| { | |
| "best_metric": 11.5, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.007275637527738368, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 3.637818763869184e-05, | |
| "grad_norm": 2.79417527053738e-05, | |
| "learning_rate": 1.003e-05, | |
| "loss": 23.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 3.637818763869184e-05, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 151.3914, | |
| "eval_samples_per_second": 76.457, | |
| "eval_steps_per_second": 19.116, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 7.275637527738368e-05, | |
| "grad_norm": 2.1153544366825372e-05, | |
| "learning_rate": 2.006e-05, | |
| "loss": 23.0, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00010913456291607552, | |
| "grad_norm": 1.7088592358049937e-05, | |
| "learning_rate": 3.0089999999999998e-05, | |
| "loss": 23.0, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00014551275055476736, | |
| "grad_norm": 2.0742239939863794e-05, | |
| "learning_rate": 4.012e-05, | |
| "loss": 23.0, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0001818909381934592, | |
| "grad_norm": 2.039460014202632e-05, | |
| "learning_rate": 5.015e-05, | |
| "loss": 23.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00021826912583215103, | |
| "grad_norm": 3.1619365472579375e-05, | |
| "learning_rate": 6.0179999999999996e-05, | |
| "loss": 23.0, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00025464731347084286, | |
| "grad_norm": 2.9644716050825082e-05, | |
| "learning_rate": 7.021e-05, | |
| "loss": 23.0, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0002910255011095347, | |
| "grad_norm": 1.7494063285994343e-05, | |
| "learning_rate": 8.024e-05, | |
| "loss": 23.0, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00032740368874822657, | |
| "grad_norm": 3.234211544622667e-05, | |
| "learning_rate": 9.027e-05, | |
| "loss": 23.0, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0003637818763869184, | |
| "grad_norm": 2.7539184884517454e-05, | |
| "learning_rate": 0.0001003, | |
| "loss": 23.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.00040016006402561027, | |
| "grad_norm": 1.9417160729062743e-05, | |
| "learning_rate": 9.97721052631579e-05, | |
| "loss": 23.0, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.00043653825166430207, | |
| "grad_norm": 2.843265792762395e-05, | |
| "learning_rate": 9.924421052631578e-05, | |
| "loss": 23.0, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0004729164393029939, | |
| "grad_norm": 4.186210935586132e-05, | |
| "learning_rate": 9.871631578947368e-05, | |
| "loss": 23.0, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0005092946269416857, | |
| "grad_norm": 4.8141664592549205e-05, | |
| "learning_rate": 9.818842105263158e-05, | |
| "loss": 23.0, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0005456728145803776, | |
| "grad_norm": 3.392771031940356e-05, | |
| "learning_rate": 9.766052631578948e-05, | |
| "loss": 23.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0005820510022190694, | |
| "grad_norm": 2.86590147879906e-05, | |
| "learning_rate": 9.713263157894736e-05, | |
| "loss": 23.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0006184291898577613, | |
| "grad_norm": 4.220665869070217e-05, | |
| "learning_rate": 9.660473684210526e-05, | |
| "loss": 23.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0006548073774964531, | |
| "grad_norm": 7.459171320078894e-05, | |
| "learning_rate": 9.607684210526316e-05, | |
| "loss": 23.0, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0006911855651351449, | |
| "grad_norm": 6.791079067625105e-05, | |
| "learning_rate": 9.554894736842104e-05, | |
| "loss": 23.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0007275637527738368, | |
| "grad_norm": 6.260742520680651e-05, | |
| "learning_rate": 9.502105263157894e-05, | |
| "loss": 23.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0007639419404125286, | |
| "grad_norm": 4.905338573735207e-05, | |
| "learning_rate": 9.449315789473684e-05, | |
| "loss": 23.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0008003201280512205, | |
| "grad_norm": 3.65267951565329e-05, | |
| "learning_rate": 9.396526315789474e-05, | |
| "loss": 23.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0008366983156899123, | |
| "grad_norm": 4.681987775256857e-05, | |
| "learning_rate": 9.343736842105264e-05, | |
| "loss": 23.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0008730765033286041, | |
| "grad_norm": 3.273553738836199e-05, | |
| "learning_rate": 9.290947368421052e-05, | |
| "loss": 23.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.000909454690967296, | |
| "grad_norm": 8.164726023096591e-05, | |
| "learning_rate": 9.238157894736842e-05, | |
| "loss": 23.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0009458328786059878, | |
| "grad_norm": 6.306690920609981e-05, | |
| "learning_rate": 9.18536842105263e-05, | |
| "loss": 23.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0009822110662446796, | |
| "grad_norm": 7.175507926149294e-05, | |
| "learning_rate": 9.132578947368422e-05, | |
| "loss": 23.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0010185892538833714, | |
| "grad_norm": 5.047854210715741e-05, | |
| "learning_rate": 9.07978947368421e-05, | |
| "loss": 23.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0010549674415220635, | |
| "grad_norm": 0.00011784725211327896, | |
| "learning_rate": 9.027e-05, | |
| "loss": 23.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0010913456291607553, | |
| "grad_norm": 0.00014080016990192235, | |
| "learning_rate": 8.97421052631579e-05, | |
| "loss": 23.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.001127723816799447, | |
| "grad_norm": 8.020276436582208e-05, | |
| "learning_rate": 8.921421052631578e-05, | |
| "loss": 23.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0011641020044381389, | |
| "grad_norm": 5.309961488819681e-05, | |
| "learning_rate": 8.868631578947368e-05, | |
| "loss": 23.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0012004801920768306, | |
| "grad_norm": 0.00015320915554184467, | |
| "learning_rate": 8.815842105263157e-05, | |
| "loss": 23.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0012368583797155227, | |
| "grad_norm": 8.746675302973017e-05, | |
| "learning_rate": 8.763052631578948e-05, | |
| "loss": 23.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0012732365673542145, | |
| "grad_norm": 0.00010071782890008762, | |
| "learning_rate": 8.710263157894737e-05, | |
| "loss": 23.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0013096147549929063, | |
| "grad_norm": 0.00013145706907380372, | |
| "learning_rate": 8.657473684210526e-05, | |
| "loss": 23.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.001345992942631598, | |
| "grad_norm": 0.00013049600238446146, | |
| "learning_rate": 8.604684210526316e-05, | |
| "loss": 23.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0013823711302702899, | |
| "grad_norm": 7.936532347230241e-05, | |
| "learning_rate": 8.551894736842105e-05, | |
| "loss": 23.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0014187493179089819, | |
| "grad_norm": 9.912149835145101e-05, | |
| "learning_rate": 8.499105263157895e-05, | |
| "loss": 23.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0014551275055476737, | |
| "grad_norm": 0.00015672955487389117, | |
| "learning_rate": 8.446315789473683e-05, | |
| "loss": 23.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0014915056931863655, | |
| "grad_norm": 0.00016603790572844446, | |
| "learning_rate": 8.393526315789474e-05, | |
| "loss": 23.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0015278838808250573, | |
| "grad_norm": 0.00018017899128608406, | |
| "learning_rate": 8.340736842105263e-05, | |
| "loss": 23.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.001564262068463749, | |
| "grad_norm": 0.00017705631034914404, | |
| "learning_rate": 8.287947368421053e-05, | |
| "loss": 23.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.001600640256102441, | |
| "grad_norm": 0.0002334263699594885, | |
| "learning_rate": 8.235157894736842e-05, | |
| "loss": 23.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0016370184437411329, | |
| "grad_norm": 0.00010936748731182888, | |
| "learning_rate": 8.182368421052631e-05, | |
| "loss": 23.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0016733966313798247, | |
| "grad_norm": 0.0001802948536351323, | |
| "learning_rate": 8.129578947368421e-05, | |
| "loss": 23.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0017097748190185165, | |
| "grad_norm": 0.00026060600066557527, | |
| "learning_rate": 8.07678947368421e-05, | |
| "loss": 23.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0017461530066572083, | |
| "grad_norm": 0.00030714573222212493, | |
| "learning_rate": 8.024e-05, | |
| "loss": 23.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0017825311942959, | |
| "grad_norm": 0.00010379219747846946, | |
| "learning_rate": 7.97121052631579e-05, | |
| "loss": 23.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.001818909381934592, | |
| "grad_norm": 0.0003222045779693872, | |
| "learning_rate": 7.918421052631579e-05, | |
| "loss": 23.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.001818909381934592, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 151.1621, | |
| "eval_samples_per_second": 76.573, | |
| "eval_steps_per_second": 19.145, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0018552875695732839, | |
| "grad_norm": 7.084694516379386e-05, | |
| "learning_rate": 7.865631578947369e-05, | |
| "loss": 23.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0018916657572119757, | |
| "grad_norm": 5.6916571338661015e-05, | |
| "learning_rate": 7.812842105263157e-05, | |
| "loss": 23.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0019280439448506675, | |
| "grad_norm": 0.00013821788888890296, | |
| "learning_rate": 7.760052631578947e-05, | |
| "loss": 23.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0019644221324893593, | |
| "grad_norm": 0.00010073825251311064, | |
| "learning_rate": 7.707263157894737e-05, | |
| "loss": 23.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0020008003201280513, | |
| "grad_norm": 3.989057950093411e-05, | |
| "learning_rate": 7.654473684210527e-05, | |
| "loss": 23.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.002037178507766743, | |
| "grad_norm": 0.00011672514665406197, | |
| "learning_rate": 7.601684210526316e-05, | |
| "loss": 23.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.002073556695405435, | |
| "grad_norm": 0.00011147064651595429, | |
| "learning_rate": 7.548894736842105e-05, | |
| "loss": 23.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.002109934883044127, | |
| "grad_norm": 0.00014741995255462825, | |
| "learning_rate": 7.496105263157895e-05, | |
| "loss": 23.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0021463130706828185, | |
| "grad_norm": 0.00019834299746435136, | |
| "learning_rate": 7.443315789473683e-05, | |
| "loss": 23.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0021826912583215105, | |
| "grad_norm": 7.386082143057138e-05, | |
| "learning_rate": 7.390526315789473e-05, | |
| "loss": 23.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.002219069445960202, | |
| "grad_norm": 0.00012008604971924797, | |
| "learning_rate": 7.337736842105263e-05, | |
| "loss": 23.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.002255447633598894, | |
| "grad_norm": 0.00010102576925419271, | |
| "learning_rate": 7.284947368421053e-05, | |
| "loss": 23.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.002291825821237586, | |
| "grad_norm": 8.834318578010425e-05, | |
| "learning_rate": 7.232157894736843e-05, | |
| "loss": 23.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0023282040088762777, | |
| "grad_norm": 0.00017303289496339858, | |
| "learning_rate": 7.179368421052631e-05, | |
| "loss": 23.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0023645821965149697, | |
| "grad_norm": 0.00021577121515292674, | |
| "learning_rate": 7.126578947368421e-05, | |
| "loss": 23.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0024009603841536613, | |
| "grad_norm": 0.00012508737563621253, | |
| "learning_rate": 7.07378947368421e-05, | |
| "loss": 23.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0024373385717923533, | |
| "grad_norm": 0.00021092577662784606, | |
| "learning_rate": 7.021e-05, | |
| "loss": 23.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0024737167594310453, | |
| "grad_norm": 9.400040289619938e-05, | |
| "learning_rate": 6.968210526315789e-05, | |
| "loss": 23.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.002510094947069737, | |
| "grad_norm": 0.00022102220100350678, | |
| "learning_rate": 6.915421052631579e-05, | |
| "loss": 23.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.002546473134708429, | |
| "grad_norm": 0.00022005224309396, | |
| "learning_rate": 6.862631578947369e-05, | |
| "loss": 23.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0025828513223471205, | |
| "grad_norm": 0.0001787764485925436, | |
| "learning_rate": 6.809842105263157e-05, | |
| "loss": 23.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0026192295099858125, | |
| "grad_norm": 0.00012445311585906893, | |
| "learning_rate": 6.757052631578947e-05, | |
| "loss": 23.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0026556076976245045, | |
| "grad_norm": 0.00024370021128561348, | |
| "learning_rate": 6.704263157894737e-05, | |
| "loss": 23.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.002691985885263196, | |
| "grad_norm": 0.00029588048346340656, | |
| "learning_rate": 6.651473684210526e-05, | |
| "loss": 23.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.002728364072901888, | |
| "grad_norm": 0.00026139113469980657, | |
| "learning_rate": 6.598684210526317e-05, | |
| "loss": 23.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0027647422605405797, | |
| "grad_norm": 9.695452899904922e-05, | |
| "learning_rate": 6.545894736842105e-05, | |
| "loss": 23.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0028011204481792717, | |
| "grad_norm": 0.00018712144810706377, | |
| "learning_rate": 6.493105263157895e-05, | |
| "loss": 23.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0028374986358179637, | |
| "grad_norm": 0.00016666650481056422, | |
| "learning_rate": 6.440315789473684e-05, | |
| "loss": 23.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0028738768234566553, | |
| "grad_norm": 0.00029639736749231815, | |
| "learning_rate": 6.387526315789473e-05, | |
| "loss": 23.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0029102550110953473, | |
| "grad_norm": 0.00020341298659332097, | |
| "learning_rate": 6.334736842105263e-05, | |
| "loss": 23.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.002946633198734039, | |
| "grad_norm": 0.0003866904880851507, | |
| "learning_rate": 6.281947368421052e-05, | |
| "loss": 23.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.002983011386372731, | |
| "grad_norm": 0.0003410225035622716, | |
| "learning_rate": 6.229157894736843e-05, | |
| "loss": 23.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.003019389574011423, | |
| "grad_norm": 0.0003634164750110358, | |
| "learning_rate": 6.176368421052631e-05, | |
| "loss": 23.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0030557677616501145, | |
| "grad_norm": 0.00019395571143832058, | |
| "learning_rate": 6.123578947368421e-05, | |
| "loss": 23.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.0030921459492888066, | |
| "grad_norm": 0.00014450523303821683, | |
| "learning_rate": 6.0707894736842105e-05, | |
| "loss": 23.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.003128524136927498, | |
| "grad_norm": 0.0003423929156269878, | |
| "learning_rate": 6.0179999999999996e-05, | |
| "loss": 23.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.00316490232456619, | |
| "grad_norm": 0.00019756775873247534, | |
| "learning_rate": 5.965210526315789e-05, | |
| "loss": 23.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.003201280512204882, | |
| "grad_norm": 0.0003216055629309267, | |
| "learning_rate": 5.912421052631578e-05, | |
| "loss": 23.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0032376586998435737, | |
| "grad_norm": 0.0005142286536283791, | |
| "learning_rate": 5.8596315789473685e-05, | |
| "loss": 23.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0032740368874822658, | |
| "grad_norm": 0.00019845775386784226, | |
| "learning_rate": 5.8068421052631583e-05, | |
| "loss": 23.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0033104150751209573, | |
| "grad_norm": 0.0002886882866732776, | |
| "learning_rate": 5.7540526315789475e-05, | |
| "loss": 23.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0033467932627596494, | |
| "grad_norm": 0.0003776240337174386, | |
| "learning_rate": 5.701263157894737e-05, | |
| "loss": 23.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.003383171450398341, | |
| "grad_norm": 0.00038095013587735593, | |
| "learning_rate": 5.648473684210526e-05, | |
| "loss": 23.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.003419549638037033, | |
| "grad_norm": 0.0003583188517950475, | |
| "learning_rate": 5.595684210526315e-05, | |
| "loss": 23.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.003455927825675725, | |
| "grad_norm": 0.0004853124264627695, | |
| "learning_rate": 5.5428947368421055e-05, | |
| "loss": 23.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0034923060133144166, | |
| "grad_norm": 0.00038572377525269985, | |
| "learning_rate": 5.490105263157895e-05, | |
| "loss": 23.0, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0035286842009531086, | |
| "grad_norm": 0.0003467136702965945, | |
| "learning_rate": 5.4373157894736846e-05, | |
| "loss": 23.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0035650623885918, | |
| "grad_norm": 0.0002495538501534611, | |
| "learning_rate": 5.384526315789474e-05, | |
| "loss": 23.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.003601440576230492, | |
| "grad_norm": 0.0007139640511013567, | |
| "learning_rate": 5.331736842105263e-05, | |
| "loss": 23.0, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.003637818763869184, | |
| "grad_norm": 0.0005742310895584524, | |
| "learning_rate": 5.278947368421052e-05, | |
| "loss": 23.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.003637818763869184, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 151.1117, | |
| "eval_samples_per_second": 76.599, | |
| "eval_steps_per_second": 19.151, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0036741969515078758, | |
| "grad_norm": 7.461366476491094e-05, | |
| "learning_rate": 5.226157894736842e-05, | |
| "loss": 23.0, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.0037105751391465678, | |
| "grad_norm": 0.00017520117398817092, | |
| "learning_rate": 5.173368421052632e-05, | |
| "loss": 23.0, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.0037469533267852594, | |
| "grad_norm": 0.00017740805924404413, | |
| "learning_rate": 5.1205789473684216e-05, | |
| "loss": 23.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0037833315144239514, | |
| "grad_norm": 0.00010646147711668164, | |
| "learning_rate": 5.067789473684211e-05, | |
| "loss": 23.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.0038197097020626434, | |
| "grad_norm": 0.0001494868629379198, | |
| "learning_rate": 5.015e-05, | |
| "loss": 23.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.003856087889701335, | |
| "grad_norm": 0.00012090586824342608, | |
| "learning_rate": 4.962210526315789e-05, | |
| "loss": 23.0, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.003892466077340027, | |
| "grad_norm": 0.00021911323710810393, | |
| "learning_rate": 4.909421052631579e-05, | |
| "loss": 23.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.003928844264978719, | |
| "grad_norm": 0.0003465862537268549, | |
| "learning_rate": 4.856631578947368e-05, | |
| "loss": 23.0, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.003965222452617411, | |
| "grad_norm": 0.0003293559711892158, | |
| "learning_rate": 4.803842105263158e-05, | |
| "loss": 23.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.004001600640256103, | |
| "grad_norm": 0.0002986920881085098, | |
| "learning_rate": 4.751052631578947e-05, | |
| "loss": 23.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.004037978827894795, | |
| "grad_norm": 0.00022487634851131588, | |
| "learning_rate": 4.698263157894737e-05, | |
| "loss": 23.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.004074357015533486, | |
| "grad_norm": 0.00037544663064181805, | |
| "learning_rate": 4.645473684210526e-05, | |
| "loss": 23.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.004110735203172178, | |
| "grad_norm": 0.0001820535253500566, | |
| "learning_rate": 4.592684210526315e-05, | |
| "loss": 23.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.00414711339081087, | |
| "grad_norm": 0.00031552044674754143, | |
| "learning_rate": 4.539894736842105e-05, | |
| "loss": 23.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.004183491578449562, | |
| "grad_norm": 8.148902998073027e-05, | |
| "learning_rate": 4.487105263157895e-05, | |
| "loss": 23.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.004219869766088254, | |
| "grad_norm": 0.00017857461352832615, | |
| "learning_rate": 4.434315789473684e-05, | |
| "loss": 23.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.004256247953726945, | |
| "grad_norm": 0.00020005616534035653, | |
| "learning_rate": 4.381526315789474e-05, | |
| "loss": 23.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.004292626141365637, | |
| "grad_norm": 0.00036300913779996336, | |
| "learning_rate": 4.328736842105263e-05, | |
| "loss": 23.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.004329004329004329, | |
| "grad_norm": 0.0003054387343581766, | |
| "learning_rate": 4.2759473684210523e-05, | |
| "loss": 23.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.004365382516643021, | |
| "grad_norm": 0.0002077149401884526, | |
| "learning_rate": 4.2231578947368415e-05, | |
| "loss": 23.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.004401760704281713, | |
| "grad_norm": 0.0002281182969454676, | |
| "learning_rate": 4.1703684210526314e-05, | |
| "loss": 23.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.004438138891920404, | |
| "grad_norm": 0.0005008115549571812, | |
| "learning_rate": 4.117578947368421e-05, | |
| "loss": 23.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.004474517079559096, | |
| "grad_norm": 0.00035556236980482936, | |
| "learning_rate": 4.0647894736842104e-05, | |
| "loss": 23.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.004510895267197788, | |
| "grad_norm": 0.0005603492027148604, | |
| "learning_rate": 4.012e-05, | |
| "loss": 23.0, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.00454727345483648, | |
| "grad_norm": 0.0004926612018607557, | |
| "learning_rate": 3.9592105263157894e-05, | |
| "loss": 23.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.004583651642475172, | |
| "grad_norm": 0.0004209111793898046, | |
| "learning_rate": 3.9064210526315785e-05, | |
| "loss": 23.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.004620029830113863, | |
| "grad_norm": 0.00046484937774948776, | |
| "learning_rate": 3.8536315789473684e-05, | |
| "loss": 23.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.004656408017752555, | |
| "grad_norm": 0.0005833171890117228, | |
| "learning_rate": 3.800842105263158e-05, | |
| "loss": 23.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.004692786205391247, | |
| "grad_norm": 0.0005073467036709189, | |
| "learning_rate": 3.7480526315789474e-05, | |
| "loss": 23.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.0047291643930299394, | |
| "grad_norm": 0.0002727651153691113, | |
| "learning_rate": 3.6952631578947366e-05, | |
| "loss": 23.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0047655425806686314, | |
| "grad_norm": 0.000585235480684787, | |
| "learning_rate": 3.6424736842105264e-05, | |
| "loss": 23.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.004801920768307323, | |
| "grad_norm": 0.0002120441640727222, | |
| "learning_rate": 3.5896842105263156e-05, | |
| "loss": 23.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.004838298955946015, | |
| "grad_norm": 0.0004325111221987754, | |
| "learning_rate": 3.536894736842105e-05, | |
| "loss": 23.0, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.004874677143584707, | |
| "grad_norm": 0.00032356969313696027, | |
| "learning_rate": 3.4841052631578946e-05, | |
| "loss": 23.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.004911055331223399, | |
| "grad_norm": 0.000456592213595286, | |
| "learning_rate": 3.4313157894736844e-05, | |
| "loss": 23.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.004947433518862091, | |
| "grad_norm": 0.0007706891628913581, | |
| "learning_rate": 3.3785263157894736e-05, | |
| "loss": 23.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.004983811706500782, | |
| "grad_norm": 0.000517175649292767, | |
| "learning_rate": 3.325736842105263e-05, | |
| "loss": 23.0, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.005020189894139474, | |
| "grad_norm": 0.000617575307842344, | |
| "learning_rate": 3.2729473684210526e-05, | |
| "loss": 23.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.005056568081778166, | |
| "grad_norm": 0.0002448149025440216, | |
| "learning_rate": 3.220157894736842e-05, | |
| "loss": 23.0, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.005092946269416858, | |
| "grad_norm": 0.0004209695034660399, | |
| "learning_rate": 3.1673684210526316e-05, | |
| "loss": 23.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.00512932445705555, | |
| "grad_norm": 0.0003657886991277337, | |
| "learning_rate": 3.1145789473684215e-05, | |
| "loss": 23.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.005165702644694241, | |
| "grad_norm": 0.0008267536177299917, | |
| "learning_rate": 3.0617894736842107e-05, | |
| "loss": 23.0, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.005202080832332933, | |
| "grad_norm": 0.0008407800341956317, | |
| "learning_rate": 3.0089999999999998e-05, | |
| "loss": 23.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.005238459019971625, | |
| "grad_norm": 0.0005053975037299097, | |
| "learning_rate": 2.956210526315789e-05, | |
| "loss": 23.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.005274837207610317, | |
| "grad_norm": 0.00048378558130934834, | |
| "learning_rate": 2.9034210526315792e-05, | |
| "loss": 23.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.005311215395249009, | |
| "grad_norm": 0.0007292490336112678, | |
| "learning_rate": 2.8506315789473683e-05, | |
| "loss": 23.0, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.0053475935828877, | |
| "grad_norm": 0.000520595523994416, | |
| "learning_rate": 2.7978421052631575e-05, | |
| "loss": 23.0, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.005383971770526392, | |
| "grad_norm": 0.0005500400438904762, | |
| "learning_rate": 2.7450526315789474e-05, | |
| "loss": 23.0, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.005420349958165084, | |
| "grad_norm": 0.00027113532996736467, | |
| "learning_rate": 2.692263157894737e-05, | |
| "loss": 23.0, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.005456728145803776, | |
| "grad_norm": 0.00021608092356473207, | |
| "learning_rate": 2.639473684210526e-05, | |
| "loss": 23.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.005456728145803776, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 151.1109, | |
| "eval_samples_per_second": 76.599, | |
| "eval_steps_per_second": 19.152, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.005493106333442468, | |
| "grad_norm": 0.0003489772789180279, | |
| "learning_rate": 2.586684210526316e-05, | |
| "loss": 23.0, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.005529484521081159, | |
| "grad_norm": 0.00012043579044984654, | |
| "learning_rate": 2.5338947368421054e-05, | |
| "loss": 23.0, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.0055658627087198514, | |
| "grad_norm": 0.00014974933583289385, | |
| "learning_rate": 2.4811052631578945e-05, | |
| "loss": 23.0, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.0056022408963585435, | |
| "grad_norm": 0.00023572241479996592, | |
| "learning_rate": 2.428315789473684e-05, | |
| "loss": 23.0, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.0056386190839972355, | |
| "grad_norm": 0.00013939540076535195, | |
| "learning_rate": 2.3755263157894736e-05, | |
| "loss": 23.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.0056749972716359275, | |
| "grad_norm": 0.00021143161575309932, | |
| "learning_rate": 2.322736842105263e-05, | |
| "loss": 23.0, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.005711375459274619, | |
| "grad_norm": 0.00038103267434053123, | |
| "learning_rate": 2.2699473684210526e-05, | |
| "loss": 23.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.005747753646913311, | |
| "grad_norm": 0.000460915471194312, | |
| "learning_rate": 2.217157894736842e-05, | |
| "loss": 23.0, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.005784131834552003, | |
| "grad_norm": 0.0002898467646446079, | |
| "learning_rate": 2.1643684210526316e-05, | |
| "loss": 23.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.005820510022190695, | |
| "grad_norm": 0.0002288204850628972, | |
| "learning_rate": 2.1115789473684208e-05, | |
| "loss": 23.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.005856888209829387, | |
| "grad_norm": 0.0002909198228735477, | |
| "learning_rate": 2.0587894736842106e-05, | |
| "loss": 23.0, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.005893266397468078, | |
| "grad_norm": 0.0002346673427382484, | |
| "learning_rate": 2.006e-05, | |
| "loss": 23.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.00592964458510677, | |
| "grad_norm": 0.0008843714604154229, | |
| "learning_rate": 1.9532105263157893e-05, | |
| "loss": 23.0, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.005966022772745462, | |
| "grad_norm": 0.0004195628280285746, | |
| "learning_rate": 1.900421052631579e-05, | |
| "loss": 23.0, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.006002400960384154, | |
| "grad_norm": 0.00046790888882242143, | |
| "learning_rate": 1.8476315789473683e-05, | |
| "loss": 23.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.006038779148022846, | |
| "grad_norm": 0.00015601440099999309, | |
| "learning_rate": 1.7948421052631578e-05, | |
| "loss": 23.0, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.006075157335661537, | |
| "grad_norm": 0.0005221933824941516, | |
| "learning_rate": 1.7420526315789473e-05, | |
| "loss": 23.0, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.006111535523300229, | |
| "grad_norm": 0.0002946684544440359, | |
| "learning_rate": 1.6892631578947368e-05, | |
| "loss": 23.0, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.006147913710938921, | |
| "grad_norm": 0.00034045460051856935, | |
| "learning_rate": 1.6364736842105263e-05, | |
| "loss": 23.0, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.006184291898577613, | |
| "grad_norm": 0.00024054516688920557, | |
| "learning_rate": 1.5836842105263158e-05, | |
| "loss": 23.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.006220670086216305, | |
| "grad_norm": 0.0005520730628632009, | |
| "learning_rate": 1.5308947368421053e-05, | |
| "loss": 23.0, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.006257048273854996, | |
| "grad_norm": 0.00023751375556457788, | |
| "learning_rate": 1.4781052631578945e-05, | |
| "loss": 23.0, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.006293426461493688, | |
| "grad_norm": 0.0005580897559411824, | |
| "learning_rate": 1.4253157894736842e-05, | |
| "loss": 23.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.00632980464913238, | |
| "grad_norm": 0.00024056828988250345, | |
| "learning_rate": 1.3725263157894737e-05, | |
| "loss": 23.0, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.006366182836771072, | |
| "grad_norm": 0.0004827072552870959, | |
| "learning_rate": 1.319736842105263e-05, | |
| "loss": 23.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.006402561024409764, | |
| "grad_norm": 0.0004384911444503814, | |
| "learning_rate": 1.2669473684210527e-05, | |
| "loss": 23.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.0064389392120484555, | |
| "grad_norm": 0.0005333396256901324, | |
| "learning_rate": 1.214157894736842e-05, | |
| "loss": 23.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.0064753173996871475, | |
| "grad_norm": 0.00027803468401543796, | |
| "learning_rate": 1.1613684210526315e-05, | |
| "loss": 23.0, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.0065116955873258395, | |
| "grad_norm": 0.0005504011642187834, | |
| "learning_rate": 1.108578947368421e-05, | |
| "loss": 23.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.0065480737749645315, | |
| "grad_norm": 0.0006978894816711545, | |
| "learning_rate": 1.0557894736842104e-05, | |
| "loss": 23.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.0065844519626032235, | |
| "grad_norm": 0.0004980647936463356, | |
| "learning_rate": 1.003e-05, | |
| "loss": 23.0, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.006620830150241915, | |
| "grad_norm": 0.0008371528820134699, | |
| "learning_rate": 9.502105263157896e-06, | |
| "loss": 23.0, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.006657208337880607, | |
| "grad_norm": 0.00043287783046253026, | |
| "learning_rate": 8.974210526315789e-06, | |
| "loss": 23.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.006693586525519299, | |
| "grad_norm": 0.00033211580011993647, | |
| "learning_rate": 8.446315789473684e-06, | |
| "loss": 23.0, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.006729964713157991, | |
| "grad_norm": 0.0016474088188260794, | |
| "learning_rate": 7.918421052631579e-06, | |
| "loss": 23.0, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.006766342900796682, | |
| "grad_norm": 0.0008112106588669121, | |
| "learning_rate": 7.3905263157894725e-06, | |
| "loss": 23.0, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.006802721088435374, | |
| "grad_norm": 0.0006704179104417562, | |
| "learning_rate": 6.862631578947368e-06, | |
| "loss": 23.0, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.006839099276074066, | |
| "grad_norm": 0.0007978321518748999, | |
| "learning_rate": 6.3347368421052634e-06, | |
| "loss": 23.0, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.006875477463712758, | |
| "grad_norm": 0.0008913553901948035, | |
| "learning_rate": 5.806842105263158e-06, | |
| "loss": 23.0, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.00691185565135145, | |
| "grad_norm": 0.0006912974058650434, | |
| "learning_rate": 5.278947368421052e-06, | |
| "loss": 23.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.006948233838990141, | |
| "grad_norm": 0.0007087139529176056, | |
| "learning_rate": 4.751052631578948e-06, | |
| "loss": 23.0, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.006984612026628833, | |
| "grad_norm": 0.00047257327241823077, | |
| "learning_rate": 4.223157894736842e-06, | |
| "loss": 23.0, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.007020990214267525, | |
| "grad_norm": 0.00034283282002434134, | |
| "learning_rate": 3.6952631578947362e-06, | |
| "loss": 23.0, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.007057368401906217, | |
| "grad_norm": 0.0011271246476098895, | |
| "learning_rate": 3.1673684210526317e-06, | |
| "loss": 23.0, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.007093746589544909, | |
| "grad_norm": 0.0007555687916465104, | |
| "learning_rate": 2.639473684210526e-06, | |
| "loss": 23.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.0071301247771836, | |
| "grad_norm": 0.0008845161064527929, | |
| "learning_rate": 2.111578947368421e-06, | |
| "loss": 23.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.007166502964822292, | |
| "grad_norm": 0.00046997558092698455, | |
| "learning_rate": 1.5836842105263159e-06, | |
| "loss": 23.0, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.007202881152460984, | |
| "grad_norm": 0.00101653509773314, | |
| "learning_rate": 1.0557894736842105e-06, | |
| "loss": 23.0, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.007239259340099676, | |
| "grad_norm": 0.00044932126183994114, | |
| "learning_rate": 5.278947368421053e-07, | |
| "loss": 23.0, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.007275637527738368, | |
| "grad_norm": 0.0005571041256189346, | |
| "learning_rate": 0.0, | |
| "loss": 23.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.007275637527738368, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 151.1327, | |
| "eval_samples_per_second": 76.588, | |
| "eval_steps_per_second": 19.149, | |
| "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": 5, | |
| "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": 8059714732032.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |