| { | |
| "best_metric": 0.0022250961046665907, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.03170577045022194, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0001585288522511097, | |
| "grad_norm": 12.08696174621582, | |
| "learning_rate": 1.008e-05, | |
| "loss": 4.2706, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0001585288522511097, | |
| "eval_loss": 4.04376220703125, | |
| "eval_runtime": 285.4706, | |
| "eval_samples_per_second": 9.304, | |
| "eval_steps_per_second": 2.326, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0003170577045022194, | |
| "grad_norm": 15.856035232543945, | |
| "learning_rate": 2.016e-05, | |
| "loss": 3.9804, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0004755865567533291, | |
| "grad_norm": 15.65820026397705, | |
| "learning_rate": 3.024e-05, | |
| "loss": 3.9763, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0006341154090044388, | |
| "grad_norm": 14.708057403564453, | |
| "learning_rate": 4.032e-05, | |
| "loss": 3.2287, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0007926442612555486, | |
| "grad_norm": 10.415152549743652, | |
| "learning_rate": 5.04e-05, | |
| "loss": 2.0174, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0009511731135066582, | |
| "grad_norm": 12.43994140625, | |
| "learning_rate": 6.048e-05, | |
| "loss": 1.2254, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0011097019657577679, | |
| "grad_norm": 10.96854019165039, | |
| "learning_rate": 7.055999999999999e-05, | |
| "loss": 0.3292, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0012682308180088776, | |
| "grad_norm": 36.675350189208984, | |
| "learning_rate": 8.064e-05, | |
| "loss": 0.2307, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0014267596702599874, | |
| "grad_norm": 2.064239501953125, | |
| "learning_rate": 9.072e-05, | |
| "loss": 0.0138, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0015852885225110971, | |
| "grad_norm": 0.1330491453409195, | |
| "learning_rate": 0.0001008, | |
| "loss": 0.001, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0017438173747622067, | |
| "grad_norm": 0.01124438177794218, | |
| "learning_rate": 0.00010026947368421052, | |
| "loss": 0.0001, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0019023462270133164, | |
| "grad_norm": 0.004640920553356409, | |
| "learning_rate": 9.973894736842104e-05, | |
| "loss": 0.0, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.002060875079264426, | |
| "grad_norm": 0.007322155870497227, | |
| "learning_rate": 9.920842105263157e-05, | |
| "loss": 0.0, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0022194039315155357, | |
| "grad_norm": 0.012465717270970345, | |
| "learning_rate": 9.86778947368421e-05, | |
| "loss": 0.0001, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0023779327837666455, | |
| "grad_norm": 0.0506725050508976, | |
| "learning_rate": 9.814736842105264e-05, | |
| "loss": 0.0001, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0025364616360177552, | |
| "grad_norm": 0.00856021698564291, | |
| "learning_rate": 9.761684210526316e-05, | |
| "loss": 0.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.002694990488268865, | |
| "grad_norm": 0.001643276889808476, | |
| "learning_rate": 9.708631578947368e-05, | |
| "loss": 0.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0028535193405199747, | |
| "grad_norm": 0.0009701295639388263, | |
| "learning_rate": 9.655578947368421e-05, | |
| "loss": 0.0, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0030120481927710845, | |
| "grad_norm": 0.0005847598076798022, | |
| "learning_rate": 9.602526315789473e-05, | |
| "loss": 0.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0031705770450221942, | |
| "grad_norm": 0.0006212879670783877, | |
| "learning_rate": 9.549473684210525e-05, | |
| "loss": 0.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0033291058972733036, | |
| "grad_norm": 0.0008075927617028356, | |
| "learning_rate": 9.496421052631579e-05, | |
| "loss": 0.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0034876347495244133, | |
| "grad_norm": 0.0006020912551321089, | |
| "learning_rate": 9.443368421052631e-05, | |
| "loss": 0.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.003646163601775523, | |
| "grad_norm": 0.0005198498256504536, | |
| "learning_rate": 9.390315789473683e-05, | |
| "loss": 0.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.003804692454026633, | |
| "grad_norm": 0.0016288729384541512, | |
| "learning_rate": 9.337263157894737e-05, | |
| "loss": 0.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.003963221306277742, | |
