| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9968, | |
| "eval_steps": 500, | |
| "global_step": 104, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0192, | |
| "grad_norm": 2.423678398132324, | |
| "learning_rate": 1.8181818181818183e-06, | |
| "loss": 0.7932, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0384, | |
| "grad_norm": 2.6566708087921143, | |
| "learning_rate": 3.6363636363636366e-06, | |
| "loss": 0.7613, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0576, | |
| "grad_norm": 2.337695837020874, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 0.8108, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0768, | |
| "grad_norm": 1.94878351688385, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 0.7694, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.096, | |
| "grad_norm": 1.4400192499160767, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 0.7506, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.1152, | |
| "grad_norm": 1.514625906944275, | |
| "learning_rate": 1.0909090909090909e-05, | |
| "loss": 0.6907, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.1344, | |
| "grad_norm": 1.185624599456787, | |
| "learning_rate": 1.2727272727272728e-05, | |
| "loss": 0.68, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.1536, | |
| "grad_norm": 1.3777049779891968, | |
| "learning_rate": 1.4545454545454546e-05, | |
| "loss": 0.6178, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.1728, | |
| "grad_norm": 1.2356456518173218, | |
| "learning_rate": 1.6363636363636366e-05, | |
| "loss": 0.6484, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.192, | |
| "grad_norm": 0.9463790059089661, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 0.6111, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.2112, | |
| "grad_norm": 1.3397246599197388, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6009, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.2304, | |
| "grad_norm": 1.0940124988555908, | |
| "learning_rate": 1.999429490929718e-05, | |
| "loss": 0.6054, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.2496, | |
| "grad_norm": 0.9236961007118225, | |
| "learning_rate": 1.9977186146800707e-05, | |
| "loss": 0.5779, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.2688, | |
| "grad_norm": 0.9376490712165833, | |
| "learning_rate": 1.994869323391895e-05, | |
| "loss": 0.5472, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.288, | |
| "grad_norm": 0.8171873688697815, | |
| "learning_rate": 1.990884868158239e-05, | |
| "loss": 0.5776, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.3072, | |
| "grad_norm": 0.5843209624290466, | |
| "learning_rate": 1.985769795314804e-05, | |
| "loss": 0.5392, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.3264, | |
| "grad_norm": 0.8455219864845276, | |
| "learning_rate": 1.9795299412524948e-05, | |
| "loss": 0.5566, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.3456, | |
| "grad_norm": 0.8293880224227905, | |
| "learning_rate": 1.9721724257579907e-05, | |
| "loss": 0.5294, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.3648, | |
| "grad_norm": 0.5752674341201782, | |
| "learning_rate": 1.963705643889941e-05, | |
| "loss": 0.5143, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.384, | |
| "grad_norm": 0.6286266446113586, | |
| "learning_rate": 1.954139256400049e-05, | |
| "loss": 0.5348, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.4032, | |
| "grad_norm": 0.6185891032218933, | |
| "learning_rate": 1.9434841787099804e-05, | |
| "loss": 0.5125, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.4224, | |
| "grad_norm": 0.6374247074127197, | |
| "learning_rate": 1.9317525684566686e-05, | |
| "loss": 0.523, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.4416, | |
| "grad_norm": 0.5338121652603149, | |
| "learning_rate": 1.918957811620231e-05, | |
| "loss": 0.5475, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.4608, | |
| "grad_norm": 0.46796715259552, | |
| "learning_rate": 1.9051145072503216e-05, | |
| "loss": 0.5069, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.5485299825668335, | |
| "learning_rate": 1.8902384508083518e-05, | |
| "loss": 0.5207, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.4992, | |
| "grad_norm": 0.4771216809749603, | |
| "learning_rate": 1.8743466161445823e-05, | |
| "loss": 0.4958, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.5184, | |
| "grad_norm": 0.41287821531295776, | |
| "learning_rate": 1.857457136130651e-05, | |
| "loss": 0.4893, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.5376, | |
| "grad_norm": 0.42082297801971436, | |
| "learning_rate": 1.839589281969639e-05, | |
| "loss": 0.5023, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.5568, | |
| "grad_norm": 0.44994717836380005, | |
| "learning_rate": 1.8207634412072765e-05, | |
| "loss": 0.4896, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.576, | |
| "grad_norm": 0.4245263934135437, | |
| "learning_rate": 1.8010010944693846e-05, | |
| "loss": 0.5175, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.5952, | |
| "grad_norm": 0.3610984981060028, | |
| "learning_rate": 1.780324790952092e-05, | |
| "loss": 0.4903, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.6144, | |
| "grad_norm": 0.32125937938690186, | |
| "learning_rate": 1.758758122692791e-05, | |
| "loss": 0.4788, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.6336, | |
| "grad_norm": 0.382164865732193, | |
| "learning_rate": 1.7363256976511972e-05, | |
| "loss": 0.4968, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.6528, | |
| "grad_norm": 0.33746933937072754, | |
| "learning_rate": 1.7130531116312202e-05, | |
| "loss": 0.4759, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.672, | |
| "grad_norm": 0.38712188601493835, | |
| "learning_rate": 1.688966919075687e-05, | |
| "loss": 0.4855, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.6912, | |
| "grad_norm": 0.3574568033218384, | |
| "learning_rate": 1.6640946027672395e-05, | |
| "loss": 0.4848, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.7104, | |
| "grad_norm": 0.33736875653266907, | |
| "learning_rate": 1.6384645424699835e-05, | |
| "loss": 0.4924, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.7296, | |
| "grad_norm": 0.3253867030143738, | |
| "learning_rate": 1.612105982547663e-05, | |
| "loss": 0.4847, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.7488, | |
| "grad_norm": 0.40177786350250244, | |
| "learning_rate": 1.5850489985953076e-05, | |
| "loss": 0.4889, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.768, | |
| "grad_norm": 0.41558143496513367, | |
| "learning_rate": 1.5573244631224364e-05, | |
| "loss": 0.4797, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.7872, | |
| "grad_norm": 0.35887712240219116, | |
| "learning_rate": 1.5289640103269626e-05, | |
| "loss": 0.4884, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.8064, | |
| "grad_norm": 0.3083953559398651, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.4464, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.8256, | |
| "grad_norm": 0.3898048996925354, | |
| "learning_rate": 1.4704654806027558e-05, | |
| "loss": 0.4762, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.8448, | |
| "grad_norm": 0.3718198239803314, | |
| "learning_rate": 1.4403941515576344e-05, | |
| "loss": 0.4554, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.864, | |
| "grad_norm": 0.4089162349700928, | |
| "learning_rate": 1.4098203247965876e-05, | |
| "loss": 0.4893, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.8832, | |
| "grad_norm": 0.3150208592414856, | |
| "learning_rate": 1.3787788856105762e-05, | |
| "loss": 0.4668, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.9024, | |
| "grad_norm": 0.3481711447238922, | |
| "learning_rate": 1.3473052528448203e-05, | |
| "loss": 0.4757, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.9216, | |
| "grad_norm": 0.343963086605072, | |
| "learning_rate": 1.3154353384852559e-05, | |
| "loss": 0.487, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.9408, | |
| "grad_norm": 0.328693151473999, | |
| "learning_rate": 1.283205506682304e-05, | |
| "loss": 0.4862, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.3076961636543274, | |
| "learning_rate": 1.2506525322587207e-05, | |
| "loss": 0.4633, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.9792, | |
| "grad_norm": 0.3493748605251312, | |
| "learning_rate": 1.2178135587488515e-05, | |
| "loss": 0.4542, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.9984, | |
| "grad_norm": 0.3048100769519806, | |
| "learning_rate": 1.1847260560171895e-05, | |
| "loss": 0.4683, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 1.0176, | |
| "grad_norm": 0.8771734237670898, | |
| "learning_rate": 1.1514277775045768e-05, | |
| "loss": 0.8037, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 1.0368, | |
| "grad_norm": 0.3285163640975952, | |
| "learning_rate": 1.1179567171508463e-05, | |
| "loss": 0.4256, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 1.056, | |
| "grad_norm": 0.3323395252227783, | |
| "learning_rate": 1.0843510660430447e-05, | |
| "loss": 0.3927, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.0752, | |
| "grad_norm": 0.3365606963634491, | |
| "learning_rate": 1.0506491688387128e-05, | |
| "loss": 0.4463, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.0944, | |
| "grad_norm": 0.31812021136283875, | |
| "learning_rate": 1.0168894800139311e-05, | |
| "loss": 0.4116, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.1136, | |
| "grad_norm": 0.38818055391311646, | |
| "learning_rate": 9.83110519986069e-06, | |
| "loss": 0.463, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.1328, | |
| "grad_norm": 0.30687442421913147, | |
| "learning_rate": 9.493508311612874e-06, | |
| "loss": 0.4379, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.152, | |
| "grad_norm": 0.2939879596233368, | |
| "learning_rate": 9.156489339569555e-06, | |
| "loss": 0.4133, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.1712, | |
| "grad_norm": 0.3425482511520386, | |
| "learning_rate": 8.820432828491542e-06, | |
| "loss": 0.4336, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.1904, | |
| "grad_norm": 0.2907508909702301, | |
| "learning_rate": 8.485722224954237e-06, | |
| "loss": 0.4075, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.2096, | |
| "grad_norm": 0.3222837448120117, | |
| "learning_rate": 8.15273943982811e-06, | |
| "loss": 0.3901, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.2288000000000001, | |
| "grad_norm": 0.31006520986557007, | |
| "learning_rate": 7.821864412511485e-06, | |
| "loss": 0.4606, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.248, | |
| "grad_norm": 0.264212042093277, | |
| "learning_rate": 7.493474677412795e-06, | |
| "loss": 0.3955, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.2671999999999999, | |
| "grad_norm": 0.27319130301475525, | |
| "learning_rate": 7.16794493317696e-06, | |
| "loss": 0.4068, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.2864, | |
| "grad_norm": 0.28153327107429504, | |
| "learning_rate": 6.845646615147445e-06, | |
| "loss": 0.4234, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.3056, | |
| "grad_norm": 0.3072468042373657, | |
| "learning_rate": 6.526947471551799e-06, | |
| "loss": 0.4408, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.3248, | |
| "grad_norm": 0.2795600891113281, | |
| "learning_rate": 6.21221114389424e-06, | |
| "loss": 0.3869, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.3439999999999999, | |
| "grad_norm": 0.2556418180465698, | |
| "learning_rate": 5.901796752034128e-06, | |
| "loss": 0.3938, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.3632, | |
| "grad_norm": 0.28127527236938477, | |
| "learning_rate": 5.5960584844236565e-06, | |
| "loss": 0.4234, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.3824, | |
| "grad_norm": 0.2638583481311798, | |
| "learning_rate": 5.295345193972445e-06, | |
| "loss": 0.4703, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.4016, | |
| "grad_norm": 0.2525731325149536, | |
| "learning_rate": 5.000000000000003e-06, | |
| "loss": 0.3932, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.4208, | |
| "grad_norm": 0.25701409578323364, | |
| "learning_rate": 4.710359896730379e-06, | |
| "loss": 0.4027, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 0.29077157378196716, | |
| "learning_rate": 4.426755368775637e-06, | |
| "loss": 0.4311, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.4592, | |
| "grad_norm": 0.2595595717430115, | |
| "learning_rate": 4.149510014046922e-06, | |
| "loss": 0.376, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.4784, | |
| "grad_norm": 0.30198609828948975, | |
| "learning_rate": 3.878940174523371e-06, | |
| "loss": 0.4595, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.4976, | |
| "grad_norm": 0.2653771936893463, | |
| "learning_rate": 3.6153545753001663e-06, | |
| "loss": 0.3897, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.5168, | |
| "grad_norm": 0.25373411178588867, | |
| "learning_rate": 3.3590539723276083e-06, | |
| "loss": 0.4137, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.536, | |
| "grad_norm": 0.22587552666664124, | |
| "learning_rate": 3.110330809243134e-06, | |
| "loss": 0.3905, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.5552000000000001, | |
| "grad_norm": 0.2660175859928131, | |
| "learning_rate": 2.869468883687798e-06, | |
| "loss": 0.4473, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.5744, | |
| "grad_norm": 0.23747386038303375, | |
| "learning_rate": 2.6367430234880286e-06, | |
| "loss": 0.3892, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.5936, | |
| "grad_norm": 0.22769302129745483, | |
| "learning_rate": 2.4124187730720916e-06, | |
| "loss": 0.4159, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.6128, | |
| "grad_norm": 0.27097997069358826, | |
| "learning_rate": 2.196752090479083e-06, | |
| "loss": 0.3836, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.6320000000000001, | |
| "grad_norm": 0.2471567988395691, | |
| "learning_rate": 1.9899890553061565e-06, | |
| "loss": 0.409, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.6512, | |
| "grad_norm": 0.263854056596756, | |
| "learning_rate": 1.7923655879272395e-06, | |
| "loss": 0.4006, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.6703999999999999, | |
| "grad_norm": 0.23444706201553345, | |
| "learning_rate": 1.60410718030361e-06, | |
| "loss": 0.3883, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.6896, | |
| "grad_norm": 0.23784613609313965, | |
| "learning_rate": 1.425428638693489e-06, | |
| "loss": 0.4117, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.7088, | |
| "grad_norm": 0.25254783034324646, | |
| "learning_rate": 1.2565338385541792e-06, | |
| "loss": 0.4345, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.728, | |
| "grad_norm": 0.2234947681427002, | |
| "learning_rate": 1.097615491916485e-06, | |
| "loss": 0.4002, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.7471999999999999, | |
| "grad_norm": 0.2332143634557724, | |
| "learning_rate": 9.488549274967873e-07, | |
| "loss": 0.4242, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.7664, | |
| "grad_norm": 0.21657587587833405, | |
| "learning_rate": 8.10421883797694e-07, | |
| "loss": 0.3806, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.7856, | |
| "grad_norm": 0.20751075446605682, | |
| "learning_rate": 6.824743154333157e-07, | |
| "loss": 0.4056, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.8048, | |
| "grad_norm": 0.2177955061197281, | |
| "learning_rate": 5.651582129001987e-07, | |
| "loss": 0.3898, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.8239999999999998, | |
| "grad_norm": 0.22828799486160278, | |
| "learning_rate": 4.5860743599951186e-07, | |
| "loss": 0.3878, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.8432, | |
| "grad_norm": 0.2466111183166504, | |
| "learning_rate": 3.629435611005916e-07, | |
| "loss": 0.4132, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.8624, | |
| "grad_norm": 0.21871791779994965, | |
| "learning_rate": 2.7827574242009434e-07, | |
| "loss": 0.3657, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.8816000000000002, | |
| "grad_norm": 0.2354569435119629, | |
| "learning_rate": 2.0470058747505516e-07, | |
| "loss": 0.4006, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.9008, | |
| "grad_norm": 0.20780694484710693, | |
| "learning_rate": 1.4230204685196202e-07, | |
| "loss": 0.3919, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 0.19831795990467072, | |
| "learning_rate": 9.11513184176116e-08, | |
| "loss": 0.4231, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.9392, | |
| "grad_norm": 0.21838048100471497, | |
| "learning_rate": 5.1306766081048456e-08, | |
| "loss": 0.3911, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.9584000000000001, | |
| "grad_norm": 0.21957120299339294, | |
| "learning_rate": 2.2813853199292745e-08, | |
| "loss": 0.4391, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.9776, | |
| "grad_norm": 0.22633568942546844, | |
| "learning_rate": 5.705090702819993e-09, | |
| "loss": 0.4069, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.9968, | |
| "grad_norm": 0.2150619626045227, | |
| "learning_rate": 0.0, | |
| "loss": 0.4599, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.9968, | |
| "step": 104, | |
| "total_flos": 356056961384448.0, | |
| "train_loss": 0.48249776661396027, | |
| "train_runtime": 36791.677, | |
| "train_samples_per_second": 0.272, | |
| "train_steps_per_second": 0.003 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 104, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 356056961384448.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |