"step": 211
},
{
"epoch": 1.0875160875160874,
"grad_norm": 6.43716287612915,
"learning_rate": 0.00019170241970443343,
"loss": 0.0559,
"step": 212
},
{
"epoch": 1.0926640926640927,
"grad_norm": 10.689008712768555,
"learning_rate": 0.00019155830203191113,
"loss": 0.1006,
"step": 213
},
{
"epoch": 1.0978120978120978,
"grad_norm": 0.15642720460891724,
"learning_rate": 0.00019141299867659036,
"loss": 0.0015,
"step": 214
},
{
"epoch": 1.102960102960103,
"grad_norm": 4.3437018394470215,
"learning_rate": 0.00019126651152015403,
"loss": 0.2429,
"step": 215
},
{
"epoch": 1.1081081081081081,
"grad_norm": 17.011978149414062,
"learning_rate": 0.00019111884245961522,
"loss": 0.7344,
"step": 216
},
{
"epoch": 1.1132561132561132,
"grad_norm": 2.5843803882598877,
"learning_rate": 0.00019096999340729282,
"loss": 0.0234,
"step": 217
},
{
"epoch": 1.1184041184041185,
"grad_norm": 495.9808349609375,
"learning_rate": 0.00019081996629078657,
"loss": 1.2841,
"step": 218
},
{
"epoch": 1.1235521235521235,
"grad_norm": 10.856950759887695,
"learning_rate": 0.00019066876305295216,
"loss": 0.1348,
"step": 219
},
{
"epoch": 1.1287001287001286,
"grad_norm": 18.06598663330078,
"learning_rate": 0.00019051638565187616,
"loss": 0.4201,
"step": 220
},
{
"epoch": 1.1338481338481339,
"grad_norm": 29.55030632019043,
"learning_rate": 0.00019036283606085053,
"loss": 0.38,
"step": 221
},
{
"epoch": 1.138996138996139,
"grad_norm": 5.556860446929932,
"learning_rate": 0.0001902081162683472,
"loss": 0.1044,
"step": 222
},
{
"epoch": 1.1441441441441442,
"grad_norm": 4.784843921661377,
"learning_rate": 0.00019005222827799212,
"loss": 0.1808,
"step": 223
},
{
"epoch": 1.1492921492921493,
"grad_norm": 8.100556373596191,
"learning_rate": 0.00018989517410853955,
"loss": 0.1281,
"step": 224
},
{
"epoch": 1.1544401544401544,
"grad_norm": 1.4319038391113281,
"learning_rate": 0.00018973695579384576,
"loss": 0.0287,
"step": 225
},
{
"epoch": 1.1595881595881596,
"grad_norm": 9.646347999572754,
"learning_rate": 0.00018957757538284273,
"loss": 0.5215,
"step": 226
},
{
"epoch": 1.1647361647361647,
"grad_norm": 8.4497652053833,
"learning_rate": 0.00018941703493951164,
"loss": 0.0688,
"step": 227
},
{
"epoch": 1.16988416988417,
"grad_norm": 4.361366271972656,
"learning_rate": 0.00018925533654285615,
"loss": 0.0714,
"step": 228
},
{
"epoch": 1.175032175032175,
"grad_norm": 1.9169917106628418,
"learning_rate": 0.00018909248228687536,
"loss": 0.0567,
"step": 229
},
{
"epoch": 1.1801801801801801,
"grad_norm": 3.0765633583068848,
"learning_rate": 0.00018892847428053693,
"loss": 0.0424,
"step": 230
},
{
"epoch": 1.1853281853281854,
"grad_norm": 0.551067590713501,
"learning_rate": 0.00018876331464774945,
"loss": 0.0153,
"step": 231
},
{
"epoch": 1.1904761904761905,
"grad_norm": 6.79862117767334,
"learning_rate": 0.00018859700552733516,
"loss": 0.1199,
"step": 232
},
{
"epoch": 1.1956241956241955,
"grad_norm": 0.6848296523094177,
"learning_rate": 0.00018842954907300236,
"loss": 0.0058,
"step": 233
},
{
"epoch": 1.2007722007722008,
"grad_norm": 1.4121878147125244,
"learning_rate": 0.00018826094745331715,
"loss": 0.0204,
"step": 234
},
{
"epoch": 1.2059202059202059,
"grad_norm": 31.389068603515625,
"learning_rate": 0.00018809120285167565,
"loss": 0.0695,
"step": 235
},
{
"epoch": 1.2110682110682112,
"grad_norm": 2.5935301780700684,
"learning_rate": 0.00018792031746627563,
"loss": 0.0301,
"step": 236
},
{
"epoch": 1.2162162162162162,
"grad_norm": 0.9850413799285889,
"learning_rate": 0.000187748293510088,
"loss": 0.0151,
"step": 237
},
{
"epoch": 1.2213642213642213,
"grad_norm": 2.5051488876342773,
"learning_rate": 0.0001875751332108283,
"loss": 0.0484,
"step": 238
},
{
"epoch": 1.2265122265122266,
"grad_norm": 0.43330660462379456,
"learning_rate": 0.0001874008388109276,
"loss": 0.0022,
"step": 239
},
{
"epoch": 1.2316602316602316,
"grad_norm": 14.615728378295898,
"learning_rate": 0.0001872254125675037,
"loss": 0.1385,
"step": 240
},
{
"epoch": 1.236808236808237,
"grad_norm": 4.9157185554504395,
"learning_rate": 0.0001870488567523318,
"loss": 0.2347,
"step": 241
},
{
"epoch": 1.241956241956242,
"grad_norm": 0.4485873579978943,
"learning_rate": 0.00018687117365181512,
"loss": 0.0024,
"step": 242
},
{
"epoch": 1.247104247104247,
"grad_norm": 6.608181476593018,
"learning_rate": 0.00018669236556695515,
"loss": 0.0602,
"step": 243
},
{
"epoch": 1.2522522522522523,
"grad_norm": 7.834930896759033,
"learning_rate": 0.00018651243481332213,
"loss": 0.4808,
"step": 244
},
{
"epoch": 1.2574002574002574,
"grad_norm": 4.652353763580322,
"learning_rate": 0.00018633138372102468,
"loss": 0.1443,
"step": 245
},
{
"epoch": 1.2625482625482625,
"grad_norm": 0.908244252204895,
"learning_rate": 0.00018614921463468002,
"loss": 0.0075,
"step": 246
},
{
"epoch": 1.2676962676962678,
"grad_norm": 4.2695536613464355,
"learning_rate": 0.00018596592991338334,
"loss": 0.0735,
"step": 247
},
{
"epoch": 1.2728442728442728,
"grad_norm": 12.879799842834473,
"learning_rate": 0.00018578153193067745,
"loss": 0.7177,
"step": 248
},
{
"epoch": 1.2779922779922779,
"grad_norm": 9.532065391540527,
"learning_rate": 0.0001855960230745217,
"loss": 0.3006,
"step": 249
},
{
"epoch": 1.2831402831402832,
"grad_norm": 6.336977958679199,
"learning_rate": 0.0001854094057472615,
"loss": 0.057,
"step": 250
},
{
"epoch": 1.2882882882882882,
"grad_norm": 4.082253456115723,
"learning_rate": 0.00018522168236559695,
"loss": 0.0638,
"step": 251
},
{
"epoch": 1.2934362934362935,
"grad_norm": 6.137638092041016,
"learning_rate": 0.00018503285536055147,
"loss": 0.0717,
"step": 252
},
{
"epoch": 1.2985842985842986,
"grad_norm": 4.905017375946045,
"learning_rate": 0.0001848429271774406,
"loss": 0.1896,
"step": 253
},
{
"epoch": 1.3037323037323039,
"grad_norm": 4.505847930908203,
"learning_rate": 0.00018465190027584005,
"loss": 0.1616,
"step": 254
},
{
"epoch": 1.308880308880309,
"grad_norm": 3.1315011978149414,
"learning_rate": 0.00018445977712955403,
"loss": 0.1974,
"step": 255
},
{
"epoch": 1.314028314028314,
"grad_norm": 2.908487319946289,
"learning_rate": 0.0001842665602265831,
"loss": 0.0241,
"step": 256
},
{
"epoch": 1.3191763191763193,
"grad_norm": 6.612605094909668,
"learning_rate": 0.00018407225206909208,
"loss": 0.2267,
"step": 257
},
{
"epoch": 1.3243243243243243,
"grad_norm": 12.149686813354492,
"learning_rate": 0.0001838768551733775,
"loss": 0.5068,
"step": 258
},
{
"epoch": 1.3294723294723294,
"grad_norm": 4.582735538482666,
"learning_rate": 0.00018368037206983515,
"loss": 0.0427,
"step": 259
},
{
"epoch": 1.3346203346203347,
"grad_norm": 2.1869008541107178,
"learning_rate": 0.00018348280530292713,
"loss": 0.0267,
"step": 260
},
{
"epoch": 1.3397683397683398,
"grad_norm": 5.434723854064941,
"learning_rate": 0.00018328415743114912,
"loss": 0.1063,
"step": 261
},
{
"epoch": 1.3449163449163448,
"grad_norm": 2.939981698989868,
"learning_rate": 0.00018308443102699713,
"loss": 0.1915,
"step": 262
},
{
"epoch": 1.35006435006435,
"grad_norm": 9.131181716918945,
"learning_rate": 0.00018288362867693414,
"loss": 0.13,
"step": 263
},
{
"epoch": 1.3552123552123552,
"grad_norm": 1.5698065757751465,
"learning_rate": 0.0001826817529813567,
"loss": 0.0232,
"step": 264
},
{
"epoch": 1.3603603603603602,
"grad_norm": 1.2605693340301514,
"learning_rate": 0.00018247880655456125,
"loss": 0.0228,
"step": 265
},
{
"epoch": 1.3655083655083655,
"grad_norm": 2.6561949253082275,
"learning_rate": 0.00018227479202471015,
"loss": 0.0501,
"step": 266
},
{
"epoch": 1.3706563706563706,
"grad_norm": 5.333164215087891,
"learning_rate": 0.00018206971203379773,
"loss": 0.1307,
"step": 267
},
{
"epoch": 1.3758043758043759,
"grad_norm": 0.7859781384468079,
"learning_rate": 0.00018186356923761626,
"loss": 0.0181,
"step": 268
},
{
"epoch": 1.380952380952381,
"grad_norm": 0.15925440192222595,
"learning_rate": 0.0001816563663057211,
"loss": 0.004,
"step": 269
},
{
"epoch": 1.3861003861003862,
"grad_norm": 5.537844181060791,
"learning_rate": 0.00018144810592139656,
"loss": 0.0719,
"step": 270
},
{
"epoch": 1.3912483912483913,
"grad_norm": 2.901510238647461,
"learning_rate": 0.00018123879078162097,
"loss": 0.0683,
"step": 271
},
{
"epoch": 1.3963963963963963,
"grad_norm": 2.2592852115631104,
"learning_rate": 0.00018102842359703176,
"loss": 0.0231,
"step": 272
},
{
"epoch": 1.4015444015444016,
"grad_norm": 5.0950798988342285,
"learning_rate": 0.0001808170070918904,
"loss": 0.3607,
"step": 273
},
{
"epoch": 1.4066924066924067,
"grad_norm": 0.20150257647037506,
"learning_rate": 0.00018060454400404695,
"loss": 0.0034,
"step": 274
},
{
"epoch": 1.4118404118404118,
"grad_norm": 11.851383209228516,
"learning_rate": 0.000180391037084905,
"loss": 0.1604,
"step": 275
},
{
"epoch": 1.416988416988417,
"grad_norm": 2.141732931137085,
"learning_rate": 0.0001801764890993856,
"loss": 0.0382,
"step": 276
},
{
"epoch": 1.422136422136422,
"grad_norm": 5.131380081176758,
"learning_rate": 0.0001799609028258917,
"loss": 0.2904,
"step": 277
},
{
"epoch": 1.4272844272844272,
"grad_norm": 5.328615188598633,
"learning_rate": 0.00017974428105627208,
"loss": 0.3075,
"step": 278
},
{
"epoch": 1.4324324324324325,
"grad_norm": 0.7733826041221619,
"learning_rate": 0.0001795266265957853,
"loss": 0.003,
"step": 279
},
{
"epoch": 1.4375804375804375,
"grad_norm": 3.2559750080108643,
"learning_rate": 0.0001793079422630632,
"loss": 0.0462,
"step": 280
},
{
"epoch": 1.4427284427284428,
"grad_norm": 7.815288066864014,
"learning_rate": 0.00017908823089007457,
"loss": 0.209,
"step": 281
},
{
"epoch": 1.4478764478764479,
"grad_norm": 9.75100040435791,
"learning_rate": 0.00017886749532208837,
"loss": 0.1576,
"step": 282
},
{
"epoch": 1.4530244530244532,
"grad_norm": 0.4934830367565155,
"learning_rate": 0.00017864573841763694,
"loss": 0.0113,
"step": 283
},
{
"epoch": 1.4581724581724582,
"grad_norm": 8.259903907775879,
"learning_rate": 0.00017842296304847893,
"loss": 0.0845,
"step": 284
},
{
"epoch": 1.4633204633204633,
"grad_norm": 1.1995900869369507,
"learning_rate": 0.00017819917209956215,
"loss": 0.005,
"step": 285
},
{
"epoch": 1.4684684684684686,
"grad_norm": 7.071656703948975,
"learning_rate": 0.00017797436846898619,
"loss": 0.1429,
"step": 286
},
{
"epoch": 1.4736164736164736,
"grad_norm": 3.0496432781219482,
"learning_rate": 0.00017774855506796496,
"loss": 0.0129,
"step": 287
},
{
"epoch": 1.4787644787644787,
"grad_norm": 13.627134323120117,
"learning_rate": 0.0001775217348207888,
"loss": 1.1236,
"step": 288
},
{
"epoch": 1.483912483912484,
"grad_norm": 7.273830890655518,
"learning_rate": 0.00017729391066478688,
"loss": 0.085,
"step": 289
},
{
"epoch": 1.489060489060489,
"grad_norm": 0.6009524464607239,
"learning_rate": 0.00017706508555028893,
"loss": 0.0052,
"step": 290
},
{
"epoch": 1.494208494208494,
"grad_norm": 13.946626663208008,
"learning_rate": 0.00017683526244058716,
"loss": 0.1716,
"step": 291
},
{
"epoch": 1.4993564993564994,
"grad_norm": 10.14864444732666,
"learning_rate": 0.0001766044443118978,
"loss": 0.0871,
"step": 292
},
{
"epoch": 1.5045045045045045,
"grad_norm": 6.030947685241699,
"learning_rate": 0.0001763726341533227,
"loss": 0.0855,
"step": 293
},
{
"epoch": 1.5096525096525095,
"grad_norm": 34.42390823364258,
"learning_rate": 0.00017613983496681044,
"loss": 0.7379,
"step": 294
},
{
"epoch": 1.5148005148005148,
"grad_norm": 1.3149932622909546,
"learning_rate": 0.00017590604976711754,
"loss": 0.0136,
"step": 295
},
{
"epoch": 1.51994851994852,
"grad_norm": 4.331245422363281,
"learning_rate": 0.00017567128158176953,
"loss": 0.1031,
"step": 296
},
{
"epoch": 1.525096525096525,
"grad_norm": 4.710329532623291,
"learning_rate": 0.00017543553345102152,
"loss": 0.1562,
"step": 297
},
{
"epoch": 1.5302445302445302,
"grad_norm": 10.789880752563477,
"learning_rate": 0.00017519880842781894,
"loss": 0.0668,
"step": 298
},
{
"epoch": 1.5353925353925355,
"grad_norm": 4.577251434326172,
"learning_rate": 0.0001749611095777581,
"loss": 0.2795,
"step": 299
},
{
"epoch": 1.5405405405405406,
"grad_norm": 3.251940965652466,
"learning_rate": 0.00017472243997904626,
"loss": 0.0476,
"step": 300
},
{
"epoch": 1.5456885456885456,
"grad_norm": 2.7281453609466553,
"learning_rate": 0.00017448280272246212,
"loss": 0.1409,
"step": 301
},
{
"epoch": 1.550836550836551,
"grad_norm": 6.675407409667969,
"learning_rate": 0.00017424220091131535,
"loss": 0.1161,
"step": 302
},
{
"epoch": 1.555984555984556,
"grad_norm": 0.6979472637176514,
"learning_rate": 0.00017400063766140678,
"loss": 0.0084,
"step": 303
},
{
"epoch": 1.561132561132561,
"grad_norm": 3.094895601272583,
"learning_rate": 0.0001737581161009878,
"loss": 0.0313,
"step": 304
},
{
"epoch": 1.5662805662805663,
"grad_norm": 7.1512861251831055,
"learning_rate": 0.00017351463937072004,
"loss": 0.2728,
"step": 305
},
{
"epoch": 1.5714285714285714,
"grad_norm": 2.706373691558838,
"learning_rate": 0.00017327021062363458,
"loss": 0.0512,
"step": 306
},
{
"epoch": 1.5765765765765765,
"grad_norm": 2.9084832668304443,
"learning_rate": 0.00017302483302509108,
"loss": 0.0743,
"step": 307
},
{
"epoch": 1.5817245817245817,
"grad_norm": 0.7898654341697693,
"learning_rate": 0.00017277850975273696,
"loss": 0.0127,
"step": 308
},
{
"epoch": 1.586872586872587,
"grad_norm": 2.7745320796966553,
"learning_rate": 0.00017253124399646606,
"loss": 0.2848,
"step": 309
},
{
"epoch": 1.5920205920205919,
"grad_norm": 0.0922975093126297,
"learning_rate": 0.00017228303895837748,
"loss": 0.0036,
"step": 310
},
{
"epoch": 1.5971685971685972,
"grad_norm": 4.047640800476074,
"learning_rate": 0.000172033897852734,
"loss": 0.0708,
"step": 311
},
{
"epoch": 1.6023166023166024,
"grad_norm": 3.7090158462524414,
"learning_rate": 0.00017178382390592057,
"loss": 0.0887,
"step": 312
},
{
"epoch": 1.6074646074646075,
"grad_norm": 4.62571382522583,
"learning_rate": 0.0001715328203564023,
"loss": 0.2746,
"step": 313
},
{
"epoch": 1.6126126126126126,
"grad_norm": 16.148685455322266,
"learning_rate": 0.00017128089045468294,
"loss": 0.1546,
"step": 314
},
{
"epoch": 1.6177606177606179,
"grad_norm": 8.929279327392578,
"learning_rate": 0.00017102803746326227,
"loss": 0.1486,
"step": 315
},
{
"epoch": 1.622908622908623,
"grad_norm": 1.5397669076919556,
"learning_rate": 0.00017077426465659433,
"loss": 0.0217,
"step": 316
},
{
"epoch": 1.628056628056628,
"grad_norm": 3.9930737018585205,
"learning_rate": 0.0001705195753210446,
"loss": 0.1983,
"step": 317
},
{
"epoch": 1.6332046332046333,
"grad_norm": 0.5112211108207703,
"learning_rate": 0.00017026397275484773,
"loss": 0.0073,
"step": 318
},
{
"epoch": 1.6383526383526383,
"grad_norm": 8.591222763061523,
"learning_rate": 0.0001700074602680648,
"loss": 0.3153,
"step": 319
},
{
"epoch": 1.6435006435006434,
"grad_norm": 3.849032163619995,
"learning_rate": 0.0001697500411825403,
"loss": 0.156,
"step": 320
},
{
"epoch": 1.6486486486486487,
"grad_norm": 3.7891454696655273,
"learning_rate": 0.00016949171883185918,
"loss": 0.2018,
"step": 321
},
{
"epoch": 1.6537966537966537,
"grad_norm": 19.799938201904297,
"learning_rate": 0.0001692324965613038,
"loss": 0.1539,
"step": 322
},
{
"epoch": 1.6589446589446588,
"grad_norm": 11.693257331848145,
"learning_rate": 0.00016897237772781044,
"loss": 0.1503,
"step": 323
},
{
"epoch": 1.664092664092664,
"grad_norm": 0.6046671271324158,
"learning_rate": 0.00016871136569992587,
"loss": 0.0164,
"step": 324
},
{
"epoch": 1.6692406692406694,
"grad_norm": 5.384612083435059,
"learning_rate": 0.00016844946385776384,
"loss": 0.2708,
"step": 325
},
{
"epoch": 1.6743886743886742,
"grad_norm": 11.719284057617188,
"learning_rate": 0.0001681866755929612,
"loss": 0.1173,
"step": 326
},
{
"epoch": 1.6795366795366795,
"grad_norm": 2.6910994052886963,
"learning_rate": 0.00016792300430863396,
"loss": 0.0728,
"step": 327
},
{
"epoch": 1.6846846846846848,
"grad_norm": 6.209327697753906,
"learning_rate": 0.0001676584534193332,
"loss": 0.0415,
"step": 328
},
{
"epoch": 1.6898326898326899,
"grad_norm": 5.639928340911865,
"learning_rate": 0.00016739302635100108,
"loss": 0.1124,
"step": 329
},
{
"epoch": 1.694980694980695,
"grad_norm": 3.1849684715270996,
"learning_rate": 0.00016712672654092622,
"loss": 0.0557,
"step": 330
},
{
"epoch": 1.7001287001287002,
"grad_norm": 4.871498107910156,
"learning_rate": 0.0001668595574376992,
"loss": 0.3408,
"step": 331
},
{
"epoch": 1.7052767052767053,
"grad_norm": 13.51830005645752,
"learning_rate": 0.00016659152250116812,
"loss": 0.1187,
"step": 332
},
{
"epoch": 1.7104247104247103,
"grad_norm": 9.41748332977295,
"learning_rate": 0.0001663226252023935,
"loss": 0.1973,
"step": 333
},
{
"epoch": 1.7155727155727156,
"grad_norm": 4.242353439331055,
"learning_rate": 0.00016605286902360357,
"loss": 0.3048,
"step": 334
},
{
"epoch": 1.7207207207207207,
"grad_norm": 10.071020126342773,
"learning_rate": 0.00016578225745814907,
"loss": 0.5854,
"step": 335
},
{
"epoch": 1.7258687258687258,
"grad_norm": 10.262407302856445,
"learning_rate": 0.000165510794010458,
"loss": 0.068,
"step": 336
},
{
"epoch": 1.731016731016731,
"grad_norm": 3.055323600769043,
"learning_rate": 0.00016523848219599023,
"loss": 0.2602,
"step": 337
},
{
"epoch": 1.7361647361647363,
"grad_norm": 6.916463851928711,
"learning_rate": 0.00016496532554119214,
"loss": 0.235,
"step": 338
},
{
"epoch": 1.7413127413127412,
"grad_norm": 4.978103160858154,
"learning_rate": 0.0001646913275834506,
"loss": 0.0821,
"step": 339
},
{
"epoch": 1.7464607464607464,
"grad_norm": 4.362355709075928,
"learning_rate": 0.00016441649187104763,
"loss": 0.0234,
"step": 340
},
{
"epoch": 1.7516087516087517,
"grad_norm": 13.96357536315918,
"learning_rate": 0.000164140821963114,
"loss": 0.1666,
"step": 341
},
{
"epoch": 1.7567567567567568,
"grad_norm": 1.1012885570526123,
"learning_rate": 0.00016386432142958342,
"loss": 0.0289,
"step": 342
},
{
"epoch": 1.7619047619047619,
"grad_norm": 4.191834449768066,
"learning_rate": 0.00016358699385114625,
"loss": 0.0573,
"step": 343
},
{
"epoch": 1.7670527670527671,
"grad_norm": 0.34135016798973083,
"learning_rate": 0.000163308842819203,
"loss": 0.0094,
"step": 344
},
{
"epoch": 1.7722007722007722,
"grad_norm": 5.802267074584961,
"learning_rate": 0.00016302987193581806,
"loss": 0.218,
"step": 345
},
{
"epoch": 1.7773487773487773,
"grad_norm": 4.298703193664551,
"learning_rate": 0.00016275008481367287,
"loss": 0.3927,
"step": 346
},
{
"epoch": 1.7824967824967826,
"grad_norm": 3.2017078399658203,
"learning_rate": 0.00016246948507601914,
"loss": 0.0525,
"step": 347
},
{
"epoch": 1.7876447876447876,
"grad_norm": 0.6790556311607361,
"learning_rate": 0.00016218807635663202,
"loss": 0.0105,
"step": 348
},
{
"epoch": 1.7927927927927927,
"grad_norm": 9.194913864135742,
"learning_rate": 0.00016190586229976304,
"loss": 0.3906,
"step": 349
},
{
"epoch": 1.797940797940798,
"grad_norm": 5.872428894042969,
"learning_rate": 0.00016162284656009274,
"loss": 0.2324,
"step": 350
},
{
"epoch": 1.803088803088803,
"grad_norm": 3.1454548835754395,
"learning_rate": 0.00016133903280268362,
"loss": 0.1475,
"step": 351
},
{
"epoch": 1.808236808236808,
"grad_norm": 3.3194046020507812,
"learning_rate": 0.0001610544247029325,
"loss": 0.1653,
"step": 352
},
{
"epoch": 1.8133848133848134,
"grad_norm": 2.6302783489227295,
"learning_rate": 0.0001607690259465229,
"loss": 0.0568,
"step": 353
},
{
"epoch": 1.8185328185328187,
"grad_norm": 0.5749773979187012,
"learning_rate": 0.0001604828402293774,
"loss": 0.0182,
"step": 354
},
{
"epoch": 1.8236808236808235,
"grad_norm": 11.607693672180176,
"learning_rate": 0.00016019587125760978,
"loss": 0.1286,
"step": 355
},
{
"epoch": 1.8288288288288288,
"grad_norm": 2.041949510574341,
"learning_rate": 0.00015990812274747692,
"loss": 0.208,
"step": 356
},
{
"epoch": 1.833976833976834,
"grad_norm": 10.237532615661621,
"learning_rate": 0.00015961959842533083,
"loss": 0.1631,
"step": 357
},
{
"epoch": 1.8391248391248392,
"grad_norm": 9.116559982299805,
"learning_rate": 0.00015933030202757022,
"loss": 0.0911,
"step": 358
},
{
"epoch": 1.8442728442728442,
"grad_norm": 6.2941203117370605,
"learning_rate": 0.00015904023730059228,
"loss": 0.2304,
"step": 359
},
{
"epoch": 1.8494208494208495,
"grad_norm": 1.2840523719787598,
"learning_rate": 0.00015874940800074402,
"loss": 0.0191,
"step": 360
},
{
"epoch": 1.8545688545688546,
"grad_norm": 1.41546630859375,
"learning_rate": 0.00015845781789427377,
"loss": 0.0073,
"step": 361
},
{
"epoch": 1.8597168597168596,
"grad_norm": 1.1912124156951904,
"learning_rate": 0.00015816547075728226,
"loss": 0.0308,
"step": 362
},
{
"epoch": 1.864864864864865,
"grad_norm": 2.443361282348633,
"learning_rate": 0.00015787237037567385,
"loss": 0.0381,
"step": 363
},
{
"epoch": 1.87001287001287,
"grad_norm": 11.012572288513184,
"learning_rate": 0.0001575785205451073,
"loss": 0.1282,
"step": 364
},
{
"epoch": 1.875160875160875,
"grad_norm": 2.7217371463775635,
"learning_rate": 0.000157283925070947,
"loss": 0.0366,
"step": 365
},
{
"epoch": 1.8803088803088803,
"grad_norm": 4.910791873931885,
"learning_rate": 0.0001569885877682132,
"loss": 0.0198,
"step": 366
},
{
"epoch": 1.8854568854568856,
"grad_norm": 9.718399047851562,
"learning_rate": 0.000156692512461533,
"loss": 0.1436,
"step": 367
},
{
"epoch": 1.8906048906048905,
"grad_norm": 8.927786827087402,
"learning_rate": 0.00015639570298509064,
"loss": 0.3022,
"step": 368
},
{
"epoch": 1.8957528957528957,
"grad_norm": 0.4057319760322571,
"learning_rate": 0.00015609816318257788,
"loss": 0.0051,
"step": 369
},
{
"epoch": 1.900900900900901,
"grad_norm": 3.1715965270996094,
"learning_rate": 0.00015579989690714423,
"loss": 0.0556,
"step": 370
},
{
"epoch": 1.906048906048906,
"grad_norm": 2.786271572113037,
"learning_rate": 0.000155500908021347,
"loss": 0.06,
"step": 371
},
{
"epoch": 1.9111969111969112,
"grad_norm": 0.1323973536491394,
"learning_rate": 0.00015520120039710139,
"loss": 0.0025,
"step": 372
},
{
"epoch": 1.9163449163449164,
"grad_norm": 1.665714979171753,
"learning_rate": 0.0001549007779156302,
"loss": 0.0251,
"step": 373
},
{
"epoch": 1.9214929214929215,
"grad_norm": 10.546355247497559,
"learning_rate": 0.00015459964446741382,
"loss": 0.3011,
"step": 374
},
{
"epoch": 1.9266409266409266,
"grad_norm": 17.591938018798828,
"learning_rate": 0.0001542978039521395,
"loss": 0.6961,
"step": 375
},
{
"epoch": 1.9317889317889319,
"grad_norm": 0.08125495910644531,
"learning_rate": 0.00015399526027865107,
"loss": 0.0018,
"step": 376
},
{
"epoch": 1.936936936936937,
"grad_norm": 4.14680290222168,
"learning_rate": 0.0001536920173648984,
"loss": 0.2298,
"step": 377
},
{
"epoch": 1.942084942084942,
"grad_norm": 3.34873104095459,
"learning_rate": 0.00015338807913788636,
"loss": 0.1337,
"step": 378
},
{
"epoch": 1.9472329472329473,
"grad_norm": 6.147327899932861,
"learning_rate": 0.0001530834495336243,
"loss": 0.2936,
"step": 379
},
{
"epoch": 1.9523809523809523,
"grad_norm": 13.515678405761719,
"learning_rate": 0.00015277813249707487,
"loss": 0.235,
"step": 380
},
{
"epoch": 1.9575289575289574,
"grad_norm": 2.5419952869415283,
"learning_rate": 0.000152472131982103,
"loss": 0.0537,
"step": 381
},
{
"epoch": 1.9626769626769627,
"grad_norm": 0.04103407263755798,
"learning_rate": 0.0001521654519514246,
"loss": 0.0013,
"step": 382
},
{
"epoch": 1.967824967824968,
"grad_norm": 0.655127227306366,
"learning_rate": 0.0001518580963765555,
"loss": 0.0136,
"step": 383
},
{
"epoch": 1.972972972972973,
"grad_norm": 1.8296195268630981,
"learning_rate": 0.00015155006923775965,
"loss": 0.0158,
"step": 384
},
{
"epoch": 1.978120978120978,
"grad_norm": 7.40383768081665,
"learning_rate": 0.00015124137452399795,
"loss": 0.4859,
"step": 385
},
{
"epoch": 1.9832689832689834,
"grad_norm": 0.8681594729423523,
"learning_rate": 0.00015093201623287631,
"loss": 0.0104,
"step": 386
},
{
"epoch": 1.9884169884169884,
"grad_norm": 15.919984817504883,
"learning_rate": 0.00015062199837059405,
"loss": 0.3156,
"step": 387
},
{
"epoch": 1.9935649935649935,
"grad_norm": 3.5267200469970703,
"learning_rate": 0.00015031132495189187,
"loss": 0.2179,
"step": 388
},
{
"epoch": 1.9987129987129988,
"grad_norm": 1.6195534467697144,
"learning_rate": 0.00015000000000000001,
"loss": 0.0381,
"step": 389
},
{
"epoch": 2.0,
"grad_norm": 0.6771867871284485,
"learning_rate": 0.00014968802754658614,
"loss": 0.0149,
"step": 390
},
{
"epoch": 2.0051480051480053,
"grad_norm": 2.9333150386810303,
"learning_rate": 0.0001493754116317029,
"loss": 0.1669,
"step": 391
},
{
"epoch": 2.01029601029601,
"grad_norm": 2.1487042903900146,
"learning_rate": 0.00014906215630373606,
"loss": 0.071,
"step": 392
},
{
"epoch": 2.0154440154440154,
"grad_norm": 1.8949365615844727,
"learning_rate": 0.00014874826561935155,
"loss": 0.0408,
"step": 393
},
{
"epoch": 2.0205920205920207,
"grad_norm": 1.5271650552749634,
"learning_rate": 0.00014843374364344333,
"loss": 0.0356,
"step": 394
},
{
"epoch": 2.0257400257400255,
"grad_norm": 2.118455410003662,
"learning_rate": 0.00014811859444908052,
"loss": 0.0673,
"step": 395
},
{
"epoch": 2.030888030888031,
"grad_norm": 3.6756184101104736,
"learning_rate": 0.0001478028221174548,
"loss": 0.0397,
"step": 396
},
{
"epoch": 2.036036036036036,
"grad_norm": 3.650200366973877,
"learning_rate": 0.00014748643073782752,
"loss": 0.0401,
"step": 397
},
{
"epoch": 2.0411840411840414,
"grad_norm": 2.751405954360962,
"learning_rate": 0.00014716942440747664,
"loss": 0.0374,
"step": 398
},
{
"epoch": 2.0463320463320462,
"grad_norm": 0.12509091198444366,
"learning_rate": 0.00014685180723164376,
"loss": 0.003,
"step": 399
},
{
"epoch": 2.0514800514800515,
"grad_norm": 1.4593161344528198,
"learning_rate": 0.000146533583323481,
"loss": 0.032,
"step": 400
},
{
"epoch": 2.056628056628057,
"grad_norm": 0.3055787980556488,
"learning_rate": 0.0001462147568039977,
"loss": 0.0077,
"step": 401
},
{
"epoch": 2.0617760617760617,
"grad_norm": 4.0168352127075195,
"learning_rate": 0.00014589533180200693,
"loss": 0.0272,
"step": 402
},
{
"epoch": 2.066924066924067,
"grad_norm": 0.4300869405269623,
"learning_rate": 0.00014557531245407225,
"loss": 0.0072,
"step": 403
},
{
"epoch": 2.0720720720720722,
"grad_norm": 0.11563416570425034,
"learning_rate": 0.00014525470290445392,
"loss": 0.0026,
"step": 404
},
{
"epoch": 2.077220077220077,
"grad_norm": 2.358717441558838,
"learning_rate": 0.00014493350730505533,
"loss": 0.0323,
"step": 405
},
{
"epoch": 2.0823680823680824,
"grad_norm": 4.539452075958252,
"learning_rate": 0.0001446117298153693,
"loss": 0.1005,
"step": 406
},
{
"epoch": 2.0875160875160876,
"grad_norm": 6.308579444885254,
"learning_rate": 0.00014428937460242417,
"loss": 0.1177,
"step": 407
},
{
"epoch": 2.0926640926640925,
"grad_norm": 1.6475294828414917,
"learning_rate": 0.00014396644584072972,
"loss": 0.0063,
"step": 408
},
{
"epoch": 2.0978120978120978,
"grad_norm": 3.9539501667022705,
"learning_rate": 0.00014364294771222324,
"loss": 0.0212,
"step": 409
},
{
"epoch": 2.102960102960103,
"grad_norm": 0.051111120730638504,
"learning_rate": 0.00014331888440621533,
"loss": 0.0013,
"step": 410
},
{
"epoch": 2.108108108108108,
"grad_norm": 0.2948058843612671,
"learning_rate": 0.00014299426011933568,
"loss": 0.0011,
"step": 411
},
{
"epoch": 2.113256113256113,
"grad_norm": 2.541996717453003,
"learning_rate": 0.0001426690790554787,
"loss": 0.0368,
"step": 412
},
{
"epoch": 2.1184041184041185,
"grad_norm": 13.898029327392578,
"learning_rate": 0.00014234334542574906,
"loss": 0.0538,
"step": 413
},
{
"epoch": 2.1235521235521237,
"grad_norm": 3.1187541484832764,
"learning_rate": 0.00014201706344840712,
"loss": 0.0364,
"step": 414
},
{
"epoch": 2.1287001287001286,
"grad_norm": 1.4270102977752686,
"learning_rate": 0.00014169023734881452,
"loss": 0.0077,
"step": 415
},
{
"epoch": 2.133848133848134,
"grad_norm": 0.07322893291711807,
"learning_rate": 0.00014136287135937915,
"loss": 0.0005,
"step": 416
},
{
"epoch": 2.138996138996139,
"grad_norm": 1.276623249053955,
"learning_rate": 0.00014103496971950053,
"loss": 0.0087,
"step": 417
},
{
"epoch": 2.144144144144144,
"grad_norm": 0.3112569749355316,
"learning_rate": 0.0001407065366755149,
"loss": 0.0013,
"step": 418
},
{
"epoch": 2.1492921492921493,
"grad_norm": 20.969881057739258,
"learning_rate": 0.00014037757648064018,
"loss": 0.5316,
"step": 419
},
{
"epoch": 2.1544401544401546,
"grad_norm": 41.78050994873047,
"learning_rate": 0.00014004809339492088,
"loss": 0.3077,
"step": 420
},
{
"epoch": 2.1595881595881594,
"grad_norm": 23.550018310546875,
"learning_rate": 0.00013971809168517298,
"loss": 0.0351,
"step": 421
},
{
"epoch": 2.1647361647361647,
"grad_norm": 0.4678838849067688,
"learning_rate": 0.00013938757562492873,
"loss": 0.0033,
"step": 422
},
{
"epoch": 2.16988416988417,
"grad_norm": 10.164952278137207,
"learning_rate": 0.0001390565494943811,
"loss": 0.0418,
"step": 423
},
{
"epoch": 2.175032175032175,
"grad_norm": 0.06984155625104904,
"learning_rate": 0.00013872501758032863,
"loss": 0.0009,
"step": 424
},
{
"epoch": 2.18018018018018,
"grad_norm": 0.4608617126941681,
"learning_rate": 0.00013839298417611963,
"loss": 0.0008,
"step": 425
},
{
"epoch": 2.1853281853281854,
"grad_norm": 19.189315795898438,
"learning_rate": 0.00013806045358159683,
"loss": 0.212,
"step": 426
},
{
"epoch": 2.1904761904761907,
"grad_norm": 0.07832283526659012,
"learning_rate": 0.00013772743010304154,
"loss": 0.0009,
"step": 427
},
{
"epoch": 2.1956241956241955,
"grad_norm": 0.24573272466659546,
"learning_rate": 0.00013739391805311793,
"loss": 0.0011,
"step": 428
},
{
"epoch": 2.200772200772201,
"grad_norm": 16.95315933227539,
"learning_rate": 0.00013705992175081728,
"loss": 0.1014,
"step": 429
},
{
"epoch": 2.205920205920206,
"grad_norm": 1.977837085723877,
"learning_rate": 0.00013672544552140176,
"loss": 0.0201,
"step": 430
},
{
"epoch": 2.211068211068211,
"grad_norm": 0.06078295037150383,
"learning_rate": 0.00013639049369634876,
"loss": 0.0004,
"step": 431
},
{
"epoch": 2.2162162162162162,
"grad_norm": 0.7487907409667969,
"learning_rate": 0.00013605507061329464,
"loss": 0.0011,
"step": 432
},
{
"epoch": 2.2213642213642215,
"grad_norm": 0.4101274311542511,
"learning_rate": 0.0001357191806159785,
"loss": 0.0005,
"step": 433
},
{
"epoch": 2.2265122265122264,
"grad_norm": 3.141564130783081,
"learning_rate": 0.0001353828280541861,
"loss": 0.008,
"step": 434
},
{
"epoch": 2.2316602316602316,
"grad_norm": 1.0309710502624512,
"learning_rate": 0.00013504601728369327,
"loss": 0.0065,
"step": 435
},
{
"epoch": 2.236808236808237,
"grad_norm": 0.02686382830142975,
"learning_rate": 0.00013470875266620978,
"loss": 0.0004,
"step": 436
},
{
"epoch": 2.2419562419562418,
"grad_norm": 2.6512622833251953,
"learning_rate": 0.00013437103856932264,
"loss": 0.0708,
"step": 437
},
{
"epoch": 2.247104247104247,
"grad_norm": 0.5304256081581116,
"learning_rate": 0.00013403287936643977,
"loss": 0.0031,
"step": 438
},
{
"epoch": 2.2522522522522523,
"grad_norm": 0.016308331862092018,
"learning_rate": 0.00013369427943673312,
"loss": 0.0002,
"step": 439
},
{
"epoch": 2.257400257400257,
"grad_norm": 13.089827537536621,
"learning_rate": 0.00013335524316508208,
"loss": 0.0738,
"step": 440
},
{
"epoch": 2.2625482625482625,
"grad_norm": 5.165826797485352,
"learning_rate": 0.00013301577494201664,
"loss": 0.0603,
"step": 441
},
{
"epoch": 2.2676962676962678,
"grad_norm": 6.9744768142700195,
"learning_rate": 0.0001326758791636607,
"loss": 0.2926,
"step": 442
},
{
"epoch": 2.2728442728442726,
"grad_norm": 2.251573085784912,
"learning_rate": 0.00013233556023167485,
"loss": 0.0886,
"step": 443
},
{
"epoch": 2.277992277992278,
"grad_norm": 0.597822904586792,
"learning_rate": 0.0001319948225531997,
"loss": 0.0075,
"step": 444
},
{
"epoch": 2.283140283140283,
"grad_norm": 14.462504386901855,
"learning_rate": 0.0001316536705407985,
"loss": 0.094,
"step": 445
},
{
"epoch": 2.2882882882882885,
"grad_norm": 8.991734504699707,
"learning_rate": 0.00013131210861240026,
"loss": 0.1248,
"step": 446
},
{
"epoch": 2.2934362934362933,
"grad_norm": 1.5384705066680908,
"learning_rate": 0.0001309701411912423,
"loss": 0.007,
"step": 447
},
{
"epoch": 2.2985842985842986,
"grad_norm": 0.03640247881412506,
"learning_rate": 0.00013062777270581312,
"loss": 0.0004,
"step": 448
},
{
"epoch": 2.303732303732304,
"grad_norm": 4.42478609085083,
"learning_rate": 0.00013028500758979506,
"loss": 0.0175,
"step": 449
},
{
"epoch": 2.3088803088803087,
"grad_norm": 2.3137266635894775,
"learning_rate": 0.00012994185028200684,
"loss": 0.0131,
"step": 450
},
{
"epoch": 2.314028314028314,
"grad_norm": 18.25682258605957,
"learning_rate": 0.00012959830522634596,
"loss": 0.2982,
"step": 451
},
{
"epoch": 2.3191763191763193,
"grad_norm": 4.840076446533203,
"learning_rate": 0.00012925437687173142,
"loss": 0.086,
"step": 452
},
{
"epoch": 2.3243243243243246,
"grad_norm": 4.567030429840088,
"learning_rate": 0.00012891006967204584,
"loss": 0.0463,
"step": 453
},
{
"epoch": 2.3294723294723294,
"grad_norm": 1.4755933284759521,
"learning_rate": 0.00012856538808607795,
"loss": 0.0101,
"step": 454
},
{
"epoch": 2.3346203346203347,
"grad_norm": 26.35512351989746,
"learning_rate": 0.00012822033657746478,
"loss": 0.0201,
"step": 455
},
{
"epoch": 2.33976833976834,
"grad_norm": 0.9548275470733643,
"learning_rate": 0.0001278749196146339,
"loss": 0.0154,
"step": 456
},
{
"epoch": 2.344916344916345,
"grad_norm": 0.14387382566928864,
"learning_rate": 0.0001275291416707454,
"loss": 0.0021,
"step": 457
},
{
"epoch": 2.35006435006435,
"grad_norm": 2.819815158843994,
"learning_rate": 0.0001271830072236343,
"loss": 0.0068,
"step": 458
},
{
"epoch": 2.3552123552123554,
"grad_norm": 18.71172523498535,
"learning_rate": 0.00012683652075575218,
"loss": 0.3091,
"step": 459
},
{
"epoch": 2.3603603603603602,
"grad_norm": 0.10403964668512344,
"learning_rate": 0.00012648968675410928,
"loss": 0.0011,
"step": 460
},
{
"epoch": 2.3655083655083655,
"grad_norm": 0.054480861872434616,
"learning_rate": 0.00012614250971021657,
"loss": 0.0009,
"step": 461
},
{
"epoch": 2.370656370656371,
"grad_norm": 11.641414642333984,
"learning_rate": 0.0001257949941200273,
"loss": 0.261,
"step": 462
},
{
"epoch": 2.3758043758043756,
"grad_norm": 9.588177680969238,
"learning_rate": 0.00012544714448387893,
"loss": 0.0243,
"step": 463
},
{
"epoch": 2.380952380952381,
"grad_norm": 8.373116493225098,
"learning_rate": 0.00012509896530643488,
"loss": 0.2887,
"step": 464
},
{
"epoch": 2.386100386100386,
"grad_norm": 0.10668466985225677,
"learning_rate": 0.00012475046109662605,
"loss": 0.0014,
"step": 465
},
{
"epoch": 2.391248391248391,
"grad_norm": 0.028492266312241554,
"learning_rate": 0.0001244016363675926,
"loss": 0.0004,
"step": 466
},
{
"epoch": 2.3963963963963963,
"grad_norm": 1.3334189653396606,
"learning_rate": 0.00012405249563662537,
"loss": 0.0147,
"step": 467
},
{
"epoch": 2.4015444015444016,
"grad_norm": 2.538148880004883,
"learning_rate": 0.00012370304342510747,
"loss": 0.1013,
"step": 468
},
{
"epoch": 2.4066924066924065,
"grad_norm": 13.441120147705078,
"learning_rate": 0.00012335328425845565,
"loss": 0.1418,
"step": 469
},
{
"epoch": 2.4118404118404118,
"grad_norm": 4.2900004386901855,
"learning_rate": 0.00012300322266606178,
"loss": 0.0108,
"step": 470
},
{
"epoch": 2.416988416988417,
"grad_norm": 6.078284740447998,
"learning_rate": 0.00012265286318123415,
"loss": 0.0329,
"step": 471
},
{
"epoch": 2.4221364221364223,
"grad_norm": 0.06496753543615341,
"learning_rate": 0.00012230221034113874,
"loss": 0.0011,
"step": 472
},
{
"epoch": 2.427284427284427,
"grad_norm": 7.0481462478637695,
"learning_rate": 0.00012195126868674051,
"loss": 0.3435,
"step": 473
},
{
"epoch": 2.4324324324324325,
"grad_norm": 3.8558356761932373,
"learning_rate": 0.00012160004276274453,
"loss": 0.0179,
"step": 474
},
{
"epoch": 2.4375804375804377,
"grad_norm": 0.058779384940862656,
"learning_rate": 0.00012124853711753727,
"loss": 0.0006,
"step": 475
},
{
"epoch": 2.4427284427284426,
"grad_norm": 3.789362668991089,
"learning_rate": 0.00012089675630312754,
"loss": 0.0516,
"step": 476
},
{
"epoch": 2.447876447876448,
"grad_norm": 2.298474073410034,
"learning_rate": 0.00012054470487508751,
"loss": 0.0111,
"step": 477
},
{
"epoch": 2.453024453024453,
"grad_norm": 9.461421012878418,
"learning_rate": 0.00012019238739249397,
"loss": 0.1299,
"step": 478
},
{
"epoch": 2.458172458172458,
"grad_norm": 4.619925022125244,
"learning_rate": 0.000119839808417869,
"loss": 0.0581,
"step": 479
},
{
"epoch": 2.4633204633204633,
"grad_norm": 5.870488166809082,
"learning_rate": 0.00011948697251712109,
"loss": 0.0322,
"step": 480
},
{
"epoch": 2.4684684684684686,
"grad_norm": 1.0087398290634155,
"learning_rate": 0.00011913388425948584,
"loss": 0.0132,
"step": 481
},
{
"epoch": 2.473616473616474,
"grad_norm": 2.685680627822876,
"learning_rate": 0.00011878054821746703,
"loss": 0.1434,
"step": 482
},
{
"epoch": 2.4787644787644787,
"grad_norm": 0.27538082003593445,
"learning_rate": 0.00011842696896677708,
"loss": 0.0038,
"step": 483
},
{
"epoch": 2.483912483912484,
"grad_norm": 0.040867727249860764,
"learning_rate": 0.00011807315108627806,
"loss": 0.0011,
"step": 484
},
{
"epoch": 2.4890604890604893,
"grad_norm": 0.03485007584095001,
"learning_rate": 0.0001177190991579223,
"loss": 0.0007,
"step": 485
},
{
"epoch": 2.494208494208494,
"grad_norm": 0.31549927592277527,
"learning_rate": 0.00011736481776669306,
"loss": 0.0049,
"step": 486
},
{
"epoch": 2.4993564993564994,
"grad_norm": 0.24435648322105408,
"learning_rate": 0.0001170103115005451,
"loss": 0.0022,
"step": 487
},
{
"epoch": 2.5045045045045047,
"grad_norm": 1.7961950302124023,
"learning_rate": 0.00011665558495034546,
"loss": 0.005,
"step": 488
},
{
"epoch": 2.5096525096525095,
"grad_norm": 5.136146545410156,
"learning_rate": 0.00011630064270981367,
"loss": 0.088,
"step": 489
},
{
"epoch": 2.514800514800515,
"grad_norm": 0.09568271785974503,
"learning_rate": 0.00011594548937546257,
"loss": 0.0008,
"step": 490
},
{
"epoch": 2.51994851994852,
"grad_norm": 0.07395853102207184,
"learning_rate": 0.00011559012954653865,
"loss": 0.0017,
"step": 491
},
{
"epoch": 2.525096525096525,
"grad_norm": 0.31645476818084717,
"learning_rate": 0.00011523456782496253,
"loss": 0.0039,
"step": 492
},
{
"epoch": 2.53024453024453,
"grad_norm": 0.14563876390457153,
"learning_rate": 0.00011487880881526934,
"loss": 0.001,
"step": 493
},
{
"epoch": 2.5353925353925355,
"grad_norm": 44.37266159057617,
"learning_rate": 0.00011452285712454904,
"loss": 0.0458,
"step": 494
},
{
"epoch": 2.5405405405405403,
"grad_norm": 2.7936692237854004,
"learning_rate": 0.00011416671736238689,
"loss": 0.0091,
"step": 495
},
{
"epoch": 2.5456885456885456,
"grad_norm": 0.1316785365343094,
"learning_rate": 0.00011381039414080365,
"loss": 0.0022,
"step": 496
},
{
"epoch": 2.550836550836551,
"grad_norm": 0.04823737591505051,
"learning_rate": 0.00011345389207419588,
"loss": 0.0008,
"step": 497
},
{
"epoch": 2.5559845559845558,
"grad_norm": 4.449501037597656,
"learning_rate": 0.00011309721577927619,
"loss": 0.0082,
"step": 498
},
{
"epoch": 2.561132561132561,
"grad_norm": 0.2923971712589264,
"learning_rate": 0.00011274036987501348,
"loss": 0.0022,
"step": 499
},
{
"epoch": 2.5662805662805663,
"grad_norm": 43.90069580078125,
"learning_rate": 0.00011238335898257304,
"loss": 0.2923,
"step": 500
},
{
"epoch": 2.571428571428571,
"grad_norm": 0.03237447142601013,
"learning_rate": 0.0001120261877252568,
"loss": 0.0004,
"step": 501
},
{
"epoch": 2.5765765765765765,
"grad_norm": 7.340906143188477,
"learning_rate": 0.00011166886072844342,
"loss": 0.0238,
"step": 502
},
{
"epoch": 2.5817245817245817,
"grad_norm": 8.266247749328613,
"learning_rate": 0.00011131138261952845,
"loss": 0.2048,
"step": 503
},
{
"epoch": 2.586872586872587,
"grad_norm": 3.1179869174957275,
"learning_rate": 0.00011095375802786419,
"loss": 0.3071,
"step": 504
},
{
"epoch": 2.592020592020592,
"grad_norm": 0.23018474876880646,
"learning_rate": 0.00011059599158470002,
"loss": 0.0006,
"step": 505
},
{
"epoch": 2.597168597168597,
"grad_norm": 2.5822629928588867,
"learning_rate": 0.00011023808792312227,
"loss": 0.017,
"step": 506
},
{
"epoch": 2.6023166023166024,
"grad_norm": 5.574324131011963,
"learning_rate": 0.00010988005167799427,
"loss": 0.2319,
"step": 507
},
{
"epoch": 2.6074646074646077,
"grad_norm": 17.87259292602539,
"learning_rate": 0.00010952188748589625,
"loss": 0.3552,
"step": 508
},
{
"epoch": 2.6126126126126126,
"grad_norm": 1.4395756721496582,
"learning_rate": 0.0001091635999850655,
"loss": 0.0075,
"step": 509
},
{
"epoch": 2.617760617760618,
"grad_norm": 0.19914308190345764,
"learning_rate": 0.00010880519381533592,
"loss": 0.002,
"step": 510
},
{
"epoch": 2.622908622908623,
"grad_norm": 27.536880493164062,
"learning_rate": 0.00010844667361807842,
"loss": 0.3869,
"step": 511
},
{
"epoch": 2.628056628056628,
"grad_norm": 1.0540646314620972,
"learning_rate": 0.00010808804403614043,
"loss": 0.0137,
"step": 512
},
{
"epoch": 2.6332046332046333,
"grad_norm": 2.812091112136841,
"learning_rate": 0.00010772930971378596,
"loss": 0.0488,
"step": 513
},
{
"epoch": 2.6383526383526386,
"grad_norm": 0.25239816308021545,
"learning_rate": 0.00010737047529663545,
"loss": 0.0029,
"step": 514
},
{
"epoch": 2.6435006435006434,
"grad_norm": 6.321053504943848,
"learning_rate": 0.00010701154543160541,
"loss": 0.0588,
"step": 515
},
{
"epoch": 2.6486486486486487,
"grad_norm": 1.5410410165786743,
"learning_rate": 0.00010665252476684864,
"loss": 0.0209,
"step": 516
},
{
"epoch": 2.653796653796654,
"grad_norm": 3.073268175125122,
"learning_rate": 0.0001062934179516936,
"loss": 0.0205,
"step": 517
},
{
"epoch": 2.658944658944659,
"grad_norm": 0.13594655692577362,
"learning_rate": 0.00010593422963658452,
"loss": 0.0023,
"step": 518
},
{
"epoch": 2.664092664092664,
"grad_norm": 3.154902458190918,
"learning_rate": 0.00010557496447302102,
"loss": 0.0198,
"step": 519
},
{
"epoch": 2.6692406692406694,
"grad_norm": 10.17590045928955,
"learning_rate": 0.00010521562711349788,
"loss": 0.2177,
"step": 520
},
{
"epoch": 2.6743886743886742,
"grad_norm": 2.1345174312591553,
"learning_rate": 0.00010485622221144484,
"loss": 0.0201,
"step": 521
},
{
"epoch": 2.6795366795366795,
"grad_norm": 8.811882972717285,
"learning_rate": 0.00010449675442116634,
"loss": 0.042,
"step": 522
},
{
"epoch": 2.684684684684685,
"grad_norm": 0.15440958738327026,
"learning_rate": 0.00010413722839778117,
"loss": 0.0012,
"step": 523
},
{
"epoch": 2.6898326898326896,
"grad_norm": 1.2474502325057983,
"learning_rate": 0.00010377764879716234,
"loss": 0.0057,
"step": 524
},
{
"epoch": 2.694980694980695,
"grad_norm": 3.4297499656677246,
"learning_rate": 0.00010341802027587659,
"loss": 0.0651,
"step": 525
},
{
"epoch": 2.7001287001287,
"grad_norm": 3.5871169567108154,
"learning_rate": 0.00010305834749112421,
"loss": 0.189,
"step": 526
},
{
"epoch": 2.705276705276705,
"grad_norm": 26.077857971191406,
"learning_rate": 0.00010269863510067872,
"loss": 0.2009,
"step": 527
},
{
"epoch": 2.7104247104247103,
"grad_norm": 0.06693233549594879,
"learning_rate": 0.00010233888776282649,
"loss": 0.0015,
"step": 528
},
{
"epoch": 2.7155727155727156,
"grad_norm": 6.119448184967041,
"learning_rate": 0.00010197911013630659,
"loss": 0.2232,
"step": 529
},
{
"epoch": 2.7207207207207205,
"grad_norm": 0.041049547493457794,
"learning_rate": 0.00010161930688025017,
"loss": 0.0009,
"step": 530
},
{
"epoch": 2.7258687258687258,
"grad_norm": 1.616286277770996,
"learning_rate": 0.00010125948265412033,
"loss": 0.1225,
"step": 531
},
{
"epoch": 2.731016731016731,
"grad_norm": 0.491202712059021,
"learning_rate": 0.0001008996421176518,
"loss": 0.0057,
"step": 532
},
{
"epoch": 2.7361647361647363,
"grad_norm": 0.2402859777212143,
"learning_rate": 0.00010053978993079045,
"loss": 0.0032,
"step": 533
},
{
"epoch": 2.741312741312741,
"grad_norm": 2.665170669555664,
"learning_rate": 0.00010017993075363305,
"loss": 0.0086,
"step": 534
},
{
"epoch": 2.7464607464607464,
"grad_norm": 8.421011924743652,
"learning_rate": 9.982006924636697e-05,
"loss": 0.0731,
"step": 535
},
{
"epoch": 2.7516087516087517,
"grad_norm": 20.673776626586914,
"learning_rate": 9.946021006920959e-05,
"loss": 0.0772,
"step": 536
},
{
"epoch": 2.756756756756757,
"grad_norm": 21.537185668945312,
"learning_rate": 9.910035788234822e-05,
"loss": 0.1713,
"step": 537
},
{
"epoch": 2.761904761904762,
"grad_norm": 2.9723384380340576,
"learning_rate": 9.874051734587968e-05,
"loss": 0.0118,
"step": 538
},
{
"epoch": 2.767052767052767,
"grad_norm": 6.648712635040283,
"learning_rate": 9.838069311974986e-05,
"loss": 0.0933,
"step": 539
},
{
"epoch": 2.7722007722007724,
"grad_norm": 14.152227401733398,
"learning_rate": 9.802088986369342e-05,
"loss": 0.2614,
"step": 540
},
{
"epoch": 2.7773487773487773,
"grad_norm": 16.849931716918945,
"learning_rate": 9.766111223717352e-05,
"loss": 0.0342,
"step": 541
},
{
"epoch": 2.7824967824967826,
"grad_norm": 1.2862924337387085,
"learning_rate": 9.730136489932133e-05,
"loss": 0.0128,
"step": 542
},
{
"epoch": 2.787644787644788,
"grad_norm": 6.020199775695801,
"learning_rate": 9.694165250887584e-05,
"loss": 0.0939,
"step": 543
},
{
"epoch": 2.7927927927927927,
"grad_norm": 0.8993662595748901,
"learning_rate": 9.658197972412345e-05,
"loss": 0.0114,
"step": 544
},
{
"epoch": 2.797940797940798,
"grad_norm": 0.16036638617515564,
"learning_rate": 9.622235120283769e-05,
"loss": 0.0036,
"step": 545
},
{
"epoch": 2.8030888030888033,
"grad_norm": 2.7636218070983887,
"learning_rate": 9.586277160221884e-05,
"loss": 0.0289,
"step": 546
},
{
"epoch": 2.808236808236808,
"grad_norm": 2.3879454135894775,
"learning_rate": 9.550324557883373e-05,
"loss": 0.2184,
"step": 547
},
{
"epoch": 2.8133848133848134,
"grad_norm": 20.0377140045166,
"learning_rate": 9.514377778855521e-05,
"loss": 0.2386,
"step": 548
},
{
"epoch": 2.8185328185328187,
"grad_norm": 0.06437746435403824,
"learning_rate": 9.478437288650213e-05,
"loss": 0.0013,
"step": 549
},
{
"epoch": 2.8236808236808235,
"grad_norm": 1.1421157121658325,
"learning_rate": 9.442503552697899e-05,
"loss": 0.0091,
"step": 550
},
{
"epoch": 2.828828828828829,
"grad_norm": 0.4120176136493683,
"learning_rate": 9.406577036341548e-05,
"loss": 0.0069,
"step": 551
},
{
"epoch": 2.833976833976834,
"grad_norm": 0.2629767060279846,
"learning_rate": 9.37065820483064e-05,
"loss": 0.003,
"step": 552
},
{
"epoch": 2.839124839124839,
"grad_norm": 1.233478307723999,
"learning_rate": 9.334747523315137e-05,
"loss": 0.012,
"step": 553
},
{
"epoch": 2.844272844272844,
"grad_norm": 0.2161356806755066,
"learning_rate": 9.298845456839459e-05,
"loss": 0.0019,
"step": 554
},
{
"epoch": 2.8494208494208495,
"grad_norm": 1.3832241296768188,
"learning_rate": 9.262952470336458e-05,
"loss": 0.0034,
"step": 555
},
{
"epoch": 2.8545688545688543,
"grad_norm": 12.501529693603516,
"learning_rate": 9.227069028621406e-05,
"loss": 0.1933,
"step": 556
},
{
"epoch": 2.8597168597168596,
"grad_norm": 16.72063446044922,
"learning_rate": 9.19119559638596e-05,
"loss": 0.264,
"step": 557
},
{
"epoch": 2.864864864864865,
"grad_norm": 7.742213726043701,
"learning_rate": 9.15533263819216e-05,
"loss": 0.1824,
"step": 558
},
{
"epoch": 2.8700128700128698,
"grad_norm": 0.7183289527893066,
"learning_rate": 9.119480618466409e-05,
"loss": 0.0063,
"step": 559
},
{
"epoch": 2.875160875160875,
"grad_norm": 27.059951782226562,
"learning_rate": 9.083640001493454e-05,
"loss": 0.0794,
"step": 560
},
{
"epoch": 2.8803088803088803,
"grad_norm": 1.6208770275115967,
"learning_rate": 9.047811251410376e-05,
"loss": 0.0223,
"step": 561
},
{
"epoch": 2.8854568854568856,
"grad_norm": 2.594089984893799,
"learning_rate": 9.011994832200577e-05,
"loss": 0.0054,
"step": 562
},
{
"epoch": 2.8906048906048905,
"grad_norm": 1.6398696899414062,
"learning_rate": 8.976191207687775e-05,
"loss": 0.0144,
"step": 563
},
{
"epoch": 2.8957528957528957,
"grad_norm": 3.3332631587982178,
"learning_rate": 8.94040084153e-05,
"loss": 0.0267,
"step": 564
},
{
"epoch": 2.900900900900901,
"grad_norm": 7.854773044586182,
"learning_rate": 8.904624197213585e-05,
"loss": 0.2081,
"step": 565
},
{
"epoch": 2.9060489060489063,
"grad_norm": 56.386634826660156,
"learning_rate": 8.868861738047158e-05,
"loss": 0.0278,
"step": 566
},
{
"epoch": 2.911196911196911,
"grad_norm": 9.046952247619629,
"learning_rate": 8.83311392715566e-05,
"loss": 0.0653,
"step": 567
},
{
"epoch": 2.9163449163449164,
"grad_norm": 0.20892998576164246,
"learning_rate": 8.797381227474324e-05,
"loss": 0.0032,
"step": 568
},
{
"epoch": 2.9214929214929217,
"grad_norm": 8.924886703491211,
"learning_rate": 8.7616641017427e-05,
"loss": 0.0671,
"step": 569
},
{
"epoch": 2.9266409266409266,
"grad_norm": 5.492880344390869,
"learning_rate": 8.725963012498657e-05,
"loss": 0.0541,
"step": 570
},
{
"epoch": 2.931788931788932,
"grad_norm": 2.715017080307007,
"learning_rate": 8.690278422072384e-05,
"loss": 0.4233,
"step": 571
},
{
"epoch": 2.936936936936937,
"grad_norm": 7.928826808929443,
"learning_rate": 8.654610792580415e-05,
"loss": 0.1993,
"step": 572
},
{
"epoch": 2.942084942084942,
"grad_norm": 17.035512924194336,
"learning_rate": 8.61896058591964e-05,
"loss": 0.1885,
"step": 573
},
{
"epoch": 2.9472329472329473,
"grad_norm": 10.870099067687988,
"learning_rate": 8.583328263761316e-05,
"loss": 0.0986,
"step": 574
},
{
"epoch": 2.9523809523809526,
"grad_norm": 11.231427192687988,
"learning_rate": 8.5477142875451e-05,
"loss": 0.3016,
"step": 575
},
{
"epoch": 2.9575289575289574,
"grad_norm": 7.302216529846191,
"learning_rate": 8.512119118473067e-05,
"loss": 0.0982,
"step": 576
},
{
"epoch": 2.9626769626769627,
"grad_norm": 1.046059489250183,
"learning_rate": 8.476543217503748e-05,
"loss": 0.0185,
"step": 577
},
{
"epoch": 2.967824967824968,
"grad_norm": 16.43110466003418,
"learning_rate": 8.440987045346134e-05,
"loss": 0.2093,
"step": 578
},
{
"epoch": 2.972972972972973,
"grad_norm": 0.18812815845012665,
"learning_rate": 8.405451062453744e-05,
"loss": 0.0027,
"step": 579
},
{
"epoch": 2.978120978120978,
"grad_norm": 2.94497013092041,
"learning_rate": 8.369935729018634e-05,
"loss": 0.0373,
"step": 580
},
{
"epoch": 2.9832689832689834,
"grad_norm": 0.6555696129798889,
"learning_rate": 8.334441504965455e-05,
"loss": 0.0089,
"step": 581
},
{
"epoch": 2.988416988416988,
"grad_norm": 1.5896837711334229,
"learning_rate": 8.29896884994549e-05,
"loss": 0.0484,
"step": 582
},
{
"epoch": 2.9935649935649935,
"grad_norm": 1.846440076828003,
"learning_rate": 8.263518223330697e-05,
"loss": 0.0245,
"step": 583
},
{
"epoch": 2.998712998712999,
"grad_norm": 0.767316997051239,
"learning_rate": 8.228090084207774e-05,
"loss": 0.0082,
"step": 584
},
{
"epoch": 3.0,
"grad_norm": 8.562474250793457,
"learning_rate": 8.192684891372198e-05,
"loss": 0.0189,
"step": 585
},
{
"epoch": 3.0051480051480053,
"grad_norm": 0.09545473754405975,
"learning_rate": 8.157303103322296e-05,
"loss": 0.0027,
"step": 586
},
{
"epoch": 3.01029601029601,
"grad_norm": 0.0682293027639389,
"learning_rate": 8.1219451782533e-05,
"loss": 0.0007,
"step": 587
},
{
"epoch": 3.0154440154440154,
"grad_norm": 2.8686673641204834,
"learning_rate": 8.086611574051417e-05,
"loss": 0.0364,
"step": 588
},
{
"epoch": 3.0205920205920207,
"grad_norm": 2.3952081203460693,
"learning_rate": 8.051302748287895e-05,
"loss": 0.0777,
"step": 589
},
{
"epoch": 3.0257400257400255,
"grad_norm": 0.2540910840034485,
"learning_rate": 8.016019158213101e-05,
"loss": 0.0042,
"step": 590
},
{
"epoch": 3.030888030888031,
"grad_norm": 0.13849835097789764,
"learning_rate": 7.980761260750607e-05,
"loss": 0.0025,
"step": 591
},
{
"epoch": 3.036036036036036,
"grad_norm": 1.6155511140823364,
"learning_rate": 7.945529512491251e-05,
"loss": 0.0397,
"step": 592
},
{
"epoch": 3.0411840411840414,
"grad_norm": 0.05959663912653923,
"learning_rate": 7.91032436968725e-05,
"loss": 0.0014,
"step": 593
},
{
"epoch": 3.0463320463320462,
"grad_norm": 0.030141429975628853,
"learning_rate": 7.875146288246275e-05,
"loss": 0.0007,
"step": 594
},
{
"epoch": 3.0514800514800515,
"grad_norm": 20.547496795654297,
"learning_rate": 7.839995723725548e-05,
"loss": 0.0046,
"step": 595
},
{
"epoch": 3.056628056628057,
"grad_norm": 0.27998486161231995,
"learning_rate": 7.804873131325954e-05,
"loss": 0.0045,
"step": 596
},
{
"epoch": 3.0617760617760617,
"grad_norm": 0.0462983138859272,
"learning_rate": 7.76977896588613e-05,
"loss": 0.0008,
"step": 597
},
{
"epoch": 3.066924066924067,
"grad_norm": 0.1543927788734436,
"learning_rate": 7.734713681876589e-05,
"loss": 0.0031,
"step": 598
},
{
"epoch": 3.0720720720720722,
"grad_norm": 0.06826331466436386,
"learning_rate": 7.699677733393826e-05,
"loss": 0.0008,
"step": 599
},
{
"epoch": 3.077220077220077,
"grad_norm": 3.4199118614196777,
"learning_rate": 7.66467157415444e-05,
"loss": 0.0433,
"step": 600
},
{
"epoch": 3.0823680823680824,
"grad_norm": 5.331698417663574,
"learning_rate": 7.629695657489257e-05,
"loss": 0.0436,
"step": 601
},
{
"epoch": 3.0875160875160876,
"grad_norm": 0.1382795125246048,
"learning_rate": 7.594750436337467e-05,
"loss": 0.0024,
"step": 602
},
{
"epoch": 3.0926640926640925,
"grad_norm": 0.4087207317352295,
"learning_rate": 7.55983636324074e-05,
"loss": 0.0053,
"step": 603
},
{
"epoch": 3.0978120978120978,
"grad_norm": 0.07870214432477951,
"learning_rate": 7.524953890337395e-05,
"loss": 0.0014,
"step": 604
},
{
"epoch": 3.102960102960103,
"grad_norm": 0.08449980616569519,
"learning_rate": 7.490103469356513e-05,
"loss": 0.0013,
"step": 605
},
{
"epoch": 3.108108108108108,
"grad_norm": 0.30085521936416626,
"learning_rate": 7.455285551612105e-05,
"loss": 0.0027,
"step": 606
},
{
"epoch": 3.113256113256113,
"grad_norm": 1.4303934574127197,
"learning_rate": 7.42050058799727e-05,
"loss": 0.0024,
"step": 607
},
{
"epoch": 3.1184041184041185,
"grad_norm": 0.21123769879341125,
"learning_rate": 7.385749028978346e-05,
"loss": 0.004,
"step": 608
},
{
"epoch": 3.1235521235521237,
"grad_norm": 0.5291070342063904,
"learning_rate": 7.351031324589074e-05,
"loss": 0.0021,
"step": 609
},
{
"epoch": 3.1287001287001286,
"grad_norm": 1.6894054412841797,
"learning_rate": 7.316347924424787e-05,
"loss": 0.0045,
"step": 610
},
{
"epoch": 3.133848133848134,
"grad_norm": 3.5198097229003906,
"learning_rate": 7.281699277636572e-05,
"loss": 0.039,
"step": 611
},
{
"epoch": 3.138996138996139,
"grad_norm": 6.83537483215332,
"learning_rate": 7.24708583292546e-05,
"loss": 0.0619,
"step": 612
},
{
"epoch": 3.144144144144144,
"grad_norm": 0.14273810386657715,
"learning_rate": 7.212508038536613e-05,
"loss": 0.0017,
"step": 613
},
{
"epoch": 3.1492921492921493,
"grad_norm": 4.771847724914551,
"learning_rate": 7.177966342253524e-05,
"loss": 0.0304,
"step": 614
},
{
"epoch": 3.1544401544401546,
"grad_norm": 0.4907284379005432,
"learning_rate": 7.143461191392206e-05,
"loss": 0.0022,
"step": 615
},
{
"epoch": 3.1595881595881594,
"grad_norm": 2.258160352706909,
"learning_rate": 7.108993032795418e-05,
"loss": 0.0068,
"step": 616
},
{
"epoch": 3.1647361647361647,
"grad_norm": 0.023620612919330597,
"learning_rate": 7.07456231282686e-05,
"loss": 0.0005,
"step": 617
},
{
"epoch": 3.16988416988417,
"grad_norm": 0.4077928960323334,
"learning_rate": 7.040169477365403e-05,
"loss": 0.0028,
"step": 618
},
{
"epoch": 3.175032175032175,
"grad_norm": 0.47410908341407776,
"learning_rate": 7.005814971799318e-05,
"loss": 0.0049,
"step": 619
},
{
"epoch": 3.18018018018018,
"grad_norm": 0.11412069201469421,
"learning_rate": 6.971499241020495e-05,
"loss": 0.0005,
"step": 620
},
{
"epoch": 3.1853281853281854,
"grad_norm": 2.9045896530151367,
"learning_rate": 6.93722272941869e-05,
"loss": 0.0025,
"step": 621
},
{
"epoch": 3.1904761904761907,
"grad_norm": 0.7240750193595886,
"learning_rate": 6.902985880875773e-05,
"loss": 0.0024,
"step": 622
},
{
"epoch": 3.1956241956241955,
"grad_norm": 2.276467800140381,
"learning_rate": 6.868789138759976e-05,
"loss": 0.0096,
"step": 623
},
{
"epoch": 3.200772200772201,
"grad_norm": 0.007870988920331001,
"learning_rate": 6.83463294592015e-05,
"loss": 0.0002,
"step": 624
},
{
"epoch": 3.205920205920206,
"grad_norm": 0.45754534006118774,
"learning_rate": 6.800517744680032e-05,
"loss": 0.0029,
"step": 625
},
{
"epoch": 3.211068211068211,
"grad_norm": 0.10642076283693314,
"learning_rate": 6.766443976832517e-05,
"loss": 0.0004,
"step": 626
},
{
"epoch": 3.2162162162162162,
"grad_norm": 0.25874975323677063,
"learning_rate": 6.732412083633936e-05,
"loss": 0.0021,
"step": 627
},
{
"epoch": 3.2213642213642215,
"grad_norm": 0.17356695234775543,
"learning_rate": 6.698422505798338e-05,
"loss": 0.0019,
"step": 628
},
{
"epoch": 3.2265122265122264,
"grad_norm": 0.5095115900039673,
"learning_rate": 6.664475683491796e-05,
"loss": 0.001,
"step": 629
},
{
"epoch": 3.2316602316602316,
"grad_norm": 0.00819287821650505,
"learning_rate": 6.630572056326687e-05,
"loss": 0.0002,
"step": 630
},
{
"epoch": 3.236808236808237,
"grad_norm": 0.18225952982902527,
"learning_rate": 6.59671206335602e-05,
"loss": 0.0008,
"step": 631
},
{
"epoch": 3.2419562419562418,
"grad_norm": 2.857295036315918,
"learning_rate": 6.562896143067734e-05,
"loss": 0.0078,
"step": 632
},
{
"epoch": 3.247104247104247,
"grad_norm": 0.20077714323997498,
"learning_rate": 6.529124733379024e-05,
"loss": 0.0016,
"step": 633
},
{
"epoch": 3.2522522522522523,
"grad_norm": 5.528991222381592,
"learning_rate": 6.495398271630675e-05,
"loss": 0.005,
"step": 634
},
{
"epoch": 3.257400257400257,
"grad_norm": 3.3158912658691406,
"learning_rate": 6.461717194581393e-05,
"loss": 0.0301,
"step": 635
},
{
"epoch": 3.2625482625482625,
"grad_norm": 3.0370380878448486,
"learning_rate": 6.428081938402149e-05,
"loss": 0.1838,
"step": 636
},
{
"epoch": 3.2676962676962678,
"grad_norm": 4.145111083984375,
"learning_rate": 6.394492938670538e-05,
"loss": 0.0183,
"step": 637
},
{
"epoch": 3.2728442728442726,
"grad_norm": 0.062105387449264526,
"learning_rate": 6.360950630365126e-05,
"loss": 0.0008,
"step": 638
},
{
"epoch": 3.277992277992278,
"grad_norm": 0.09910746663808823,
"learning_rate": 6.327455447859827e-05,
"loss": 0.0004,
"step": 639
},
{
"epoch": 3.283140283140283,
"grad_norm": 0.20502103865146637,
"learning_rate": 6.294007824918276e-05,
"loss": 0.0008,
"step": 640
},
{
"epoch": 3.2882882882882885,
"grad_norm": 0.01120681967586279,
"learning_rate": 6.260608194688206e-05,
"loss": 0.0002,
"step": 641
},
{
"epoch": 3.2934362934362933,
"grad_norm": 0.03213547170162201,
"learning_rate": 6.227256989695848e-05,
"loss": 0.0006,
"step": 642
},
{
"epoch": 3.2985842985842986,
"grad_norm": 3.5040829181671143,
"learning_rate": 6.193954641840318e-05,
"loss": 0.0137,
"step": 643
},
{
"epoch": 3.303732303732304,
"grad_norm": 0.07908669859170914,
"learning_rate": 6.160701582388038e-05,
"loss": 0.0005,
"step": 644
},
{
"epoch": 3.3088803088803087,
"grad_norm": 0.010519213043153286,
"learning_rate": 6.12749824196714e-05,
"loss": 0.0002,
"step": 645
},
{
"epoch": 3.314028314028314,
"grad_norm": 2.3272342681884766,
"learning_rate": 6.0943450505618917e-05,
"loss": 0.0544,
"step": 646
},
{
"epoch": 3.3191763191763193,
"grad_norm": 0.023254504427313805,
"learning_rate": 6.061242437507131e-05,
"loss": 0.0002,
"step": 647
},
{
"epoch": 3.3243243243243246,
"grad_norm": 1.5178560018539429,
"learning_rate": 6.028190831482703e-05,
"loss": 0.0067,
"step": 648
},
{
"epoch": 3.3294723294723294,
"grad_norm": 0.134837806224823,
"learning_rate": 5.995190660507915e-05,
"loss": 0.0007,
"step": 649
},
{
"epoch": 3.3346203346203347,
"grad_norm": 0.13291309773921967,
"learning_rate": 5.962242351935985e-05,
"loss": 0.0008,
"step": 650
},
{
"epoch": 3.33976833976834,
"grad_norm": 9.101471900939941,
"learning_rate": 5.929346332448511e-05,
"loss": 0.0783,
"step": 651
},
{
"epoch": 3.344916344916345,
"grad_norm": 22.862743377685547,
"learning_rate": 5.89650302804995e-05,
"loss": 0.0973,
"step": 652
},
{
"epoch": 3.35006435006435,
"grad_norm": 0.17167998850345612,
"learning_rate": 5.863712864062089e-05,
"loss": 0.0011,
"step": 653
},
{
"epoch": 3.3552123552123554,
"grad_norm": 2.473607063293457,
"learning_rate": 5.8309762651185484e-05,
"loss": 0.0137,
"step": 654
},
{
"epoch": 3.3603603603603602,
"grad_norm": 0.6087380647659302,
"learning_rate": 5.7982936551592906e-05,
"loss": 0.0031,
"step": 655
},
{
"epoch": 3.3655083655083655,
"grad_norm": 30.847816467285156,
"learning_rate": 5.765665457425102e-05,
"loss": 0.0382,
"step": 656
},
{
"epoch": 3.370656370656371,
"grad_norm": 0.022877676412463188,
"learning_rate": 5.733092094452135e-05,
"loss": 0.0003,
"step": 657
},
{
"epoch": 3.3758043758043756,
"grad_norm": 0.014584055170416832,
"learning_rate": 5.700573988066433e-05,
"loss": 0.0003,
"step": 658
},
{
"epoch": 3.380952380952381,
"grad_norm": 0.603316068649292,
"learning_rate": 5.668111559378471e-05,
"loss": 0.0046,
"step": 659
},
{
"epoch": 3.386100386100386,
"grad_norm": 0.07768981903791428,
"learning_rate": 5.6357052287776765e-05,
"loss": 0.0005,
"step": 660
},
{
"epoch": 3.391248391248391,
"grad_norm": 4.15887451171875,
"learning_rate": 5.6033554159270294e-05,
"loss": 0.0182,
"step": 661
},
{
"epoch": 3.3963963963963963,
"grad_norm": 19.61903190612793,
"learning_rate": 5.571062539757581e-05,
"loss": 0.1167,
"step": 662
},
{
"epoch": 3.4015444015444016,
"grad_norm": 0.6196698546409607,
"learning_rate": 5.538827018463069e-05,
"loss": 0.0022,
"step": 663
},
{
"epoch": 3.4066924066924065,
"grad_norm": 0.08763101696968079,
"learning_rate": 5.50664926949447e-05,
"loss": 0.0013,
"step": 664
},
{
"epoch": 3.4118404118404118,
"grad_norm": 0.6720811128616333,
"learning_rate": 5.474529709554612e-05,
"loss": 0.0044,
"step": 665
},
{
"epoch": 3.416988416988417,
"grad_norm": 0.03270460665225983,
"learning_rate": 5.4424687545927776e-05,
"loss": 0.0005,
"step": 666
},
{
"epoch": 3.4221364221364223,
"grad_norm": 1.180560827255249,
"learning_rate": 5.410466819799306e-05,
"loss": 0.0038,
"step": 667
},
{
"epoch": 3.427284427284427,
"grad_norm": 0.8589619398117065,
"learning_rate": 5.378524319600231e-05,
"loss": 0.0039,
"step": 668
},
{
"epoch": 3.4324324324324325,
"grad_norm": 0.002127144020050764,
"learning_rate": 5.346641667651897e-05,
"loss": 0.0,
"step": 669
},
{
"epoch": 3.4375804375804377,
"grad_norm": 0.09254667907953262,
"learning_rate": 5.314819276835625e-05,
"loss": 0.0006,
"step": 670
},
{
"epoch": 3.4427284427284426,
"grad_norm": 0.47161731123924255,
"learning_rate": 5.283057559252341e-05,
"loss": 0.001,
"step": 671
},
{
"epoch": 3.447876447876448,
"grad_norm": 4.138031005859375,
"learning_rate": 5.25135692621725e-05,
"loss": 0.0122,
"step": 672
},
{
"epoch": 3.453024453024453,
"grad_norm": 0.8576433062553406,
"learning_rate": 5.219717788254521e-05,
"loss": 0.0028,
"step": 673
},
{
"epoch": 3.458172458172458,
"grad_norm": 1.9877179861068726,
"learning_rate": 5.1881405550919493e-05,
"loss": 0.0289,
"step": 674
},
{
"epoch": 3.4633204633204633,
"grad_norm": 0.49565389752388,
"learning_rate": 5.156625635655672e-05,
"loss": 0.0015,
"step": 675
},
{
"epoch": 3.4684684684684686,
"grad_norm": 0.15118490159511566,
"learning_rate": 5.12517343806485e-05,
"loss": 0.0007,
"step": 676
},
{
"epoch": 3.473616473616474,
"grad_norm": 0.00682939775288105,
"learning_rate": 5.0937843696263966e-05,
"loss": 0.0001,
"step": 677
},
{
"epoch": 3.4787644787644787,
"grad_norm": 0.019901292398571968,
"learning_rate": 5.062458836829711e-05,
"loss": 0.0001,
"step": 678
},
{
"epoch": 3.483912483912484,
"grad_norm": 0.02060600183904171,
"learning_rate": 5.03119724534139e-05,
"loss": 0.0002,
"step": 679
},
{
"epoch": 3.4890604890604893,
"grad_norm": 0.13615114986896515,
"learning_rate": 5.000000000000002e-05,
"loss": 0.0011,
"step": 680
},
{
"epoch": 3.494208494208494,
"grad_norm": 11.097320556640625,
"learning_rate": 4.96886750481082e-05,
"loss": 0.1335,
"step": 681
},
{
"epoch": 3.4993564993564994,
"grad_norm": 0.10474572330713272,
"learning_rate": 4.9378001629406e-05,
"loss": 0.0012,
"step": 682
},
{
"epoch": 3.5045045045045047,
"grad_norm": 0.0054645356722176075,
"learning_rate": 4.9067983767123736e-05,
"loss": 0.0001,
"step": 683
},
{
"epoch": 3.5096525096525095,
"grad_norm": 0.12902982532978058,
"learning_rate": 4.875862547600207e-05,
"loss": 0.0012,
"step": 684
},
{
"epoch": 3.514800514800515,
"grad_norm": 0.052395693957805634,
"learning_rate": 4.8449930762240355e-05,
"loss": 0.0005,
"step": 685
},
{
"epoch": 3.51994851994852,
"grad_norm": 2.1456692218780518,
"learning_rate": 4.814190362344454e-05,
"loss": 0.0016,
"step": 686
},
{
"epoch": 3.525096525096525,
"grad_norm": 4.447965621948242,
"learning_rate": 4.783454804857539e-05,
"loss": 0.015,
"step": 687
},
{
"epoch": 3.53024453024453,
"grad_norm": 0.04698798060417175,
"learning_rate": 4.752786801789703e-05,
"loss": 0.0002,
"step": 688
},
{
"epoch": 3.5353925353925355,
"grad_norm": 0.08583614230155945,
"learning_rate": 4.722186750292511e-05,
"loss": 0.0006,
"step": 689
},
{
"epoch": 3.5405405405405403,
"grad_norm": 1.175262689590454,
"learning_rate": 4.6916550466375684e-05,
"loss": 0.0065,
"step": 690
},
{
"epoch": 3.5456885456885456,
"grad_norm": 1.9470089673995972,
"learning_rate": 4.661192086211366e-05,
"loss": 0.0018,
"step": 691
},
{
"epoch": 3.550836550836551,
"grad_norm": 0.17172235250473022,
"learning_rate": 4.630798263510162e-05,
"loss": 0.0004,
"step": 692
},
{
"epoch": 3.5559845559845558,
"grad_norm": 0.006491504609584808,
"learning_rate": 4.600473972134894e-05,
"loss": 0.0001,
"step": 693
},
{
"epoch": 3.561132561132561,
"grad_norm": 4.946051597595215,
"learning_rate": 4.570219604786051e-05,
"loss": 0.0263,
"step": 694
},
{
"epoch": 3.5662805662805663,
"grad_norm": 1.0091392993927002,
"learning_rate": 4.540035553258619e-05,
"loss": 0.0029,
"step": 695
},
{
"epoch": 3.571428571428571,
"grad_norm": 0.05569381266832352,
"learning_rate": 4.5099222084369805e-05,
"loss": 0.0004,
"step": 696
},
{
"epoch": 3.5765765765765765,
"grad_norm": 8.457650184631348,
"learning_rate": 4.479879960289863e-05,
"loss": 0.0159,
"step": 697
},
{
"epoch": 3.5817245817245817,
"grad_norm": 0.4417124390602112,
"learning_rate": 4.449909197865303e-05,
"loss": 0.0016,
"step": 698
},
{
"epoch": 3.586872586872587,
"grad_norm": 1.062852382659912,
"learning_rate": 4.420010309285577e-05,
"loss": 0.0005,
"step": 699
},
{
"epoch": 3.592020592020592,
"grad_norm": 3.5031871795654297,
"learning_rate": 4.3901836817422124e-05,
"loss": 0.0277,
"step": 700
},
{
"epoch": 3.597168597168597,
"grad_norm": 0.006413920782506466,
"learning_rate": 4.360429701490934e-05,
"loss": 0.0001,
"step": 701
},
{
"epoch": 3.6023166023166024,
"grad_norm": 0.06792295724153519,
"learning_rate": 4.3307487538467006e-05,
"loss": 0.0001,
"step": 702
},
{
"epoch": 3.6074646074646077,
"grad_norm": 0.5535875558853149,
"learning_rate": 4.301141223178684e-05,
"loss": 0.0013,
"step": 703
},
{
"epoch": 3.6126126126126126,
"grad_norm": 0.549075186252594,
"learning_rate": 4.271607492905303e-05,
"loss": 0.0017,
"step": 704
},
{
"epoch": 3.617760617760618,
"grad_norm": 10.674209594726562,
"learning_rate": 4.242147945489272e-05,
"loss": 0.0436,
"step": 705
},
{
"epoch": 3.622908622908623,
"grad_norm": 4.2363691329956055,
"learning_rate": 4.212762962432619e-05,
"loss": 0.0335,
"step": 706
},
{
"epoch": 3.628056628056628,
"grad_norm": 12.446212768554688,
"learning_rate": 4.183452924271776e-05,
"loss": 0.0713,
"step": 707
},
{
"epoch": 3.6332046332046333,
"grad_norm": 8.711955070495605,
"learning_rate": 4.154218210572627e-05,
"loss": 0.3495,
"step": 708
},
{
"epoch": 3.6383526383526386,
"grad_norm": 0.001297969138249755,
"learning_rate": 4.125059199925599e-05,
"loss": 0.0,
"step": 709
},
{
"epoch": 3.6435006435006434,
"grad_norm": 0.005227327346801758,
"learning_rate": 4.0959762699407766e-05,
"loss": 0.0001,
"step": 710
},
{
"epoch": 3.6486486486486487,
"grad_norm": 0.10913996398448944,
"learning_rate": 4.06696979724298e-05,
"loss": 0.0008,
"step": 711
},
{
"epoch": 3.653796653796654,
"grad_norm": 0.015117711387574673,
"learning_rate": 4.038040157466918e-05,
"loss": 0.0002,
"step": 712
},
{
"epoch": 3.658944658944659,
"grad_norm": 0.010094228200614452,
"learning_rate": 4.009187725252309e-05,
"loss": 0.0001,
"step": 713
},
{
"epoch": 3.664092664092664,
"grad_norm": 1.2325109243392944,
"learning_rate": 3.980412874239021e-05,
"loss": 0.0031,
"step": 714
},
{
"epoch": 3.6692406692406694,
"grad_norm": 0.5344968438148499,
"learning_rate": 3.95171597706226e-05,
"loss": 0.0038,
"step": 715
},
{
"epoch": 3.6743886743886742,
"grad_norm": 0.1807209551334381,
"learning_rate": 3.9230974053477086e-05,
"loss": 0.0002,
"step": 716
},
{
"epoch": 3.6795366795366795,
"grad_norm": 0.053924065083265305,
"learning_rate": 3.8945575297067506e-05,
"loss": 0.0001,
"step": 717
},
{
"epoch": 3.684684684684685,
"grad_norm": 0.037016600370407104,
"learning_rate": 3.866096719731639e-05,
"loss": 0.0002,
"step": 718
},
{
"epoch": 3.6898326898326896,
"grad_norm": 6.042881965637207,
"learning_rate": 3.8377153439907266e-05,
"loss": 0.0047,
"step": 719
},
{
"epoch": 3.694980694980695,
"grad_norm": 1.9064173698425293,
"learning_rate": 3.809413770023701e-05,
"loss": 0.0026,
"step": 720
},
{
"epoch": 3.7001287001287,
"grad_norm": 0.002955307951197028,
"learning_rate": 3.7811923643367974e-05,
"loss": 0.0,
"step": 721
},
{
"epoch": 3.705276705276705,
"grad_norm": 0.019066477194428444,
"learning_rate": 3.7530514923980884e-05,
"loss": 0.0001,
"step": 722
},
{
"epoch": 3.7104247104247103,
"grad_norm": 6.450963020324707,
"learning_rate": 3.724991518632717e-05,
"loss": 0.0218,
"step": 723
},
{
"epoch": 3.7155727155727156,
"grad_norm": 0.00752641586586833,
"learning_rate": 3.697012806418194e-05,
"loss": 0.0001,
"step": 724
},
{
"epoch": 3.7207207207207205,
"grad_norm": 0.03809618204832077,
"learning_rate": 3.669115718079702e-05,
"loss": 0.0002,
"step": 725
},
{
"epoch": 3.7258687258687258,
"grad_norm": 0.00477253645658493,
"learning_rate": 3.641300614885378e-05,
"loss": 0.0001,
"step": 726
},
{
"epoch": 3.731016731016731,
"grad_norm": 3.528430700302124,
"learning_rate": 3.61356785704166e-05,
"loss": 0.0632,
"step": 727
},
{
"epoch": 3.7361647361647363,
"grad_norm": 0.03414354845881462,
"learning_rate": 3.585917803688603e-05,
"loss": 0.0002,
"step": 728
},
{
"epoch": 3.741312741312741,
"grad_norm": 0.7383514046669006,
"learning_rate": 3.558350812895238e-05,
"loss": 0.0013,
"step": 729
},
{
"epoch": 3.7464607464607464,
"grad_norm": 0.032647211104631424,
"learning_rate": 3.530867241654942e-05,
"loss": 0.0002,
"step": 730
},
{
"epoch": 3.7516087516087517,
"grad_norm": 0.053170569241046906,
"learning_rate": 3.503467445880789e-05,
"loss": 0.0002,
"step": 731
},
{
"epoch": 3.756756756756757,
"grad_norm": 8.38620662689209,
"learning_rate": 3.476151780400979e-05,
"loss": 0.0201,
"step": 732
},
{
"epoch": 3.761904761904762,
"grad_norm": 0.003682285314425826,
"learning_rate": 3.448920598954203e-05,
"loss": 0.0001,
"step": 733
},
{
"epoch": 3.767052767052767,
"grad_norm": 0.00584920309484005,
"learning_rate": 3.421774254185096e-05,
"loss": 0.0001,
"step": 734
},
{
"epoch": 3.7722007722007724,
"grad_norm": 0.004093759693205357,
"learning_rate": 3.394713097639647e-05,
"loss": 0.0001,
"step": 735
},
{
"epoch": 3.7773487773487773,
"grad_norm": 1.4704846143722534,
"learning_rate": 3.367737479760652e-05,
"loss": 0.008,
"step": 736
},
{
"epoch": 3.7824967824967826,
"grad_norm": 0.07209017872810364,
"learning_rate": 3.340847749883191e-05,
"loss": 0.0004,
"step": 737
},
{
"epoch": 3.787644787644788,
"grad_norm": 15.315757751464844,
"learning_rate": 3.31404425623008e-05,
"loss": 0.141,
"step": 738
},
{
"epoch": 3.7927927927927927,
"grad_norm": 2.056124210357666,
"learning_rate": 3.287327345907381e-05,
"loss": 0.0031,
"step": 739
},
{
"epoch": 3.797940797940798,
"grad_norm": 5.367823600769043,
"learning_rate": 3.2606973648998915e-05,
"loss": 0.0807,
"step": 740
},
{
"epoch": 3.8030888030888033,
"grad_norm": 0.12702032923698425,
"learning_rate": 3.2341546580666796e-05,
"loss": 0.0004,
"step": 741
},
{
"epoch": 3.808236808236808,
"grad_norm": 10.258810997009277,
"learning_rate": 3.207699569136608e-05,
"loss": 0.009,
"step": 742
},
{
"epoch": 3.8133848133848134,
"grad_norm": 0.15279260277748108,
"learning_rate": 3.1813324407038825e-05,
"loss": 0.0005,
"step": 743
},
{
"epoch": 3.8185328185328187,
"grad_norm": 11.908256530761719,
"learning_rate": 3.1550536142236145e-05,
"loss": 0.1451,
"step": 744
},
{
"epoch": 3.8236808236808235,
"grad_norm": 1.814513087272644,
"learning_rate": 3.128863430007414e-05,
"loss": 0.0034,
"step": 745
},
{
"epoch": 3.828828828828829,
"grad_norm": 0.08874842524528503,
"learning_rate": 3.102762227218957e-05,
"loss": 0.0006,
"step": 746
},
{
"epoch": 3.833976833976834,
"grad_norm": 0.9989665746688843,
"learning_rate": 3.0767503438696213e-05,
"loss": 0.0032,
"step": 747
},
{
"epoch": 3.839124839124839,
"grad_norm": 0.012054390273988247,
"learning_rate": 3.0508281168140806e-05,
"loss": 0.0001,
"step": 748
},
{
"epoch": 3.844272844272844,
"grad_norm": 0.04495546594262123,
"learning_rate": 3.0249958817459722e-05,
"loss": 0.0002,
"step": 749
},
{
"epoch": 3.8494208494208495,
"grad_norm": 0.21954287588596344,
"learning_rate": 2.999253973193522e-05,
"loss": 0.0021,
"step": 750
},
{
"epoch": 3.8545688545688543,
"grad_norm": 5.57328462600708,
"learning_rate": 2.9736027245152275e-05,
"loss": 0.0154,
"step": 751
},
{
"epoch": 3.8597168597168596,
"grad_norm": 0.032941389828920364,
"learning_rate": 2.9480424678955443e-05,
"loss": 0.0003,
"step": 752
},
{
"epoch": 3.864864864864865,
"grad_norm": 0.16867409646511078,
"learning_rate": 2.9225735343405693e-05,
"loss": 0.001,
"step": 753
},
{
"epoch": 3.8700128700128698,
"grad_norm": 21.949522018432617,
"learning_rate": 2.897196253673773e-05,
"loss": 0.0023,
"step": 754
},
{
"epoch": 3.875160875160875,
"grad_norm": 4.997512340545654,
"learning_rate": 2.8719109545317103e-05,
"loss": 0.0887,
"step": 755
},
{
"epoch": 3.8803088803088803,
"grad_norm": 0.009235148318111897,
"learning_rate": 2.8467179643597697e-05,
"loss": 0.0001,
"step": 756
},
{
"epoch": 3.8854568854568856,
"grad_norm": 0.01561771985143423,
"learning_rate": 2.8216176094079482e-05,
"loss": 0.0001,
"step": 757
},
{
"epoch": 3.8906048906048905,
"grad_norm": 0.32219594717025757,
"learning_rate": 2.7966102147265994e-05,
"loss": 0.0012,
"step": 758
},
{
"epoch": 3.8957528957528957,
"grad_norm": 2.680638313293457,
"learning_rate": 2.7716961041622534e-05,
"loss": 0.1297,
"step": 759
},
{
"epoch": 3.900900900900901,
"grad_norm": 0.014756716787815094,
"learning_rate": 2.746875600353398e-05,
"loss": 0.0002,
"step": 760
},
{
"epoch": 3.9060489060489063,
"grad_norm": 0.15277861058712006,
"learning_rate": 2.722149024726307e-05,
"loss": 0.0003,
"step": 761
},
{
"epoch": 3.911196911196911,
"grad_norm": 0.018368825316429138,
"learning_rate": 2.697516697490896e-05,
"loss": 0.0002,
"step": 762
},
{
"epoch": 3.9163449163449164,
"grad_norm": 2.6372954845428467,
"learning_rate": 2.6729789376365456e-05,
"loss": 0.0082,
"step": 763
},
{
"epoch": 3.9214929214929217,
"grad_norm": 2.936980724334717,
"learning_rate": 2.6485360629279987e-05,
"loss": 0.1449,
"step": 764
},
{
"epoch": 3.9266409266409266,
"grad_norm": 0.8106775283813477,
"learning_rate": 2.624188389901221e-05,
"loss": 0.0046,
"step": 765
},
{
"epoch": 3.931788931788932,
"grad_norm": 0.43472757935523987,
"learning_rate": 2.599936233859326e-05,
"loss": 0.0012,
"step": 766
},
{
"epoch": 3.936936936936937,
"grad_norm": 1.5690232515335083,
"learning_rate": 2.5757799088684654e-05,
"loss": 0.0049,
"step": 767
},
{
"epoch": 3.942084942084942,
"grad_norm": 0.010339402593672276,
"learning_rate": 2.5517197277537886e-05,
"loss": 0.0002,
"step": 768
},
{
"epoch": 3.9472329472329473,
"grad_norm": 0.03642657771706581,
"learning_rate": 2.527756002095373e-05,
"loss": 0.0003,
"step": 769
},
{
"epoch": 3.9523809523809526,
"grad_norm": 0.8840352892875671,
"learning_rate": 2.5038890422241958e-05,
"loss": 0.0066,
"step": 770
},
{
"epoch": 3.9575289575289574,
"grad_norm": 0.15709532797336578,
"learning_rate": 2.480119157218108e-05,
"loss": 0.0012,
"step": 771
},
{
"epoch": 3.9626769626769627,
"grad_norm": 7.115312099456787,
"learning_rate": 2.4564466548978525e-05,
"loss": 0.0164,
"step": 772
},
{
"epoch": 3.967824967824968,
"grad_norm": 7.106594085693359,
"learning_rate": 2.432871841823047e-05,
"loss": 0.0072,
"step": 773
},
{
"epoch": 3.972972972972973,
"grad_norm": 0.10783769935369492,
"learning_rate": 2.4093950232882456e-05,
"loss": 0.0007,
"step": 774
},
{
"epoch": 3.978120978120978,
"grad_norm": 0.24498294293880463,
"learning_rate": 2.3860165033189587e-05,
"loss": 0.001,
"step": 775
},
{
"epoch": 3.9832689832689834,
"grad_norm": 18.48102569580078,
"learning_rate": 2.3627365846677306e-05,
"loss": 0.3756,
"step": 776
},
{
"epoch": 3.988416988416988,
"grad_norm": 2.5727248191833496,
"learning_rate": 2.339555568810221e-05,
"loss": 0.006,
"step": 777
},
{
"epoch": 3.9935649935649935,
"grad_norm": 0.28317496180534363,
"learning_rate": 2.3164737559412854e-05,
"loss": 0.0011,
"step": 778
},
{
"epoch": 3.998712998712999,
"grad_norm": 1.873467206954956,
"learning_rate": 2.2934914449711087e-05,
"loss": 0.0048,
"step": 779
},
{
"epoch": 4.0,
"grad_norm": 0.006583232432603836,
"learning_rate": 2.2706089335213122e-05,
"loss": 0.0001,
"step": 780
}
],
"logging_steps": 1,
"max_steps": 970,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 2.0784491173701962e+17,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}