| "grad_norm": 0.0007833559648133814, | |
| "learning_rate": 9.28421052631579e-05, | |
| "loss": 0.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.004121750158528852, | |
| "grad_norm": 0.0014727013185620308, | |
| "learning_rate": 9.231157894736842e-05, | |
| "loss": 0.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.004280279010779962, | |
| "grad_norm": 0.0016179227968677878, | |
| "learning_rate": 9.178105263157895e-05, | |
| "loss": 0.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.004438807863031071, | |
| "grad_norm": 0.0005032668123021722, | |
| "learning_rate": 9.125052631578948e-05, | |
| "loss": 0.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.004597336715282181, | |
| "grad_norm": 0.0005934814107604325, | |
| "learning_rate": 9.072e-05, | |
| "loss": 0.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.004755865567533291, | |
| "grad_norm": 0.00047520003863610327, | |
| "learning_rate": 9.018947368421052e-05, | |
| "loss": 0.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.004914394419784401, | |
| "grad_norm": 0.0007378292502835393, | |
| "learning_rate": 8.965894736842104e-05, | |
| "loss": 0.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0050729232720355105, | |
| "grad_norm": 0.0007299358258023858, | |
| "learning_rate": 8.912842105263157e-05, | |
| "loss": 0.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.00523145212428662, | |
| "grad_norm": 0.0005630860105156898, | |
| "learning_rate": 8.85978947368421e-05, | |
| "loss": 0.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.00538998097653773, | |
| "grad_norm": 0.0009994081920012832, | |
| "learning_rate": 8.806736842105264e-05, | |
| "loss": 0.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.00554850982878884, | |
| "grad_norm": 0.0004544692055787891, | |
| "learning_rate": 8.753684210526316e-05, | |
| "loss": 0.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0057070386810399495, | |
| "grad_norm": 0.00069584691664204, | |
| "learning_rate": 8.700631578947369e-05, | |
| "loss": 0.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.005865567533291059, | |
| "grad_norm": 0.0019399194279685616, | |
| "learning_rate": 8.647578947368421e-05, | |
| "loss": 0.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.006024096385542169, | |
| "grad_norm": 0.0005930989282205701, | |
| "learning_rate": 8.594526315789473e-05, | |
| "loss": 0.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.006182625237793279, | |
| "grad_norm": 0.0007997532375156879, | |
| "learning_rate": 8.541473684210525e-05, | |
| "loss": 0.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0063411540900443885, | |
| "grad_norm": 0.0013047147076576948, | |
| "learning_rate": 8.488421052631578e-05, | |
| "loss": 0.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.006499682942295497, | |
| "grad_norm": 0.00036864462890662253, | |
| "learning_rate": 8.435368421052631e-05, | |
| "loss": 0.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.006658211794546607, | |
| "grad_norm": 0.0003495727141853422, | |
| "learning_rate": 8.382315789473684e-05, | |
| "loss": 0.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.006816740646797717, | |
| "grad_norm": 0.0003555091971065849, | |
| "learning_rate": 8.329263157894737e-05, | |
| "loss": 0.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.006975269499048827, | |
| "grad_norm": 0.0002377888304181397, | |
| "learning_rate": 8.27621052631579e-05, | |
| "loss": 0.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.007133798351299936, | |
| "grad_norm": 0.0002825877454597503, | |
| "learning_rate": 8.223157894736842e-05, | |
| "loss": 0.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.007292327203551046, | |
| "grad_norm": 0.0002043453569058329, | |
| "learning_rate": 8.170105263157894e-05, | |
| "loss": 0.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.007450856055802156, | |
| "grad_norm": 0.0002539000706747174, | |
| "learning_rate": 8.117052631578946e-05, | |
| "loss": 0.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.007609384908053266, | |
| "grad_norm": 0.00020019305520690978, | |
| "learning_rate": 8.064e-05, | |
| "loss": 0.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0077679137603043754, | |
| "grad_norm": 0.00020320792100392282, | |
| "learning_rate": 8.010947368421052e-05, | |
| "loss": 0.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.007926442612555484, | |
| "grad_norm": 0.00017714353452902287, | |
| "learning_rate": 7.957894736842105e-05, | |
| "loss": 0.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.007926442612555484, | |
| "eval_loss": 0.0037753561045974493, | |
| "eval_runtime": 285.8239, | |
| "eval_samples_per_second": 9.292, | |
| "eval_steps_per_second": 2.323, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.008084971464806594, | |
| "grad_norm": 11.202030181884766, | |
| "learning_rate": 7.904842105263158e-05, | |
| "loss": 0.3948, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.008243500317057704, | |
| "grad_norm": 4.6145090891513973e-05, | |
| "learning_rate": 7.85178947368421e-05, | |
| "loss": 0.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.008402029169308814, | |
| "grad_norm": 5.335342575563118e-05, | |
| "learning_rate": 7.798736842105263e-05, | |
| "loss": 0.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.008560558021559923, | |
| "grad_norm": 6.71695961500518e-05, | |
| "learning_rate": 7.745684210526315e-05, | |
| "loss": 0.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.008719086873811033, | |
| "grad_norm": 7.083545642672107e-05, | |
| "learning_rate": 7.692631578947369e-05, | |
| "loss": 0.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.008877615726062143, | |
| "grad_norm": 0.00011363301746314391, | |
| "learning_rate": 7.639578947368421e-05, | |
| "loss": 0.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.009036144578313253, | |
| "grad_norm": 0.00015090873057488352, | |
| "learning_rate": 7.586526315789473e-05, | |
| "loss": 0.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.009194673430564362, | |
| "grad_norm": 0.00017162153380922973, | |
| "learning_rate": 7.533473684210526e-05, | |
| "loss": 0.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.009353202282815472, | |
| "grad_norm": 0.00037030354724265635, | |
| "learning_rate": 7.480421052631578e-05, | |
| "loss": 0.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.009511731135066582, | |
| "grad_norm": 0.00045603603939525783, | |
| "learning_rate": 7.427368421052632e-05, | |
| "loss": 0.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.009670259987317692, | |
| "grad_norm": 0.000606843619607389, | |
| "learning_rate": 7.374315789473685e-05, | |
| "loss": 0.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.009828788839568801, | |
| "grad_norm": 0.0012902193702757359, | |
| "learning_rate": 7.321263157894737e-05, | |
| "loss": 0.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.009987317691819911, | |
| "grad_norm": 0.0015680071664974093, | |
| "learning_rate": 7.26821052631579e-05, | |
| "loss": 0.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.010145846544071021, | |
| "grad_norm": 0.0017102680867537856, | |
| "learning_rate": 7.215157894736842e-05, | |
| "loss": 0.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.01030437539632213, | |
| "grad_norm": 0.001034354092553258, | |
| "learning_rate": 7.162105263157894e-05, | |
| "loss": 0.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.01046290424857324, | |
| "grad_norm": 0.0008957475074566901, | |
| "learning_rate": 7.109052631578947e-05, | |
| "loss": 0.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.01062143310082435, | |
| "grad_norm": 0.000811917707324028, | |
| "learning_rate": 7.055999999999999e-05, | |
| "loss": 0.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.01077996195307546, | |
| "grad_norm": 0.0010698516853153706, | |
| "learning_rate": 7.002947368421052e-05, | |
| "loss": 0.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.01093849080532657, | |
| "grad_norm": 0.0007929237326607108, | |
| "learning_rate": 6.949894736842105e-05, | |
| "loss": 0.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.01109701965757768, | |
| "grad_norm": 0.0005297662573866546, | |
| "learning_rate": 6.896842105263158e-05, | |
| "loss": 0.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.01125554850982879, | |
| "grad_norm": 0.0004970860900357366, | |
| "learning_rate": 6.843789473684211e-05, | |
| "loss": 0.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.011414077362079899, | |
| "grad_norm": 0.00042307560215704143, | |
| "learning_rate": 6.790736842105263e-05, | |
| "loss": 0.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.011572606214331009, | |
| "grad_norm": 0.0003402529109735042, | |
| "learning_rate": 6.737684210526315e-05, | |
| "loss": 0.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.011731135066582118, | |
| "grad_norm": 0.00037735735531896353, | |
| "learning_rate": 6.684631578947368e-05, | |
| "loss": 0.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.011889663918833228, | |
| "grad_norm": 0.0004949852591380477, | |
| "learning_rate": 6.631578947368421e-05, | |
| "loss": 0.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.012048192771084338, | |
| "grad_norm": 0.00039665098302066326, | |
| "learning_rate": 6.578526315789473e-05, | |
| "loss": 0.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.012206721623335448, | |
| "grad_norm": 0.0003759284154511988, | |
| "learning_rate": 6.525473684210526e-05, | |
| "loss": 0.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.012365250475586557, | |
| "grad_norm": 0.0007349163061007857, | |
| "learning_rate": 6.47242105263158e-05, | |
| "loss": 0.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.012523779327837667, | |
| "grad_norm": 0.00038908098940737545, | |
| "learning_rate": 6.419368421052632e-05, | |
| "loss": 0.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.012682308180088777, | |
| "grad_norm": 0.00039207786903716624, | |
| "learning_rate": 6.366315789473684e-05, | |
| "loss": 0.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.012840837032339885, | |
| "grad_norm": 0.00031621192465536296, | |
| "learning_rate": 6.313263157894736e-05, | |
| "loss": 0.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.012999365884590995, | |
| "grad_norm": 0.001207337947562337, | |
| "learning_rate": 6.26021052631579e-05, | |
| "loss": 0.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.013157894736842105, | |
| "grad_norm": 0.0003833776863757521, | |
| "learning_rate": 6.207157894736842e-05, | |
| "loss": 0.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.013316423589093214, | |
| "grad_norm": 0.0002807167184073478, | |
| "learning_rate": 6.154105263157894e-05, | |
| "loss": 0.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.013474952441344324, | |
| "grad_norm": 0.00025957514299079776, | |
| "learning_rate": 6.1010526315789474e-05, | |
| "loss": 0.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.013633481293595434, | |
| "grad_norm": 0.00022323857410810888, | |
| "learning_rate": 6.048e-05, | |
| "loss": 0.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.013792010145846544, | |
| "grad_norm": 0.00020459384541027248, | |
| "learning_rate": 5.994947368421052e-05, | |
| "loss": 0.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.013950538998097653, | |
| "grad_norm": 0.00021896703401580453, | |
| "learning_rate": 5.941894736842104e-05, | |
| "loss": 0.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.014109067850348763, | |
| "grad_norm": 0.0003069478552788496, | |
| "learning_rate": 5.888842105263158e-05, | |
| "loss": 0.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.014267596702599873, | |
| "grad_norm": 0.0003965249052271247, | |
| "learning_rate": 5.835789473684211e-05, | |
| "loss": 0.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.014426125554850983, | |
| "grad_norm": 0.00017867452697828412, | |
| "learning_rate": 5.782736842105263e-05, | |
| "loss": 0.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.014584654407102092, | |
| "grad_norm": 0.00016691711789462715, | |
| "learning_rate": 5.7296842105263154e-05, | |
| "loss": 0.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.014743183259353202, | |
| "grad_norm": 0.0001635922526475042, | |
| "learning_rate": 5.676631578947368e-05, | |
| "loss": 0.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.014901712111604312, | |
| "grad_norm": 0.00017515213403385133, | |
| "learning_rate": 5.623578947368421e-05, | |
| "loss": 0.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.015060240963855422, | |
| "grad_norm": 0.00022856853320263326, | |
| "learning_rate": 5.570526315789474e-05, | |
| "loss": 0.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.015218769816106531, | |
| "grad_norm": 0.00013485149247571826, | |
| "learning_rate": 5.5174736842105266e-05, | |
| "loss": 0.0, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.015377298668357641, | |
| "grad_norm": 0.00022567079577129334, | |
| "learning_rate": 5.464421052631579e-05, | |
| "loss": 0.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.015535827520608751, | |
| "grad_norm": 0.00031975016463547945, | |
| "learning_rate": 5.411368421052631e-05, | |
| "loss": 0.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.01569435637285986, | |
| "grad_norm": 0.00018277288472745568, | |
| "learning_rate": 5.358315789473684e-05, | |
| "loss": 0.0, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.01585288522511097, | |
| "grad_norm": 0.00018042400188278407, | |
| "learning_rate": 5.3052631578947364e-05, | |
| "loss": 0.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01585288522511097, | |
| "eval_loss": 0.0034066014923155308, | |
| "eval_runtime": 285.7064, | |
| "eval_samples_per_second": 9.296, | |
| "eval_steps_per_second": 2.324, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01601141407736208, | |
| "grad_norm": 0.0001420114713255316, | |
| "learning_rate": 5.252210526315789e-05, | |
| "loss": 0.0, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.016169942929613188, | |
| "grad_norm": 0.00012088767834939063, | |
| "learning_rate": 5.199157894736842e-05, | |
| "loss": 0.0, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.0163284717818643, | |
| "grad_norm": 0.00011178933345945552, | |
| "learning_rate": 5.1461052631578946e-05, | |
| "loss": 0.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.016487000634115408, | |
| "grad_norm": 0.00010531868610996753, | |
| "learning_rate": 5.0930526315789476e-05, | |
| "loss": 0.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.01664552948636652, | |
| "grad_norm": 0.00011901999096153304, | |
| "learning_rate": 5.04e-05, | |
| "loss": 0.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.016804058338617627, | |
| "grad_norm": 9.147950186161324e-05, | |
| "learning_rate": 4.986947368421052e-05, | |
| "loss": 0.0, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.01696258719086874, | |
| "grad_norm": 7.989244477357715e-05, | |
| "learning_rate": 4.933894736842105e-05, | |
| "loss": 0.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.017121116043119847, | |
| "grad_norm": 7.46723817428574e-05, | |
| "learning_rate": 4.880842105263158e-05, | |
| "loss": 0.0, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.017279644895370958, | |
| "grad_norm": 8.380914368899539e-05, | |
| "learning_rate": 4.8277894736842103e-05, | |
| "loss": 0.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.017438173747622066, | |
| "grad_norm": 7.211839692899957e-05, | |
| "learning_rate": 4.7747368421052626e-05, | |
| "loss": 0.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.017596702599873178, | |
| "grad_norm": 7.725647446932271e-05, | |
| "learning_rate": 4.7216842105263156e-05, | |
| "loss": 0.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.017755231452124286, | |
| "grad_norm": 8.656168211018667e-05, | |
| "learning_rate": 4.6686315789473686e-05, | |
| "loss": 0.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.017913760304375397, | |
| "grad_norm": 7.190388714661822e-05, | |
| "learning_rate": 4.615578947368421e-05, | |
| "loss": 0.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.018072289156626505, | |
| "grad_norm": 7.352729880949482e-05, | |
| "learning_rate": 4.562526315789474e-05, | |
| "loss": 0.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.018230818008877617, | |
| "grad_norm": 6.0772359574912116e-05, | |
| "learning_rate": 4.509473684210526e-05, | |
| "loss": 0.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.018389346861128725, | |
| "grad_norm": 5.866353239980526e-05, | |
| "learning_rate": 4.4564210526315784e-05, | |
| "loss": 0.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.018547875713379836, | |
| "grad_norm": 6.619851046707481e-05, | |
| "learning_rate": 4.403368421052632e-05, | |
| "loss": 0.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.018706404565630944, | |
| "grad_norm": 6.62534948787652e-05, | |
| "learning_rate": 4.350315789473684e-05, | |
| "loss": 0.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.018864933417882056, | |
| "grad_norm": 6.304805719992146e-05, | |
| "learning_rate": 4.2972631578947366e-05, | |
| "loss": 0.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.019023462270133164, | |
| "grad_norm": 6.0336358728818595e-05, | |
| "learning_rate": 4.244210526315789e-05, | |
| "loss": 0.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.019181991122384275, | |
| "grad_norm": 5.7237852161051705e-05, | |
| "learning_rate": 4.191157894736842e-05, | |
| "loss": 0.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.019340519974635383, | |
| "grad_norm": 5.5000280553940684e-05, | |
| "learning_rate": 4.138105263157895e-05, | |
| "loss": 0.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.019499048826886495, | |
| "grad_norm": 6.333758938126266e-05, | |
| "learning_rate": 4.085052631578947e-05, | |
| "loss": 0.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.019657577679137603, | |
| "grad_norm": 6.331897748168558e-05, | |
| "learning_rate": 4.032e-05, | |
| "loss": 0.0, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.019816106531388714, | |
| "grad_norm": 5.774655437562615e-05, | |
| "learning_rate": 3.978947368421052e-05, | |
| "loss": 0.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.019974635383639822, | |
| "grad_norm": 5.671928738593124e-05, | |
| "learning_rate": 3.925894736842105e-05, | |
| "loss": 0.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.020133164235890934, | |
| "grad_norm": 6.89297157805413e-05, | |
| "learning_rate": 3.8728421052631575e-05, | |
| "loss": 0.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.020291693088142042, | |
| "grad_norm": 6.829660560470074e-05, | |
| "learning_rate": 3.8197894736842105e-05, | |
| "loss": 0.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.02045022194039315, | |
| "grad_norm": 5.680310641764663e-05, | |
| "learning_rate": 3.766736842105263e-05, | |
| "loss": 0.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.02060875079264426, | |
| "grad_norm": 4.718761192634702e-05, | |
| "learning_rate": 3.713684210526316e-05, | |
| "loss": 0.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.02076727964489537, | |
| "grad_norm": 5.037297523813322e-05, | |
| "learning_rate": 3.660631578947369e-05, | |
| "loss": 0.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.02092580849714648, | |
| "grad_norm": 5.18505803484004e-05, | |
| "learning_rate": 3.607578947368421e-05, | |
| "loss": 0.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.02108433734939759, | |
| "grad_norm": 5.0042519433191046e-05, | |
| "learning_rate": 3.554526315789473e-05, | |
| "loss": 0.0, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.0212428662016487, | |
| "grad_norm": 5.112058715894818e-05, | |
| "learning_rate": 3.501473684210526e-05, | |
| "loss": 0.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.02140139505389981, | |
| "grad_norm": 5.2210583817213774e-05, | |
| "learning_rate": 3.448421052631579e-05, | |
| "loss": 0.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.02155992390615092, | |
| "grad_norm": 5.472183329402469e-05, | |
| "learning_rate": 3.3953684210526315e-05, | |
| "loss": 0.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.021718452758402028, | |
| "grad_norm": 5.2417293773032725e-05, | |
| "learning_rate": 3.342315789473684e-05, | |
| "loss": 0.0, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.02187698161065314, | |
| "grad_norm": 4.7179757530102506e-05, | |
| "learning_rate": 3.289263157894737e-05, | |
| "loss": 0.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.022035510462904247, | |
| "grad_norm": 5.0374997954349965e-05, | |
| "learning_rate": 3.23621052631579e-05, | |
| "loss": 0.0, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.02219403931515536, | |
| "grad_norm": 5.104386946186423e-05, | |
| "learning_rate": 3.183157894736842e-05, | |
| "loss": 0.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.022352568167406467, | |
| "grad_norm": 4.212348358123563e-05, | |
| "learning_rate": 3.130105263157895e-05, | |
| "loss": 0.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.02251109701965758, | |
| "grad_norm": 6.773701898055151e-05, | |
| "learning_rate": 3.077052631578947e-05, | |
| "loss": 0.0, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.022669625871908686, | |
| "grad_norm": 4.5040869736112654e-05, | |
| "learning_rate": 3.024e-05, | |
| "loss": 0.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.022828154724159798, | |
| "grad_norm": 4.242918294039555e-05, | |
| "learning_rate": 2.970947368421052e-05, | |
| "loss": 0.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.022986683576410906, | |
| "grad_norm": 7.274608651641756e-05, | |
| "learning_rate": 2.9178947368421054e-05, | |
| "loss": 0.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.023145212428662017, | |
| "grad_norm": 7.292564259842038e-05, | |
| "learning_rate": 2.8648421052631577e-05, | |
| "loss": 0.0, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.023303741280913125, | |
| "grad_norm": 8.248071389971301e-05, | |
| "learning_rate": 2.8117894736842103e-05, | |
| "loss": 0.0, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.023462270133164237, | |
| "grad_norm": 7.596130308229476e-05, | |
| "learning_rate": 2.7587368421052633e-05, | |
| "loss": 0.0, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.023620798985415345, | |
| "grad_norm": 8.240217721322551e-05, | |
| "learning_rate": 2.7056842105263156e-05, | |
| "loss": 0.0, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.023779327837666456, | |
| "grad_norm": 0.0001047488913172856, | |
| "learning_rate": 2.6526315789473682e-05, | |
| "loss": 0.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.023779327837666456, | |
| "eval_loss": 0.003376348875463009, | |
| "eval_runtime": 285.9072, | |
| "eval_samples_per_second": 9.29, | |
| "eval_steps_per_second": 2.322, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.023937856689917564, | |
| "grad_norm": 5.650425434112549, | |
| "learning_rate": 2.599578947368421e-05, | |
| "loss": 0.1783, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.024096385542168676, | |
| "grad_norm": 0.00012569170212373137, | |
| "learning_rate": 2.5465263157894738e-05, | |
| "loss": 0.0, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.024254914394419784, | |
| "grad_norm": 0.00018095099949277937, | |
| "learning_rate": 2.493473684210526e-05, | |
| "loss": 0.0, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.024413443246670895, | |
| "grad_norm": 0.0001806170621421188, | |
| "learning_rate": 2.440421052631579e-05, | |
| "loss": 0.0, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.024571972098922003, | |
| "grad_norm": 0.00021117427968420088, | |
| "learning_rate": 2.3873684210526313e-05, | |
| "loss": 0.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.024730500951173115, | |
| "grad_norm": 0.00030483287991955876, | |
| "learning_rate": 2.3343157894736843e-05, | |
| "loss": 0.0, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.024889029803424223, | |
| "grad_norm": 0.00029481080127879977, | |
| "learning_rate": 2.281263157894737e-05, | |
| "loss": 0.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.025047558655675334, | |
| "grad_norm": 0.0004884671652689576, | |
| "learning_rate": 2.2282105263157892e-05, | |
| "loss": 0.0, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.025206087507926443, | |
| "grad_norm": 0.0002942118444480002, | |
| "learning_rate": 2.175157894736842e-05, | |
| "loss": 0.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.025364616360177554, | |
| "grad_norm": 0.000287515576928854, | |
| "learning_rate": 2.1221052631578944e-05, | |
| "loss": 0.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.025523145212428662, | |
| "grad_norm": 0.00042692935676313937, | |
| "learning_rate": 2.0690526315789474e-05, | |
| "loss": 0.0, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.02568167406467977, | |
| "grad_norm": 0.0005435315542854369, | |
| "learning_rate": 2.016e-05, | |
| "loss": 0.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.02584020291693088, | |
| "grad_norm": 0.00035297736758366227, | |
| "learning_rate": 1.9629473684210526e-05, | |
| "loss": 0.0, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.02599873176918199, | |
| "grad_norm": 0.0005012881010770798, | |
| "learning_rate": 1.9098947368421053e-05, | |
| "loss": 0.0, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.0261572606214331, | |
| "grad_norm": 0.00044648078619502485, | |
| "learning_rate": 1.856842105263158e-05, | |
| "loss": 0.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.02631578947368421, | |
| "grad_norm": 0.0004886375973001122, | |
| "learning_rate": 1.8037894736842105e-05, | |
| "loss": 0.0, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.02647431832593532, | |
| "grad_norm": 0.0005168926436454058, | |
| "learning_rate": 1.750736842105263e-05, | |
| "loss": 0.0, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.02663284717818643, | |
| "grad_norm": 0.0004425595107022673, | |
| "learning_rate": 1.6976842105263157e-05, | |
| "loss": 0.0, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.02679137603043754, | |
| "grad_norm": 0.0005173716926947236, | |
| "learning_rate": 1.6446315789473684e-05, | |
| "loss": 0.0, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.026949904882688648, | |
| "grad_norm": 0.00048417490324936807, | |
| "learning_rate": 1.591578947368421e-05, | |
| "loss": 0.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.02710843373493976, | |
| "grad_norm": 0.0006713734474033117, | |
| "learning_rate": 1.5385263157894736e-05, | |
| "loss": 0.0, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.027266962587190868, | |
| "grad_norm": 0.0004995199851691723, | |
| "learning_rate": 1.485473684210526e-05, | |
| "loss": 0.0, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.02742549143944198, | |
| "grad_norm": 0.0004266517935320735, | |
| "learning_rate": 1.4324210526315789e-05, | |
| "loss": 0.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.027584020291693087, | |
| "grad_norm": 0.0004872908757533878, | |
| "learning_rate": 1.3793684210526316e-05, | |
| "loss": 0.0, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0277425491439442, | |
| "grad_norm": 0.00042201511678285897, | |
| "learning_rate": 1.3263157894736841e-05, | |
| "loss": 0.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.027901077996195307, | |
| "grad_norm": 0.0004310183576308191, | |
| "learning_rate": 1.2732631578947369e-05, | |
| "loss": 0.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.028059606848446418, | |
| "grad_norm": 0.00046641906374134123, | |
| "learning_rate": 1.2202105263157895e-05, | |
| "loss": 0.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.028218135700697526, | |
| "grad_norm": 0.0004791380779352039, | |
| "learning_rate": 1.1671578947368421e-05, | |
| "loss": 0.0, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.028376664552948638, | |
| "grad_norm": 0.000661531463265419, | |
| "learning_rate": 1.1141052631578946e-05, | |
| "loss": 0.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.028535193405199746, | |
| "grad_norm": 0.0005801509832963347, | |
| "learning_rate": 1.0610526315789472e-05, | |
| "loss": 0.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.028693722257450857, | |
| "grad_norm": 0.00038415539893321693, | |
| "learning_rate": 1.008e-05, | |
| "loss": 0.0, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.028852251109701965, | |
| "grad_norm": 0.00040587177500128746, | |
| "learning_rate": 9.549473684210526e-06, | |
| "loss": 0.0, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.029010779961953077, | |
| "grad_norm": 0.00045143821625970304, | |
| "learning_rate": 9.018947368421052e-06, | |
| "loss": 0.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.029169308814204185, | |
| "grad_norm": 0.0005063486169092357, | |
| "learning_rate": 8.488421052631579e-06, | |
| "loss": 0.0, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.029327837666455296, | |
| "grad_norm": 0.0005101272254250944, | |
| "learning_rate": 7.957894736842105e-06, | |
| "loss": 0.0, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.029486366518706404, | |
| "grad_norm": 0.00046848724014125764, | |
| "learning_rate": 7.42736842105263e-06, | |
| "loss": 0.0, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.029644895370957516, | |
| "grad_norm": 0.000389144872315228, | |
| "learning_rate": 6.896842105263158e-06, | |
| "loss": 0.0, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.029803424223208624, | |
| "grad_norm": 0.00045255443546921015, | |
| "learning_rate": 6.3663157894736845e-06, | |
| "loss": 0.0, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.029961953075459735, | |
| "grad_norm": 0.00038248361670412123, | |
| "learning_rate": 5.835789473684211e-06, | |
| "loss": 0.0, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.030120481927710843, | |
| "grad_norm": 0.00043523870408535004, | |
| "learning_rate": 5.305263157894736e-06, | |
| "loss": 0.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.030279010779961955, | |
| "grad_norm": 0.00042310234857723117, | |
| "learning_rate": 4.774736842105263e-06, | |
| "loss": 0.0, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.030437539632213063, | |
| "grad_norm": 0.00040010226075537503, | |
| "learning_rate": 4.244210526315789e-06, | |
| "loss": 0.0, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.030596068484464174, | |
| "grad_norm": 0.0003844479797407985, | |
| "learning_rate": 3.713684210526315e-06, | |
| "loss": 0.0, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.030754597336715282, | |
| "grad_norm": 0.0004438844043761492, | |
| "learning_rate": 3.1831578947368422e-06, | |
| "loss": 0.0, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.03091312618896639, | |
| "grad_norm": 0.0004899434861727059, | |
| "learning_rate": 2.652631578947368e-06, | |
| "loss": 0.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.031071655041217502, | |
| "grad_norm": 0.0006046644994057715, | |
| "learning_rate": 2.1221052631578947e-06, | |
| "loss": 0.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.03123018389346861, | |
| "grad_norm": 0.0007566219428554177, | |
| "learning_rate": 1.5915789473684211e-06, | |
| "loss": 0.0, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.03138871274571972, | |
| "grad_norm": 0.0006015272228978574, | |
| "learning_rate": 1.0610526315789473e-06, | |
| "loss": 0.0, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.03154724159797083, | |
| "grad_norm": 0.0004844815412070602, | |
| "learning_rate": 5.305263157894737e-07, | |
| "loss": 0.0, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.03170577045022194, | |
| "grad_norm": 0.0004923184169456363, | |
| "learning_rate": 0.0, | |
| "loss": 0.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.03170577045022194, | |
| "eval_loss": 0.0022250961046665907, | |
| "eval_runtime": 285.0318, | |
| "eval_samples_per_second": 9.318, | |
| "eval_steps_per_second": 2.33, | |
| "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": 7.162764252413952e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |