| { | |
| "best_metric": 4.415204048156738, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 0.0379416646905383, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.000379416646905383, | |
| "grad_norm": 4.444539546966553, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 3.6119, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.000379416646905383, | |
| "eval_loss": 5.630036354064941, | |
| "eval_runtime": 70.5773, | |
| "eval_samples_per_second": 62.896, | |
| "eval_steps_per_second": 15.727, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.000758833293810766, | |
| "grad_norm": 10.298806190490723, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 4.4198, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0011382499407161489, | |
| "grad_norm": 10.508999824523926, | |
| "learning_rate": 1.5e-06, | |
| "loss": 4.3823, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.001517666587621532, | |
| "grad_norm": 14.867856979370117, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 5.3149, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.001897083234526915, | |
| "grad_norm": 11.521822929382324, | |
| "learning_rate": 2.5e-06, | |
| "loss": 4.957, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0022764998814322978, | |
| "grad_norm": 10.65507984161377, | |
| "learning_rate": 3e-06, | |
| "loss": 5.0594, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.002655916528337681, | |
| "grad_norm": 14.449377059936523, | |
| "learning_rate": 3.5e-06, | |
| "loss": 5.4046, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.003035333175243064, | |
| "grad_norm": 15.257190704345703, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 5.4988, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.003414749822148447, | |
| "grad_norm": 15.488490104675293, | |
| "learning_rate": 4.5e-06, | |
| "loss": 4.9476, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00379416646905383, | |
| "grad_norm": 15.310659408569336, | |
| "learning_rate": 5e-06, | |
| "loss": 5.2732, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0041735831159592125, | |
| "grad_norm": 10.856189727783203, | |
| "learning_rate": 4.99847706754774e-06, | |
| "loss": 5.0276, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0045529997628645956, | |
| "grad_norm": 17.822044372558594, | |
| "learning_rate": 4.993910125649561e-06, | |
| "loss": 5.2223, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.004932416409769979, | |
| "grad_norm": 11.17652702331543, | |
| "learning_rate": 4.986304738420684e-06, | |
| "loss": 4.818, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.005311833056675362, | |
| "grad_norm": 16.337657928466797, | |
| "learning_rate": 4.975670171853926e-06, | |
| "loss": 5.4286, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.005691249703580745, | |
| "grad_norm": 16.31676483154297, | |
| "learning_rate": 4.962019382530521e-06, | |
| "loss": 5.0845, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.006070666350486128, | |
| "grad_norm": 13.25898551940918, | |
| "learning_rate": 4.9453690018345144e-06, | |
| "loss": 5.1326, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.006450082997391511, | |
| "grad_norm": 15.131325721740723, | |
| "learning_rate": 4.925739315689991e-06, | |
| "loss": 5.0989, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.006829499644296894, | |
| "grad_norm": 14.353851318359375, | |
| "learning_rate": 4.903154239845798e-06, | |
| "loss": 5.0136, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.007208916291202277, | |
| "grad_norm": 13.283928871154785, | |
| "learning_rate": 4.8776412907378845e-06, | |
| "loss": 5.5043, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.00758833293810766, | |
| "grad_norm": 13.317709922790527, | |
| "learning_rate": 4.849231551964771e-06, | |
| "loss": 5.1231, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.007967749585013043, | |
| "grad_norm": 10.19308853149414, | |
| "learning_rate": 4.817959636416969e-06, | |
| "loss": 5.0222, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.008347166231918425, | |
| "grad_norm": 12.509997367858887, | |
| "learning_rate": 4.783863644106502e-06, | |
| "loss": 5.0395, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.008726582878823809, | |
| "grad_norm": 13.469011306762695, | |
| "learning_rate": 4.746985115747918e-06, | |
| "loss": 4.9736, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.009105999525729191, | |
| "grad_norm": 13.559428215026855, | |
| "learning_rate": 4.707368982147318e-06, | |
| "loss": 5.1422, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.009485416172634575, | |
| "grad_norm": 12.164199829101562, | |
| "learning_rate": 4.665063509461098e-06, | |
| "loss": 5.0459, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.009864832819539957, | |
| "grad_norm": 12.601916313171387, | |
| "learning_rate": 4.620120240391065e-06, | |
| "loss": 5.5161, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.010244249466445341, | |
| "grad_norm": 10.389106750488281, | |
| "learning_rate": 4.572593931387604e-06, | |
| "loss": 4.7012, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.010623666113350723, | |
| "grad_norm": 11.981220245361328, | |
| "learning_rate": 4.522542485937369e-06, | |
| "loss": 5.2306, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.011003082760256105, | |
| "grad_norm": 8.926562309265137, | |
| "learning_rate": 4.470026884016805e-06, | |
| "loss": 5.2015, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01138249940716149, | |
| "grad_norm": 11.07724666595459, | |
| "learning_rate": 4.415111107797445e-06, | |
| "loss": 5.3189, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.011761916054066872, | |
| "grad_norm": 23.328981399536133, | |
| "learning_rate": 4.357862063693486e-06, | |
| "loss": 5.2879, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.012141332700972255, | |
| "grad_norm": 9.831049919128418, | |
| "learning_rate": 4.2983495008466285e-06, | |
| "loss": 5.2354, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.012520749347877638, | |
| "grad_norm": 10.883379936218262, | |
| "learning_rate": 4.236645926147493e-06, | |
| "loss": 5.101, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.012900165994783021, | |
| "grad_norm": 13.54520034790039, | |
| "learning_rate": 4.172826515897146e-06, | |
| "loss": 5.3399, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.013279582641688404, | |
| "grad_norm": 9.359816551208496, | |
| "learning_rate": 4.106969024216348e-06, | |
| "loss": 5.0393, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.013658999288593788, | |
| "grad_norm": 8.04438304901123, | |
| "learning_rate": 4.039153688314146e-06, | |
| "loss": 5.1888, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.01403841593549917, | |
| "grad_norm": 7.696422100067139, | |
| "learning_rate": 3.969463130731183e-06, | |
| "loss": 5.0935, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.014417832582404554, | |
| "grad_norm": 9.850431442260742, | |
| "learning_rate": 3.897982258676867e-06, | |
| "loss": 5.3628, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.014797249229309936, | |
| "grad_norm": 17.82948112487793, | |
| "learning_rate": 3.824798160583012e-06, | |
| "loss": 5.2962, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.01517666587621532, | |
| "grad_norm": 8.378523826599121, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 4.873, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.015556082523120702, | |
| "grad_norm": 9.744098663330078, | |
| "learning_rate": 3.6736789069647273e-06, | |
| "loss": 5.1868, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.015935499170026086, | |
| "grad_norm": 8.492013931274414, | |
| "learning_rate": 3.595927866972694e-06, | |
| "loss": 5.3123, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.016314915816931466, | |
| "grad_norm": 9.203422546386719, | |
| "learning_rate": 3.516841607689501e-06, | |
| "loss": 5.1639, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.01669433246383685, | |
| "grad_norm": 8.369019508361816, | |
| "learning_rate": 3.436516483539781e-06, | |
| "loss": 5.0934, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.017073749110742234, | |
| "grad_norm": 8.947461128234863, | |
| "learning_rate": 3.3550503583141726e-06, | |
| "loss": 5.1892, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.017453165757647618, | |
| "grad_norm": 10.578983306884766, | |
| "learning_rate": 3.272542485937369e-06, | |
| "loss": 5.6706, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.017832582404553, | |
| "grad_norm": 9.356407165527344, | |
| "learning_rate": 3.189093389542498e-06, | |
| "loss": 5.6261, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.018211999051458382, | |
| "grad_norm": 9.004945755004883, | |
| "learning_rate": 3.1048047389991693e-06, | |
| "loss": 5.3589, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.018591415698363766, | |
| "grad_norm": 8.144246101379395, | |
| "learning_rate": 3.019779227044398e-06, | |
| "loss": 5.4095, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.01897083234526915, | |
| "grad_norm": 9.677080154418945, | |
| "learning_rate": 2.9341204441673267e-06, | |
| "loss": 5.7475, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01897083234526915, | |
| "eval_loss": 4.675076484680176, | |
| "eval_runtime": 70.0014, | |
| "eval_samples_per_second": 63.413, | |
| "eval_steps_per_second": 15.857, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01935024899217453, | |
| "grad_norm": 3.7669692039489746, | |
| "learning_rate": 2.847932752400164e-06, | |
| "loss": 4.0896, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.019729665639079914, | |
| "grad_norm": 4.779418468475342, | |
| "learning_rate": 2.761321158169134e-06, | |
| "loss": 3.8922, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0201090822859853, | |
| "grad_norm": 5.031223297119141, | |
| "learning_rate": 2.6743911843603134e-06, | |
| "loss": 4.1811, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.020488498932890682, | |
| "grad_norm": 4.953243255615234, | |
| "learning_rate": 2.587248741756253e-06, | |
| "loss": 3.7881, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.020867915579796063, | |
| "grad_norm": 5.7064208984375, | |
| "learning_rate": 2.5e-06, | |
| "loss": 4.0777, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.021247332226701447, | |
| "grad_norm": 5.718916893005371, | |
| "learning_rate": 2.4127512582437486e-06, | |
| "loss": 4.1119, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.02162674887360683, | |
| "grad_norm": 6.402388572692871, | |
| "learning_rate": 2.325608815639687e-06, | |
| "loss": 4.0965, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.02200616552051221, | |
| "grad_norm": 6.114238739013672, | |
| "learning_rate": 2.238678841830867e-06, | |
| "loss": 4.1586, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.022385582167417595, | |
| "grad_norm": 5.461117267608643, | |
| "learning_rate": 2.1520672475998374e-06, | |
| "loss": 3.6969, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.02276499881432298, | |
| "grad_norm": 6.01952600479126, | |
| "learning_rate": 2.0658795558326745e-06, | |
| "loss": 4.0374, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.023144415461228363, | |
| "grad_norm": 6.108837604522705, | |
| "learning_rate": 1.9802207729556023e-06, | |
| "loss": 4.3124, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.023523832108133743, | |
| "grad_norm": 6.704663276672363, | |
| "learning_rate": 1.895195261000831e-06, | |
| "loss": 3.8151, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.023903248755039127, | |
| "grad_norm": 6.462284088134766, | |
| "learning_rate": 1.8109066104575023e-06, | |
| "loss": 4.3228, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.02428266540194451, | |
| "grad_norm": 7.066595554351807, | |
| "learning_rate": 1.7274575140626318e-06, | |
| "loss": 4.1603, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.024662082048849895, | |
| "grad_norm": 6.231473922729492, | |
| "learning_rate": 1.6449496416858285e-06, | |
| "loss": 4.3381, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.025041498695755275, | |
| "grad_norm": 6.865777492523193, | |
| "learning_rate": 1.56348351646022e-06, | |
| "loss": 3.7183, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.02542091534266066, | |
| "grad_norm": 6.609790802001953, | |
| "learning_rate": 1.4831583923105e-06, | |
| "loss": 4.5205, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.025800331989566043, | |
| "grad_norm": 6.334497928619385, | |
| "learning_rate": 1.4040721330273063e-06, | |
| "loss": 4.3837, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.026179748636471427, | |
| "grad_norm": 7.466740608215332, | |
| "learning_rate": 1.3263210930352737e-06, | |
| "loss": 4.3184, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.026559165283376807, | |
| "grad_norm": 7.18544340133667, | |
| "learning_rate": 1.2500000000000007e-06, | |
| "loss": 4.3352, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.02693858193028219, | |
| "grad_norm": 6.762673377990723, | |
| "learning_rate": 1.1752018394169882e-06, | |
| "loss": 4.504, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.027317998577187575, | |
| "grad_norm": 5.527563095092773, | |
| "learning_rate": 1.1020177413231334e-06, | |
| "loss": 4.1255, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.027697415224092956, | |
| "grad_norm": 6.735006332397461, | |
| "learning_rate": 1.0305368692688175e-06, | |
| "loss": 4.3378, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.02807683187099834, | |
| "grad_norm": 6.313520431518555, | |
| "learning_rate": 9.608463116858544e-07, | |
| "loss": 3.8856, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.028456248517903723, | |
| "grad_norm": 6.748076915740967, | |
| "learning_rate": 8.930309757836517e-07, | |
| "loss": 4.1941, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.028835665164809107, | |
| "grad_norm": 7.661444187164307, | |
| "learning_rate": 8.271734841028553e-07, | |
| "loss": 4.2481, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.029215081811714488, | |
| "grad_norm": 7.217004299163818, | |
| "learning_rate": 7.633540738525066e-07, | |
| "loss": 4.3925, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.02959449845861987, | |
| "grad_norm": 7.152622222900391, | |
| "learning_rate": 7.016504991533727e-07, | |
| "loss": 3.8941, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.029973915105525255, | |
| "grad_norm": 6.765809059143066, | |
| "learning_rate": 6.421379363065142e-07, | |
| "loss": 4.5282, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.03035333175243064, | |
| "grad_norm": 6.723090171813965, | |
| "learning_rate": 5.848888922025553e-07, | |
| "loss": 4.3739, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03073274839933602, | |
| "grad_norm": 7.00840950012207, | |
| "learning_rate": 5.299731159831953e-07, | |
| "loss": 4.2958, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.031112165046241404, | |
| "grad_norm": 6.291935443878174, | |
| "learning_rate": 4.774575140626317e-07, | |
| "loss": 3.9951, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.03149158169314679, | |
| "grad_norm": 6.735934734344482, | |
| "learning_rate": 4.27406068612396e-07, | |
| "loss": 4.4475, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.03187099834005217, | |
| "grad_norm": 7.057698726654053, | |
| "learning_rate": 3.798797596089351e-07, | |
| "loss": 4.248, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.032250414986957555, | |
| "grad_norm": 7.530543804168701, | |
| "learning_rate": 3.3493649053890325e-07, | |
| "loss": 4.8054, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03262983163386293, | |
| "grad_norm": 11.086931228637695, | |
| "learning_rate": 2.9263101785268253e-07, | |
| "loss": 4.5513, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.033009248280768316, | |
| "grad_norm": 6.664738655090332, | |
| "learning_rate": 2.53014884252083e-07, | |
| "loss": 4.3811, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.0333886649276737, | |
| "grad_norm": 7.505542278289795, | |
| "learning_rate": 2.1613635589349756e-07, | |
| "loss": 4.5977, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.033768081574579084, | |
| "grad_norm": 7.716992378234863, | |
| "learning_rate": 1.8204036358303173e-07, | |
| "loss": 4.8648, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.03414749822148447, | |
| "grad_norm": 9.030020713806152, | |
| "learning_rate": 1.507684480352292e-07, | |
| "loss": 4.5738, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03452691486838985, | |
| "grad_norm": 7.875784873962402, | |
| "learning_rate": 1.223587092621162e-07, | |
| "loss": 4.5702, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.034906331515295236, | |
| "grad_norm": 6.821437358856201, | |
| "learning_rate": 9.684576015420277e-08, | |
| "loss": 4.7577, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.03528574816220062, | |
| "grad_norm": 7.495672225952148, | |
| "learning_rate": 7.426068431000883e-08, | |
| "loss": 4.5927, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.035665164809106, | |
| "grad_norm": 8.056374549865723, | |
| "learning_rate": 5.463099816548578e-08, | |
| "loss": 4.8237, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.03604458145601138, | |
| "grad_norm": 8.059727668762207, | |
| "learning_rate": 3.798061746947995e-08, | |
| "loss": 5.0902, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.036423998102916764, | |
| "grad_norm": 8.202387809753418, | |
| "learning_rate": 2.4329828146074096e-08, | |
| "loss": 4.8697, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.03680341474982215, | |
| "grad_norm": 8.158020973205566, | |
| "learning_rate": 1.3695261579316776e-08, | |
| "loss": 4.7059, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.03718283139672753, | |
| "grad_norm": 8.748244285583496, | |
| "learning_rate": 6.089874350439507e-09, | |
| "loss": 4.7764, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.037562248043632916, | |
| "grad_norm": 8.805473327636719, | |
| "learning_rate": 1.5229324522605949e-09, | |
| "loss": 5.3301, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0379416646905383, | |
| "grad_norm": 8.97905158996582, | |
| "learning_rate": 0.0, | |
| "loss": 5.5488, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0379416646905383, | |
| "eval_loss": 4.415204048156738, | |
| "eval_runtime": 70.0282, | |
| "eval_samples_per_second": 63.389, | |
| "eval_steps_per_second": 15.851, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 100, | |
| "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": 6212314516684800.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |