| { | |
| "best_metric": 11.866828918457031, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.006207613638127163, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 3.103806819063581e-05, | |
| "grad_norm": 0.06145656853914261, | |
| "learning_rate": 1.009e-05, | |
| "loss": 11.9393, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 3.103806819063581e-05, | |
| "eval_loss": 11.937196731567383, | |
| "eval_runtime": 414.1282, | |
| "eval_samples_per_second": 32.758, | |
| "eval_steps_per_second": 8.191, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 6.207613638127163e-05, | |
| "grad_norm": 0.10884089767932892, | |
| "learning_rate": 2.018e-05, | |
| "loss": 11.9275, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 9.311420457190745e-05, | |
| "grad_norm": 0.11498255282640457, | |
| "learning_rate": 3.027e-05, | |
| "loss": 11.9388, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00012415227276254325, | |
| "grad_norm": 0.088991180062294, | |
| "learning_rate": 4.036e-05, | |
| "loss": 11.9282, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00015519034095317908, | |
| "grad_norm": 0.07985679805278778, | |
| "learning_rate": 5.045e-05, | |
| "loss": 11.9416, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0001862284091438149, | |
| "grad_norm": 0.07948625832796097, | |
| "learning_rate": 6.054e-05, | |
| "loss": 11.9386, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0002172664773344507, | |
| "grad_norm": 0.08249641954898834, | |
| "learning_rate": 7.062999999999999e-05, | |
| "loss": 11.9342, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0002483045455250865, | |
| "grad_norm": 0.06686617434024811, | |
| "learning_rate": 8.072e-05, | |
| "loss": 11.9395, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0002793426137157223, | |
| "grad_norm": 0.07867742329835892, | |
| "learning_rate": 9.081e-05, | |
| "loss": 11.9405, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00031038068190635817, | |
| "grad_norm": 0.09347719699144363, | |
| "learning_rate": 0.0001009, | |
| "loss": 11.9347, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.000341418750096994, | |
| "grad_norm": 0.09216347336769104, | |
| "learning_rate": 0.00010036894736842106, | |
| "loss": 11.9372, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0003724568182876298, | |
| "grad_norm": 0.08478887379169464, | |
| "learning_rate": 9.98378947368421e-05, | |
| "loss": 11.9306, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0004034948864782656, | |
| "grad_norm": 0.10052121430635452, | |
| "learning_rate": 9.930684210526315e-05, | |
| "loss": 11.9362, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0004345329546689014, | |
| "grad_norm": 0.09549523890018463, | |
| "learning_rate": 9.877578947368421e-05, | |
| "loss": 11.933, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0004655710228595372, | |
| "grad_norm": 0.10671835392713547, | |
| "learning_rate": 9.824473684210527e-05, | |
| "loss": 11.9354, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.000496609091050173, | |
| "grad_norm": 0.12740136682987213, | |
| "learning_rate": 9.771368421052632e-05, | |
| "loss": 11.9261, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0005276471592408088, | |
| "grad_norm": 0.10865606367588043, | |
| "learning_rate": 9.718263157894736e-05, | |
| "loss": 11.9368, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0005586852274314446, | |
| "grad_norm": 0.15083859860897064, | |
| "learning_rate": 9.665157894736842e-05, | |
| "loss": 11.9356, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0005897232956220805, | |
| "grad_norm": 0.18175148963928223, | |
| "learning_rate": 9.612052631578948e-05, | |
| "loss": 11.9251, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0006207613638127163, | |
| "grad_norm": 0.1715611070394516, | |
| "learning_rate": 9.558947368421052e-05, | |
| "loss": 11.927, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0006517994320033521, | |
| "grad_norm": 0.17699895799160004, | |
| "learning_rate": 9.505842105263159e-05, | |
| "loss": 11.9289, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.000682837500193988, | |
| "grad_norm": 0.17426027357578278, | |
| "learning_rate": 9.452736842105263e-05, | |
| "loss": 11.9333, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0007138755683846238, | |
| "grad_norm": 0.20959268510341644, | |
| "learning_rate": 9.399631578947368e-05, | |
| "loss": 11.9314, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0007449136365752596, | |
| "grad_norm": 0.22871772944927216, | |
| "learning_rate": 9.346526315789474e-05, | |
| "loss": 11.9286, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0007759517047658954, | |
| "grad_norm": 0.24363191425800323, | |
| "learning_rate": 9.293421052631578e-05, | |
| "loss": 11.9192, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0008069897729565312, | |
| "grad_norm": 0.28882232308387756, | |
| "learning_rate": 9.240315789473684e-05, | |
| "loss": 11.9225, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.000838027841147167, | |
| "grad_norm": 0.3017212152481079, | |
| "learning_rate": 9.18721052631579e-05, | |
| "loss": 11.9349, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0008690659093378028, | |
| "grad_norm": 0.36739078164100647, | |
| "learning_rate": 9.134105263157895e-05, | |
| "loss": 11.9307, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0009001039775284386, | |
| "grad_norm": 0.35056230425834656, | |
| "learning_rate": 9.081e-05, | |
| "loss": 11.9151, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0009311420457190744, | |
| "grad_norm": 0.3288112282752991, | |
| "learning_rate": 9.027894736842105e-05, | |
| "loss": 11.9204, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0009621801139097103, | |
| "grad_norm": 0.43591663241386414, | |
| "learning_rate": 8.97478947368421e-05, | |
| "loss": 11.9073, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.000993218182100346, | |
| "grad_norm": 0.36357542872428894, | |
| "learning_rate": 8.921684210526316e-05, | |
| "loss": 11.9184, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0010242562502909818, | |
| "grad_norm": 0.4034142792224884, | |
| "learning_rate": 8.86857894736842e-05, | |
| "loss": 11.9087, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0010552943184816176, | |
| "grad_norm": 0.33256056904792786, | |
| "learning_rate": 8.815473684210527e-05, | |
| "loss": 11.9187, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0010863323866722534, | |
| "grad_norm": 0.3923008143901825, | |
| "learning_rate": 8.762368421052631e-05, | |
| "loss": 11.9074, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0011173704548628892, | |
| "grad_norm": 0.4441995322704315, | |
| "learning_rate": 8.709263157894737e-05, | |
| "loss": 11.9153, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0011484085230535253, | |
| "grad_norm": 0.3605518639087677, | |
| "learning_rate": 8.656157894736843e-05, | |
| "loss": 11.912, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.001179446591244161, | |
| "grad_norm": 0.4256209135055542, | |
| "learning_rate": 8.603052631578947e-05, | |
| "loss": 11.9072, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0012104846594347969, | |
| "grad_norm": 0.35672155022621155, | |
| "learning_rate": 8.549947368421052e-05, | |
| "loss": 11.9033, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0012415227276254327, | |
| "grad_norm": 0.3789241909980774, | |
| "learning_rate": 8.496842105263158e-05, | |
| "loss": 11.8991, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0012725607958160685, | |
| "grad_norm": 0.353971928358078, | |
| "learning_rate": 8.443736842105264e-05, | |
| "loss": 11.9023, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0013035988640067043, | |
| "grad_norm": 0.36267516016960144, | |
| "learning_rate": 8.390631578947369e-05, | |
| "loss": 11.8962, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.00133463693219734, | |
| "grad_norm": 0.30361491441726685, | |
| "learning_rate": 8.337526315789473e-05, | |
| "loss": 11.8893, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.001365675000387976, | |
| "grad_norm": 0.3609680235385895, | |
| "learning_rate": 8.284421052631579e-05, | |
| "loss": 11.898, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0013967130685786117, | |
| "grad_norm": 0.3087683618068695, | |
| "learning_rate": 8.231315789473685e-05, | |
| "loss": 11.89, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0014277511367692475, | |
| "grad_norm": 0.3304439187049866, | |
| "learning_rate": 8.178210526315789e-05, | |
| "loss": 11.8939, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0014587892049598833, | |
| "grad_norm": 0.29173874855041504, | |
| "learning_rate": 8.125105263157894e-05, | |
| "loss": 11.8919, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0014898272731505191, | |
| "grad_norm": 0.23176749050617218, | |
| "learning_rate": 8.072e-05, | |
| "loss": 11.895, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.001520865341341155, | |
| "grad_norm": 0.29861122369766235, | |
| "learning_rate": 8.018894736842106e-05, | |
| "loss": 11.8894, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0015519034095317907, | |
| "grad_norm": 0.22058968245983124, | |
| "learning_rate": 7.965789473684211e-05, | |
| "loss": 11.8917, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0015519034095317907, | |
| "eval_loss": 11.89046573638916, | |
| "eval_runtime": 412.729, | |
| "eval_samples_per_second": 32.869, | |
| "eval_steps_per_second": 8.218, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0015829414777224265, | |
| "grad_norm": 0.28083565831184387, | |
| "learning_rate": 7.912684210526315e-05, | |
| "loss": 11.8962, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0016139795459130623, | |
| "grad_norm": 0.40294113755226135, | |
| "learning_rate": 7.859578947368421e-05, | |
| "loss": 11.8997, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0016450176141036982, | |
| "grad_norm": 0.3762059807777405, | |
| "learning_rate": 7.806473684210527e-05, | |
| "loss": 11.8922, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.001676055682294334, | |
| "grad_norm": 0.3565743863582611, | |
| "learning_rate": 7.753368421052631e-05, | |
| "loss": 11.8875, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0017070937504849698, | |
| "grad_norm": 0.23867182433605194, | |
| "learning_rate": 7.700263157894738e-05, | |
| "loss": 11.8867, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0017381318186756056, | |
| "grad_norm": 0.25642549991607666, | |
| "learning_rate": 7.647157894736842e-05, | |
| "loss": 11.8881, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0017691698868662414, | |
| "grad_norm": 0.21004194021224976, | |
| "learning_rate": 7.594052631578948e-05, | |
| "loss": 11.8872, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0018002079550568772, | |
| "grad_norm": 0.21973472833633423, | |
| "learning_rate": 7.540947368421053e-05, | |
| "loss": 11.8852, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.001831246023247513, | |
| "grad_norm": 0.17524276673793793, | |
| "learning_rate": 7.487842105263157e-05, | |
| "loss": 11.8839, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0018622840914381488, | |
| "grad_norm": 0.18540893495082855, | |
| "learning_rate": 7.434736842105263e-05, | |
| "loss": 11.8891, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0018933221596287846, | |
| "grad_norm": 0.19046209752559662, | |
| "learning_rate": 7.381631578947368e-05, | |
| "loss": 11.8908, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0019243602278194206, | |
| "grad_norm": 0.1590886116027832, | |
| "learning_rate": 7.328526315789474e-05, | |
| "loss": 11.8905, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0019553982960100564, | |
| "grad_norm": 0.188175231218338, | |
| "learning_rate": 7.27542105263158e-05, | |
| "loss": 11.8886, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.001986436364200692, | |
| "grad_norm": 0.1609676331281662, | |
| "learning_rate": 7.222315789473684e-05, | |
| "loss": 11.8729, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.002017474432391328, | |
| "grad_norm": 0.13807262480258942, | |
| "learning_rate": 7.16921052631579e-05, | |
| "loss": 11.8826, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0020485125005819636, | |
| "grad_norm": 0.14472399652004242, | |
| "learning_rate": 7.116105263157895e-05, | |
| "loss": 11.8801, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0020795505687725997, | |
| "grad_norm": 0.17055800557136536, | |
| "learning_rate": 7.062999999999999e-05, | |
| "loss": 11.8832, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0021105886369632352, | |
| "grad_norm": 0.1315099149942398, | |
| "learning_rate": 7.009894736842106e-05, | |
| "loss": 11.8775, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0021416267051538713, | |
| "grad_norm": 0.10270003229379654, | |
| "learning_rate": 6.95678947368421e-05, | |
| "loss": 11.8751, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.002172664773344507, | |
| "grad_norm": 0.13660825788974762, | |
| "learning_rate": 6.903684210526316e-05, | |
| "loss": 11.8779, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.002203702841535143, | |
| "grad_norm": 0.11236346513032913, | |
| "learning_rate": 6.850578947368422e-05, | |
| "loss": 11.8768, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0022347409097257785, | |
| "grad_norm": 0.20397527515888214, | |
| "learning_rate": 6.797473684210526e-05, | |
| "loss": 11.8695, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0022657789779164145, | |
| "grad_norm": 0.186505526304245, | |
| "learning_rate": 6.744368421052631e-05, | |
| "loss": 11.8669, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0022968170461070505, | |
| "grad_norm": 0.12963062524795532, | |
| "learning_rate": 6.691263157894736e-05, | |
| "loss": 11.8862, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.002327855114297686, | |
| "grad_norm": 0.10962740331888199, | |
| "learning_rate": 6.638157894736843e-05, | |
| "loss": 11.8836, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.002358893182488322, | |
| "grad_norm": 0.13987334072589874, | |
| "learning_rate": 6.585052631578948e-05, | |
| "loss": 11.8744, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0023899312506789577, | |
| "grad_norm": 0.19806340336799622, | |
| "learning_rate": 6.531947368421052e-05, | |
| "loss": 11.8869, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0024209693188695937, | |
| "grad_norm": 0.11947502940893173, | |
| "learning_rate": 6.478842105263158e-05, | |
| "loss": 11.8842, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0024520073870602293, | |
| "grad_norm": 0.17792446911334991, | |
| "learning_rate": 6.425736842105264e-05, | |
| "loss": 11.8844, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0024830454552508653, | |
| "grad_norm": 0.13691994547843933, | |
| "learning_rate": 6.372631578947368e-05, | |
| "loss": 11.8785, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.002514083523441501, | |
| "grad_norm": 0.19070185720920563, | |
| "learning_rate": 6.319526315789473e-05, | |
| "loss": 11.8824, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.002545121591632137, | |
| "grad_norm": 0.13721631467342377, | |
| "learning_rate": 6.266421052631579e-05, | |
| "loss": 11.8779, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0025761596598227725, | |
| "grad_norm": 0.11055205017328262, | |
| "learning_rate": 6.213315789473685e-05, | |
| "loss": 11.8808, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0026071977280134086, | |
| "grad_norm": 0.1432357281446457, | |
| "learning_rate": 6.16021052631579e-05, | |
| "loss": 11.8835, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.002638235796204044, | |
| "grad_norm": 0.15524394810199738, | |
| "learning_rate": 6.107105263157894e-05, | |
| "loss": 11.8805, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.00266927386439468, | |
| "grad_norm": 0.12444309145212173, | |
| "learning_rate": 6.054e-05, | |
| "loss": 11.8804, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.0027003119325853158, | |
| "grad_norm": 0.19381719827651978, | |
| "learning_rate": 6.000894736842105e-05, | |
| "loss": 11.8824, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.002731350000775952, | |
| "grad_norm": 0.1044355109333992, | |
| "learning_rate": 5.94778947368421e-05, | |
| "loss": 11.8786, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0027623880689665874, | |
| "grad_norm": 0.1657666563987732, | |
| "learning_rate": 5.894684210526316e-05, | |
| "loss": 11.8843, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0027934261371572234, | |
| "grad_norm": 0.15279187262058258, | |
| "learning_rate": 5.841578947368421e-05, | |
| "loss": 11.8778, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.002824464205347859, | |
| "grad_norm": 0.13506744801998138, | |
| "learning_rate": 5.7884736842105265e-05, | |
| "loss": 11.8795, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.002855502273538495, | |
| "grad_norm": 0.1315164864063263, | |
| "learning_rate": 5.7353684210526314e-05, | |
| "loss": 11.8806, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0028865403417291306, | |
| "grad_norm": 0.1755332350730896, | |
| "learning_rate": 5.6822631578947364e-05, | |
| "loss": 11.8803, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0029175784099197666, | |
| "grad_norm": 0.11907346546649933, | |
| "learning_rate": 5.629157894736842e-05, | |
| "loss": 11.8779, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0029486164781104022, | |
| "grad_norm": 0.1171933189034462, | |
| "learning_rate": 5.576052631578948e-05, | |
| "loss": 11.874, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0029796545463010382, | |
| "grad_norm": 0.13140052556991577, | |
| "learning_rate": 5.522947368421053e-05, | |
| "loss": 11.8762, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.003010692614491674, | |
| "grad_norm": 0.12108182907104492, | |
| "learning_rate": 5.469842105263158e-05, | |
| "loss": 11.8796, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.00304173068268231, | |
| "grad_norm": 0.10808394849300385, | |
| "learning_rate": 5.416736842105263e-05, | |
| "loss": 11.8786, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.003072768750872946, | |
| "grad_norm": 0.08503346145153046, | |
| "learning_rate": 5.3636315789473685e-05, | |
| "loss": 11.8801, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0031038068190635815, | |
| "grad_norm": 0.1425827592611313, | |
| "learning_rate": 5.3105263157894734e-05, | |
| "loss": 11.8832, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0031038068190635815, | |
| "eval_loss": 11.877439498901367, | |
| "eval_runtime": 413.0154, | |
| "eval_samples_per_second": 32.846, | |
| "eval_steps_per_second": 8.213, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0031348448872542175, | |
| "grad_norm": 0.2032780945301056, | |
| "learning_rate": 5.257421052631578e-05, | |
| "loss": 11.8815, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.003165882955444853, | |
| "grad_norm": 0.234865203499794, | |
| "learning_rate": 5.2043157894736846e-05, | |
| "loss": 11.8732, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.003196921023635489, | |
| "grad_norm": 0.2010481357574463, | |
| "learning_rate": 5.1512105263157895e-05, | |
| "loss": 11.8762, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0032279590918261247, | |
| "grad_norm": 0.30129870772361755, | |
| "learning_rate": 5.098105263157895e-05, | |
| "loss": 11.8719, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.0032589971600167607, | |
| "grad_norm": 0.1704847365617752, | |
| "learning_rate": 5.045e-05, | |
| "loss": 11.8783, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.0032900352282073963, | |
| "grad_norm": 0.125693678855896, | |
| "learning_rate": 4.991894736842105e-05, | |
| "loss": 11.8782, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.0033210732963980323, | |
| "grad_norm": 0.1328500360250473, | |
| "learning_rate": 4.9387894736842105e-05, | |
| "loss": 11.8734, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.003352111364588668, | |
| "grad_norm": 0.10017025470733643, | |
| "learning_rate": 4.885684210526316e-05, | |
| "loss": 11.8779, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.003383149432779304, | |
| "grad_norm": 0.1521725356578827, | |
| "learning_rate": 4.832578947368421e-05, | |
| "loss": 11.8802, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.0034141875009699395, | |
| "grad_norm": 0.12134622782468796, | |
| "learning_rate": 4.779473684210526e-05, | |
| "loss": 11.8718, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0034452255691605756, | |
| "grad_norm": 0.13203729689121246, | |
| "learning_rate": 4.7263684210526315e-05, | |
| "loss": 11.8839, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.003476263637351211, | |
| "grad_norm": 0.14758096635341644, | |
| "learning_rate": 4.673263157894737e-05, | |
| "loss": 11.8757, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.003507301705541847, | |
| "grad_norm": 0.14375559985637665, | |
| "learning_rate": 4.620157894736842e-05, | |
| "loss": 11.8748, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0035383397737324828, | |
| "grad_norm": 0.10907881706953049, | |
| "learning_rate": 4.5670526315789475e-05, | |
| "loss": 11.8673, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0035693778419231188, | |
| "grad_norm": 0.11960776150226593, | |
| "learning_rate": 4.5139473684210524e-05, | |
| "loss": 11.8824, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.0036004159101137544, | |
| "grad_norm": 0.12177892029285431, | |
| "learning_rate": 4.460842105263158e-05, | |
| "loss": 11.8694, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.0036314539783043904, | |
| "grad_norm": 0.11586112529039383, | |
| "learning_rate": 4.4077368421052636e-05, | |
| "loss": 11.8777, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.003662492046495026, | |
| "grad_norm": 0.10951674729585648, | |
| "learning_rate": 4.3546315789473685e-05, | |
| "loss": 11.8792, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.003693530114685662, | |
| "grad_norm": 0.13929174840450287, | |
| "learning_rate": 4.3015263157894734e-05, | |
| "loss": 11.8713, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.0037245681828762976, | |
| "grad_norm": 0.14048022031784058, | |
| "learning_rate": 4.248421052631579e-05, | |
| "loss": 11.8762, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0037556062510669336, | |
| "grad_norm": 0.1154230386018753, | |
| "learning_rate": 4.1953157894736846e-05, | |
| "loss": 11.8679, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.003786644319257569, | |
| "grad_norm": 0.11130271852016449, | |
| "learning_rate": 4.1422105263157895e-05, | |
| "loss": 11.8732, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0038176823874482052, | |
| "grad_norm": 0.19456033408641815, | |
| "learning_rate": 4.0891052631578944e-05, | |
| "loss": 11.8726, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0038487204556388412, | |
| "grad_norm": 0.17270055413246155, | |
| "learning_rate": 4.036e-05, | |
| "loss": 11.8599, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.003879758523829477, | |
| "grad_norm": 0.1157611832022667, | |
| "learning_rate": 3.9828947368421056e-05, | |
| "loss": 11.8741, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.003910796592020113, | |
| "grad_norm": 0.13942551612854004, | |
| "learning_rate": 3.9297894736842105e-05, | |
| "loss": 11.8738, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.003941834660210749, | |
| "grad_norm": 0.1275920271873474, | |
| "learning_rate": 3.8766842105263154e-05, | |
| "loss": 11.8615, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.003972872728401384, | |
| "grad_norm": 0.09613733738660812, | |
| "learning_rate": 3.823578947368421e-05, | |
| "loss": 11.8738, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.00400391079659202, | |
| "grad_norm": 0.16122451424598694, | |
| "learning_rate": 3.7704736842105265e-05, | |
| "loss": 11.8692, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.004034948864782656, | |
| "grad_norm": 0.13367600739002228, | |
| "learning_rate": 3.7173684210526315e-05, | |
| "loss": 11.8725, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.004065986932973292, | |
| "grad_norm": 0.11731020361185074, | |
| "learning_rate": 3.664263157894737e-05, | |
| "loss": 11.8773, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.004097025001163927, | |
| "grad_norm": 0.13289429247379303, | |
| "learning_rate": 3.611157894736842e-05, | |
| "loss": 11.8661, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.004128063069354563, | |
| "grad_norm": 0.13751353323459625, | |
| "learning_rate": 3.5580526315789475e-05, | |
| "loss": 11.8654, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.004159101137545199, | |
| "grad_norm": 0.13276422023773193, | |
| "learning_rate": 3.504947368421053e-05, | |
| "loss": 11.8733, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.004190139205735835, | |
| "grad_norm": 0.12328977137804031, | |
| "learning_rate": 3.451842105263158e-05, | |
| "loss": 11.8791, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.0042211772739264705, | |
| "grad_norm": 0.14081747829914093, | |
| "learning_rate": 3.398736842105263e-05, | |
| "loss": 11.8727, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.0042522153421171065, | |
| "grad_norm": 0.13730375468730927, | |
| "learning_rate": 3.345631578947368e-05, | |
| "loss": 11.8715, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.0042832534103077425, | |
| "grad_norm": 0.1453617960214615, | |
| "learning_rate": 3.292526315789474e-05, | |
| "loss": 11.8744, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.0043142914784983786, | |
| "grad_norm": 0.18549758195877075, | |
| "learning_rate": 3.239421052631579e-05, | |
| "loss": 11.8698, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.004345329546689014, | |
| "grad_norm": 0.13958542048931122, | |
| "learning_rate": 3.186315789473684e-05, | |
| "loss": 11.8703, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.00437636761487965, | |
| "grad_norm": 0.1116572916507721, | |
| "learning_rate": 3.1332105263157895e-05, | |
| "loss": 11.8733, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.004407405683070286, | |
| "grad_norm": 0.15264497697353363, | |
| "learning_rate": 3.080105263157895e-05, | |
| "loss": 11.8744, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.004438443751260922, | |
| "grad_norm": 0.16946198046207428, | |
| "learning_rate": 3.027e-05, | |
| "loss": 11.8748, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.004469481819451557, | |
| "grad_norm": 0.17827944457530975, | |
| "learning_rate": 2.973894736842105e-05, | |
| "loss": 11.8683, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.004500519887642193, | |
| "grad_norm": 0.15590260922908783, | |
| "learning_rate": 2.9207894736842105e-05, | |
| "loss": 11.8667, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.004531557955832829, | |
| "grad_norm": 0.17276710271835327, | |
| "learning_rate": 2.8676842105263157e-05, | |
| "loss": 11.8716, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.004562596024023465, | |
| "grad_norm": 0.15102654695510864, | |
| "learning_rate": 2.814578947368421e-05, | |
| "loss": 11.8671, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.004593634092214101, | |
| "grad_norm": 0.18881015479564667, | |
| "learning_rate": 2.7614736842105266e-05, | |
| "loss": 11.8761, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.004624672160404736, | |
| "grad_norm": 0.20375801622867584, | |
| "learning_rate": 2.7083684210526315e-05, | |
| "loss": 11.8732, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.004655710228595372, | |
| "grad_norm": 0.221470445394516, | |
| "learning_rate": 2.6552631578947367e-05, | |
| "loss": 11.8677, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.004655710228595372, | |
| "eval_loss": 11.868966102600098, | |
| "eval_runtime": 414.3737, | |
| "eval_samples_per_second": 32.739, | |
| "eval_steps_per_second": 8.186, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.004686748296786008, | |
| "grad_norm": 0.1980566531419754, | |
| "learning_rate": 2.6021578947368423e-05, | |
| "loss": 11.8806, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.004717786364976644, | |
| "grad_norm": 0.19020256400108337, | |
| "learning_rate": 2.5490526315789475e-05, | |
| "loss": 11.8685, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.004748824433167279, | |
| "grad_norm": 0.24149803817272186, | |
| "learning_rate": 2.4959473684210524e-05, | |
| "loss": 11.857, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.004779862501357915, | |
| "grad_norm": 0.2381637543439865, | |
| "learning_rate": 2.442842105263158e-05, | |
| "loss": 11.8716, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.0048109005695485514, | |
| "grad_norm": 0.18365250527858734, | |
| "learning_rate": 2.389736842105263e-05, | |
| "loss": 11.8579, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.0048419386377391875, | |
| "grad_norm": 0.13402819633483887, | |
| "learning_rate": 2.3366315789473685e-05, | |
| "loss": 11.8672, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.004872976705929823, | |
| "grad_norm": 0.13781289756298065, | |
| "learning_rate": 2.2835263157894738e-05, | |
| "loss": 11.869, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.004904014774120459, | |
| "grad_norm": 0.1478467583656311, | |
| "learning_rate": 2.230421052631579e-05, | |
| "loss": 11.8676, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.004935052842311095, | |
| "grad_norm": 0.12892381846904755, | |
| "learning_rate": 2.1773157894736843e-05, | |
| "loss": 11.8707, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.004966090910501731, | |
| "grad_norm": 0.1606762856245041, | |
| "learning_rate": 2.1242105263157895e-05, | |
| "loss": 11.8716, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.004997128978692366, | |
| "grad_norm": 0.07333854585886002, | |
| "learning_rate": 2.0711052631578947e-05, | |
| "loss": 11.8637, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.005028167046883002, | |
| "grad_norm": 0.15114228427410126, | |
| "learning_rate": 2.018e-05, | |
| "loss": 11.8736, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.005059205115073638, | |
| "grad_norm": 0.0825541689991951, | |
| "learning_rate": 1.9648947368421052e-05, | |
| "loss": 11.859, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.005090243183264274, | |
| "grad_norm": 0.14535300433635712, | |
| "learning_rate": 1.9117894736842105e-05, | |
| "loss": 11.8684, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.005121281251454909, | |
| "grad_norm": 0.10561253130435944, | |
| "learning_rate": 1.8586842105263157e-05, | |
| "loss": 11.8702, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.005152319319645545, | |
| "grad_norm": 0.1263810098171234, | |
| "learning_rate": 1.805578947368421e-05, | |
| "loss": 11.8643, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.005183357387836181, | |
| "grad_norm": 0.07599103450775146, | |
| "learning_rate": 1.7524736842105266e-05, | |
| "loss": 11.8712, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.005214395456026817, | |
| "grad_norm": 0.10940805822610855, | |
| "learning_rate": 1.6993684210526315e-05, | |
| "loss": 11.8615, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.005245433524217452, | |
| "grad_norm": 0.07466328144073486, | |
| "learning_rate": 1.646263157894737e-05, | |
| "loss": 11.8694, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.005276471592408088, | |
| "grad_norm": 0.16310299932956696, | |
| "learning_rate": 1.593157894736842e-05, | |
| "loss": 11.8719, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.005307509660598724, | |
| "grad_norm": 0.09811323136091232, | |
| "learning_rate": 1.5400526315789475e-05, | |
| "loss": 11.8716, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.00533854772878936, | |
| "grad_norm": 0.07701898366212845, | |
| "learning_rate": 1.4869473684210524e-05, | |
| "loss": 11.8667, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.005369585796979996, | |
| "grad_norm": 0.09958688914775848, | |
| "learning_rate": 1.4338421052631579e-05, | |
| "loss": 11.869, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.0054006238651706315, | |
| "grad_norm": 0.10621592402458191, | |
| "learning_rate": 1.3807368421052633e-05, | |
| "loss": 11.8738, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.005431661933361268, | |
| "grad_norm": 0.09998718649148941, | |
| "learning_rate": 1.3276315789473684e-05, | |
| "loss": 11.8742, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.005462700001551904, | |
| "grad_norm": 0.08477786928415298, | |
| "learning_rate": 1.2745263157894738e-05, | |
| "loss": 11.8695, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.00549373806974254, | |
| "grad_norm": 0.0795200765132904, | |
| "learning_rate": 1.221421052631579e-05, | |
| "loss": 11.8666, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.005524776137933175, | |
| "grad_norm": 0.13940560817718506, | |
| "learning_rate": 1.1683157894736843e-05, | |
| "loss": 11.8669, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.005555814206123811, | |
| "grad_norm": 0.09820152819156647, | |
| "learning_rate": 1.1152105263157895e-05, | |
| "loss": 11.8619, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.005586852274314447, | |
| "grad_norm": 0.10361889004707336, | |
| "learning_rate": 1.0621052631578948e-05, | |
| "loss": 11.8694, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.005617890342505083, | |
| "grad_norm": 0.1097809374332428, | |
| "learning_rate": 1.009e-05, | |
| "loss": 11.8732, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.005648928410695718, | |
| "grad_norm": 0.18127013742923737, | |
| "learning_rate": 9.558947368421052e-06, | |
| "loss": 11.8678, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.005679966478886354, | |
| "grad_norm": 0.12430532276630402, | |
| "learning_rate": 9.027894736842105e-06, | |
| "loss": 11.8681, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.00571100454707699, | |
| "grad_norm": 0.17203256487846375, | |
| "learning_rate": 8.496842105263157e-06, | |
| "loss": 11.8631, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.005742042615267626, | |
| "grad_norm": 0.09748750180006027, | |
| "learning_rate": 7.96578947368421e-06, | |
| "loss": 11.8646, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.005773080683458261, | |
| "grad_norm": 0.14566555619239807, | |
| "learning_rate": 7.434736842105262e-06, | |
| "loss": 11.8662, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.005804118751648897, | |
| "grad_norm": 0.13249988853931427, | |
| "learning_rate": 6.903684210526316e-06, | |
| "loss": 11.8669, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.005835156819839533, | |
| "grad_norm": 0.14271654188632965, | |
| "learning_rate": 6.372631578947369e-06, | |
| "loss": 11.8662, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.005866194888030169, | |
| "grad_norm": 0.10937763005495071, | |
| "learning_rate": 5.841578947368421e-06, | |
| "loss": 11.8672, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.0058972329562208044, | |
| "grad_norm": 0.1206103041768074, | |
| "learning_rate": 5.310526315789474e-06, | |
| "loss": 11.862, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.0059282710244114405, | |
| "grad_norm": 0.12608934938907623, | |
| "learning_rate": 4.779473684210526e-06, | |
| "loss": 11.8674, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.0059593090926020765, | |
| "grad_norm": 0.1504194289445877, | |
| "learning_rate": 4.248421052631579e-06, | |
| "loss": 11.8704, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.0059903471607927125, | |
| "grad_norm": 0.19314518570899963, | |
| "learning_rate": 3.717368421052631e-06, | |
| "loss": 11.8675, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.006021385228983348, | |
| "grad_norm": 0.1470961570739746, | |
| "learning_rate": 3.1863157894736844e-06, | |
| "loss": 11.8726, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.006052423297173984, | |
| "grad_norm": 0.17263321578502655, | |
| "learning_rate": 2.655263157894737e-06, | |
| "loss": 11.8699, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.00608346136536462, | |
| "grad_norm": 0.18947991728782654, | |
| "learning_rate": 2.1242105263157893e-06, | |
| "loss": 11.8713, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.006114499433555256, | |
| "grad_norm": 0.19662337005138397, | |
| "learning_rate": 1.5931578947368422e-06, | |
| "loss": 11.873, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.006145537501745892, | |
| "grad_norm": 0.1542241871356964, | |
| "learning_rate": 1.0621052631578947e-06, | |
| "loss": 11.8692, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.006176575569936527, | |
| "grad_norm": 0.20754170417785645, | |
| "learning_rate": 5.310526315789473e-07, | |
| "loss": 11.8685, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.006207613638127163, | |
| "grad_norm": 0.20448267459869385, | |
| "learning_rate": 0.0, | |
| "loss": 11.8738, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.006207613638127163, | |
| "eval_loss": 11.866828918457031, | |
| "eval_runtime": 412.6675, | |
| "eval_samples_per_second": 32.874, | |
| "eval_steps_per_second": 8.22, | |
| "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": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 261529927680.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |