Files
llama3-3B-sft/checkpoint-368/trainer_state.json

2611 lines
63 KiB
JSON
Raw Normal View History

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.4163855421686746,
"eval_steps": 500,
"global_step": 368,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0038554216867469878,
"grad_norm": 1.2289724349975586,
"learning_rate": 0.0,
"loss": 0.0363,
"step": 1
},
{
"epoch": 0.0077108433734939755,
"grad_norm": 1.148133635520935,
"learning_rate": 5.128205128205128e-07,
"loss": 0.042,
"step": 2
},
{
"epoch": 0.011566265060240964,
"grad_norm": 1.251704216003418,
"learning_rate": 1.0256410256410257e-06,
"loss": 0.0442,
"step": 3
},
{
"epoch": 0.015421686746987951,
"grad_norm": 1.1709717512130737,
"learning_rate": 1.5384615384615387e-06,
"loss": 0.0363,
"step": 4
},
{
"epoch": 0.01927710843373494,
"grad_norm": 1.1885852813720703,
"learning_rate": 2.0512820512820513e-06,
"loss": 0.0281,
"step": 5
},
{
"epoch": 0.02313253012048193,
"grad_norm": 0.7513725161552429,
"learning_rate": 2.564102564102564e-06,
"loss": 0.0456,
"step": 6
},
{
"epoch": 0.026987951807228915,
"grad_norm": 0.6604480147361755,
"learning_rate": 3.0769230769230774e-06,
"loss": 0.0336,
"step": 7
},
{
"epoch": 0.030843373493975902,
"grad_norm": 0.5245765447616577,
"learning_rate": 3.58974358974359e-06,
"loss": 0.0351,
"step": 8
},
{
"epoch": 0.03469879518072289,
"grad_norm": 0.35446059703826904,
"learning_rate": 4.102564102564103e-06,
"loss": 0.022,
"step": 9
},
{
"epoch": 0.03855421686746988,
"grad_norm": 0.40557414293289185,
"learning_rate": 4.615384615384616e-06,
"loss": 0.0248,
"step": 10
},
{
"epoch": 0.042409638554216866,
"grad_norm": 0.3844715356826782,
"learning_rate": 5.128205128205128e-06,
"loss": 0.0289,
"step": 11
},
{
"epoch": 0.04626506024096386,
"grad_norm": 0.31521302461624146,
"learning_rate": 5.641025641025641e-06,
"loss": 0.0221,
"step": 12
},
{
"epoch": 0.05012048192771084,
"grad_norm": 0.40441566705703735,
"learning_rate": 6.153846153846155e-06,
"loss": 0.0226,
"step": 13
},
{
"epoch": 0.05397590361445783,
"grad_norm": 0.30345258116722107,
"learning_rate": 6.666666666666667e-06,
"loss": 0.0181,
"step": 14
},
{
"epoch": 0.05783132530120482,
"grad_norm": 0.32419148087501526,
"learning_rate": 7.17948717948718e-06,
"loss": 0.0217,
"step": 15
},
{
"epoch": 0.061686746987951804,
"grad_norm": 0.29662227630615234,
"learning_rate": 7.692307692307694e-06,
"loss": 0.0216,
"step": 16
},
{
"epoch": 0.0655421686746988,
"grad_norm": 0.24562332034111023,
"learning_rate": 8.205128205128205e-06,
"loss": 0.0163,
"step": 17
},
{
"epoch": 0.06939759036144579,
"grad_norm": 0.21251554787158966,
"learning_rate": 8.717948717948719e-06,
"loss": 0.0176,
"step": 18
},
{
"epoch": 0.07325301204819278,
"grad_norm": 0.195358544588089,
"learning_rate": 9.230769230769232e-06,
"loss": 0.0121,
"step": 19
},
{
"epoch": 0.07710843373493977,
"grad_norm": 0.1519210785627365,
"learning_rate": 9.743589743589744e-06,
"loss": 0.0103,
"step": 20
},
{
"epoch": 0.08096385542168674,
"grad_norm": 0.29832565784454346,
"learning_rate": 1.0256410256410256e-05,
"loss": 0.0197,
"step": 21
},
{
"epoch": 0.08481927710843373,
"grad_norm": 0.24802711606025696,
"learning_rate": 1.076923076923077e-05,
"loss": 0.0207,
"step": 22
},
{
"epoch": 0.08867469879518072,
"grad_norm": 0.17878903448581696,
"learning_rate": 1.1282051282051283e-05,
"loss": 0.014,
"step": 23
},
{
"epoch": 0.09253012048192771,
"grad_norm": 0.23687325417995453,
"learning_rate": 1.1794871794871796e-05,
"loss": 0.0153,
"step": 24
},
{
"epoch": 0.0963855421686747,
"grad_norm": 0.29058969020843506,
"learning_rate": 1.230769230769231e-05,
"loss": 0.0149,
"step": 25
},
{
"epoch": 0.10024096385542168,
"grad_norm": 0.1624310314655304,
"learning_rate": 1.2820512820512823e-05,
"loss": 0.0127,
"step": 26
},
{
"epoch": 0.10409638554216867,
"grad_norm": 0.27256354689598083,
"learning_rate": 1.3333333333333333e-05,
"loss": 0.0173,
"step": 27
},
{
"epoch": 0.10795180722891566,
"grad_norm": 0.24831782281398773,
"learning_rate": 1.3846153846153847e-05,
"loss": 0.0161,
"step": 28
},
{
"epoch": 0.11180722891566265,
"grad_norm": 0.20417729020118713,
"learning_rate": 1.435897435897436e-05,
"loss": 0.0127,
"step": 29
},
{
"epoch": 0.11566265060240964,
"grad_norm": 0.22253485023975372,
"learning_rate": 1.4871794871794874e-05,
"loss": 0.018,
"step": 30
},
{
"epoch": 0.11951807228915663,
"grad_norm": 0.1634419709444046,
"learning_rate": 1.5384615384615387e-05,
"loss": 0.0092,
"step": 31
},
{
"epoch": 0.12337349397590361,
"grad_norm": 0.16349917650222778,
"learning_rate": 1.5897435897435897e-05,
"loss": 0.0151,
"step": 32
},
{
"epoch": 0.1272289156626506,
"grad_norm": 0.2271573394536972,
"learning_rate": 1.641025641025641e-05,
"loss": 0.0174,
"step": 33
},
{
"epoch": 0.1310843373493976,
"grad_norm": 0.1916988492012024,
"learning_rate": 1.6923076923076924e-05,
"loss": 0.0126,
"step": 34
},
{
"epoch": 0.13493975903614458,
"grad_norm": 0.23479051887989044,
"learning_rate": 1.7435897435897438e-05,
"loss": 0.0116,
"step": 35
},
{
"epoch": 0.13879518072289157,
"grad_norm": 0.18296493589878082,
"learning_rate": 1.794871794871795e-05,
"loss": 0.0133,
"step": 36
},
{
"epoch": 0.14265060240963856,
"grad_norm": 0.12307147681713104,
"learning_rate": 1.8461538461538465e-05,
"loss": 0.0093,
"step": 37
},
{
"epoch": 0.14650602409638555,
"grad_norm": 0.1736001819372177,
"learning_rate": 1.8974358974358975e-05,
"loss": 0.0097,
"step": 38
},
{
"epoch": 0.15036144578313254,
"grad_norm": 0.21048344671726227,
"learning_rate": 1.9487179487179488e-05,
"loss": 0.0121,
"step": 39
},
{
"epoch": 0.15421686746987953,
"grad_norm": 0.18844228982925415,
"learning_rate": 2e-05,
"loss": 0.0145,
"step": 40
},
{
"epoch": 0.1580722891566265,
"grad_norm": 0.24053065478801727,
"learning_rate": 2.0512820512820512e-05,
"loss": 0.0233,
"step": 41
},
{
"epoch": 0.16192771084337348,
"grad_norm": 0.16952425241470337,
"learning_rate": 2.102564102564103e-05,
"loss": 0.024,
"step": 42
},
{
"epoch": 0.16578313253012048,
"grad_norm": 0.1816851794719696,
"learning_rate": 2.153846153846154e-05,
"loss": 0.0165,
"step": 43
},
{
"epoch": 0.16963855421686747,
"grad_norm": 0.17718464136123657,
"learning_rate": 2.2051282051282056e-05,
"loss": 0.0171,
"step": 44
},
{
"epoch": 0.17349397590361446,
"grad_norm": 0.44719773530960083,
"learning_rate": 2.2564102564102566e-05,
"loss": 0.0357,
"step": 45
},
{
"epoch": 0.17734939759036145,
"grad_norm": 0.1649208813905716,
"learning_rate": 2.3076923076923076e-05,
"loss": 0.016,
"step": 46
},
{
"epoch": 0.18120481927710844,
"grad_norm": 0.17772121727466583,
"learning_rate": 2.3589743589743593e-05,
"loss": 0.0106,
"step": 47
},
{
"epoch": 0.18506024096385543,
"grad_norm": 0.30155256390571594,
"learning_rate": 2.4102564102564103e-05,
"loss": 0.0132,
"step": 48
},
{
"epoch": 0.18891566265060242,
"grad_norm": 0.1777803599834442,
"learning_rate": 2.461538461538462e-05,
"loss": 0.0089,
"step": 49
},
{
"epoch": 0.1927710843373494,
"grad_norm": 0.23959968984127045,
"learning_rate": 2.512820512820513e-05,
"loss": 0.0137,
"step": 50
},
{
"epoch": 0.1966265060240964,
"grad_norm": 0.2473219782114029,
"learning_rate": 2.5641025641025646e-05,
"loss": 0.0241,
"step": 51
},
{
"epoch": 0.20048192771084336,
"grad_norm": 0.14216162264347076,
"learning_rate": 2.6153846153846157e-05,
"loss": 0.0089,
"step": 52
},
{
"epoch": 0.20433734939759035,
"grad_norm": 0.16024824976921082,
"learning_rate": 2.6666666666666667e-05,
"loss": 0.0105,
"step": 53
},
{
"epoch": 0.20819277108433734,
"grad_norm": 0.19614927470684052,
"learning_rate": 2.7179487179487183e-05,
"loss": 0.0102,
"step": 54
},
{
"epoch": 0.21204819277108433,
"grad_norm": 0.18651551008224487,
"learning_rate": 2.7692307692307694e-05,
"loss": 0.0109,
"step": 55
},
{
"epoch": 0.21590361445783132,
"grad_norm": 0.1358720362186432,
"learning_rate": 2.820512820512821e-05,
"loss": 0.0082,
"step": 56
},
{
"epoch": 0.2197590361445783,
"grad_norm": 0.13998733460903168,
"learning_rate": 2.871794871794872e-05,
"loss": 0.0081,
"step": 57
},
{
"epoch": 0.2236144578313253,
"grad_norm": 0.1299574077129364,
"learning_rate": 2.923076923076923e-05,
"loss": 0.0067,
"step": 58
},
{
"epoch": 0.2274698795180723,
"grad_norm": 0.17477169632911682,
"learning_rate": 2.9743589743589747e-05,
"loss": 0.0091,
"step": 59
},
{
"epoch": 0.23132530120481928,
"grad_norm": 0.27078455686569214,
"learning_rate": 3.0256410256410257e-05,
"loss": 0.0128,
"step": 60
},
{
"epoch": 0.23518072289156627,
"grad_norm": 0.2184700220823288,
"learning_rate": 3.0769230769230774e-05,
"loss": 0.0236,
"step": 61
},
{
"epoch": 0.23903614457831326,
"grad_norm": 0.21013228595256805,
"learning_rate": 3.128205128205129e-05,
"loss": 0.0114,
"step": 62
},
{
"epoch": 0.24289156626506025,
"grad_norm": 0.4988560974597931,
"learning_rate": 3.1794871794871795e-05,
"loss": 0.0111,
"step": 63
},
{
"epoch": 0.24674698795180722,
"grad_norm": 0.18484456837177277,
"learning_rate": 3.230769230769231e-05,
"loss": 0.0104,
"step": 64
},
{
"epoch": 0.25060240963855424,
"grad_norm": 0.19871847331523895,
"learning_rate": 3.282051282051282e-05,
"loss": 0.0094,
"step": 65
},
{
"epoch": 0.2544578313253012,
"grad_norm": 0.17952671647071838,
"learning_rate": 3.3333333333333335e-05,
"loss": 0.0127,
"step": 66
},
{
"epoch": 0.2583132530120482,
"grad_norm": 0.29055604338645935,
"learning_rate": 3.384615384615385e-05,
"loss": 0.023,
"step": 67
},
{
"epoch": 0.2621686746987952,
"grad_norm": 0.2033795416355133,
"learning_rate": 3.435897435897436e-05,
"loss": 0.0091,
"step": 68
},
{
"epoch": 0.26602409638554214,
"grad_norm": 0.2499230057001114,
"learning_rate": 3.4871794871794875e-05,
"loss": 0.0221,
"step": 69
},
{
"epoch": 0.26987951807228916,
"grad_norm": 0.22038578987121582,
"learning_rate": 3.538461538461539e-05,
"loss": 0.0139,
"step": 70
},
{
"epoch": 0.2737349397590361,
"grad_norm": 0.14657297730445862,
"learning_rate": 3.58974358974359e-05,
"loss": 0.0104,
"step": 71
},
{
"epoch": 0.27759036144578314,
"grad_norm": 0.23861773312091827,
"learning_rate": 3.6410256410256416e-05,
"loss": 0.0126,
"step": 72
},
{
"epoch": 0.2814457831325301,
"grad_norm": 0.21007320284843445,
"learning_rate": 3.692307692307693e-05,
"loss": 0.0133,
"step": 73
},
{
"epoch": 0.2853012048192771,
"grad_norm": 0.20938654243946075,
"learning_rate": 3.7435897435897436e-05,
"loss": 0.0116,
"step": 74
},
{
"epoch": 0.2891566265060241,
"grad_norm": 0.27394789457321167,
"learning_rate": 3.794871794871795e-05,
"loss": 0.0158,
"step": 75
},
{
"epoch": 0.2930120481927711,
"grad_norm": 0.1749437153339386,
"learning_rate": 3.846153846153846e-05,
"loss": 0.0097,
"step": 76
},
{
"epoch": 0.29686746987951806,
"grad_norm": 0.2208889275789261,
"learning_rate": 3.8974358974358976e-05,
"loss": 0.024,
"step": 77
},
{
"epoch": 0.3007228915662651,
"grad_norm": 0.32802650332450867,
"learning_rate": 3.948717948717949e-05,
"loss": 0.015,
"step": 78
},
{
"epoch": 0.30457831325301205,
"grad_norm": 0.21527938544750214,
"learning_rate": 4e-05,
"loss": 0.0137,
"step": 79
},
{
"epoch": 0.30843373493975906,
"grad_norm": 0.2894026041030884,
"learning_rate": 3.999979972589652e-05,
"loss": 0.0177,
"step": 80
},
{
"epoch": 0.312289156626506,
"grad_norm": 0.21148355305194855,
"learning_rate": 3.9999198907597046e-05,
"loss": 0.0146,
"step": 81
},
{
"epoch": 0.316144578313253,
"grad_norm": 0.1950712502002716,
"learning_rate": 3.999819755713442e-05,
"loss": 0.0174,
"step": 82
},
{
"epoch": 0.32,
"grad_norm": 0.23408709466457367,
"learning_rate": 3.9996795694563096e-05,
"loss": 0.0142,
"step": 83
},
{
"epoch": 0.32385542168674697,
"grad_norm": 0.4009031355381012,
"learning_rate": 3.999499334795875e-05,
"loss": 0.014,
"step": 84
},
{
"epoch": 0.327710843373494,
"grad_norm": 0.2507084012031555,
"learning_rate": 3.999279055341771e-05,
"loss": 0.013,
"step": 85
},
{
"epoch": 0.33156626506024095,
"grad_norm": 0.2922132909297943,
"learning_rate": 3.999018735505626e-05,
"loss": 0.013,
"step": 86
},
{
"epoch": 0.33542168674698797,
"grad_norm": 0.2073613405227661,
"learning_rate": 3.998718380500971e-05,
"loss": 0.0107,
"step": 87
},
{
"epoch": 0.33927710843373493,
"grad_norm": 0.3113904595375061,
"learning_rate": 3.998377996343139e-05,
"loss": 0.0337,
"step": 88
},
{
"epoch": 0.34313253012048195,
"grad_norm": 0.18452070653438568,
"learning_rate": 3.997997589849145e-05,
"loss": 0.0137,
"step": 89
},
{
"epoch": 0.3469879518072289,
"grad_norm": 0.24420331418514252,
"learning_rate": 3.9975771686375434e-05,
"loss": 0.0104,
"step": 90
},
{
"epoch": 0.35084337349397593,
"grad_norm": 0.3163854479789734,
"learning_rate": 3.9971167411282835e-05,
"loss": 0.0133,
"step": 91
},
{
"epoch": 0.3546987951807229,
"grad_norm": 0.11558058857917786,
"learning_rate": 3.996616316542537e-05,
"loss": 0.0069,
"step": 92
},
{
"epoch": 0.35855421686746985,
"grad_norm": 2.0870206356048584,
"learning_rate": 3.99607590490251e-05,
"loss": 0.0227,
"step": 93
},
{
"epoch": 0.3624096385542169,
"grad_norm": 0.23557044565677643,
"learning_rate": 3.9954955170312504e-05,
"loss": 0.0175,
"step": 94
},
{
"epoch": 0.36626506024096384,
"grad_norm": 0.2593333423137665,
"learning_rate": 3.9948751645524235e-05,
"loss": 0.0184,
"step": 95
},
{
"epoch": 0.37012048192771085,
"grad_norm": 0.5161668658256531,
"learning_rate": 3.994214859890083e-05,
"loss": 0.0265,
"step": 96
},
{
"epoch": 0.3739759036144578,
"grad_norm": 0.5731181502342224,
"learning_rate": 3.9935146162684206e-05,
"loss": 0.0167,
"step": 97
},
{
"epoch": 0.37783132530120483,
"grad_norm": 1.9357824325561523,
"learning_rate": 3.992774447711503e-05,
"loss": 0.0321,
"step": 98
},
{
"epoch": 0.3816867469879518,
"grad_norm": 4.194522857666016,
"learning_rate": 3.9919943690429906e-05,
"loss": 0.086,
"step": 99
},
{
"epoch": 0.3855421686746988,
"grad_norm": 0.3377005457878113,
"learning_rate": 3.9911743958858376e-05,
"loss": 0.0206,
"step": 100
},
{
"epoch": 0.3893975903614458,
"grad_norm": 0.3599908649921417,
"learning_rate": 3.9903145446619837e-05,
"loss": 0.0207,
"step": 101
},
{
"epoch": 0.3932530120481928,
"grad_norm": 0.3402753174304962,
"learning_rate": 3.9894148325920226e-05,
"loss": 0.0241,
"step": 102
},
{
"epoch": 0.39710843373493976,
"grad_norm": 0.6041432023048401,
"learning_rate": 3.9884752776948564e-05,
"loss": 0.0138,
"step": 103
},
{
"epoch": 0.4009638554216867,
"grad_norm": 4.082062721252441,
"learning_rate": 3.987495898787336e-05,
"loss": 0.1119,
"step": 104
},
{
"epoch": 0.40481927710843374,
"grad_norm": 6.899234771728516,
"learning_rate": 3.9864767154838864e-05,
"loss": 0.1046,
"step": 105
},
{
"epoch": 0.4086746987951807,
"grad_norm": 3.1316874027252197,
"learning_rate": 3.985417748196108e-05,
"loss": 0.0404,
"step": 106
},
{
"epoch": 0.4125301204819277,
"grad_norm": 1.8661688566207886,
"learning_rate": 3.9843190181323744e-05,
"loss": 0.0348,
"step": 107
},
{
"epoch": 0.4163855421686747,
"grad_norm": 0.5149962306022644,
"learning_rate": 3.983180547297404e-05,
"loss": 0.0284,
"step": 108
},
{
"epoch": 0.4202409638554217,
"grad_norm": 0.3855750858783722,
"learning_rate": 3.982002358491817e-05,
"loss": 0.0242,
"step": 109
},
{
"epoch": 0.42409638554216866,
"grad_norm": 0.35508987307548523,
"learning_rate": 3.9807844753116864e-05,
"loss": 0.0179,
"step": 110
},
{
"epoch": 0.4279518072289157,
"grad_norm": 1.477452278137207,
"learning_rate": 3.979526922148058e-05,
"loss": 0.0436,
"step": 111
},
{
"epoch": 0.43180722891566264,
"grad_norm": 0.41172751784324646,
"learning_rate": 3.978229724186464e-05,
"loss": 0.026,
"step": 112
},
{
"epoch": 0.43566265060240966,
"grad_norm": 0.5889139771461487,
"learning_rate": 3.9768929074064206e-05,
"loss": 0.0249,
"step": 113
},
{
"epoch": 0.4395180722891566,
"grad_norm": 0.38035663962364197,
"learning_rate": 3.975516498580907e-05,
"loss": 0.0257,
"step": 114
},
{
"epoch": 0.4433734939759036,
"grad_norm": 0.35500895977020264,
"learning_rate": 3.9741005252758255e-05,
"loss": 0.026,
"step": 115
},
{
"epoch": 0.4472289156626506,
"grad_norm": 0.6464815735816956,
"learning_rate": 3.972645015849457e-05,
"loss": 0.0499,
"step": 116
},
{
"epoch": 0.45108433734939757,
"grad_norm": 0.20825283229351044,
"learning_rate": 3.971149999451886e-05,
"loss": 0.034,
"step": 117
},
{
"epoch": 0.4549397590361446,
"grad_norm": 0.34638360142707825,
"learning_rate": 3.9696155060244166e-05,
"loss": 0.0226,
"step": 118
},
{
"epoch": 0.45879518072289155,
"grad_norm": 0.27782803773880005,
"learning_rate": 3.9680415662989806e-05,
"loss": 0.0197,
"step": 119
},
{
"epoch": 0.46265060240963857,
"grad_norm": 0.39609628915786743,
"learning_rate": 3.966428211797514e-05,
"loss": 0.0229,
"step": 120
},
{
"epoch": 0.46650602409638553,
"grad_norm": 0.30235278606414795,
"learning_rate": 3.9647754748313294e-05,
"loss": 0.046,
"step": 121
},
{
"epoch": 0.47036144578313255,
"grad_norm": 0.2774790823459625,
"learning_rate": 3.963083388500469e-05,
"loss": 0.0208,
"step": 122
},
{
"epoch": 0.4742168674698795,
"grad_norm": 0.24587354063987732,
"learning_rate": 3.96135198669304e-05,
"loss": 0.026,
"step": 123
},
{
"epoch": 0.47807228915662653,
"grad_norm": 0.2603200376033783,
"learning_rate": 3.959581304084536e-05,
"loss": 0.0189,
"step": 124
},
{
"epoch": 0.4819277108433735,
"grad_norm": 0.33505868911743164,
"learning_rate": 3.957771376137144e-05,
"loss": 0.0227,
"step": 125
},
{
"epoch": 0.4857831325301205,
"grad_norm": 0.32992950081825256,
"learning_rate": 3.955922239099035e-05,
"loss": 0.0207,
"step": 126
},
{
"epoch": 0.48963855421686747,
"grad_norm": 0.2520611882209778,
"learning_rate": 3.954033930003634e-05,
"loss": 0.0308,
"step": 127
},
{
"epoch": 0.49349397590361443,
"grad_norm": 0.3344120383262634,
"learning_rate": 3.952106486668884e-05,
"loss": 0.0187,
"step": 128
},
{
"epoch": 0.49734939759036145,
"grad_norm": 0.46383559703826904,
"learning_rate": 3.9501399476964806e-05,
"loss": 0.0174,
"step": 129
},
{
"epoch": 0.5012048192771085,
"grad_norm": 0.2674378752708435,
"learning_rate": 3.9481343524711095e-05,
"loss": 0.0164,
"step": 130
},
{
"epoch": 0.5050602409638554,
"grad_norm": 0.21262381970882416,
"learning_rate": 3.946089741159648e-05,
"loss": 0.0088,
"step": 131
},
{
"epoch": 0.5089156626506024,
"grad_norm": 0.2904111444950104,
"learning_rate": 3.944006154710366e-05,
"loss": 0.0195,
"step": 132
},
{
"epoch": 0.5127710843373494,
"grad_norm": 0.42615893483161926,
"learning_rate": 3.9418836348521045e-05,
"loss": 0.0172,
"step": 133
},
{
"epoch": 0.5166265060240964,
"grad_norm": 0.25124824047088623,
"learning_rate": 3.939722224093439e-05,
"loss": 0.0151,
"step": 134
},
{
"epoch": 0.5204819277108433,
"grad_norm": 1.9261884689331055,
"learning_rate": 3.937521965721831e-05,
"loss": 0.0247,
"step": 135
},
{
"epoch": 0.5243373493975904,
"grad_norm": 0.24449346959590912,
"learning_rate": 3.9352829038027564e-05,
"loss": 0.0113,
"step": 136
},
{
"epoch": 0.5281927710843374,
"grad_norm": 2.4259743690490723,
"learning_rate": 3.933005083178828e-05,
"loss": 0.0533,
"step": 137
},
{
"epoch": 0.5320481927710843,
"grad_norm": 0.3187374472618103,
"learning_rate": 3.930688549468894e-05,
"loss": 0.0179,
"step": 138
},
{
"epoch": 0.5359036144578313,
"grad_norm": 2.9974002838134766,
"learning_rate": 3.928333349067125e-05,
"loss": 0.023,
"step": 139
},
{
"epoch": 0.5397590361445783,
"grad_norm": 0.7724501490592957,
"learning_rate": 3.9259395291420864e-05,
"loss": 0.0179,
"step": 140
},
{
"epoch": 0.5436144578313253,
"grad_norm": 0.2811576724052429,
"learning_rate": 3.923507137635792e-05,
"loss": 0.0118,
"step": 141
},
{
"epoch": 0.5474698795180722,
"grad_norm": 0.301895409822464,
"learning_rate": 3.921036223262745e-05,
"loss": 0.0152,
"step": 142
},
{
"epoch": 0.5513253012048193,
"grad_norm": 0.32171159982681274,
"learning_rate": 3.9185268355089606e-05,
"loss": 0.0297,
"step": 143
},
{
"epoch": 0.5551807228915663,
"grad_norm": 0.16498999297618866,
"learning_rate": 3.915979024630978e-05,
"loss": 0.0134,
"step": 144
},
{
"epoch": 0.5590361445783133,
"grad_norm": 0.27746957540512085,
"learning_rate": 3.913392841654851e-05,
"loss": 0.013,
"step": 145
},
{
"epoch": 0.5628915662650602,
"grad_norm": 0.2298375517129898,
"learning_rate": 3.9107683383751264e-05,
"loss": 0.0106,
"step": 146
},
{
"epoch": 0.5667469879518072,
"grad_norm": 0.2801443636417389,
"learning_rate": 3.9081055673538093e-05,
"loss": 0.0214,
"step": 147
},
{
"epoch": 0.5706024096385542,
"grad_norm": 0.22764678299427032,
"learning_rate": 3.9054045819193074e-05,
"loss": 0.0148,
"step": 148
},
{
"epoch": 0.5744578313253013,
"grad_norm": 0.1658417284488678,
"learning_rate": 3.902665436165364e-05,
"loss": 0.0333,
"step": 149
},
{
"epoch": 0.5783132530120482,
"grad_norm": 0.1666492223739624,
"learning_rate": 3.8998881849499756e-05,
"loss": 0.0162,
"step": 150
},
{
"epoch": 0.5821686746987952,
"grad_norm": 0.20655786991119385,
"learning_rate": 3.897072883894291e-05,
"loss": 0.0152,
"step": 151
},
{
"epoch": 0.5860240963855422,
"grad_norm": 0.25162509083747864,
"learning_rate": 3.894219589381501e-05,
"loss": 0.0422,
"step": 152
},
{
"epoch": 0.5898795180722891,
"grad_norm": 0.17071524262428284,
"learning_rate": 3.8913283585557054e-05,
"loss": 0.0082,
"step": 153
},
{
"epoch": 0.5937349397590361,
"grad_norm": 0.20576122403144836,
"learning_rate": 3.8883992493207696e-05,
"loss": 0.0111,
"step": 154
},
{
"epoch": 0.5975903614457831,
"grad_norm": 0.19477005302906036,
"learning_rate": 3.885432320339167e-05,
"loss": 0.0084,
"step": 155
},
{
"epoch": 0.6014457831325302,
"grad_norm": 0.26111698150634766,
"learning_rate": 3.8824276310308e-05,
"loss": 0.0172,
"step": 156
},
{
"epoch": 0.6053012048192771,
"grad_norm": 0.2800676226615906,
"learning_rate": 3.879385241571817e-05,
"loss": 0.0238,
"step": 157
},
{
"epoch": 0.6091566265060241,
"grad_norm": 0.2781737148761749,
"learning_rate": 3.876305212893399e-05,
"loss": 0.0272,
"step": 158
},
{
"epoch": 0.6130120481927711,
"grad_norm": 0.6798153519630432,
"learning_rate": 3.873187606680543e-05,
"loss": 0.0091,
"step": 159
},
{
"epoch": 0.6168674698795181,
"grad_norm": 0.1639501005411148,
"learning_rate": 3.8700324853708304e-05,
"loss": 0.0233,
"step": 160
},
{
"epoch": 0.620722891566265,
"grad_norm": 0.39862754940986633,
"learning_rate": 3.866839912153168e-05,
"loss": 0.016,
"step": 161
},
{
"epoch": 0.624578313253012,
"grad_norm": 0.48113104701042175,
"learning_rate": 3.863609950966531e-05,
"loss": 0.016,
"step": 162
},
{
"epoch": 0.6284337349397591,
"grad_norm": 0.17467358708381653,
"learning_rate": 3.860342666498677e-05,
"loss": 0.0103,
"step": 163
},
{
"epoch": 0.632289156626506,
"grad_norm": 0.5942115783691406,
"learning_rate": 3.8570381241848534e-05,
"loss": 0.0158,
"step": 164
},
{
"epoch": 0.636144578313253,
"grad_norm": 0.2018311470746994,
"learning_rate": 3.853696390206484e-05,
"loss": 0.0193,
"step": 165
},
{
"epoch": 0.64,
"grad_norm": 0.5149483680725098,
"learning_rate": 3.850317531489848e-05,
"loss": 0.0268,
"step": 166
},
{
"epoch": 0.643855421686747,
"grad_norm": 0.16745764017105103,
"learning_rate": 3.846901615704734e-05,
"loss": 0.0132,
"step": 167
},
{
"epoch": 0.6477108433734939,
"grad_norm": 1.0696477890014648,
"learning_rate": 3.843448711263089e-05,
"loss": 0.0228,
"step": 168
},
{
"epoch": 0.651566265060241,
"grad_norm": 0.27323034405708313,
"learning_rate": 3.839958887317649e-05,
"loss": 0.0309,
"step": 169
},
{
"epoch": 0.655421686746988,
"grad_norm": 0.32417500019073486,
"learning_rate": 3.8364322137605484e-05,
"loss": 0.0282,
"step": 170
},
{
"epoch": 0.659277108433735,
"grad_norm": 0.24416105449199677,
"learning_rate": 3.832868761221926e-05,
"loss": 0.0149,
"step": 171
},
{
"epoch": 0.6631325301204819,
"grad_norm": 0.40103256702423096,
"learning_rate": 3.829268601068509e-05,
"loss": 0.0196,
"step": 172
},
{
"epoch": 0.6669879518072289,
"grad_norm": 0.4967587888240814,
"learning_rate": 3.825631805402182e-05,
"loss": 0.027,
"step": 173
},
{
"epoch": 0.6708433734939759,
"grad_norm": 0.32446494698524475,
"learning_rate": 3.821958447058543e-05,
"loss": 0.0186,
"step": 174
},
{
"epoch": 0.6746987951807228,
"grad_norm": 0.17924204468727112,
"learning_rate": 3.818248599605448e-05,
"loss": 0.0158,
"step": 175
},
{
"epoch": 0.6785542168674699,
"grad_norm": 0.32113662362098694,
"learning_rate": 3.8145023373415326e-05,
"loss": 0.0186,
"step": 176
},
{
"epoch": 0.6824096385542169,
"grad_norm": 0.2955350875854492,
"learning_rate": 3.810719735294731e-05,
"loss": 0.0254,
"step": 177
},
{
"epoch": 0.6862650602409639,
"grad_norm": 0.3955155611038208,
"learning_rate": 3.806900869220765e-05,
"loss": 0.0247,
"step": 178
},
{
"epoch": 0.6901204819277108,
"grad_norm": 0.28007134795188904,
"learning_rate": 3.8030458156016326e-05,
"loss": 0.0159,
"step": 179
},
{
"epoch": 0.6939759036144578,
"grad_norm": 0.3042643666267395,
"learning_rate": 3.799154651644075e-05,
"loss": 0.0163,
"step": 180
},
{
"epoch": 0.6978313253012048,
"grad_norm": 0.18276739120483398,
"learning_rate": 3.795227455278029e-05,
"loss": 0.0088,
"step": 181
},
{
"epoch": 0.7016867469879519,
"grad_norm": 0.19548135995864868,
"learning_rate": 3.7912643051550674e-05,
"loss": 0.0115,
"step": 182
},
{
"epoch": 0.7055421686746988,
"grad_norm": 0.2164657711982727,
"learning_rate": 3.787265280646825e-05,
"loss": 0.0152,
"step": 183
},
{
"epoch": 0.7093975903614458,
"grad_norm": 0.19452515244483948,
"learning_rate": 3.783230461843406e-05,
"loss": 0.008,
"step": 184
},
{
"epoch": 0.7132530120481928,
"grad_norm": 0.27214348316192627,
"learning_rate": 3.7791599295517825e-05,
"loss": 0.0127,
"step": 185
},
{
"epoch": 0.7171084337349397,
"grad_norm": 0.6383033990859985,
"learning_rate": 3.775053765294175e-05,
"loss": 0.0151,
"step": 186
},
{
"epoch": 0.7209638554216867,
"grad_norm": 0.17882253229618073,
"learning_rate": 3.7709120513064196e-05,
"loss": 0.0092,
"step": 187
},
{
"epoch": 0.7248192771084337,
"grad_norm": 0.2078862190246582,
"learning_rate": 3.7667348705363227e-05,
"loss": 0.0133,
"step": 188
},
{
"epoch": 0.7286746987951808,
"grad_norm": 0.1957724541425705,
"learning_rate": 3.762522306641998e-05,
"loss": 0.0108,
"step": 189
},
{
"epoch": 0.7325301204819277,
"grad_norm": 0.16418907046318054,
"learning_rate": 3.75827444399019e-05,
"loss": 0.0137,
"step": 190
},
{
"epoch": 0.7363855421686747,
"grad_norm": 0.45467960834503174,
"learning_rate": 3.7539913676545874e-05,
"loss": 0.0142,
"step": 191
},
{
"epoch": 0.7402409638554217,
"grad_norm": 1.2079054117202759,
"learning_rate": 3.749673163414117e-05,
"loss": 0.019,
"step": 192
},
{
"epoch": 0.7440963855421687,
"grad_norm": 0.11845115572214127,
"learning_rate": 3.745319917751229e-05,
"loss": 0.0098,
"step": 193
},
{
"epoch": 0.7479518072289156,
"grad_norm": 0.18299613893032074,
"learning_rate": 3.740931717850159e-05,
"loss": 0.0132,
"step": 194
},
{
"epoch": 0.7518072289156627,
"grad_norm": 0.225267231464386,
"learning_rate": 3.736508651595188e-05,
"loss": 0.0148,
"step": 195
},
{
"epoch": 0.7556626506024097,
"grad_norm": 0.49103838205337524,
"learning_rate": 3.732050807568878e-05,
"loss": 0.0132,
"step": 196
},
{
"epoch": 0.7595180722891566,
"grad_norm": 0.19766205549240112,
"learning_rate": 3.727558275050301e-05,
"loss": 0.02,
"step": 197
},
{
"epoch": 0.7633734939759036,
"grad_norm": 0.28864291310310364,
"learning_rate": 3.7230311440132494e-05,
"loss": 0.0222,
"step": 198
},
{
"epoch": 0.7672289156626506,
"grad_norm": 0.28816428780555725,
"learning_rate": 3.718469505124434e-05,
"loss": 0.0256,
"step": 199
},
{
"epoch": 0.7710843373493976,
"grad_norm": 0.20701606571674347,
"learning_rate": 3.713873449741669e-05,
"loss": 0.0114,
"step": 200
},
{
"epoch": 0.7749397590361445,
"grad_norm": 0.5334108471870422,
"learning_rate": 3.709243069912041e-05,
"loss": 0.0119,
"step": 201
},
{
"epoch": 0.7787951807228916,
"grad_norm": 0.7352932095527649,
"learning_rate": 3.704578458370067e-05,
"loss": 0.0148,
"step": 202
},
{
"epoch": 0.7826506024096386,
"grad_norm": 0.22625291347503662,
"learning_rate": 3.699879708535838e-05,
"loss": 0.0164,
"step": 203
},
{
"epoch": 0.7865060240963856,
"grad_norm": 0.2132428139448166,
"learning_rate": 3.695146914513142e-05,
"loss": 0.0083,
"step": 204
},
{
"epoch": 0.7903614457831325,
"grad_norm": 0.15848131477832794,
"learning_rate": 3.69038017108759e-05,
"loss": 0.0098,
"step": 205
},
{
"epoch": 0.7942168674698795,
"grad_norm": 1.2738759517669678,
"learning_rate": 3.685579573724707e-05,
"loss": 0.0137,
"step": 206
},
{
"epoch": 0.7980722891566265,
"grad_norm": 0.2131299078464508,
"learning_rate": 3.680745218568026e-05,
"loss": 0.019,
"step": 207
},
{
"epoch": 0.8019277108433734,
"grad_norm": 0.24905090034008026,
"learning_rate": 3.6758772024371626e-05,
"loss": 0.0242,
"step": 208
},
{
"epoch": 0.8057831325301205,
"grad_norm": 0.36632227897644043,
"learning_rate": 3.6709756228258735e-05,
"loss": 0.021,
"step": 209
},
{
"epoch": 0.8096385542168675,
"grad_norm": 0.19981122016906738,
"learning_rate": 3.6660405779001036e-05,
"loss": 0.0174,
"step": 210
},
{
"epoch": 0.8134939759036145,
"grad_norm": 0.13923773169517517,
"learning_rate": 3.6610721664960236e-05,
"loss": 0.0088,
"step": 211
},
{
"epoch": 0.8173493975903614,
"grad_norm": 0.27602291107177734,
"learning_rate": 3.656070488118047e-05,
"loss": 0.0391,
"step": 212
},
{
"epoch": 0.8212048192771084,
"grad_norm": 1.3190431594848633,
"learning_rate": 3.65103564293684e-05,
"loss": 0.0255,
"step": 213
},
{
"epoch": 0.8250602409638554,
"grad_norm": 0.15323355793952942,
"learning_rate": 3.645967731787313e-05,
"loss": 0.0164,
"step": 214
},
{
"epoch": 0.8289156626506025,
"grad_norm": 0.15326736867427826,
"learning_rate": 3.640866856166601e-05,
"loss": 0.0109,
"step": 215
},
{
"epoch": 0.8327710843373494,
"grad_norm": 0.2949405312538147,
"learning_rate": 3.635733118232034e-05,
"loss": 0.0117,
"step": 216
},
{
"epoch": 0.8366265060240964,
"grad_norm": 0.243217334151268,
"learning_rate": 3.6305666207990886e-05,
"loss": 0.0191,
"step": 217
},
{
"epoch": 0.8404819277108434,
"grad_norm": 0.43892231583595276,
"learning_rate": 3.625367467339329e-05,
"loss": 0.0174,
"step": 218
},
{
"epoch": 0.8443373493975903,
"grad_norm": 0.36832818388938904,
"learning_rate": 3.6201357619783336e-05,
"loss": 0.0201,
"step": 219
},
{
"epoch": 0.8481927710843373,
"grad_norm": 0.2650613486766815,
"learning_rate": 3.614871609493614e-05,
"loss": 0.0163,
"step": 220
},
{
"epoch": 0.8520481927710843,
"grad_norm": 0.1648036539554596,
"learning_rate": 3.609575115312511e-05,
"loss": 0.0111,
"step": 221
},
{
"epoch": 0.8559036144578314,
"grad_norm": 0.2738659083843231,
"learning_rate": 3.604246385510088e-05,
"loss": 0.0063,
"step": 222
},
{
"epoch": 0.8597590361445783,
"grad_norm": 0.22977620363235474,
"learning_rate": 3.598885526807003e-05,
"loss": 0.015,
"step": 223
},
{
"epoch": 0.8636144578313253,
"grad_norm": 0.535546064376831,
"learning_rate": 3.593492646567372e-05,
"loss": 0.0102,
"step": 224
},
{
"epoch": 0.8674698795180723,
"grad_norm": 0.494342565536499,
"learning_rate": 3.5880678527966224e-05,
"loss": 0.0102,
"step": 225
},
{
"epoch": 0.8713253012048193,
"grad_norm": 0.17625126242637634,
"learning_rate": 3.582611254139324e-05,
"loss": 0.0259,
"step": 226
},
{
"epoch": 0.8751807228915662,
"grad_norm": 0.20568431913852692,
"learning_rate": 3.577122959877017e-05,
"loss": 0.0237,
"step": 227
},
{
"epoch": 0.8790361445783132,
"grad_norm": 0.31444859504699707,
"learning_rate": 3.571603079926024e-05,
"loss": 0.0171,
"step": 228
},
{
"epoch": 0.8828915662650603,
"grad_norm": 0.48810216784477234,
"learning_rate": 3.566051724835245e-05,
"loss": 0.01,
"step": 229
},
{
"epoch": 0.8867469879518072,
"grad_norm": 0.4229196608066559,
"learning_rate": 3.5604690057839456e-05,
"loss": 0.0111,
"step": 230
},
{
"epoch": 0.8906024096385542,
"grad_norm": 0.6600443124771118,
"learning_rate": 3.554855034579532e-05,
"loss": 0.0208,
"step": 231
},
{
"epoch": 0.8944578313253012,
"grad_norm": 0.2891610860824585,
"learning_rate": 3.549209923655309e-05,
"loss": 0.0252,
"step": 232
},
{
"epoch": 0.8983132530120482,
"grad_norm": 0.20314915478229523,
"learning_rate": 3.5435337860682304e-05,
"loss": 0.0164,
"step": 233
},
{
"epoch": 0.9021686746987951,
"grad_norm": 0.19718259572982788,
"learning_rate": 3.537826735496632e-05,
"loss": 0.0125,
"step": 234
},
{
"epoch": 0.9060240963855422,
"grad_norm": 1.074033498764038,
"learning_rate": 3.532088886237956e-05,
"loss": 0.015,
"step": 235
},
{
"epoch": 0.9098795180722892,
"grad_norm": 0.7579637169837952,
"learning_rate": 3.5263203532064674e-05,
"loss": 0.0153,
"step": 236
},
{
"epoch": 0.9137349397590362,
"grad_norm": 0.15700305998325348,
"learning_rate": 3.520521251930941e-05,
"loss": 0.0109,
"step": 237
},
{
"epoch": 0.9175903614457831,
"grad_norm": 0.3394688069820404,
"learning_rate": 3.5146916985523604e-05,
"loss": 0.0156,
"step": 238
},
{
"epoch": 0.9214457831325301,
"grad_norm": 1.3616353273391724,
"learning_rate": 3.5088318098215805e-05,
"loss": 0.0194,
"step": 239
},
{
"epoch": 0.9253012048192771,
"grad_norm": 0.4036584198474884,
"learning_rate": 3.5029417030970005e-05,
"loss": 0.012,
"step": 240
},
{
"epoch": 0.929156626506024,
"grad_norm": 0.18816007673740387,
"learning_rate": 3.497021496342203e-05,
"loss": 0.0146,
"step": 241
},
{
"epoch": 0.9330120481927711,
"grad_norm": 0.17888401448726654,
"learning_rate": 3.4910713081235975e-05,
"loss": 0.0095,
"step": 242
},
{
"epoch": 0.9368674698795181,
"grad_norm": 0.22190752625465393,
"learning_rate": 3.485091257608047e-05,
"loss": 0.018,
"step": 243
},
{
"epoch": 0.9407228915662651,
"grad_norm": 0.48926958441734314,
"learning_rate": 3.479081464560475e-05,
"loss": 0.0407,
"step": 244
},
{
"epoch": 0.944578313253012,
"grad_norm": 0.236537367105484,
"learning_rate": 3.473042049341474e-05,
"loss": 0.0152,
"step": 245
},
{
"epoch": 0.948433734939759,
"grad_norm": 0.13449200987815857,
"learning_rate": 3.466973132904892e-05,
"loss": 0.0105,
"step": 246
},
{
"epoch": 0.952289156626506,
"grad_norm": 0.23878124356269836,
"learning_rate": 3.4608748367954064e-05,
"loss": 0.0154,
"step": 247
},
{
"epoch": 0.9561445783132531,
"grad_norm": 0.8225765824317932,
"learning_rate": 3.4547472831460976e-05,
"loss": 0.0258,
"step": 248
},
{
"epoch": 0.96,
"grad_norm": 0.2288682758808136,
"learning_rate": 3.4485905946759965e-05,
"loss": 0.0137,
"step": 249
},
{
"epoch": 0.963855421686747,
"grad_norm": 3.1449129581451416,
"learning_rate": 3.442404894687629e-05,
"loss": 0.009,
"step": 250
},
{
"epoch": 0.967710843373494,
"grad_norm": 0.36979448795318604,
"learning_rate": 3.4361903070645484e-05,
"loss": 0.0135,
"step": 251
},
{
"epoch": 0.971566265060241,
"grad_norm": 0.434368759393692,
"learning_rate": 3.4299469562688494e-05,
"loss": 0.0207,
"step": 252
},
{
"epoch": 0.9754216867469879,
"grad_norm": 0.20272351801395416,
"learning_rate": 3.423674967338681e-05,
"loss": 0.0166,
"step": 253
},
{
"epoch": 0.9792771084337349,
"grad_norm": 0.30525901913642883,
"learning_rate": 3.4173744658857403e-05,
"loss": 0.0088,
"step": 254
},
{
"epoch": 0.983132530120482,
"grad_norm": 0.8094114661216736,
"learning_rate": 3.411045578092754e-05,
"loss": 0.0246,
"step": 255
},
{
"epoch": 0.9869879518072289,
"grad_norm": 0.19253239035606384,
"learning_rate": 3.404688430710955e-05,
"loss": 0.0158,
"step": 256
},
{
"epoch": 0.9908433734939759,
"grad_norm": 0.18241408467292786,
"learning_rate": 3.398303151057543e-05,
"loss": 0.0208,
"step": 257
},
{
"epoch": 0.9946987951807229,
"grad_norm": 0.15632668137550354,
"learning_rate": 3.391889867013134e-05,
"loss": 0.014,
"step": 258
},
{
"epoch": 0.9985542168674699,
"grad_norm": 0.15484513342380524,
"learning_rate": 3.385448707019199e-05,
"loss": 0.0098,
"step": 259
},
{
"epoch": 1.0,
"grad_norm": 0.44285017251968384,
"learning_rate": 3.3789798000754924e-05,
"loss": 0.0181,
"step": 260
},
{
"epoch": 1.003855421686747,
"grad_norm": 0.22358648478984833,
"learning_rate": 3.372483275737468e-05,
"loss": 0.0123,
"step": 261
},
{
"epoch": 1.007710843373494,
"grad_norm": 0.26415324211120605,
"learning_rate": 3.365959264113683e-05,
"loss": 0.0097,
"step": 262
},
{
"epoch": 1.011566265060241,
"grad_norm": 0.16320234537124634,
"learning_rate": 3.359407895863199e-05,
"loss": 0.0168,
"step": 263
},
{
"epoch": 1.0154216867469879,
"grad_norm": 0.11904699355363846,
"learning_rate": 3.3528293021929523e-05,
"loss": 0.011,
"step": 264
},
{
"epoch": 1.0192771084337349,
"grad_norm": 0.19491620361804962,
"learning_rate": 3.34622361485514e-05,
"loss": 0.0203,
"step": 265
},
{
"epoch": 1.0231325301204819,
"grad_norm": 0.11890948563814163,
"learning_rate": 3.339590966144574e-05,
"loss": 0.0039,
"step": 266
},
{
"epoch": 1.026987951807229,
"grad_norm": 0.375266969203949,
"learning_rate": 3.332931488896029e-05,
"loss": 0.0108,
"step": 267
},
{
"epoch": 1.030843373493976,
"grad_norm": 0.2763790488243103,
"learning_rate": 3.326245316481591e-05,
"loss": 0.0231,
"step": 268
},
{
"epoch": 1.034698795180723,
"grad_norm": 0.13364073634147644,
"learning_rate": 3.319532582807977e-05,
"loss": 0.0096,
"step": 269
},
{
"epoch": 1.03855421686747,
"grad_norm": 0.14633706212043762,
"learning_rate": 3.31279342231386e-05,
"loss": 0.0085,
"step": 270
},
{
"epoch": 1.0424096385542168,
"grad_norm": 0.11920950561761856,
"learning_rate": 3.30602796996717e-05,
"loss": 0.0056,
"step": 271
},
{
"epoch": 1.0462650602409638,
"grad_norm": 0.2369781881570816,
"learning_rate": 3.299236361262401e-05,
"loss": 0.0092,
"step": 272
},
{
"epoch": 1.0501204819277108,
"grad_norm": 0.13494941592216492,
"learning_rate": 3.2924187322178865e-05,
"loss": 0.0084,
"step": 273
},
{
"epoch": 1.0539759036144578,
"grad_norm": 0.11123005300760269,
"learning_rate": 3.285575219373079e-05,
"loss": 0.0058,
"step": 274
},
{
"epoch": 1.0578313253012048,
"grad_norm": 0.18400225043296814,
"learning_rate": 3.278705959785821e-05,
"loss": 0.0113,
"step": 275
},
{
"epoch": 1.0616867469879518,
"grad_norm": 0.19578677415847778,
"learning_rate": 3.27181109102959e-05,
"loss": 0.008,
"step": 276
},
{
"epoch": 1.0655421686746989,
"grad_norm": 0.15978166460990906,
"learning_rate": 3.2648907511907544e-05,
"loss": 0.0074,
"step": 277
},
{
"epoch": 1.0693975903614459,
"grad_norm": 0.22295606136322021,
"learning_rate": 3.2579450788657997e-05,
"loss": 0.033,
"step": 278
},
{
"epoch": 1.0732530120481927,
"grad_norm": 0.11566543579101562,
"learning_rate": 3.250974213158555e-05,
"loss": 0.0097,
"step": 279
},
{
"epoch": 1.0771084337349397,
"grad_norm": 0.14439509809017181,
"learning_rate": 3.243978293677408e-05,
"loss": 0.0099,
"step": 280
},
{
"epoch": 1.0809638554216867,
"grad_norm": 0.21390187740325928,
"learning_rate": 3.23695746053251e-05,
"loss": 0.0115,
"step": 281
},
{
"epoch": 1.0848192771084337,
"grad_norm": 0.1340283751487732,
"learning_rate": 3.229911854332967e-05,
"loss": 0.0112,
"step": 282
},
{
"epoch": 1.0886746987951808,
"grad_norm": 0.08028711378574371,
"learning_rate": 3.222841616184025e-05,
"loss": 0.0044,
"step": 283
},
{
"epoch": 1.0925301204819278,
"grad_norm": 0.0875745564699173,
"learning_rate": 3.215746887684244e-05,
"loss": 0.005,
"step": 284
},
{
"epoch": 1.0963855421686748,
"grad_norm": 0.12632474303245544,
"learning_rate": 3.208627810922665e-05,
"loss": 0.0102,
"step": 285
},
{
"epoch": 1.1002409638554216,
"grad_norm": 0.07411441951990128,
"learning_rate": 3.201484528475958e-05,
"loss": 0.0051,
"step": 286
},
{
"epoch": 1.1040963855421686,
"grad_norm": 0.188252255320549,
"learning_rate": 3.194317183405573e-05,
"loss": 0.0088,
"step": 287
},
{
"epoch": 1.1079518072289156,
"grad_norm": 1.1099765300750732,
"learning_rate": 3.187125919254869e-05,
"loss": 0.0182,
"step": 288
},
{
"epoch": 1.1118072289156626,
"grad_norm": 0.09729217737913132,
"learning_rate": 3.1799108800462466e-05,
"loss": 0.0058,
"step": 289
},
{
"epoch": 1.1156626506024097,
"grad_norm": 0.25208383798599243,
"learning_rate": 3.172672210278254e-05,
"loss": 0.0104,
"step": 290
},
{
"epoch": 1.1195180722891567,
"grad_norm": 0.1512216329574585,
"learning_rate": 3.1654100549227024e-05,
"loss": 0.0058,
"step": 291
},
{
"epoch": 1.1233734939759037,
"grad_norm": 0.17274995148181915,
"learning_rate": 3.158124559421758e-05,
"loss": 0.0059,
"step": 292
},
{
"epoch": 1.1272289156626507,
"grad_norm": 0.32907241582870483,
"learning_rate": 3.1508158696850275e-05,
"loss": 0.0119,
"step": 293
},
{
"epoch": 1.1310843373493975,
"grad_norm": 0.1462509036064148,
"learning_rate": 3.14348413208664e-05,
"loss": 0.0093,
"step": 294
},
{
"epoch": 1.1349397590361445,
"grad_norm": 0.23990927636623383,
"learning_rate": 3.136129493462312e-05,
"loss": 0.0095,
"step": 295
},
{
"epoch": 1.1387951807228915,
"grad_norm": 0.13981395959854126,
"learning_rate": 3.12875210110641e-05,
"loss": 0.006,
"step": 296
},
{
"epoch": 1.1426506024096386,
"grad_norm": 0.10694962739944458,
"learning_rate": 3.121352102768998e-05,
"loss": 0.005,
"step": 297
},
{
"epoch": 1.1465060240963856,
"grad_norm": 0.12599286437034607,
"learning_rate": 3.113929646652879e-05,
"loss": 0.0057,
"step": 298
},
{
"epoch": 1.1503614457831326,
"grad_norm": 0.09407994896173477,
"learning_rate": 3.106484881410628e-05,
"loss": 0.0056,
"step": 299
},
{
"epoch": 1.1542168674698796,
"grad_norm": 0.7781950831413269,
"learning_rate": 3.0990179561416124e-05,
"loss": 0.017,
"step": 300
},
{
"epoch": 1.1580722891566264,
"grad_norm": 0.12835298478603363,
"learning_rate": 3.091529020389009e-05,
"loss": 0.0053,
"step": 301
},
{
"epoch": 1.1619277108433734,
"grad_norm": 0.7488322854042053,
"learning_rate": 3.0840182241368083e-05,
"loss": 0.011,
"step": 302
},
{
"epoch": 1.1657831325301204,
"grad_norm": 0.16700711846351624,
"learning_rate": 3.076485717806808e-05,
"loss": 0.0082,
"step": 303
},
{
"epoch": 1.1696385542168675,
"grad_norm": 0.17361144721508026,
"learning_rate": 3.0689316522556026e-05,
"loss": 0.0097,
"step": 304
},
{
"epoch": 1.1734939759036145,
"grad_norm": 0.08619613200426102,
"learning_rate": 3.061356178771564e-05,
"loss": 0.0082,
"step": 305
},
{
"epoch": 1.1773493975903615,
"grad_norm": 0.15666592121124268,
"learning_rate": 3.053759449071807e-05,
"loss": 0.0106,
"step": 306
},
{
"epoch": 1.1812048192771085,
"grad_norm": 0.2268570214509964,
"learning_rate": 3.0461416152991555e-05,
"loss": 0.0082,
"step": 307
},
{
"epoch": 1.1850602409638555,
"grad_norm": 0.17297455668449402,
"learning_rate": 3.038502830019092e-05,
"loss": 0.0105,
"step": 308
},
{
"epoch": 1.1889156626506023,
"grad_norm": 0.18929991126060486,
"learning_rate": 3.0308432462167045e-05,
"loss": 0.0076,
"step": 309
},
{
"epoch": 1.1927710843373494,
"grad_norm": 0.25152456760406494,
"learning_rate": 3.0231630172936205e-05,
"loss": 0.0072,
"step": 310
},
{
"epoch": 1.1966265060240964,
"grad_norm": 0.6995307207107544,
"learning_rate": 3.015462297064936e-05,
"loss": 0.0155,
"step": 311
},
{
"epoch": 1.2004819277108434,
"grad_norm": 0.14346441626548767,
"learning_rate": 3.0077412397561345e-05,
"loss": 0.0194,
"step": 312
},
{
"epoch": 1.2043373493975904,
"grad_norm": 0.2826595604419708,
"learning_rate": 3.0000000000000004e-05,
"loss": 0.0198,
"step": 313
},
{
"epoch": 1.2081927710843374,
"grad_norm": 0.1361214816570282,
"learning_rate": 2.992238732833518e-05,
"loss": 0.007,
"step": 314
},
{
"epoch": 1.2120481927710842,
"grad_norm": 0.13326340913772583,
"learning_rate": 2.98445759369477e-05,
"loss": 0.0116,
"step": 315
},
{
"epoch": 1.2159036144578312,
"grad_norm": 0.09617264568805695,
"learning_rate": 2.9766567384198228e-05,
"loss": 0.0082,
"step": 316
},
{
"epoch": 1.2197590361445783,
"grad_norm": 0.1366315484046936,
"learning_rate": 2.9688363232396056e-05,
"loss": 0.0075,
"step": 317
},
{
"epoch": 1.2236144578313253,
"grad_norm": 0.696820080280304,
"learning_rate": 2.960996504776783e-05,
"loss": 0.0087,
"step": 318
},
{
"epoch": 1.2274698795180723,
"grad_norm": 0.42925387620925903,
"learning_rate": 2.9531374400426158e-05,
"loss": 0.0277,
"step": 319
},
{
"epoch": 1.2313253012048193,
"grad_norm": 0.16131658852100372,
"learning_rate": 2.945259286433819e-05,
"loss": 0.0097,
"step": 320
},
{
"epoch": 1.2351807228915663,
"grad_norm": 0.1256534606218338,
"learning_rate": 2.9373622017294075e-05,
"loss": 0.0086,
"step": 321
},
{
"epoch": 1.2390361445783133,
"grad_norm": 0.15791697800159454,
"learning_rate": 2.9294463440875375e-05,
"loss": 0.0119,
"step": 322
},
{
"epoch": 1.2428915662650604,
"grad_norm": 0.12754394114017487,
"learning_rate": 2.9215118720423375e-05,
"loss": 0.0071,
"step": 323
},
{
"epoch": 1.2467469879518072,
"grad_norm": 0.9191262125968933,
"learning_rate": 2.913558944500736e-05,
"loss": 0.0097,
"step": 324
},
{
"epoch": 1.2506024096385542,
"grad_norm": 0.10129745304584503,
"learning_rate": 2.9055877207392752e-05,
"loss": 0.0162,
"step": 325
},
{
"epoch": 1.2544578313253012,
"grad_norm": 0.12669669091701508,
"learning_rate": 2.897598360400925e-05,
"loss": 0.0051,
"step": 326
},
{
"epoch": 1.2583132530120482,
"grad_norm": 0.1282287985086441,
"learning_rate": 2.8895910234918828e-05,
"loss": 0.0069,
"step": 327
},
{
"epoch": 1.2621686746987952,
"grad_norm": 0.10596070438623428,
"learning_rate": 2.8815658703783715e-05,
"loss": 0.0044,
"step": 328
},
{
"epoch": 1.266024096385542,
"grad_norm": 0.3934219777584076,
"learning_rate": 2.873523061783426e-05,
"loss": 0.0122,
"step": 329
},
{
"epoch": 1.269879518072289,
"grad_norm": 0.13733914494514465,
"learning_rate": 2.8654627587836724e-05,
"loss": 0.0052,
"step": 330
},
{
"epoch": 1.273734939759036,
"grad_norm": 0.21395160257816315,
"learning_rate": 2.8573851228061084e-05,
"loss": 0.0067,
"step": 331
},
{
"epoch": 1.277590361445783,
"grad_norm": 0.23374490439891815,
"learning_rate": 2.8492903156248642e-05,
"loss": 0.0111,
"step": 332
},
{
"epoch": 1.28144578313253,
"grad_norm": 0.14450624585151672,
"learning_rate": 2.8411784993579633e-05,
"loss": 0.0097,
"step": 333
},
{
"epoch": 1.2853012048192771,
"grad_norm": 0.12868821620941162,
"learning_rate": 2.8330498364640803e-05,
"loss": 0.0088,
"step": 334
},
{
"epoch": 1.2891566265060241,
"grad_norm": 0.2879028618335724,
"learning_rate": 2.8249044897392814e-05,
"loss": 0.0191,
"step": 335
},
{
"epoch": 1.2930120481927712,
"grad_norm": 0.10924389213323593,
"learning_rate": 2.8167426223137682e-05,
"loss": 0.0063,
"step": 336
},
{
"epoch": 1.2968674698795182,
"grad_norm": 0.14800770580768585,
"learning_rate": 2.80856439764861e-05,
"loss": 0.01,
"step": 337
},
{
"epoch": 1.3007228915662652,
"grad_norm": 0.1235174834728241,
"learning_rate": 2.8003699795324674e-05,
"loss": 0.0044,
"step": 338
},
{
"epoch": 1.304578313253012,
"grad_norm": 0.2660425305366516,
"learning_rate": 2.792159532078314e-05,
"loss": 0.0068,
"step": 339
},
{
"epoch": 1.308433734939759,
"grad_norm": 0.10823825001716614,
"learning_rate": 2.7839332197201505e-05,
"loss": 0.0041,
"step": 340
},
{
"epoch": 1.312289156626506,
"grad_norm": 0.9430158734321594,
"learning_rate": 2.77569120720971e-05,
"loss": 0.0141,
"step": 341
},
{
"epoch": 1.316144578313253,
"grad_norm": 0.1147020012140274,
"learning_rate": 2.7674336596131596e-05,
"loss": 0.0075,
"step": 342
},
{
"epoch": 1.32,
"grad_norm": 0.1446715146303177,
"learning_rate": 2.7591607423077932e-05,
"loss": 0.0094,
"step": 343
},
{
"epoch": 1.3238554216867469,
"grad_norm": 0.08478965610265732,
"learning_rate": 2.75087262097872e-05,
"loss": 0.0034,
"step": 344
},
{
"epoch": 1.3277108433734939,
"grad_norm": 0.08715037256479263,
"learning_rate": 2.7425694616155474e-05,
"loss": 0.0052,
"step": 345
},
{
"epoch": 1.331566265060241,
"grad_norm": 2.43072772026062,
"learning_rate": 2.7342514305090554e-05,
"loss": 0.0223,
"step": 346
},
{
"epoch": 1.335421686746988,
"grad_norm": 0.1595911830663681,
"learning_rate": 2.7259186942478656e-05,
"loss": 0.0042,
"step": 347
},
{
"epoch": 1.339277108433735,
"grad_norm": 0.2817286550998688,
"learning_rate": 2.717571419715107e-05,
"loss": 0.027,
"step": 348
},
{
"epoch": 1.343132530120482,
"grad_norm": 0.5103950500488281,
"learning_rate": 2.7092097740850712e-05,
"loss": 0.0085,
"step": 349
},
{
"epoch": 1.346987951807229,
"grad_norm": 1.2923753261566162,
"learning_rate": 2.700833924819868e-05,
"loss": 0.01,
"step": 350
},
{
"epoch": 1.350843373493976,
"grad_norm": 0.16906492412090302,
"learning_rate": 2.692444039666066e-05,
"loss": 0.0109,
"step": 351
},
{
"epoch": 1.354698795180723,
"grad_norm": 0.37144047021865845,
"learning_rate": 2.684040286651338e-05,
"loss": 0.0113,
"step": 352
},
{
"epoch": 1.3585542168674698,
"grad_norm": 0.3601152300834656,
"learning_rate": 2.6756228340810946e-05,
"loss": 0.0114,
"step": 353
},
{
"epoch": 1.3624096385542168,
"grad_norm": 0.5444064140319824,
"learning_rate": 2.6671918505351132e-05,
"loss": 0.0086,
"step": 354
},
{
"epoch": 1.3662650602409638,
"grad_norm": 0.14150206744670868,
"learning_rate": 2.6587475048641596e-05,
"loss": 0.0043,
"step": 355
},
{
"epoch": 1.3701204819277109,
"grad_norm": 0.09511060267686844,
"learning_rate": 2.6502899661866104e-05,
"loss": 0.0062,
"step": 356
},
{
"epoch": 1.3739759036144579,
"grad_norm": 0.16923631727695465,
"learning_rate": 2.6418194038850634e-05,
"loss": 0.0157,
"step": 357
},
{
"epoch": 1.377831325301205,
"grad_norm": 0.13884925842285156,
"learning_rate": 2.6333359876029455e-05,
"loss": 0.0058,
"step": 358
},
{
"epoch": 1.3816867469879517,
"grad_norm": 0.1783420443534851,
"learning_rate": 2.624839887241115e-05,
"loss": 0.0071,
"step": 359
},
{
"epoch": 1.3855421686746987,
"grad_norm": 0.15586194396018982,
"learning_rate": 2.6163312729544616e-05,
"loss": 0.0079,
"step": 360
},
{
"epoch": 1.3893975903614457,
"grad_norm": 0.22028474509716034,
"learning_rate": 2.607810315148494e-05,
"loss": 0.0078,
"step": 361
},
{
"epoch": 1.3932530120481927,
"grad_norm": 0.14147606492042542,
"learning_rate": 2.5992771844759315e-05,
"loss": 0.0078,
"step": 362
},
{
"epoch": 1.3971084337349398,
"grad_norm": 0.11384939402341843,
"learning_rate": 2.5907320518332827e-05,
"loss": 0.0067,
"step": 363
},
{
"epoch": 1.4009638554216868,
"grad_norm": 0.6917237639427185,
"learning_rate": 2.582175088357426e-05,
"loss": 0.0097,
"step": 364
},
{
"epoch": 1.4048192771084338,
"grad_norm": 0.26896387338638306,
"learning_rate": 2.5736064654221808e-05,
"loss": 0.0109,
"step": 365
},
{
"epoch": 1.4086746987951808,
"grad_norm": 0.1673029661178589,
"learning_rate": 2.5650263546348743e-05,
"loss": 0.0082,
"step": 366
},
{
"epoch": 1.4125301204819278,
"grad_norm": 0.1272229552268982,
"learning_rate": 2.5564349278329056e-05,
"loss": 0.0101,
"step": 367
},
{
"epoch": 1.4163855421686746,
"grad_norm": 0.1690957248210907,
"learning_rate": 2.547832357080305e-05,
"loss": 0.0125,
"step": 368
}
],
"logging_steps": 1,
"max_steps": 780,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 92,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 2.801896417658929e+18,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}