| { | |
| "best_metric": 11.5, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.13499831252109348, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0006749915626054674, | |
| "grad_norm": 1.9157516362611204e-05, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 46.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0006749915626054674, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 11.0558, | |
| "eval_samples_per_second": 225.674, | |
| "eval_steps_per_second": 56.441, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0013499831252109348, | |
| "grad_norm": 2.201150346081704e-05, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 46.0, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0020249746878164025, | |
| "grad_norm": 1.953383798536379e-05, | |
| "learning_rate": 2.4e-05, | |
| "loss": 46.0, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0026999662504218697, | |
| "grad_norm": 1.975701525225304e-05, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 46.0, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0033749578130273373, | |
| "grad_norm": 2.441896685922984e-05, | |
| "learning_rate": 4e-05, | |
| "loss": 46.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.004049949375632805, | |
| "grad_norm": 2.1600873878924176e-05, | |
| "learning_rate": 4.8e-05, | |
| "loss": 46.0, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.004724940938238272, | |
| "grad_norm": 2.630273229442537e-05, | |
| "learning_rate": 5.6e-05, | |
| "loss": 46.0, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.005399932500843739, | |
| "grad_norm": 2.1237785404082388e-05, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 46.0, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.006074924063449207, | |
| "grad_norm": 2.6038042051368393e-05, | |
| "learning_rate": 7.2e-05, | |
| "loss": 46.0, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.006749915626054675, | |
| "grad_norm": 1.9523216906236485e-05, | |
| "learning_rate": 8e-05, | |
| "loss": 46.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.007424907188660142, | |
| "grad_norm": 3.139385080430657e-05, | |
| "learning_rate": 7.999453219969877e-05, | |
| "loss": 46.0, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.00809989875126561, | |
| "grad_norm": 2.8158918212284334e-05, | |
| "learning_rate": 7.997813029363704e-05, | |
| "loss": 46.0, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.008774890313871077, | |
| "grad_norm": 1.8800117686623707e-05, | |
| "learning_rate": 7.99507987659322e-05, | |
| "loss": 46.0, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.009449881876476544, | |
| "grad_norm": 2.5126875698333606e-05, | |
| "learning_rate": 7.991254508875098e-05, | |
| "loss": 46.0, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.010124873439082012, | |
| "grad_norm": 2.2397709471988492e-05, | |
| "learning_rate": 7.98633797202668e-05, | |
| "loss": 46.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.010799865001687479, | |
| "grad_norm": 1.7211687008966692e-05, | |
| "learning_rate": 7.980331610180046e-05, | |
| "loss": 46.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.011474856564292946, | |
| "grad_norm": 2.2995038307271898e-05, | |
| "learning_rate": 7.973237065414553e-05, | |
| "loss": 46.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.012149848126898413, | |
| "grad_norm": 2.6865302061196417e-05, | |
| "learning_rate": 7.965056277307902e-05, | |
| "loss": 46.0, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.01282483968950388, | |
| "grad_norm": 3.483371983747929e-05, | |
| "learning_rate": 7.955791482405875e-05, | |
| "loss": 46.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01349983125210935, | |
| "grad_norm": 3.103768176515587e-05, | |
| "learning_rate": 7.94544521361089e-05, | |
| "loss": 46.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.014174822814714817, | |
| "grad_norm": 4.7862751671345904e-05, | |
| "learning_rate": 7.93402029948953e-05, | |
| "loss": 46.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.014849814377320284, | |
| "grad_norm": 3.231195660191588e-05, | |
| "learning_rate": 7.921519863499239e-05, | |
| "loss": 46.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.015524805939925751, | |
| "grad_norm": 3.36208613589406e-05, | |
| "learning_rate": 7.907947323134398e-05, | |
| "loss": 46.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.01619979750253122, | |
| "grad_norm": 3.122496127616614e-05, | |
| "learning_rate": 7.893306388992023e-05, | |
| "loss": 46.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.016874789065136685, | |
| "grad_norm": 4.556611384032294e-05, | |
| "learning_rate": 7.877601063757323e-05, | |
| "loss": 46.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.017549780627742154, | |
| "grad_norm": 3.778899917961098e-05, | |
| "learning_rate": 7.860835641109395e-05, | |
| "loss": 46.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01822477219034762, | |
| "grad_norm": 5.3873991419095546e-05, | |
| "learning_rate": 7.843014704547393e-05, | |
| "loss": 46.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.01889976375295309, | |
| "grad_norm": 4.7375739086419344e-05, | |
| "learning_rate": 7.824143126137431e-05, | |
| "loss": 46.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.019574755315558554, | |
| "grad_norm": 5.459750900627114e-05, | |
| "learning_rate": 7.804226065180615e-05, | |
| "loss": 46.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.020249746878164023, | |
| "grad_norm": 5.2733059419551864e-05, | |
| "learning_rate": 7.783268966802539e-05, | |
| "loss": 46.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.020924738440769492, | |
| "grad_norm": 6.106657383497804e-05, | |
| "learning_rate": 7.761277560464645e-05, | |
| "loss": 46.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.021599730003374958, | |
| "grad_norm": 5.731414785259403e-05, | |
| "learning_rate": 7.738257858397844e-05, | |
| "loss": 46.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.022274721565980427, | |
| "grad_norm": 5.25766663486138e-05, | |
| "learning_rate": 7.71421615395883e-05, | |
| "loss": 46.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.022949713128585892, | |
| "grad_norm": 6.520126044051722e-05, | |
| "learning_rate": 7.68915901990954e-05, | |
| "loss": 46.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.02362470469119136, | |
| "grad_norm": 6.199046765686944e-05, | |
| "learning_rate": 7.663093306620231e-05, | |
| "loss": 46.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.024299696253796826, | |
| "grad_norm": 6.82024983689189e-05, | |
| "learning_rate": 7.636026140196651e-05, | |
| "loss": 46.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.024974687816402295, | |
| "grad_norm": 5.999802306178026e-05, | |
| "learning_rate": 7.607964920531837e-05, | |
| "loss": 46.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02564967937900776, | |
| "grad_norm": 6.861635483801365e-05, | |
| "learning_rate": 7.578917319283055e-05, | |
| "loss": 46.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.02632467094161323, | |
| "grad_norm": 6.790792394895107e-05, | |
| "learning_rate": 7.548891277774448e-05, | |
| "loss": 46.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0269996625042187, | |
| "grad_norm": 5.505201261257753e-05, | |
| "learning_rate": 7.517895004825956e-05, | |
| "loss": 46.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.027674654066824164, | |
| "grad_norm": 0.00010721717990236357, | |
| "learning_rate": 7.48593697450911e-05, | |
| "loss": 46.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.028349645629429633, | |
| "grad_norm": 7.804264168953523e-05, | |
| "learning_rate": 7.453025923830296e-05, | |
| "loss": 46.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0290246371920351, | |
| "grad_norm": 0.00016989179130177945, | |
| "learning_rate": 7.419170850342156e-05, | |
| "loss": 46.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.029699628754640568, | |
| "grad_norm": 0.00015660426288377494, | |
| "learning_rate": 7.384381009683742e-05, | |
| "loss": 46.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.030374620317246033, | |
| "grad_norm": 0.0001554411428514868, | |
| "learning_rate": 7.348665913050115e-05, | |
| "loss": 46.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.031049611879851502, | |
| "grad_norm": 9.427656186744571e-05, | |
| "learning_rate": 7.312035324592081e-05, | |
| "loss": 46.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.03172460344245697, | |
| "grad_norm": 0.00013080888311378658, | |
| "learning_rate": 7.274499258746771e-05, | |
| "loss": 46.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.03239959500506244, | |
| "grad_norm": 0.00012277769565116614, | |
| "learning_rate": 7.236067977499791e-05, | |
| "loss": 46.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0330745865676679, | |
| "grad_norm": 0.0002552400401327759, | |
| "learning_rate": 7.196751987579699e-05, | |
| "loss": 46.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.03374957813027337, | |
| "grad_norm": 0.0002983387385029346, | |
| "learning_rate": 7.156562037585576e-05, | |
| "loss": 46.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03374957813027337, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 11.3501, | |
| "eval_samples_per_second": 219.821, | |
| "eval_steps_per_second": 54.977, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03442456969287884, | |
| "grad_norm": 5.4881169489817694e-05, | |
| "learning_rate": 7.11550911504845e-05, | |
| "loss": 46.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.03509956125548431, | |
| "grad_norm": 7.321939483517781e-05, | |
| "learning_rate": 7.073604443427437e-05, | |
| "loss": 46.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.03577455281808977, | |
| "grad_norm": 7.501189247705042e-05, | |
| "learning_rate": 7.03085947904134e-05, | |
| "loss": 46.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.03644954438069524, | |
| "grad_norm": 7.780554005876184e-05, | |
| "learning_rate": 6.987285907936617e-05, | |
| "loss": 46.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03712453594330071, | |
| "grad_norm": 9.891616355162114e-05, | |
| "learning_rate": 6.942895642692527e-05, | |
| "loss": 46.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03779952750590618, | |
| "grad_norm": 0.00010379558807471767, | |
| "learning_rate": 6.897700819164357e-05, | |
| "loss": 46.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.038474519068511646, | |
| "grad_norm": 6.376452802214772e-05, | |
| "learning_rate": 6.851713793165589e-05, | |
| "loss": 46.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.03914951063111711, | |
| "grad_norm": 0.00011976366658927873, | |
| "learning_rate": 6.804947137089955e-05, | |
| "loss": 46.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03982450219372258, | |
| "grad_norm": 0.00011366497346898541, | |
| "learning_rate": 6.757413636474263e-05, | |
| "loss": 46.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.040499493756328046, | |
| "grad_norm": 0.00012407948088366538, | |
| "learning_rate": 6.709126286502965e-05, | |
| "loss": 46.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.041174485318933515, | |
| "grad_norm": 0.00011083694698754698, | |
| "learning_rate": 6.660098288455393e-05, | |
| "loss": 46.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.041849476881538984, | |
| "grad_norm": 0.00012434367090463638, | |
| "learning_rate": 6.610343046096674e-05, | |
| "loss": 46.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.042524468444144446, | |
| "grad_norm": 9.564027277519926e-05, | |
| "learning_rate": 6.559874162013267e-05, | |
| "loss": 46.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.043199460006749915, | |
| "grad_norm": 8.99869337445125e-05, | |
| "learning_rate": 6.508705433894149e-05, | |
| "loss": 46.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.043874451569355384, | |
| "grad_norm": 0.00011574227391975, | |
| "learning_rate": 6.456850850758673e-05, | |
| "loss": 46.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04454944313196085, | |
| "grad_norm": 0.00011398899368941784, | |
| "learning_rate": 6.404324589132101e-05, | |
| "loss": 46.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.045224434694566315, | |
| "grad_norm": 8.823705138638616e-05, | |
| "learning_rate": 6.351141009169893e-05, | |
| "loss": 46.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.045899426257171784, | |
| "grad_norm": 0.0001269930216949433, | |
| "learning_rate": 6.297314650731775e-05, | |
| "loss": 46.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.04657441781977725, | |
| "grad_norm": 0.0001396976731484756, | |
| "learning_rate": 6.242860229406692e-05, | |
| "loss": 46.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.04724940938238272, | |
| "grad_norm": 0.0001476786273997277, | |
| "learning_rate": 6.18779263248971e-05, | |
| "loss": 46.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04792440094498819, | |
| "grad_norm": 0.0001579802919877693, | |
| "learning_rate": 6.132126914911976e-05, | |
| "loss": 46.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.04859939250759365, | |
| "grad_norm": 0.00014366269169840962, | |
| "learning_rate": 6.075878295124861e-05, | |
| "loss": 46.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.04927438407019912, | |
| "grad_norm": 0.00015545522910542786, | |
| "learning_rate": 6.019062150939376e-05, | |
| "loss": 46.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.04994937563280459, | |
| "grad_norm": 0.00011848592112073675, | |
| "learning_rate": 5.9616940153220336e-05, | |
| "loss": 46.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.05062436719541006, | |
| "grad_norm": 0.00015544149209745228, | |
| "learning_rate": 5.903789572148295e-05, | |
| "loss": 46.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05129935875801552, | |
| "grad_norm": 0.00016012511332519352, | |
| "learning_rate": 5.845364651914752e-05, | |
| "loss": 46.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.05197435032062099, | |
| "grad_norm": 0.00024290069995913655, | |
| "learning_rate": 5.786435227411227e-05, | |
| "loss": 46.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.05264934188322646, | |
| "grad_norm": 0.00016308229533024132, | |
| "learning_rate": 5.727017409353971e-05, | |
| "loss": 46.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.05332433344583193, | |
| "grad_norm": 0.0001312433450948447, | |
| "learning_rate": 5.667127441981162e-05, | |
| "loss": 46.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0539993250084374, | |
| "grad_norm": 0.00020814283925574273, | |
| "learning_rate": 5.606781698611879e-05, | |
| "loss": 46.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05467431657104286, | |
| "grad_norm": 0.00017544587899465114, | |
| "learning_rate": 5.5459966771698096e-05, | |
| "loss": 46.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.05534930813364833, | |
| "grad_norm": 0.00021598087914753705, | |
| "learning_rate": 5.4847889956728834e-05, | |
| "loss": 46.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0560242996962538, | |
| "grad_norm": 0.00022435332357417792, | |
| "learning_rate": 5.423175387690067e-05, | |
| "loss": 46.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.056699291258859266, | |
| "grad_norm": 0.00025759119307622313, | |
| "learning_rate": 5.361172697766573e-05, | |
| "loss": 46.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.05737428282146473, | |
| "grad_norm": 0.00027369000599719584, | |
| "learning_rate": 5.298797876818735e-05, | |
| "loss": 46.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0580492743840702, | |
| "grad_norm": 0.00021083872707094997, | |
| "learning_rate": 5.23606797749979e-05, | |
| "loss": 46.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.058724265946675666, | |
| "grad_norm": 0.00017781296628527343, | |
| "learning_rate": 5.17300014953786e-05, | |
| "loss": 46.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.059399257509281135, | |
| "grad_norm": 0.0002701023768167943, | |
| "learning_rate": 5.109611635047379e-05, | |
| "loss": 46.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.060074249071886604, | |
| "grad_norm": 0.00020761927589774132, | |
| "learning_rate": 5.04591976381528e-05, | |
| "loss": 46.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.060749240634492066, | |
| "grad_norm": 0.0002119742421200499, | |
| "learning_rate": 4.981941948563197e-05, | |
| "loss": 46.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.061424232197097535, | |
| "grad_norm": 0.0003119312459602952, | |
| "learning_rate": 4.9176956801870065e-05, | |
| "loss": 46.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.062099223759703004, | |
| "grad_norm": 0.00029339047614485025, | |
| "learning_rate": 4.853198522974988e-05, | |
| "loss": 46.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.06277421532230847, | |
| "grad_norm": 0.00024452415527775884, | |
| "learning_rate": 4.788468109805921e-05, | |
| "loss": 46.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.06344920688491394, | |
| "grad_norm": 0.00024892069632187486, | |
| "learning_rate": 4.7235221373284407e-05, | |
| "loss": 46.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0641241984475194, | |
| "grad_norm": 0.00032684431062079966, | |
| "learning_rate": 4.658378361122936e-05, | |
| "loss": 46.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.06479919001012488, | |
| "grad_norm": 0.00033167307265102863, | |
| "learning_rate": 4.593054590847368e-05, | |
| "loss": 46.0, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.06547418157273034, | |
| "grad_norm": 0.00021435694361571223, | |
| "learning_rate": 4.5275686853682765e-05, | |
| "loss": 46.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0661491731353358, | |
| "grad_norm": 0.00033286900725215673, | |
| "learning_rate": 4.4619385478783456e-05, | |
| "loss": 46.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.06682416469794128, | |
| "grad_norm": 0.0003271261812187731, | |
| "learning_rate": 4.396182121001852e-05, | |
| "loss": 46.0, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.06749915626054674, | |
| "grad_norm": 0.0008189138607122004, | |
| "learning_rate": 4.33031738188933e-05, | |
| "loss": 46.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06749915626054674, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 11.1305, | |
| "eval_samples_per_second": 224.16, | |
| "eval_steps_per_second": 56.062, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06817414782315222, | |
| "grad_norm": 0.00011564007581910118, | |
| "learning_rate": 4.264362337302798e-05, | |
| "loss": 46.0, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.06884913938575768, | |
| "grad_norm": 0.00014632352394983172, | |
| "learning_rate": 4.1983350186928894e-05, | |
| "loss": 46.0, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.06952413094836314, | |
| "grad_norm": 0.00017013819888234138, | |
| "learning_rate": 4.132253477269233e-05, | |
| "loss": 46.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.07019912251096862, | |
| "grad_norm": 0.0001804150379030034, | |
| "learning_rate": 4.0661357790654345e-05, | |
| "loss": 46.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.07087411407357408, | |
| "grad_norm": 0.00017299810133408755, | |
| "learning_rate": 4e-05, | |
| "loss": 46.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.07154910563617954, | |
| "grad_norm": 0.00026783038629218936, | |
| "learning_rate": 3.933864220934566e-05, | |
| "loss": 46.0, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.07222409719878502, | |
| "grad_norm": 0.00013552827294915915, | |
| "learning_rate": 3.8677465227307676e-05, | |
| "loss": 46.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.07289908876139048, | |
| "grad_norm": 0.00019998128118459135, | |
| "learning_rate": 3.8016649813071106e-05, | |
| "loss": 46.0, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.07357408032399596, | |
| "grad_norm": 0.00025718274991959333, | |
| "learning_rate": 3.735637662697203e-05, | |
| "loss": 46.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.07424907188660142, | |
| "grad_norm": 0.00015579417231492698, | |
| "learning_rate": 3.669682618110671e-05, | |
| "loss": 46.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.07492406344920688, | |
| "grad_norm": 0.0001929922727867961, | |
| "learning_rate": 3.6038178789981494e-05, | |
| "loss": 46.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.07559905501181235, | |
| "grad_norm": 0.00021357124205678701, | |
| "learning_rate": 3.538061452121656e-05, | |
| "loss": 46.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.07627404657441782, | |
| "grad_norm": 0.00018953881226480007, | |
| "learning_rate": 3.472431314631724e-05, | |
| "loss": 46.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.07694903813702329, | |
| "grad_norm": 0.00020883440447505563, | |
| "learning_rate": 3.406945409152632e-05, | |
| "loss": 46.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.07762402969962875, | |
| "grad_norm": 0.000206172582693398, | |
| "learning_rate": 3.341621638877064e-05, | |
| "loss": 46.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.07829902126223422, | |
| "grad_norm": 0.00018206911045126617, | |
| "learning_rate": 3.276477862671562e-05, | |
| "loss": 46.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.07897401282483969, | |
| "grad_norm": 0.0003483205218799412, | |
| "learning_rate": 3.21153189019408e-05, | |
| "loss": 46.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.07964900438744515, | |
| "grad_norm": 0.00022339758288580924, | |
| "learning_rate": 3.146801477025013e-05, | |
| "loss": 46.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.08032399595005063, | |
| "grad_norm": 0.00033201611950062215, | |
| "learning_rate": 3.082304319812994e-05, | |
| "loss": 46.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.08099898751265609, | |
| "grad_norm": 0.0002913949720095843, | |
| "learning_rate": 3.0180580514368037e-05, | |
| "loss": 46.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08167397907526155, | |
| "grad_norm": 0.00019378158322069794, | |
| "learning_rate": 2.9540802361847212e-05, | |
| "loss": 46.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.08234897063786703, | |
| "grad_norm": 0.00023098803649190813, | |
| "learning_rate": 2.890388364952623e-05, | |
| "loss": 46.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.08302396220047249, | |
| "grad_norm": 0.00033675433951430023, | |
| "learning_rate": 2.8269998504621416e-05, | |
| "loss": 46.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.08369895376307797, | |
| "grad_norm": 0.00025172470486722887, | |
| "learning_rate": 2.7639320225002108e-05, | |
| "loss": 46.0, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.08437394532568343, | |
| "grad_norm": 0.0002851325261872262, | |
| "learning_rate": 2.7012021231812666e-05, | |
| "loss": 46.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.08504893688828889, | |
| "grad_norm": 0.00030112077365629375, | |
| "learning_rate": 2.638827302233428e-05, | |
| "loss": 46.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.08572392845089437, | |
| "grad_norm": 0.00021006348833907396, | |
| "learning_rate": 2.576824612309934e-05, | |
| "loss": 46.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.08639892001349983, | |
| "grad_norm": 0.00036132585955783725, | |
| "learning_rate": 2.5152110043271166e-05, | |
| "loss": 46.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.08707391157610529, | |
| "grad_norm": 0.0002813682076521218, | |
| "learning_rate": 2.454003322830192e-05, | |
| "loss": 46.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.08774890313871077, | |
| "grad_norm": 0.00021099702280480415, | |
| "learning_rate": 2.393218301388123e-05, | |
| "loss": 46.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.08842389470131623, | |
| "grad_norm": 0.000335617340169847, | |
| "learning_rate": 2.3328725580188395e-05, | |
| "loss": 46.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.0890988862639217, | |
| "grad_norm": 0.0002918387181125581, | |
| "learning_rate": 2.272982590646029e-05, | |
| "loss": 46.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.08977387782652717, | |
| "grad_norm": 0.0003133406862616539, | |
| "learning_rate": 2.2135647725887744e-05, | |
| "loss": 46.0, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.09044886938913263, | |
| "grad_norm": 0.0002820506924763322, | |
| "learning_rate": 2.1546353480852495e-05, | |
| "loss": 46.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.0911238609517381, | |
| "grad_norm": 0.0003138439788017422, | |
| "learning_rate": 2.096210427851706e-05, | |
| "loss": 46.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.09179885251434357, | |
| "grad_norm": 0.0002928907924797386, | |
| "learning_rate": 2.038305984677969e-05, | |
| "loss": 46.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.09247384407694904, | |
| "grad_norm": 0.00030657005845569074, | |
| "learning_rate": 1.9809378490606264e-05, | |
| "loss": 46.0, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.0931488356395545, | |
| "grad_norm": 0.0002871578617487103, | |
| "learning_rate": 1.9241217048751406e-05, | |
| "loss": 46.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.09382382720215997, | |
| "grad_norm": 0.0004345919587649405, | |
| "learning_rate": 1.867873085088026e-05, | |
| "loss": 46.0, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.09449881876476544, | |
| "grad_norm": 0.0002062419371213764, | |
| "learning_rate": 1.8122073675102935e-05, | |
| "loss": 46.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0951738103273709, | |
| "grad_norm": 0.00035626027965918183, | |
| "learning_rate": 1.75713977059331e-05, | |
| "loss": 46.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.09584880188997638, | |
| "grad_norm": 0.00038553011836484075, | |
| "learning_rate": 1.702685349268226e-05, | |
| "loss": 46.0, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.09652379345258184, | |
| "grad_norm": 0.0004723902966361493, | |
| "learning_rate": 1.648858990830108e-05, | |
| "loss": 46.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.0971987850151873, | |
| "grad_norm": 0.00033361100940965116, | |
| "learning_rate": 1.5956754108678996e-05, | |
| "loss": 46.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.09787377657779278, | |
| "grad_norm": 0.00029087415896356106, | |
| "learning_rate": 1.5431491492413288e-05, | |
| "loss": 46.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.09854876814039824, | |
| "grad_norm": 0.0003720764070749283, | |
| "learning_rate": 1.491294566105852e-05, | |
| "loss": 46.0, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.09922375970300372, | |
| "grad_norm": 0.0003948597877752036, | |
| "learning_rate": 1.4401258379867335e-05, | |
| "loss": 46.0, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.09989875126560918, | |
| "grad_norm": 0.000279300264082849, | |
| "learning_rate": 1.3896569539033253e-05, | |
| "loss": 46.0, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.10057374282821464, | |
| "grad_norm": 0.000632896029856056, | |
| "learning_rate": 1.3399017115446067e-05, | |
| "loss": 46.0, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.10124873439082012, | |
| "grad_norm": 0.0006425530300475657, | |
| "learning_rate": 1.2908737134970367e-05, | |
| "loss": 46.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.10124873439082012, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 11.0746, | |
| "eval_samples_per_second": 225.291, | |
| "eval_steps_per_second": 56.345, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.10192372595342558, | |
| "grad_norm": 0.00019876357691828161, | |
| "learning_rate": 1.242586363525737e-05, | |
| "loss": 46.0, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.10259871751603104, | |
| "grad_norm": 0.0002352362935198471, | |
| "learning_rate": 1.1950528629100457e-05, | |
| "loss": 46.0, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.10327370907863652, | |
| "grad_norm": 0.00019797831191681325, | |
| "learning_rate": 1.1482862068344121e-05, | |
| "loss": 46.0, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.10394870064124198, | |
| "grad_norm": 0.0002398581855231896, | |
| "learning_rate": 1.1022991808356442e-05, | |
| "loss": 46.0, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.10462369220384746, | |
| "grad_norm": 0.00019246955343987793, | |
| "learning_rate": 1.0571043573074737e-05, | |
| "loss": 46.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.10529868376645292, | |
| "grad_norm": 0.00017457427748013288, | |
| "learning_rate": 1.0127140920633857e-05, | |
| "loss": 46.0, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.10597367532905838, | |
| "grad_norm": 0.00029743704362772405, | |
| "learning_rate": 9.69140520958662e-06, | |
| "loss": 46.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.10664866689166386, | |
| "grad_norm": 0.00017338975158054382, | |
| "learning_rate": 9.263955565725648e-06, | |
| "loss": 46.0, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.10732365845426932, | |
| "grad_norm": 0.00029915120103396475, | |
| "learning_rate": 8.844908849515509e-06, | |
| "loss": 46.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.1079986500168748, | |
| "grad_norm": 0.00023515670909546316, | |
| "learning_rate": 8.434379624144261e-06, | |
| "loss": 46.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.10867364157948026, | |
| "grad_norm": 0.00025672902120277286, | |
| "learning_rate": 8.032480124203013e-06, | |
| "loss": 46.0, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.10934863314208572, | |
| "grad_norm": 0.00027126428904011846, | |
| "learning_rate": 7.639320225002106e-06, | |
| "loss": 46.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.1100236247046912, | |
| "grad_norm": 0.0002725074300542474, | |
| "learning_rate": 7.255007412532307e-06, | |
| "loss": 46.0, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.11069861626729666, | |
| "grad_norm": 0.0002880211395677179, | |
| "learning_rate": 6.8796467540791986e-06, | |
| "loss": 46.0, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.11137360782990213, | |
| "grad_norm": 0.00027329163276590407, | |
| "learning_rate": 6.513340869498859e-06, | |
| "loss": 46.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.1120485993925076, | |
| "grad_norm": 0.00027789725572802126, | |
| "learning_rate": 6.1561899031625794e-06, | |
| "loss": 46.0, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.11272359095511306, | |
| "grad_norm": 0.0002505785960238427, | |
| "learning_rate": 5.808291496578435e-06, | |
| "loss": 46.0, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.11339858251771853, | |
| "grad_norm": 0.00036295506288297474, | |
| "learning_rate": 5.469740761697044e-06, | |
| "loss": 46.0, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.114073574080324, | |
| "grad_norm": 0.000231773461564444, | |
| "learning_rate": 5.140630254908905e-06, | |
| "loss": 46.0, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.11474856564292946, | |
| "grad_norm": 0.00036393426125869155, | |
| "learning_rate": 4.821049951740442e-06, | |
| "loss": 46.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.11542355720553493, | |
| "grad_norm": 0.0003143355716019869, | |
| "learning_rate": 4.511087222255528e-06, | |
| "loss": 46.0, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.1160985487681404, | |
| "grad_norm": 0.00033845927100628614, | |
| "learning_rate": 4.2108268071694616e-06, | |
| "loss": 46.0, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.11677354033074587, | |
| "grad_norm": 0.0003002511220984161, | |
| "learning_rate": 3.9203507946816445e-06, | |
| "loss": 46.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.11744853189335133, | |
| "grad_norm": 0.00032554613426327705, | |
| "learning_rate": 3.6397385980335e-06, | |
| "loss": 46.0, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.1181235234559568, | |
| "grad_norm": 0.0002870889729820192, | |
| "learning_rate": 3.3690669337977e-06, | |
| "loss": 46.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.11879851501856227, | |
| "grad_norm": 0.0003786945599131286, | |
| "learning_rate": 3.1084098009046106e-06, | |
| "loss": 46.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.11947350658116773, | |
| "grad_norm": 0.0003410293720662594, | |
| "learning_rate": 2.8578384604117217e-06, | |
| "loss": 46.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.12014849814377321, | |
| "grad_norm": 0.0002813494356814772, | |
| "learning_rate": 2.6174214160215704e-06, | |
| "loss": 46.0, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.12082348970637867, | |
| "grad_norm": 0.00023148763284552842, | |
| "learning_rate": 2.3872243953535535e-06, | |
| "loss": 46.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.12149848126898413, | |
| "grad_norm": 0.00044364703353494406, | |
| "learning_rate": 2.1673103319746146e-06, | |
| "loss": 46.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.12217347283158961, | |
| "grad_norm": 0.00044808979146182537, | |
| "learning_rate": 1.957739348193859e-06, | |
| "loss": 46.0, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.12284846439419507, | |
| "grad_norm": 0.0001999161613639444, | |
| "learning_rate": 1.7585687386256944e-06, | |
| "loss": 46.0, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.12352345595680055, | |
| "grad_norm": 0.00022155142505653203, | |
| "learning_rate": 1.5698529545260744e-06, | |
| "loss": 46.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.12419844751940601, | |
| "grad_norm": 0.0003148863324895501, | |
| "learning_rate": 1.3916435889060575e-06, | |
| "loss": 46.0, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.12487343908201147, | |
| "grad_norm": 0.0004106830747332424, | |
| "learning_rate": 1.2239893624267852e-06, | |
| "loss": 46.0, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.12554843064461693, | |
| "grad_norm": 0.00034603834501467645, | |
| "learning_rate": 1.0669361100797704e-06, | |
| "loss": 46.0, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.1262234222072224, | |
| "grad_norm": 0.0004213521024212241, | |
| "learning_rate": 9.205267686560293e-07, | |
| "loss": 46.0, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.12689841376982788, | |
| "grad_norm": 0.0002621794701553881, | |
| "learning_rate": 7.848013650076258e-07, | |
| "loss": 46.0, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.12757340533243333, | |
| "grad_norm": 0.000253551232162863, | |
| "learning_rate": 6.597970051047053e-07, | |
| "loss": 46.0, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.1282483968950388, | |
| "grad_norm": 0.00030335516203194857, | |
| "learning_rate": 5.455478638911071e-07, | |
| "loss": 46.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.12892338845764428, | |
| "grad_norm": 0.00027419746038503945, | |
| "learning_rate": 4.420851759412603e-07, | |
| "loss": 46.0, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.12959838002024976, | |
| "grad_norm": 0.0003773905918933451, | |
| "learning_rate": 3.4943722692099224e-07, | |
| "loss": 46.0, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.1302733715828552, | |
| "grad_norm": 0.00046274266787804663, | |
| "learning_rate": 2.676293458544743e-07, | |
| "loss": 46.0, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.13094836314546068, | |
| "grad_norm": 0.00036215598811395466, | |
| "learning_rate": 1.9668389819954338e-07, | |
| "loss": 46.0, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.13162335470806616, | |
| "grad_norm": 0.00038238154957070947, | |
| "learning_rate": 1.3662027973320614e-07, | |
| "loss": 46.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1322983462706716, | |
| "grad_norm": 0.0004448361578397453, | |
| "learning_rate": 8.745491124901861e-08, | |
| "loss": 46.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.13297333783327708, | |
| "grad_norm": 0.0004605569993145764, | |
| "learning_rate": 4.920123406781052e-08, | |
| "loss": 46.0, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.13364832939588256, | |
| "grad_norm": 0.0005355703178793192, | |
| "learning_rate": 2.1869706362958044e-08, | |
| "loss": 46.0, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.134323320958488, | |
| "grad_norm": 0.00035198143450543284, | |
| "learning_rate": 5.467800301239834e-09, | |
| "loss": 46.0, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.13499831252109348, | |
| "grad_norm": 0.0006307726143859327, | |
| "learning_rate": 0.0, | |
| "loss": 46.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.13499831252109348, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 11.0674, | |
| "eval_samples_per_second": 225.436, | |
| "eval_steps_per_second": 56.382, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 4, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 3 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 32158462771200.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |