| { | |
| "best_metric": 1.6925917863845825, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 1.9138755980861244, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.009569377990430622, | |
| "grad_norm": 0.38082700967788696, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.6292, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.009569377990430622, | |
| "eval_loss": 1.831426739692688, | |
| "eval_runtime": 5.8102, | |
| "eval_samples_per_second": 30.464, | |
| "eval_steps_per_second": 7.745, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.019138755980861243, | |
| "grad_norm": 0.3949052691459656, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 1.5218, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.028708133971291867, | |
| "grad_norm": 0.4035767614841461, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.5025, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03827751196172249, | |
| "grad_norm": 0.41593310236930847, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 1.5595, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.04784688995215311, | |
| "grad_norm": 0.40817272663116455, | |
| "learning_rate": 4e-05, | |
| "loss": 1.4547, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.05741626794258373, | |
| "grad_norm": 0.42838597297668457, | |
| "learning_rate": 4.8e-05, | |
| "loss": 1.6265, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.06698564593301436, | |
| "grad_norm": 0.45393508672714233, | |
| "learning_rate": 5.6e-05, | |
| "loss": 1.6049, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.07655502392344497, | |
| "grad_norm": 0.43397629261016846, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 1.5147, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0861244019138756, | |
| "grad_norm": 0.45276832580566406, | |
| "learning_rate": 7.2e-05, | |
| "loss": 1.5187, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.09569377990430622, | |
| "grad_norm": 0.4829266369342804, | |
| "learning_rate": 8e-05, | |
| "loss": 1.5023, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.10526315789473684, | |
| "grad_norm": 0.5055172443389893, | |
| "learning_rate": 7.999453219969877e-05, | |
| "loss": 1.5217, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.11483253588516747, | |
| "grad_norm": 0.5195921659469604, | |
| "learning_rate": 7.997813029363704e-05, | |
| "loss": 1.4627, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.12440191387559808, | |
| "grad_norm": 0.5627723932266235, | |
| "learning_rate": 7.99507987659322e-05, | |
| "loss": 1.5943, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.1339712918660287, | |
| "grad_norm": 0.561619758605957, | |
| "learning_rate": 7.991254508875098e-05, | |
| "loss": 1.633, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.14354066985645933, | |
| "grad_norm": 0.5991867780685425, | |
| "learning_rate": 7.98633797202668e-05, | |
| "loss": 1.7317, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.15311004784688995, | |
| "grad_norm": 0.6589527130126953, | |
| "learning_rate": 7.980331610180046e-05, | |
| "loss": 1.7673, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.16267942583732056, | |
| "grad_norm": 0.6777735948562622, | |
| "learning_rate": 7.973237065414553e-05, | |
| "loss": 1.7703, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.1722488038277512, | |
| "grad_norm": 0.7253565192222595, | |
| "learning_rate": 7.965056277307902e-05, | |
| "loss": 1.887, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.18181818181818182, | |
| "grad_norm": 0.7007018327713013, | |
| "learning_rate": 7.955791482405875e-05, | |
| "loss": 1.8466, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.19138755980861244, | |
| "grad_norm": 0.7932828068733215, | |
| "learning_rate": 7.94544521361089e-05, | |
| "loss": 1.9501, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.20095693779904306, | |
| "grad_norm": 0.8723483085632324, | |
| "learning_rate": 7.93402029948953e-05, | |
| "loss": 1.8275, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.21052631578947367, | |
| "grad_norm": 0.9875766038894653, | |
| "learning_rate": 7.921519863499239e-05, | |
| "loss": 1.9756, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.22009569377990432, | |
| "grad_norm": 1.1991780996322632, | |
| "learning_rate": 7.907947323134398e-05, | |
| "loss": 2.0454, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.22966507177033493, | |
| "grad_norm": 1.3916829824447632, | |
| "learning_rate": 7.893306388992023e-05, | |
| "loss": 2.1839, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.23923444976076555, | |
| "grad_norm": 2.0087339878082275, | |
| "learning_rate": 7.877601063757323e-05, | |
| "loss": 1.8768, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.24880382775119617, | |
| "grad_norm": 3.617227554321289, | |
| "learning_rate": 7.860835641109395e-05, | |
| "loss": 2.2618, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.2583732057416268, | |
| "grad_norm": 0.3412221074104309, | |
| "learning_rate": 7.843014704547393e-05, | |
| "loss": 1.3234, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.2679425837320574, | |
| "grad_norm": 0.42941951751708984, | |
| "learning_rate": 7.824143126137431e-05, | |
| "loss": 1.4562, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.27751196172248804, | |
| "grad_norm": 0.44896551966667175, | |
| "learning_rate": 7.804226065180615e-05, | |
| "loss": 1.6191, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.28708133971291866, | |
| "grad_norm": 0.4686514437198639, | |
| "learning_rate": 7.783268966802539e-05, | |
| "loss": 1.5335, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.2966507177033493, | |
| "grad_norm": 0.41287559270858765, | |
| "learning_rate": 7.761277560464645e-05, | |
| "loss": 1.5145, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.3062200956937799, | |
| "grad_norm": 0.3899138271808624, | |
| "learning_rate": 7.738257858397844e-05, | |
| "loss": 1.5435, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.3157894736842105, | |
| "grad_norm": 0.38428395986557007, | |
| "learning_rate": 7.71421615395883e-05, | |
| "loss": 1.4655, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.3253588516746411, | |
| "grad_norm": 0.4115948975086212, | |
| "learning_rate": 7.68915901990954e-05, | |
| "loss": 1.4021, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.3349282296650718, | |
| "grad_norm": 0.428444504737854, | |
| "learning_rate": 7.663093306620231e-05, | |
| "loss": 1.5792, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.3444976076555024, | |
| "grad_norm": 0.4278987646102905, | |
| "learning_rate": 7.636026140196651e-05, | |
| "loss": 1.4603, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.35406698564593303, | |
| "grad_norm": 0.45422014594078064, | |
| "learning_rate": 7.607964920531837e-05, | |
| "loss": 1.5074, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.36363636363636365, | |
| "grad_norm": 0.4825778007507324, | |
| "learning_rate": 7.578917319283055e-05, | |
| "loss": 1.6137, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.37320574162679426, | |
| "grad_norm": 0.49149173498153687, | |
| "learning_rate": 7.548891277774448e-05, | |
| "loss": 1.5871, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.3827751196172249, | |
| "grad_norm": 0.5421543121337891, | |
| "learning_rate": 7.517895004825956e-05, | |
| "loss": 1.6206, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.3923444976076555, | |
| "grad_norm": 0.5325545072555542, | |
| "learning_rate": 7.48593697450911e-05, | |
| "loss": 1.5887, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.4019138755980861, | |
| "grad_norm": 0.583401620388031, | |
| "learning_rate": 7.453025923830296e-05, | |
| "loss": 1.8583, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.41148325358851673, | |
| "grad_norm": 0.618493378162384, | |
| "learning_rate": 7.419170850342156e-05, | |
| "loss": 1.7723, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.42105263157894735, | |
| "grad_norm": 0.6902540922164917, | |
| "learning_rate": 7.384381009683742e-05, | |
| "loss": 1.8642, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.430622009569378, | |
| "grad_norm": 0.6965833902359009, | |
| "learning_rate": 7.348665913050115e-05, | |
| "loss": 1.8881, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.44019138755980863, | |
| "grad_norm": 0.7248061299324036, | |
| "learning_rate": 7.312035324592081e-05, | |
| "loss": 1.8599, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.44976076555023925, | |
| "grad_norm": 0.7599021196365356, | |
| "learning_rate": 7.274499258746771e-05, | |
| "loss": 1.919, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.45933014354066987, | |
| "grad_norm": 0.8597122430801392, | |
| "learning_rate": 7.236067977499791e-05, | |
| "loss": 1.8464, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.4688995215311005, | |
| "grad_norm": 0.9372475147247314, | |
| "learning_rate": 7.196751987579699e-05, | |
| "loss": 1.8919, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.4784688995215311, | |
| "grad_norm": 1.1424167156219482, | |
| "learning_rate": 7.156562037585576e-05, | |
| "loss": 1.875, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4784688995215311, | |
| "eval_loss": 1.7351869344711304, | |
| "eval_runtime": 5.7379, | |
| "eval_samples_per_second": 30.848, | |
| "eval_steps_per_second": 7.843, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4880382775119617, | |
| "grad_norm": 1.656938076019287, | |
| "learning_rate": 7.11550911504845e-05, | |
| "loss": 1.9099, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.49760765550239233, | |
| "grad_norm": 2.4753801822662354, | |
| "learning_rate": 7.073604443427437e-05, | |
| "loss": 1.6779, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.507177033492823, | |
| "grad_norm": 0.29961133003234863, | |
| "learning_rate": 7.03085947904134e-05, | |
| "loss": 1.4577, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.5167464114832536, | |
| "grad_norm": 0.3351958394050598, | |
| "learning_rate": 6.987285907936617e-05, | |
| "loss": 1.5448, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.5263157894736842, | |
| "grad_norm": 0.3630099296569824, | |
| "learning_rate": 6.942895642692527e-05, | |
| "loss": 1.425, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5358851674641149, | |
| "grad_norm": 0.3872605860233307, | |
| "learning_rate": 6.897700819164357e-05, | |
| "loss": 1.5064, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.5454545454545454, | |
| "grad_norm": 0.3808247148990631, | |
| "learning_rate": 6.851713793165589e-05, | |
| "loss": 1.5932, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.5550239234449761, | |
| "grad_norm": 0.39085471630096436, | |
| "learning_rate": 6.804947137089955e-05, | |
| "loss": 1.4699, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.5645933014354066, | |
| "grad_norm": 0.38267335295677185, | |
| "learning_rate": 6.757413636474263e-05, | |
| "loss": 1.4573, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.5741626794258373, | |
| "grad_norm": 0.3885556757450104, | |
| "learning_rate": 6.709126286502965e-05, | |
| "loss": 1.477, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.583732057416268, | |
| "grad_norm": 0.40890273451805115, | |
| "learning_rate": 6.660098288455393e-05, | |
| "loss": 1.4652, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.5933014354066986, | |
| "grad_norm": 0.4209669828414917, | |
| "learning_rate": 6.610343046096674e-05, | |
| "loss": 1.5038, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.6028708133971292, | |
| "grad_norm": 0.420719176530838, | |
| "learning_rate": 6.559874162013267e-05, | |
| "loss": 1.5592, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.6124401913875598, | |
| "grad_norm": 0.4601193070411682, | |
| "learning_rate": 6.508705433894149e-05, | |
| "loss": 1.6004, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.6220095693779905, | |
| "grad_norm": 0.4896854758262634, | |
| "learning_rate": 6.456850850758673e-05, | |
| "loss": 1.6569, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.631578947368421, | |
| "grad_norm": 0.4936942756175995, | |
| "learning_rate": 6.404324589132101e-05, | |
| "loss": 1.6673, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.6411483253588517, | |
| "grad_norm": 0.5432229042053223, | |
| "learning_rate": 6.351141009169893e-05, | |
| "loss": 1.6781, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.6507177033492823, | |
| "grad_norm": 0.5662530064582825, | |
| "learning_rate": 6.297314650731775e-05, | |
| "loss": 1.7307, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.6602870813397129, | |
| "grad_norm": 0.6015027761459351, | |
| "learning_rate": 6.242860229406692e-05, | |
| "loss": 1.7204, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.6698564593301436, | |
| "grad_norm": 0.632143497467041, | |
| "learning_rate": 6.18779263248971e-05, | |
| "loss": 1.7661, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.6794258373205742, | |
| "grad_norm": 0.6799759268760681, | |
| "learning_rate": 6.132126914911976e-05, | |
| "loss": 1.8595, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.6889952153110048, | |
| "grad_norm": 0.7476505041122437, | |
| "learning_rate": 6.075878295124861e-05, | |
| "loss": 1.7854, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.6985645933014354, | |
| "grad_norm": 0.8099026083946228, | |
| "learning_rate": 6.019062150939376e-05, | |
| "loss": 1.8917, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.7081339712918661, | |
| "grad_norm": 0.8964723348617554, | |
| "learning_rate": 5.9616940153220336e-05, | |
| "loss": 1.9968, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.7177033492822966, | |
| "grad_norm": 0.9372945427894592, | |
| "learning_rate": 5.903789572148295e-05, | |
| "loss": 2.0208, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 1.2482417821884155, | |
| "learning_rate": 5.845364651914752e-05, | |
| "loss": 2.1952, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.7368421052631579, | |
| "grad_norm": 1.5300464630126953, | |
| "learning_rate": 5.786435227411227e-05, | |
| "loss": 1.7757, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.7464114832535885, | |
| "grad_norm": 2.1026394367218018, | |
| "learning_rate": 5.727017409353971e-05, | |
| "loss": 1.6011, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.7559808612440191, | |
| "grad_norm": 0.2756762206554413, | |
| "learning_rate": 5.667127441981162e-05, | |
| "loss": 1.4248, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.7655502392344498, | |
| "grad_norm": 0.3412489891052246, | |
| "learning_rate": 5.606781698611879e-05, | |
| "loss": 1.5389, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.7751196172248804, | |
| "grad_norm": 0.3573220670223236, | |
| "learning_rate": 5.5459966771698096e-05, | |
| "loss": 1.506, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.784688995215311, | |
| "grad_norm": 0.40763863921165466, | |
| "learning_rate": 5.4847889956728834e-05, | |
| "loss": 1.5875, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.7942583732057417, | |
| "grad_norm": 0.40273529291152954, | |
| "learning_rate": 5.423175387690067e-05, | |
| "loss": 1.4032, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.8038277511961722, | |
| "grad_norm": 0.41646698117256165, | |
| "learning_rate": 5.361172697766573e-05, | |
| "loss": 1.6233, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.8133971291866029, | |
| "grad_norm": 0.40466055274009705, | |
| "learning_rate": 5.298797876818735e-05, | |
| "loss": 1.4895, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.8229665071770335, | |
| "grad_norm": 0.42662107944488525, | |
| "learning_rate": 5.23606797749979e-05, | |
| "loss": 1.482, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.8325358851674641, | |
| "grad_norm": 0.44496554136276245, | |
| "learning_rate": 5.17300014953786e-05, | |
| "loss": 1.5595, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.8421052631578947, | |
| "grad_norm": 0.41382166743278503, | |
| "learning_rate": 5.109611635047379e-05, | |
| "loss": 1.3534, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.8516746411483254, | |
| "grad_norm": 0.44714534282684326, | |
| "learning_rate": 5.04591976381528e-05, | |
| "loss": 1.4973, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.861244019138756, | |
| "grad_norm": 0.4594970643520355, | |
| "learning_rate": 4.981941948563197e-05, | |
| "loss": 1.7519, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.8708133971291866, | |
| "grad_norm": 0.4741780161857605, | |
| "learning_rate": 4.9176956801870065e-05, | |
| "loss": 1.5083, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.8803827751196173, | |
| "grad_norm": 0.49775177240371704, | |
| "learning_rate": 4.853198522974988e-05, | |
| "loss": 1.6772, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.8899521531100478, | |
| "grad_norm": 0.5491089820861816, | |
| "learning_rate": 4.788468109805921e-05, | |
| "loss": 1.7706, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.8995215311004785, | |
| "grad_norm": 0.5795297622680664, | |
| "learning_rate": 4.7235221373284407e-05, | |
| "loss": 1.7992, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.9090909090909091, | |
| "grad_norm": 0.5943451523780823, | |
| "learning_rate": 4.658378361122936e-05, | |
| "loss": 1.7066, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.9186602870813397, | |
| "grad_norm": 0.6468049883842468, | |
| "learning_rate": 4.593054590847368e-05, | |
| "loss": 1.8196, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.9282296650717703, | |
| "grad_norm": 0.7615320086479187, | |
| "learning_rate": 4.5275686853682765e-05, | |
| "loss": 2.0393, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.937799043062201, | |
| "grad_norm": 0.7791320085525513, | |
| "learning_rate": 4.4619385478783456e-05, | |
| "loss": 1.9799, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.9473684210526315, | |
| "grad_norm": 0.870513379573822, | |
| "learning_rate": 4.396182121001852e-05, | |
| "loss": 2.058, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.9569377990430622, | |
| "grad_norm": 0.9276971220970154, | |
| "learning_rate": 4.33031738188933e-05, | |
| "loss": 1.8242, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9569377990430622, | |
| "eval_loss": 1.6937828063964844, | |
| "eval_runtime": 5.7937, | |
| "eval_samples_per_second": 30.551, | |
| "eval_steps_per_second": 7.767, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9665071770334929, | |
| "grad_norm": 1.0401110649108887, | |
| "learning_rate": 4.264362337302798e-05, | |
| "loss": 1.9696, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.9760765550239234, | |
| "grad_norm": 1.2967274188995361, | |
| "learning_rate": 4.1983350186928894e-05, | |
| "loss": 2.0238, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.9856459330143541, | |
| "grad_norm": 1.6171687841415405, | |
| "learning_rate": 4.132253477269233e-05, | |
| "loss": 1.7819, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.9952153110047847, | |
| "grad_norm": 2.1571907997131348, | |
| "learning_rate": 4.0661357790654345e-05, | |
| "loss": 1.4887, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.0047846889952152, | |
| "grad_norm": 0.6212150454521179, | |
| "learning_rate": 4e-05, | |
| "loss": 2.2897, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.014354066985646, | |
| "grad_norm": 0.2727871239185333, | |
| "learning_rate": 3.933864220934566e-05, | |
| "loss": 1.3308, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.0239234449760766, | |
| "grad_norm": 0.3046855330467224, | |
| "learning_rate": 3.8677465227307676e-05, | |
| "loss": 1.4615, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.0334928229665072, | |
| "grad_norm": 0.31469711661338806, | |
| "learning_rate": 3.8016649813071106e-05, | |
| "loss": 1.3841, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.0430622009569377, | |
| "grad_norm": 0.32190945744514465, | |
| "learning_rate": 3.735637662697203e-05, | |
| "loss": 1.3789, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.0526315789473684, | |
| "grad_norm": 0.3380140960216522, | |
| "learning_rate": 3.669682618110671e-05, | |
| "loss": 1.437, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.062200956937799, | |
| "grad_norm": 0.3436225652694702, | |
| "learning_rate": 3.6038178789981494e-05, | |
| "loss": 1.2847, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.0717703349282297, | |
| "grad_norm": 0.36676591634750366, | |
| "learning_rate": 3.538061452121656e-05, | |
| "loss": 1.297, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.0813397129186604, | |
| "grad_norm": 0.39319589734077454, | |
| "learning_rate": 3.472431314631724e-05, | |
| "loss": 1.489, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.0909090909090908, | |
| "grad_norm": 0.38813480734825134, | |
| "learning_rate": 3.406945409152632e-05, | |
| "loss": 1.4377, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.1004784688995215, | |
| "grad_norm": 0.41558539867401123, | |
| "learning_rate": 3.341621638877064e-05, | |
| "loss": 1.4287, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.1100478468899522, | |
| "grad_norm": 0.40909650921821594, | |
| "learning_rate": 3.276477862671562e-05, | |
| "loss": 1.441, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.1196172248803828, | |
| "grad_norm": 0.45959237217903137, | |
| "learning_rate": 3.21153189019408e-05, | |
| "loss": 1.6774, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.1291866028708135, | |
| "grad_norm": 0.46516451239585876, | |
| "learning_rate": 3.146801477025013e-05, | |
| "loss": 1.4372, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.138755980861244, | |
| "grad_norm": 0.5069877505302429, | |
| "learning_rate": 3.082304319812994e-05, | |
| "loss": 1.642, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.1483253588516746, | |
| "grad_norm": 0.5341264009475708, | |
| "learning_rate": 3.0180580514368037e-05, | |
| "loss": 1.6585, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.1578947368421053, | |
| "grad_norm": 0.56055748462677, | |
| "learning_rate": 2.9540802361847212e-05, | |
| "loss": 1.6557, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.167464114832536, | |
| "grad_norm": 0.5781103372573853, | |
| "learning_rate": 2.890388364952623e-05, | |
| "loss": 1.6671, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.1770334928229664, | |
| "grad_norm": 0.6097198724746704, | |
| "learning_rate": 2.8269998504621416e-05, | |
| "loss": 1.6451, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.186602870813397, | |
| "grad_norm": 0.6518456339836121, | |
| "learning_rate": 2.7639320225002108e-05, | |
| "loss": 1.6557, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.1961722488038278, | |
| "grad_norm": 0.6572650671005249, | |
| "learning_rate": 2.7012021231812666e-05, | |
| "loss": 1.4739, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.2057416267942584, | |
| "grad_norm": 0.783046543598175, | |
| "learning_rate": 2.638827302233428e-05, | |
| "loss": 1.8914, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.215311004784689, | |
| "grad_norm": 0.808749794960022, | |
| "learning_rate": 2.576824612309934e-05, | |
| "loss": 1.6836, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.2248803827751196, | |
| "grad_norm": 0.9802379012107849, | |
| "learning_rate": 2.5152110043271166e-05, | |
| "loss": 1.6596, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.2344497607655502, | |
| "grad_norm": 1.2099603414535522, | |
| "learning_rate": 2.454003322830192e-05, | |
| "loss": 1.5137, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.244019138755981, | |
| "grad_norm": 1.4862251281738281, | |
| "learning_rate": 2.393218301388123e-05, | |
| "loss": 1.1418, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.2535885167464116, | |
| "grad_norm": 1.0337613821029663, | |
| "learning_rate": 2.3328725580188395e-05, | |
| "loss": 1.2111, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.263157894736842, | |
| "grad_norm": 0.33979716897010803, | |
| "learning_rate": 2.272982590646029e-05, | |
| "loss": 1.4224, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.2727272727272727, | |
| "grad_norm": 0.37942126393318176, | |
| "learning_rate": 2.2135647725887744e-05, | |
| "loss": 1.4692, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.2822966507177034, | |
| "grad_norm": 0.3820892572402954, | |
| "learning_rate": 2.1546353480852495e-05, | |
| "loss": 1.3997, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.291866028708134, | |
| "grad_norm": 0.37589502334594727, | |
| "learning_rate": 2.096210427851706e-05, | |
| "loss": 1.44, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.3014354066985647, | |
| "grad_norm": 0.39906907081604004, | |
| "learning_rate": 2.038305984677969e-05, | |
| "loss": 1.4383, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.3110047846889952, | |
| "grad_norm": 0.3934566378593445, | |
| "learning_rate": 1.9809378490606264e-05, | |
| "loss": 1.3633, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.3205741626794258, | |
| "grad_norm": 0.4008135497570038, | |
| "learning_rate": 1.9241217048751406e-05, | |
| "loss": 1.4147, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.3301435406698565, | |
| "grad_norm": 0.4096076488494873, | |
| "learning_rate": 1.867873085088026e-05, | |
| "loss": 1.3079, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.339712918660287, | |
| "grad_norm": 0.4406437873840332, | |
| "learning_rate": 1.8122073675102935e-05, | |
| "loss": 1.4559, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.3492822966507176, | |
| "grad_norm": 0.42370909452438354, | |
| "learning_rate": 1.75713977059331e-05, | |
| "loss": 1.3705, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.3588516746411483, | |
| "grad_norm": 0.4583252966403961, | |
| "learning_rate": 1.702685349268226e-05, | |
| "loss": 1.4909, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.368421052631579, | |
| "grad_norm": 0.44213107228279114, | |
| "learning_rate": 1.648858990830108e-05, | |
| "loss": 1.3054, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.3779904306220097, | |
| "grad_norm": 0.46722662448883057, | |
| "learning_rate": 1.5956754108678996e-05, | |
| "loss": 1.443, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.38755980861244, | |
| "grad_norm": 0.48926499485969543, | |
| "learning_rate": 1.5431491492413288e-05, | |
| "loss": 1.4217, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.3971291866028708, | |
| "grad_norm": 0.5186731219291687, | |
| "learning_rate": 1.491294566105852e-05, | |
| "loss": 1.567, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.4066985645933014, | |
| "grad_norm": 0.5692178010940552, | |
| "learning_rate": 1.4401258379867335e-05, | |
| "loss": 1.6171, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.4162679425837321, | |
| "grad_norm": 0.5609941482543945, | |
| "learning_rate": 1.3896569539033253e-05, | |
| "loss": 1.5476, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.4258373205741628, | |
| "grad_norm": 0.6115127801895142, | |
| "learning_rate": 1.3399017115446067e-05, | |
| "loss": 1.5292, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.4354066985645932, | |
| "grad_norm": 0.6445640325546265, | |
| "learning_rate": 1.2908737134970367e-05, | |
| "loss": 1.6103, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.4354066985645932, | |
| "eval_loss": 1.6925917863845825, | |
| "eval_runtime": 5.7417, | |
| "eval_samples_per_second": 30.827, | |
| "eval_steps_per_second": 7.837, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.444976076555024, | |
| "grad_norm": 0.6974514722824097, | |
| "learning_rate": 1.242586363525737e-05, | |
| "loss": 1.7167, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.4545454545454546, | |
| "grad_norm": 0.7227963805198669, | |
| "learning_rate": 1.1950528629100457e-05, | |
| "loss": 1.6717, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.464114832535885, | |
| "grad_norm": 0.8344351053237915, | |
| "learning_rate": 1.1482862068344121e-05, | |
| "loss": 1.686, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.4736842105263157, | |
| "grad_norm": 0.9203869104385376, | |
| "learning_rate": 1.1022991808356442e-05, | |
| "loss": 1.5159, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.4832535885167464, | |
| "grad_norm": 1.1346211433410645, | |
| "learning_rate": 1.0571043573074737e-05, | |
| "loss": 1.5358, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.492822966507177, | |
| "grad_norm": 1.6541485786437988, | |
| "learning_rate": 1.0127140920633857e-05, | |
| "loss": 1.1705, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.5023923444976077, | |
| "grad_norm": 1.685089349746704, | |
| "learning_rate": 9.69140520958662e-06, | |
| "loss": 1.1268, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.5119617224880382, | |
| "grad_norm": 0.30736255645751953, | |
| "learning_rate": 9.263955565725648e-06, | |
| "loss": 1.3637, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.5215311004784688, | |
| "grad_norm": 0.3349371552467346, | |
| "learning_rate": 8.844908849515509e-06, | |
| "loss": 1.474, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.5311004784688995, | |
| "grad_norm": 0.35800015926361084, | |
| "learning_rate": 8.434379624144261e-06, | |
| "loss": 1.5093, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.5406698564593302, | |
| "grad_norm": 0.37739279866218567, | |
| "learning_rate": 8.032480124203013e-06, | |
| "loss": 1.4268, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.5502392344497609, | |
| "grad_norm": 0.40975770354270935, | |
| "learning_rate": 7.639320225002106e-06, | |
| "loss": 1.5343, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.5598086124401913, | |
| "grad_norm": 0.3935057520866394, | |
| "learning_rate": 7.255007412532307e-06, | |
| "loss": 1.4853, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.569377990430622, | |
| "grad_norm": 0.4058459401130676, | |
| "learning_rate": 6.8796467540791986e-06, | |
| "loss": 1.3923, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.5789473684210527, | |
| "grad_norm": 0.4203774631023407, | |
| "learning_rate": 6.513340869498859e-06, | |
| "loss": 1.3381, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.588516746411483, | |
| "grad_norm": 0.42906367778778076, | |
| "learning_rate": 6.1561899031625794e-06, | |
| "loss": 1.4494, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.598086124401914, | |
| "grad_norm": 0.43578189611434937, | |
| "learning_rate": 5.808291496578435e-06, | |
| "loss": 1.2433, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.6076555023923444, | |
| "grad_norm": 0.4657846987247467, | |
| "learning_rate": 5.469740761697044e-06, | |
| "loss": 1.5992, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.6172248803827751, | |
| "grad_norm": 0.49452725052833557, | |
| "learning_rate": 5.140630254908905e-06, | |
| "loss": 1.392, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.6267942583732058, | |
| "grad_norm": 0.5063107013702393, | |
| "learning_rate": 4.821049951740442e-06, | |
| "loss": 1.6072, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.6363636363636362, | |
| "grad_norm": 0.5106216073036194, | |
| "learning_rate": 4.511087222255528e-06, | |
| "loss": 1.3298, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.6459330143540671, | |
| "grad_norm": 0.5595565438270569, | |
| "learning_rate": 4.2108268071694616e-06, | |
| "loss": 1.5702, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.6555023923444976, | |
| "grad_norm": 0.5791851282119751, | |
| "learning_rate": 3.9203507946816445e-06, | |
| "loss": 1.505, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.6650717703349283, | |
| "grad_norm": 0.6414560079574585, | |
| "learning_rate": 3.6397385980335e-06, | |
| "loss": 1.6376, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.674641148325359, | |
| "grad_norm": 0.6914544105529785, | |
| "learning_rate": 3.3690669337977e-06, | |
| "loss": 1.7964, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.6842105263157894, | |
| "grad_norm": 0.7928903698921204, | |
| "learning_rate": 3.1084098009046106e-06, | |
| "loss": 1.8557, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.69377990430622, | |
| "grad_norm": 0.8021854162216187, | |
| "learning_rate": 2.8578384604117217e-06, | |
| "loss": 1.6362, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.7033492822966507, | |
| "grad_norm": 0.7553149461746216, | |
| "learning_rate": 2.6174214160215704e-06, | |
| "loss": 1.342, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.7129186602870812, | |
| "grad_norm": 0.9696720242500305, | |
| "learning_rate": 2.3872243953535535e-06, | |
| "loss": 1.7634, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.722488038277512, | |
| "grad_norm": 1.1475272178649902, | |
| "learning_rate": 2.1673103319746146e-06, | |
| "loss": 1.7356, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.7320574162679425, | |
| "grad_norm": 1.3815126419067383, | |
| "learning_rate": 1.957739348193859e-06, | |
| "loss": 1.6326, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.7416267942583732, | |
| "grad_norm": 2.003146171569824, | |
| "learning_rate": 1.7585687386256944e-06, | |
| "loss": 1.5472, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.7511961722488039, | |
| "grad_norm": 1.2215324640274048, | |
| "learning_rate": 1.5698529545260744e-06, | |
| "loss": 1.1225, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.7607655502392343, | |
| "grad_norm": 0.32720524072647095, | |
| "learning_rate": 1.3916435889060575e-06, | |
| "loss": 1.4342, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.7703349282296652, | |
| "grad_norm": 0.34399574995040894, | |
| "learning_rate": 1.2239893624267852e-06, | |
| "loss": 1.512, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.7799043062200957, | |
| "grad_norm": 0.3406524956226349, | |
| "learning_rate": 1.0669361100797704e-06, | |
| "loss": 1.2191, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.7894736842105263, | |
| "grad_norm": 0.3624950349330902, | |
| "learning_rate": 9.205267686560293e-07, | |
| "loss": 1.4136, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.799043062200957, | |
| "grad_norm": 0.4109777808189392, | |
| "learning_rate": 7.848013650076258e-07, | |
| "loss": 1.5427, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.8086124401913874, | |
| "grad_norm": 0.39658841490745544, | |
| "learning_rate": 6.597970051047053e-07, | |
| "loss": 1.3736, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.8181818181818183, | |
| "grad_norm": 0.41146281361579895, | |
| "learning_rate": 5.455478638911071e-07, | |
| "loss": 1.445, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.8277511961722488, | |
| "grad_norm": 0.410113126039505, | |
| "learning_rate": 4.420851759412603e-07, | |
| "loss": 1.2919, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.8373205741626795, | |
| "grad_norm": 0.43137213587760925, | |
| "learning_rate": 3.4943722692099224e-07, | |
| "loss": 1.3197, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.8468899521531101, | |
| "grad_norm": 0.43355488777160645, | |
| "learning_rate": 2.676293458544743e-07, | |
| "loss": 1.3578, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.8564593301435406, | |
| "grad_norm": 0.4816869795322418, | |
| "learning_rate": 1.9668389819954338e-07, | |
| "loss": 1.4214, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.8660287081339713, | |
| "grad_norm": 0.48422351479530334, | |
| "learning_rate": 1.3662027973320614e-07, | |
| "loss": 1.3847, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.875598086124402, | |
| "grad_norm": 0.522202730178833, | |
| "learning_rate": 8.745491124901861e-08, | |
| "loss": 1.5238, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.8851674641148324, | |
| "grad_norm": 0.5350428223609924, | |
| "learning_rate": 4.920123406781052e-08, | |
| "loss": 1.5597, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.8947368421052633, | |
| "grad_norm": 0.5734149813652039, | |
| "learning_rate": 2.1869706362958044e-08, | |
| "loss": 1.5932, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.9043062200956937, | |
| "grad_norm": 0.6013189554214478, | |
| "learning_rate": 5.467800301239834e-09, | |
| "loss": 1.6291, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.9138755980861244, | |
| "grad_norm": 0.6234821677207947, | |
| "learning_rate": 0.0, | |
| "loss": 1.4807, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.9138755980861244, | |
| "eval_loss": 1.7030987739562988, | |
| "eval_runtime": 5.7485, | |
| "eval_samples_per_second": 30.79, | |
| "eval_steps_per_second": 7.828, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 4, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.57896697905152e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |