|
{ |
|
"best_global_step": null, |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.18066847335140018, |
|
"eval_steps": 500, |
|
"global_step": 100, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0018066847335140017, |
|
"grad_norm": 2.6597820350894357, |
|
"learning_rate": 0.0, |
|
"loss": 0.8251, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0036133694670280035, |
|
"grad_norm": 2.7266510344811494, |
|
"learning_rate": 1.0714285714285714e-06, |
|
"loss": 0.8284, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.005420054200542005, |
|
"grad_norm": 2.7200407943823968, |
|
"learning_rate": 2.1428571428571427e-06, |
|
"loss": 0.8429, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.007226738934056007, |
|
"grad_norm": 2.6398267415203076, |
|
"learning_rate": 3.2142857142857143e-06, |
|
"loss": 0.8381, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.009033423667570008, |
|
"grad_norm": 2.450058379558589, |
|
"learning_rate": 4.2857142857142855e-06, |
|
"loss": 0.8245, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.01084010840108401, |
|
"grad_norm": 1.9536840033460428, |
|
"learning_rate": 5.357142857142857e-06, |
|
"loss": 0.814, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.012646793134598013, |
|
"grad_norm": 1.2180340367347295, |
|
"learning_rate": 6.428571428571429e-06, |
|
"loss": 0.7845, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.014453477868112014, |
|
"grad_norm": 1.1514973199014618, |
|
"learning_rate": 7.5e-06, |
|
"loss": 0.7874, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.016260162601626018, |
|
"grad_norm": 1.1427251731602237, |
|
"learning_rate": 8.571428571428571e-06, |
|
"loss": 0.7793, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.018066847335140017, |
|
"grad_norm": 2.4537001853666673, |
|
"learning_rate": 9.642857142857144e-06, |
|
"loss": 0.7624, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.01987353206865402, |
|
"grad_norm": 2.474412722550665, |
|
"learning_rate": 1.0714285714285714e-05, |
|
"loss": 0.7491, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.02168021680216802, |
|
"grad_norm": 2.333456673377541, |
|
"learning_rate": 1.1785714285714286e-05, |
|
"loss": 0.762, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.023486901535682024, |
|
"grad_norm": 1.6274020508818596, |
|
"learning_rate": 1.2857142857142857e-05, |
|
"loss": 0.7368, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.025293586269196026, |
|
"grad_norm": 1.563869927563279, |
|
"learning_rate": 1.3928571428571429e-05, |
|
"loss": 0.7378, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.02710027100271003, |
|
"grad_norm": 1.253626680340628, |
|
"learning_rate": 1.5e-05, |
|
"loss": 0.7271, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.028906955736224028, |
|
"grad_norm": 0.9176330164857595, |
|
"learning_rate": 1.6071428571428572e-05, |
|
"loss": 0.7131, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.03071364046973803, |
|
"grad_norm": 0.7185343991261351, |
|
"learning_rate": 1.7142857142857142e-05, |
|
"loss": 0.7131, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.032520325203252036, |
|
"grad_norm": 0.7884609645105337, |
|
"learning_rate": 1.8214285714285712e-05, |
|
"loss": 0.7045, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.03432700993676603, |
|
"grad_norm": 0.7399830971167167, |
|
"learning_rate": 1.928571428571429e-05, |
|
"loss": 0.6995, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.036133694670280034, |
|
"grad_norm": 0.7376328500886337, |
|
"learning_rate": 2.0357142857142858e-05, |
|
"loss": 0.6779, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.037940379403794036, |
|
"grad_norm": 0.6439873145292982, |
|
"learning_rate": 2.1428571428571428e-05, |
|
"loss": 0.6834, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.03974706413730804, |
|
"grad_norm": 0.5790542088592184, |
|
"learning_rate": 2.25e-05, |
|
"loss": 0.6721, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.04155374887082204, |
|
"grad_norm": 0.5191304250088061, |
|
"learning_rate": 2.357142857142857e-05, |
|
"loss": 0.6707, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.04336043360433604, |
|
"grad_norm": 0.4488946670945292, |
|
"learning_rate": 2.464285714285714e-05, |
|
"loss": 0.676, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.045167118337850046, |
|
"grad_norm": 0.4798647764154148, |
|
"learning_rate": 2.5714285714285714e-05, |
|
"loss": 0.6585, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.04697380307136405, |
|
"grad_norm": 0.4204330566493729, |
|
"learning_rate": 2.6785714285714288e-05, |
|
"loss": 0.673, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.04878048780487805, |
|
"grad_norm": 0.47130112684574876, |
|
"learning_rate": 2.7857142857142858e-05, |
|
"loss": 0.6516, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.05058717253839205, |
|
"grad_norm": 0.4362815754533529, |
|
"learning_rate": 2.892857142857143e-05, |
|
"loss": 0.6545, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.052393857271906055, |
|
"grad_norm": 0.43822648560890193, |
|
"learning_rate": 3e-05, |
|
"loss": 0.6647, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.05420054200542006, |
|
"grad_norm": 0.3853311574389301, |
|
"learning_rate": 2.9999731440137413e-05, |
|
"loss": 0.6644, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.05600722673893405, |
|
"grad_norm": 0.4221785516753897, |
|
"learning_rate": 2.9998925770166232e-05, |
|
"loss": 0.6534, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.057813911472448055, |
|
"grad_norm": 0.3980570059327311, |
|
"learning_rate": 2.9997583018935875e-05, |
|
"loss": 0.6416, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.05962059620596206, |
|
"grad_norm": 0.35571676272289743, |
|
"learning_rate": 2.9995703234527553e-05, |
|
"loss": 0.6513, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.06142728093947606, |
|
"grad_norm": 0.38024957472516396, |
|
"learning_rate": 2.999328648425255e-05, |
|
"loss": 0.6385, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.06323396567299007, |
|
"grad_norm": 0.2924858728569485, |
|
"learning_rate": 2.999033285464982e-05, |
|
"loss": 0.644, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.06504065040650407, |
|
"grad_norm": 0.3482798299572746, |
|
"learning_rate": 2.9986842451482876e-05, |
|
"loss": 0.6387, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.06684733514001806, |
|
"grad_norm": 0.3705337140725527, |
|
"learning_rate": 2.9982815399736008e-05, |
|
"loss": 0.6432, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.06865401987353206, |
|
"grad_norm": 0.3364721163164359, |
|
"learning_rate": 2.9978251843609816e-05, |
|
"loss": 0.6351, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.07046070460704607, |
|
"grad_norm": 0.3272631660096153, |
|
"learning_rate": 2.9973151946516027e-05, |
|
"loss": 0.6364, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.07226738934056007, |
|
"grad_norm": 0.29708127282462954, |
|
"learning_rate": 2.996751589107167e-05, |
|
"loss": 0.6252, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.07407407407407407, |
|
"grad_norm": 0.24921077158970448, |
|
"learning_rate": 2.9961343879092512e-05, |
|
"loss": 0.6308, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.07588075880758807, |
|
"grad_norm": 0.3479569257927118, |
|
"learning_rate": 2.9954636131585845e-05, |
|
"loss": 0.6234, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.07768744354110207, |
|
"grad_norm": 0.28917511643628596, |
|
"learning_rate": 2.9947392888742566e-05, |
|
"loss": 0.6233, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.07949412827461608, |
|
"grad_norm": 0.34737822266968965, |
|
"learning_rate": 2.993961440992859e-05, |
|
"loss": 0.6171, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.08130081300813008, |
|
"grad_norm": 0.3247689228102166, |
|
"learning_rate": 2.993130097367553e-05, |
|
"loss": 0.6238, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.08310749774164408, |
|
"grad_norm": 0.30719902123424325, |
|
"learning_rate": 2.9922452877670775e-05, |
|
"loss": 0.6252, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.08491418247515808, |
|
"grad_norm": 0.3175141153740525, |
|
"learning_rate": 2.991307043874677e-05, |
|
"loss": 0.6281, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.08672086720867209, |
|
"grad_norm": 0.31244488771705314, |
|
"learning_rate": 2.9903153992869734e-05, |
|
"loss": 0.6148, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.08852755194218609, |
|
"grad_norm": 0.3429131319420423, |
|
"learning_rate": 2.989270389512756e-05, |
|
"loss": 0.6211, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.09033423667570009, |
|
"grad_norm": 0.3104062918234721, |
|
"learning_rate": 2.988172051971717e-05, |
|
"loss": 0.6204, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.0921409214092141, |
|
"grad_norm": 0.31461103786584654, |
|
"learning_rate": 2.9870204259931062e-05, |
|
"loss": 0.6097, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.0939476061427281, |
|
"grad_norm": 0.3151625720082357, |
|
"learning_rate": 2.9858155528143256e-05, |
|
"loss": 0.6268, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.0957542908762421, |
|
"grad_norm": 0.2860126481493585, |
|
"learning_rate": 2.9845574755794522e-05, |
|
"loss": 0.6125, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.0975609756097561, |
|
"grad_norm": 0.28841121592431124, |
|
"learning_rate": 2.9832462393376926e-05, |
|
"loss": 0.6018, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.0993676603432701, |
|
"grad_norm": 0.23534181927847642, |
|
"learning_rate": 2.9818818910417706e-05, |
|
"loss": 0.6169, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.1011743450767841, |
|
"grad_norm": 0.28661471046317555, |
|
"learning_rate": 2.9804644795462437e-05, |
|
"loss": 0.6212, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.10298102981029811, |
|
"grad_norm": 0.2443857675436309, |
|
"learning_rate": 2.9789940556057574e-05, |
|
"loss": 0.6065, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.10478771454381211, |
|
"grad_norm": 0.292418070185767, |
|
"learning_rate": 2.9774706718732255e-05, |
|
"loss": 0.6152, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.10659439927732611, |
|
"grad_norm": 0.24742182455944323, |
|
"learning_rate": 2.9758943828979444e-05, |
|
"loss": 0.6009, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.10840108401084012, |
|
"grad_norm": 0.25007211477600044, |
|
"learning_rate": 2.9742652451236414e-05, |
|
"loss": 0.6023, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.1102077687443541, |
|
"grad_norm": 0.24539117288225054, |
|
"learning_rate": 2.972583316886451e-05, |
|
"loss": 0.6059, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.1120144534778681, |
|
"grad_norm": 0.2544907417129908, |
|
"learning_rate": 2.9708486584128303e-05, |
|
"loss": 0.6107, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.11382113821138211, |
|
"grad_norm": 0.25269030392560393, |
|
"learning_rate": 2.9690613318173966e-05, |
|
"loss": 0.611, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.11562782294489611, |
|
"grad_norm": 0.20651691272156247, |
|
"learning_rate": 2.9672214011007087e-05, |
|
"loss": 0.6091, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.11743450767841011, |
|
"grad_norm": 0.25759031036445135, |
|
"learning_rate": 2.9653289321469715e-05, |
|
"loss": 0.6031, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.11924119241192412, |
|
"grad_norm": 0.27004948462624506, |
|
"learning_rate": 2.9633839927216793e-05, |
|
"loss": 0.6061, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.12104787714543812, |
|
"grad_norm": 0.23899477664304233, |
|
"learning_rate": 2.9613866524691867e-05, |
|
"loss": 0.5958, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.12285456187895212, |
|
"grad_norm": 0.3109356502387666, |
|
"learning_rate": 2.9593369829102173e-05, |
|
"loss": 0.6119, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.12466124661246612, |
|
"grad_norm": 0.23973937288759506, |
|
"learning_rate": 2.957235057439301e-05, |
|
"loss": 0.6135, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.12646793134598014, |
|
"grad_norm": 0.29402451520285133, |
|
"learning_rate": 2.955080951322147e-05, |
|
"loss": 0.6178, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.12827461607949414, |
|
"grad_norm": 0.29085498534508775, |
|
"learning_rate": 2.9528747416929467e-05, |
|
"loss": 0.5907, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.13008130081300814, |
|
"grad_norm": 0.25477514255751904, |
|
"learning_rate": 2.9506165075516148e-05, |
|
"loss": 0.6073, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.13188798554652212, |
|
"grad_norm": 0.28366229733780235, |
|
"learning_rate": 2.9483063297609577e-05, |
|
"loss": 0.6063, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.13369467028003612, |
|
"grad_norm": 0.26317649422081363, |
|
"learning_rate": 2.9459442910437798e-05, |
|
"loss": 0.598, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.13550135501355012, |
|
"grad_norm": 0.28039599919104496, |
|
"learning_rate": 2.94353047597992e-05, |
|
"loss": 0.5971, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.13730803974706413, |
|
"grad_norm": 0.271745198157929, |
|
"learning_rate": 2.941064971003224e-05, |
|
"loss": 0.593, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.13911472448057813, |
|
"grad_norm": 0.24998502389958926, |
|
"learning_rate": 2.9385478643984484e-05, |
|
"loss": 0.5878, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.14092140921409213, |
|
"grad_norm": 0.30322559816019845, |
|
"learning_rate": 2.9359792462981007e-05, |
|
"loss": 0.599, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.14272809394760613, |
|
"grad_norm": 0.29351194680487536, |
|
"learning_rate": 2.9333592086792113e-05, |
|
"loss": 0.6075, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.14453477868112014, |
|
"grad_norm": 0.2670107767188747, |
|
"learning_rate": 2.9306878453600382e-05, |
|
"loss": 0.6161, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.14634146341463414, |
|
"grad_norm": 0.2927366904918854, |
|
"learning_rate": 2.9279652519967105e-05, |
|
"loss": 0.6004, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.14814814814814814, |
|
"grad_norm": 0.28777141201555767, |
|
"learning_rate": 2.9251915260798024e-05, |
|
"loss": 0.593, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.14995483288166214, |
|
"grad_norm": 0.282262421377073, |
|
"learning_rate": 2.9223667669308395e-05, |
|
"loss": 0.6038, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.15176151761517614, |
|
"grad_norm": 0.24811154699761442, |
|
"learning_rate": 2.9194910756987464e-05, |
|
"loss": 0.5976, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.15356820234869015, |
|
"grad_norm": 0.27670190484589513, |
|
"learning_rate": 2.9165645553562215e-05, |
|
"loss": 0.5997, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.15537488708220415, |
|
"grad_norm": 0.24338487756299684, |
|
"learning_rate": 2.9135873106960525e-05, |
|
"loss": 0.5994, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.15718157181571815, |
|
"grad_norm": 0.3040191021283788, |
|
"learning_rate": 2.9105594483273603e-05, |
|
"loss": 0.6012, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.15898825654923215, |
|
"grad_norm": 0.26703924304748566, |
|
"learning_rate": 2.9074810766717865e-05, |
|
"loss": 0.6058, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.16079494128274616, |
|
"grad_norm": 0.2694312104706441, |
|
"learning_rate": 2.904352305959606e-05, |
|
"loss": 0.602, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.16260162601626016, |
|
"grad_norm": 0.29517928394037635, |
|
"learning_rate": 2.9011732482257835e-05, |
|
"loss": 0.5944, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.16440831074977416, |
|
"grad_norm": 0.25118801401605734, |
|
"learning_rate": 2.89794401730596e-05, |
|
"loss": 0.5923, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.16621499548328816, |
|
"grad_norm": 0.3348680034328015, |
|
"learning_rate": 2.894664728832377e-05, |
|
"loss": 0.5958, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.16802168021680217, |
|
"grad_norm": 0.25945936645525364, |
|
"learning_rate": 2.8913355002297367e-05, |
|
"loss": 0.6025, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.16982836495031617, |
|
"grad_norm": 0.31672412557275315, |
|
"learning_rate": 2.887956450710995e-05, |
|
"loss": 0.602, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.17163504968383017, |
|
"grad_norm": 0.2631935513788724, |
|
"learning_rate": 2.8845277012730963e-05, |
|
"loss": 0.5939, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.17344173441734417, |
|
"grad_norm": 0.28309967073656406, |
|
"learning_rate": 2.8810493746926364e-05, |
|
"loss": 0.5932, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.17524841915085818, |
|
"grad_norm": 0.28939178950900774, |
|
"learning_rate": 2.87752159552147e-05, |
|
"loss": 0.5993, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.17705510388437218, |
|
"grad_norm": 0.3021445789644071, |
|
"learning_rate": 2.87394449008225e-05, |
|
"loss": 0.5924, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.17886178861788618, |
|
"grad_norm": 0.2648161341085879, |
|
"learning_rate": 2.8703181864639013e-05, |
|
"loss": 0.6058, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.18066847335140018, |
|
"grad_norm": 0.28484263640616314, |
|
"learning_rate": 2.8666428145170385e-05, |
|
"loss": 0.5789, |
|
"step": 100 |
|
} |
|
], |
|
"logging_steps": 1, |
|
"max_steps": 553, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 100, |
|
"stateful_callbacks": { |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": false |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 115440686923776.0, |
|
"train_batch_size": 2, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|