| { | |
| "best_metric": 0.0005893517518416047, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 0.10787486515641856, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0005393743257820927, | |
| "grad_norm": 0.9549857378005981, | |
| "learning_rate": 1.001e-05, | |
| "loss": 0.2799, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0005393743257820927, | |
| "eval_loss": 0.4979684054851532, | |
| "eval_runtime": 45.1708, | |
| "eval_samples_per_second": 17.29, | |
| "eval_steps_per_second": 4.339, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0010787486515641855, | |
| "grad_norm": 0.6719993948936462, | |
| "learning_rate": 2.002e-05, | |
| "loss": 0.1665, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0016181229773462784, | |
| "grad_norm": 0.5584933757781982, | |
| "learning_rate": 3.0029999999999995e-05, | |
| "loss": 0.1785, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.002157497303128371, | |
| "grad_norm": 0.6560541391372681, | |
| "learning_rate": 4.004e-05, | |
| "loss": 0.1566, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.002696871628910464, | |
| "grad_norm": 0.8838739991188049, | |
| "learning_rate": 5.005e-05, | |
| "loss": 0.1356, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.003236245954692557, | |
| "grad_norm": 0.6256913542747498, | |
| "learning_rate": 6.005999999999999e-05, | |
| "loss": 0.0931, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0037756202804746495, | |
| "grad_norm": 0.6155531406402588, | |
| "learning_rate": 7.006999999999998e-05, | |
| "loss": 0.0667, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.004314994606256742, | |
| "grad_norm": 0.584822416305542, | |
| "learning_rate": 8.008e-05, | |
| "loss": 0.0352, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0048543689320388345, | |
| "grad_norm": 1.1850945949554443, | |
| "learning_rate": 9.009e-05, | |
| "loss": 0.0139, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.005393743257820928, | |
| "grad_norm": 0.36779069900512695, | |
| "learning_rate": 0.0001001, | |
| "loss": 0.0119, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.005933117583603021, | |
| "grad_norm": 1.6826813220977783, | |
| "learning_rate": 9.957315789473684e-05, | |
| "loss": 0.0115, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.006472491909385114, | |
| "grad_norm": 0.3343207538127899, | |
| "learning_rate": 9.904631578947367e-05, | |
| "loss": 0.0034, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.007011866235167206, | |
| "grad_norm": 0.03047369234263897, | |
| "learning_rate": 9.851947368421052e-05, | |
| "loss": 0.0005, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.007551240560949299, | |
| "grad_norm": 0.04409059137105942, | |
| "learning_rate": 9.799263157894736e-05, | |
| "loss": 0.0003, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.008090614886731391, | |
| "grad_norm": 3.0516834259033203, | |
| "learning_rate": 9.746578947368421e-05, | |
| "loss": 0.0131, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.008629989212513484, | |
| "grad_norm": 0.3956446945667267, | |
| "learning_rate": 9.693894736842104e-05, | |
| "loss": 0.001, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.009169363538295576, | |
| "grad_norm": 1.5148561000823975, | |
| "learning_rate": 9.641210526315789e-05, | |
| "loss": 0.0061, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.009708737864077669, | |
| "grad_norm": 0.49977800250053406, | |
| "learning_rate": 9.588526315789473e-05, | |
| "loss": 0.0017, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.010248112189859764, | |
| "grad_norm": 0.012113417498767376, | |
| "learning_rate": 9.535842105263157e-05, | |
| "loss": 0.0001, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.010787486515641856, | |
| "grad_norm": 0.008402293547987938, | |
| "learning_rate": 9.483157894736841e-05, | |
| "loss": 0.0002, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.011326860841423949, | |
| "grad_norm": 0.002001185668632388, | |
| "learning_rate": 9.430473684210526e-05, | |
| "loss": 0.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.011866235167206042, | |
| "grad_norm": 0.5545786619186401, | |
| "learning_rate": 9.37778947368421e-05, | |
| "loss": 0.0068, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.012405609492988134, | |
| "grad_norm": 0.06889030337333679, | |
| "learning_rate": 9.325105263157894e-05, | |
| "loss": 0.0004, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.012944983818770227, | |
| "grad_norm": 0.15830865502357483, | |
| "learning_rate": 9.272421052631578e-05, | |
| "loss": 0.0011, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01348435814455232, | |
| "grad_norm": 0.7862672805786133, | |
| "learning_rate": 9.219736842105263e-05, | |
| "loss": 0.0063, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.014023732470334413, | |
| "grad_norm": 0.009982357732951641, | |
| "learning_rate": 9.167052631578946e-05, | |
| "loss": 0.0001, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.014563106796116505, | |
| "grad_norm": 0.009793877601623535, | |
| "learning_rate": 9.114368421052632e-05, | |
| "loss": 0.0001, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.015102481121898598, | |
| "grad_norm": 0.006467557977885008, | |
| "learning_rate": 9.061684210526315e-05, | |
| "loss": 0.0001, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.01564185544768069, | |
| "grad_norm": 0.012111790478229523, | |
| "learning_rate": 9.009e-05, | |
| "loss": 0.0001, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.016181229773462782, | |
| "grad_norm": 0.06000563129782677, | |
| "learning_rate": 8.956315789473683e-05, | |
| "loss": 0.0007, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.016720604099244876, | |
| "grad_norm": 0.03677884489297867, | |
| "learning_rate": 8.903631578947368e-05, | |
| "loss": 0.0007, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.017259978425026967, | |
| "grad_norm": 0.03280596062541008, | |
| "learning_rate": 8.850947368421052e-05, | |
| "loss": 0.0003, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.01779935275080906, | |
| "grad_norm": 0.002885522786527872, | |
| "learning_rate": 8.798263157894736e-05, | |
| "loss": 0.0001, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.018338727076591153, | |
| "grad_norm": 0.4575079679489136, | |
| "learning_rate": 8.745578947368422e-05, | |
| "loss": 0.0156, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.018878101402373247, | |
| "grad_norm": 0.0030563841573894024, | |
| "learning_rate": 8.692894736842105e-05, | |
| "loss": 0.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.019417475728155338, | |
| "grad_norm": 0.0014249957166612148, | |
| "learning_rate": 8.64021052631579e-05, | |
| "loss": 0.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.019956850053937433, | |
| "grad_norm": 0.0230252668261528, | |
| "learning_rate": 8.587526315789473e-05, | |
| "loss": 0.0001, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.020496224379719527, | |
| "grad_norm": 0.005644981749355793, | |
| "learning_rate": 8.534842105263157e-05, | |
| "loss": 0.0001, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.021035598705501618, | |
| "grad_norm": 0.02447853423655033, | |
| "learning_rate": 8.482157894736842e-05, | |
| "loss": 0.0001, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.021574973031283712, | |
| "grad_norm": 0.004401453770697117, | |
| "learning_rate": 8.429473684210525e-05, | |
| "loss": 0.0001, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.022114347357065803, | |
| "grad_norm": 0.05064772441983223, | |
| "learning_rate": 8.376789473684211e-05, | |
| "loss": 0.0004, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.022653721682847898, | |
| "grad_norm": 1.4466474056243896, | |
| "learning_rate": 8.324105263157894e-05, | |
| "loss": 0.0453, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.02319309600862999, | |
| "grad_norm": 0.0030545571353286505, | |
| "learning_rate": 8.271421052631579e-05, | |
| "loss": 0.0001, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.023732470334412083, | |
| "grad_norm": 0.45695409178733826, | |
| "learning_rate": 8.218736842105262e-05, | |
| "loss": 0.0142, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.024271844660194174, | |
| "grad_norm": 0.45072630047798157, | |
| "learning_rate": 8.166052631578947e-05, | |
| "loss": 0.0007, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02481121898597627, | |
| "grad_norm": 0.20009317994117737, | |
| "learning_rate": 8.113368421052631e-05, | |
| "loss": 0.0005, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02535059331175836, | |
| "grad_norm": 0.01844087988138199, | |
| "learning_rate": 8.060684210526315e-05, | |
| "loss": 0.0001, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.025889967637540454, | |
| "grad_norm": 0.009140203706920147, | |
| "learning_rate": 8.008e-05, | |
| "loss": 0.0002, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.026429341963322545, | |
| "grad_norm": 1.665168046951294, | |
| "learning_rate": 7.955315789473684e-05, | |
| "loss": 0.0267, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.02696871628910464, | |
| "grad_norm": 1.6538095474243164, | |
| "learning_rate": 7.902631578947368e-05, | |
| "loss": 0.0311, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02696871628910464, | |
| "eval_loss": 0.05669600889086723, | |
| "eval_runtime": 43.6492, | |
| "eval_samples_per_second": 17.893, | |
| "eval_steps_per_second": 4.49, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02750809061488673, | |
| "grad_norm": 2.739877223968506, | |
| "learning_rate": 7.849947368421052e-05, | |
| "loss": 0.0604, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.028047464940668825, | |
| "grad_norm": 0.010129177011549473, | |
| "learning_rate": 7.797263157894736e-05, | |
| "loss": 0.0001, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.028586839266450916, | |
| "grad_norm": 0.03367248922586441, | |
| "learning_rate": 7.744578947368421e-05, | |
| "loss": 0.0002, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.02912621359223301, | |
| "grad_norm": 1.0684995651245117, | |
| "learning_rate": 7.691894736842104e-05, | |
| "loss": 0.0011, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0296655879180151, | |
| "grad_norm": 0.006163384765386581, | |
| "learning_rate": 7.63921052631579e-05, | |
| "loss": 0.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.030204962243797196, | |
| "grad_norm": 0.004382474347949028, | |
| "learning_rate": 7.586526315789473e-05, | |
| "loss": 0.0001, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.030744336569579287, | |
| "grad_norm": 0.006588236894458532, | |
| "learning_rate": 7.533842105263158e-05, | |
| "loss": 0.0001, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.03128371089536138, | |
| "grad_norm": 0.019274592399597168, | |
| "learning_rate": 7.481157894736841e-05, | |
| "loss": 0.0001, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03182308522114347, | |
| "grad_norm": 0.006948168855160475, | |
| "learning_rate": 7.428473684210526e-05, | |
| "loss": 0.0001, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.032362459546925564, | |
| "grad_norm": 0.001927727716974914, | |
| "learning_rate": 7.375789473684209e-05, | |
| "loss": 0.0001, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03290183387270766, | |
| "grad_norm": 0.05395427718758583, | |
| "learning_rate": 7.323105263157895e-05, | |
| "loss": 0.0004, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.03344120819848975, | |
| "grad_norm": 0.14390406012535095, | |
| "learning_rate": 7.270421052631578e-05, | |
| "loss": 0.0006, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03398058252427184, | |
| "grad_norm": 0.04436345770955086, | |
| "learning_rate": 7.217736842105263e-05, | |
| "loss": 0.0004, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.034519956850053934, | |
| "grad_norm": 0.005591034423559904, | |
| "learning_rate": 7.165052631578947e-05, | |
| "loss": 0.0001, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03505933117583603, | |
| "grad_norm": 1.5156598091125488, | |
| "learning_rate": 7.11236842105263e-05, | |
| "loss": 0.0033, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.03559870550161812, | |
| "grad_norm": 0.20944921672344208, | |
| "learning_rate": 7.059684210526315e-05, | |
| "loss": 0.0015, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.036138079827400214, | |
| "grad_norm": 0.02203325368463993, | |
| "learning_rate": 7.006999999999998e-05, | |
| "loss": 0.0002, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.036677454153182305, | |
| "grad_norm": 0.012391135096549988, | |
| "learning_rate": 6.954315789473684e-05, | |
| "loss": 0.0001, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0372168284789644, | |
| "grad_norm": 0.003814356168732047, | |
| "learning_rate": 6.901631578947368e-05, | |
| "loss": 0.0001, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.037756202804746494, | |
| "grad_norm": 0.43400076031684875, | |
| "learning_rate": 6.848947368421052e-05, | |
| "loss": 0.0108, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.038295577130528585, | |
| "grad_norm": 0.001942815724760294, | |
| "learning_rate": 6.796263157894737e-05, | |
| "loss": 0.0001, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.038834951456310676, | |
| "grad_norm": 0.007045631296932697, | |
| "learning_rate": 6.74357894736842e-05, | |
| "loss": 0.0001, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.039374325782092774, | |
| "grad_norm": 0.0019868926610797644, | |
| "learning_rate": 6.690894736842105e-05, | |
| "loss": 0.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.039913700107874865, | |
| "grad_norm": 0.3928805887699127, | |
| "learning_rate": 6.638210526315788e-05, | |
| "loss": 0.001, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.040453074433656956, | |
| "grad_norm": 0.00091419683303684, | |
| "learning_rate": 6.585526315789474e-05, | |
| "loss": 0.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.040992448759439054, | |
| "grad_norm": 0.00108022999484092, | |
| "learning_rate": 6.532842105263157e-05, | |
| "loss": 0.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.041531823085221145, | |
| "grad_norm": 0.004014655947685242, | |
| "learning_rate": 6.480157894736842e-05, | |
| "loss": 0.0001, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.042071197411003236, | |
| "grad_norm": 0.0030816688667982817, | |
| "learning_rate": 6.427473684210526e-05, | |
| "loss": 0.0001, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.04261057173678533, | |
| "grad_norm": 0.0016585150733590126, | |
| "learning_rate": 6.37478947368421e-05, | |
| "loss": 0.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.043149946062567425, | |
| "grad_norm": 0.00667704688385129, | |
| "learning_rate": 6.322105263157894e-05, | |
| "loss": 0.0001, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.043689320388349516, | |
| "grad_norm": 0.007074718829244375, | |
| "learning_rate": 6.269421052631577e-05, | |
| "loss": 0.0001, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.04422869471413161, | |
| "grad_norm": 0.002394700888544321, | |
| "learning_rate": 6.216736842105263e-05, | |
| "loss": 0.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0447680690399137, | |
| "grad_norm": 0.0016268681501969695, | |
| "learning_rate": 6.164052631578947e-05, | |
| "loss": 0.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.045307443365695796, | |
| "grad_norm": 0.00599514739587903, | |
| "learning_rate": 6.111368421052631e-05, | |
| "loss": 0.0001, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.04584681769147789, | |
| "grad_norm": 0.0022963311057537794, | |
| "learning_rate": 6.058684210526315e-05, | |
| "loss": 0.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.04638619201725998, | |
| "grad_norm": 0.007355900481343269, | |
| "learning_rate": 6.005999999999999e-05, | |
| "loss": 0.0001, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.04692556634304207, | |
| "grad_norm": 0.02201763354241848, | |
| "learning_rate": 5.953315789473684e-05, | |
| "loss": 0.0002, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.04746494066882417, | |
| "grad_norm": 0.005624725949019194, | |
| "learning_rate": 5.9006315789473676e-05, | |
| "loss": 0.0001, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.04800431499460626, | |
| "grad_norm": 0.04811863973736763, | |
| "learning_rate": 5.847947368421053e-05, | |
| "loss": 0.0002, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.04854368932038835, | |
| "grad_norm": 0.0029467041604220867, | |
| "learning_rate": 5.795263157894737e-05, | |
| "loss": 0.0001, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.04908306364617044, | |
| "grad_norm": 0.03302125632762909, | |
| "learning_rate": 5.742578947368421e-05, | |
| "loss": 0.0001, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.04962243797195254, | |
| "grad_norm": 0.0017132150242105126, | |
| "learning_rate": 5.6898947368421046e-05, | |
| "loss": 0.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.05016181229773463, | |
| "grad_norm": 0.0011448146542534232, | |
| "learning_rate": 5.6372105263157886e-05, | |
| "loss": 0.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.05070118662351672, | |
| "grad_norm": 0.0019763971213251352, | |
| "learning_rate": 5.584526315789473e-05, | |
| "loss": 0.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.05124056094929881, | |
| "grad_norm": 0.0009319214150309563, | |
| "learning_rate": 5.531842105263158e-05, | |
| "loss": 0.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.05177993527508091, | |
| "grad_norm": 0.03375813364982605, | |
| "learning_rate": 5.4791578947368424e-05, | |
| "loss": 0.0005, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.052319309600863, | |
| "grad_norm": 0.16154968738555908, | |
| "learning_rate": 5.426473684210526e-05, | |
| "loss": 0.0025, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.05285868392664509, | |
| "grad_norm": 0.0034327851608395576, | |
| "learning_rate": 5.37378947368421e-05, | |
| "loss": 0.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.05339805825242718, | |
| "grad_norm": 0.0067018670961260796, | |
| "learning_rate": 5.321105263157894e-05, | |
| "loss": 0.0001, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.05393743257820928, | |
| "grad_norm": 0.21445177495479584, | |
| "learning_rate": 5.268421052631578e-05, | |
| "loss": 0.0017, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05393743257820928, | |
| "eval_loss": 0.0010083840461447835, | |
| "eval_runtime": 43.6657, | |
| "eval_samples_per_second": 17.886, | |
| "eval_steps_per_second": 4.489, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05447680690399137, | |
| "grad_norm": 0.459598571062088, | |
| "learning_rate": 5.2157368421052626e-05, | |
| "loss": 0.0081, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.05501618122977346, | |
| "grad_norm": 0.0010549342259764671, | |
| "learning_rate": 5.163052631578947e-05, | |
| "loss": 0.0, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.05555555555555555, | |
| "grad_norm": 0.002258673310279846, | |
| "learning_rate": 5.110368421052632e-05, | |
| "loss": 0.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.05609492988133765, | |
| "grad_norm": 0.0001912364095915109, | |
| "learning_rate": 5.057684210526316e-05, | |
| "loss": 0.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.05663430420711974, | |
| "grad_norm": 0.0003946495126001537, | |
| "learning_rate": 5.005e-05, | |
| "loss": 0.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.05717367853290183, | |
| "grad_norm": 0.0004527179116848856, | |
| "learning_rate": 4.9523157894736836e-05, | |
| "loss": 0.0, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.05771305285868392, | |
| "grad_norm": 0.0007224463624879718, | |
| "learning_rate": 4.899631578947368e-05, | |
| "loss": 0.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.05825242718446602, | |
| "grad_norm": 0.0027744832914322615, | |
| "learning_rate": 4.846947368421052e-05, | |
| "loss": 0.0001, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.05879180151024811, | |
| "grad_norm": 0.0003102615592069924, | |
| "learning_rate": 4.794263157894737e-05, | |
| "loss": 0.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.0593311758360302, | |
| "grad_norm": 0.00021459658455569297, | |
| "learning_rate": 4.7415789473684206e-05, | |
| "loss": 0.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.059870550161812294, | |
| "grad_norm": 0.00045238586608320475, | |
| "learning_rate": 4.688894736842105e-05, | |
| "loss": 0.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.06040992448759439, | |
| "grad_norm": 0.00048040415276773274, | |
| "learning_rate": 4.636210526315789e-05, | |
| "loss": 0.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.06094929881337648, | |
| "grad_norm": 0.00028796499827876687, | |
| "learning_rate": 4.583526315789473e-05, | |
| "loss": 0.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.061488673139158574, | |
| "grad_norm": 0.0009007256594486535, | |
| "learning_rate": 4.530842105263158e-05, | |
| "loss": 0.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.06202804746494067, | |
| "grad_norm": 0.001204671454615891, | |
| "learning_rate": 4.4781578947368416e-05, | |
| "loss": 0.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06256742179072276, | |
| "grad_norm": 0.007920761592686176, | |
| "learning_rate": 4.425473684210526e-05, | |
| "loss": 0.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.06310679611650485, | |
| "grad_norm": 0.000548014766536653, | |
| "learning_rate": 4.372789473684211e-05, | |
| "loss": 0.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06364617044228695, | |
| "grad_norm": 0.0007022125646471977, | |
| "learning_rate": 4.320105263157895e-05, | |
| "loss": 0.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.06418554476806904, | |
| "grad_norm": 0.000792205857578665, | |
| "learning_rate": 4.2674210526315786e-05, | |
| "loss": 0.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.06472491909385113, | |
| "grad_norm": 0.0005556220421567559, | |
| "learning_rate": 4.2147368421052626e-05, | |
| "loss": 0.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06526429341963323, | |
| "grad_norm": 0.0013536950573325157, | |
| "learning_rate": 4.162052631578947e-05, | |
| "loss": 0.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.06580366774541532, | |
| "grad_norm": 0.001718906918540597, | |
| "learning_rate": 4.109368421052631e-05, | |
| "loss": 0.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.06634304207119741, | |
| "grad_norm": 0.016501938924193382, | |
| "learning_rate": 4.056684210526316e-05, | |
| "loss": 0.0001, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0668824163969795, | |
| "grad_norm": 0.2795630395412445, | |
| "learning_rate": 4.004e-05, | |
| "loss": 0.0005, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0674217907227616, | |
| "grad_norm": 0.035502709448337555, | |
| "learning_rate": 3.951315789473684e-05, | |
| "loss": 0.0001, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.06796116504854369, | |
| "grad_norm": 0.002328946255147457, | |
| "learning_rate": 3.898631578947368e-05, | |
| "loss": 0.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.06850053937432578, | |
| "grad_norm": 0.0006106572109274566, | |
| "learning_rate": 3.845947368421052e-05, | |
| "loss": 0.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.06903991370010787, | |
| "grad_norm": 0.0013282396830618382, | |
| "learning_rate": 3.7932631578947367e-05, | |
| "loss": 0.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.06957928802588997, | |
| "grad_norm": 0.41567087173461914, | |
| "learning_rate": 3.7405789473684206e-05, | |
| "loss": 0.0035, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.07011866235167206, | |
| "grad_norm": 0.0007774877594783902, | |
| "learning_rate": 3.6878947368421045e-05, | |
| "loss": 0.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07065803667745416, | |
| "grad_norm": 0.05387866869568825, | |
| "learning_rate": 3.635210526315789e-05, | |
| "loss": 0.0006, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.07119741100323625, | |
| "grad_norm": 0.0006187845719978213, | |
| "learning_rate": 3.582526315789474e-05, | |
| "loss": 0.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.07173678532901834, | |
| "grad_norm": 0.3249104917049408, | |
| "learning_rate": 3.5298421052631576e-05, | |
| "loss": 0.0058, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.07227615965480043, | |
| "grad_norm": 0.0009034304530359805, | |
| "learning_rate": 3.477157894736842e-05, | |
| "loss": 0.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.07281553398058252, | |
| "grad_norm": 0.0008392046438530087, | |
| "learning_rate": 3.424473684210526e-05, | |
| "loss": 0.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07335490830636461, | |
| "grad_norm": 0.001984312431886792, | |
| "learning_rate": 3.37178947368421e-05, | |
| "loss": 0.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.07389428263214672, | |
| "grad_norm": 0.05136461928486824, | |
| "learning_rate": 3.319105263157894e-05, | |
| "loss": 0.0003, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.0744336569579288, | |
| "grad_norm": 0.0010661619016900659, | |
| "learning_rate": 3.2664210526315786e-05, | |
| "loss": 0.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.0749730312837109, | |
| "grad_norm": 0.007086303550750017, | |
| "learning_rate": 3.213736842105263e-05, | |
| "loss": 0.0001, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.07551240560949299, | |
| "grad_norm": 0.0012311493046581745, | |
| "learning_rate": 3.161052631578947e-05, | |
| "loss": 0.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07605177993527508, | |
| "grad_norm": 0.0016970309661701322, | |
| "learning_rate": 3.108368421052632e-05, | |
| "loss": 0.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.07659115426105717, | |
| "grad_norm": 0.6121507883071899, | |
| "learning_rate": 3.0556842105263156e-05, | |
| "loss": 0.0016, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.07713052858683926, | |
| "grad_norm": 0.0011852330062538385, | |
| "learning_rate": 3.0029999999999995e-05, | |
| "loss": 0.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.07766990291262135, | |
| "grad_norm": 0.0008150951471179724, | |
| "learning_rate": 2.9503157894736838e-05, | |
| "loss": 0.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.07820927723840346, | |
| "grad_norm": 0.0009472601232118905, | |
| "learning_rate": 2.8976315789473684e-05, | |
| "loss": 0.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.07874865156418555, | |
| "grad_norm": 0.003503723070025444, | |
| "learning_rate": 2.8449473684210523e-05, | |
| "loss": 0.0001, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.07928802588996764, | |
| "grad_norm": 0.0022005646023899317, | |
| "learning_rate": 2.7922631578947366e-05, | |
| "loss": 0.0001, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.07982740021574973, | |
| "grad_norm": 0.5109080672264099, | |
| "learning_rate": 2.7395789473684212e-05, | |
| "loss": 0.0101, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.08036677454153182, | |
| "grad_norm": 0.0015276124468073249, | |
| "learning_rate": 2.686894736842105e-05, | |
| "loss": 0.0001, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.08090614886731391, | |
| "grad_norm": 0.0025751348584890366, | |
| "learning_rate": 2.634210526315789e-05, | |
| "loss": 0.0001, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08090614886731391, | |
| "eval_loss": 0.0005893517518416047, | |
| "eval_runtime": 43.6869, | |
| "eval_samples_per_second": 17.877, | |
| "eval_steps_per_second": 4.486, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.081445523193096, | |
| "grad_norm": 0.0056745377369225025, | |
| "learning_rate": 2.5815263157894736e-05, | |
| "loss": 0.0, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.08198489751887811, | |
| "grad_norm": 0.00784875638782978, | |
| "learning_rate": 2.528842105263158e-05, | |
| "loss": 0.0001, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.0825242718446602, | |
| "grad_norm": 0.06929171085357666, | |
| "learning_rate": 2.4761578947368418e-05, | |
| "loss": 0.0015, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.08306364617044229, | |
| "grad_norm": 0.0055921077728271484, | |
| "learning_rate": 2.423473684210526e-05, | |
| "loss": 0.0001, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.08360302049622438, | |
| "grad_norm": 0.002456638030707836, | |
| "learning_rate": 2.3707894736842103e-05, | |
| "loss": 0.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.08414239482200647, | |
| "grad_norm": 0.016230836510658264, | |
| "learning_rate": 2.3181052631578946e-05, | |
| "loss": 0.0002, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.08468176914778856, | |
| "grad_norm": 0.0014073759084567428, | |
| "learning_rate": 2.265421052631579e-05, | |
| "loss": 0.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.08522114347357065, | |
| "grad_norm": 0.18957898020744324, | |
| "learning_rate": 2.212736842105263e-05, | |
| "loss": 0.002, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.08576051779935275, | |
| "grad_norm": 0.001173275290057063, | |
| "learning_rate": 2.1600526315789474e-05, | |
| "loss": 0.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.08629989212513485, | |
| "grad_norm": 0.0007326776976697147, | |
| "learning_rate": 2.1073684210526313e-05, | |
| "loss": 0.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.08683926645091694, | |
| "grad_norm": 0.008999370969831944, | |
| "learning_rate": 2.0546842105263155e-05, | |
| "loss": 0.0001, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.08737864077669903, | |
| "grad_norm": 0.0007513080490753055, | |
| "learning_rate": 2.002e-05, | |
| "loss": 0.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.08791801510248112, | |
| "grad_norm": 0.0003164597728755325, | |
| "learning_rate": 1.949315789473684e-05, | |
| "loss": 0.0, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.08845738942826321, | |
| "grad_norm": 0.0006100442842580378, | |
| "learning_rate": 1.8966315789473683e-05, | |
| "loss": 0.0, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.0889967637540453, | |
| "grad_norm": 0.002358382800593972, | |
| "learning_rate": 1.8439473684210522e-05, | |
| "loss": 0.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.0895361380798274, | |
| "grad_norm": 0.0008702076738700271, | |
| "learning_rate": 1.791263157894737e-05, | |
| "loss": 0.0, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.09007551240560949, | |
| "grad_norm": 0.0013868837850168347, | |
| "learning_rate": 1.738578947368421e-05, | |
| "loss": 0.0, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.09061488673139159, | |
| "grad_norm": 0.0006682085804641247, | |
| "learning_rate": 1.685894736842105e-05, | |
| "loss": 0.0, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.09115426105717368, | |
| "grad_norm": 0.0009037026320584118, | |
| "learning_rate": 1.6332105263157893e-05, | |
| "loss": 0.0, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.09169363538295577, | |
| "grad_norm": 0.0003754556237254292, | |
| "learning_rate": 1.5805263157894735e-05, | |
| "loss": 0.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09223300970873786, | |
| "grad_norm": 0.0007883550715632737, | |
| "learning_rate": 1.5278421052631578e-05, | |
| "loss": 0.0, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.09277238403451996, | |
| "grad_norm": 0.00029830558924004436, | |
| "learning_rate": 1.4751578947368419e-05, | |
| "loss": 0.0, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.09331175836030205, | |
| "grad_norm": 0.00025384570471942425, | |
| "learning_rate": 1.4224736842105262e-05, | |
| "loss": 0.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.09385113268608414, | |
| "grad_norm": 0.000792273145634681, | |
| "learning_rate": 1.3697894736842106e-05, | |
| "loss": 0.0, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.09439050701186623, | |
| "grad_norm": 0.00023794101434759796, | |
| "learning_rate": 1.3171052631578945e-05, | |
| "loss": 0.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09492988133764833, | |
| "grad_norm": 0.0003109975659754127, | |
| "learning_rate": 1.264421052631579e-05, | |
| "loss": 0.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.09546925566343042, | |
| "grad_norm": 0.0019688678439706564, | |
| "learning_rate": 1.211736842105263e-05, | |
| "loss": 0.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.09600862998921252, | |
| "grad_norm": 0.00809526164084673, | |
| "learning_rate": 1.1590526315789473e-05, | |
| "loss": 0.0001, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.0965480043149946, | |
| "grad_norm": 0.0008922089473344386, | |
| "learning_rate": 1.1063684210526316e-05, | |
| "loss": 0.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.0970873786407767, | |
| "grad_norm": 0.0005603457102552056, | |
| "learning_rate": 1.0536842105263156e-05, | |
| "loss": 0.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.09762675296655879, | |
| "grad_norm": 0.001602335018105805, | |
| "learning_rate": 1.001e-05, | |
| "loss": 0.0, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.09816612729234088, | |
| "grad_norm": 0.000997183844447136, | |
| "learning_rate": 9.483157894736842e-06, | |
| "loss": 0.0, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.09870550161812297, | |
| "grad_norm": 0.0009953331900760531, | |
| "learning_rate": 8.956315789473684e-06, | |
| "loss": 0.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.09924487594390508, | |
| "grad_norm": 0.0015611740527674556, | |
| "learning_rate": 8.429473684210525e-06, | |
| "loss": 0.0, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.09978425026968717, | |
| "grad_norm": 0.08797088265419006, | |
| "learning_rate": 7.902631578947368e-06, | |
| "loss": 0.0011, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.10032362459546926, | |
| "grad_norm": 0.0015272749587893486, | |
| "learning_rate": 7.3757894736842095e-06, | |
| "loss": 0.0, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.10086299892125135, | |
| "grad_norm": 0.007411181926727295, | |
| "learning_rate": 6.848947368421053e-06, | |
| "loss": 0.0001, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.10140237324703344, | |
| "grad_norm": 0.001111446414142847, | |
| "learning_rate": 6.322105263157895e-06, | |
| "loss": 0.0, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.10194174757281553, | |
| "grad_norm": 0.0008587180054746568, | |
| "learning_rate": 5.7952631578947365e-06, | |
| "loss": 0.0, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.10248112189859762, | |
| "grad_norm": 0.008806944824755192, | |
| "learning_rate": 5.268421052631578e-06, | |
| "loss": 0.0001, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.10302049622437973, | |
| "grad_norm": 0.059028804302215576, | |
| "learning_rate": 4.741578947368421e-06, | |
| "loss": 0.0008, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.10355987055016182, | |
| "grad_norm": 0.0017502488335594535, | |
| "learning_rate": 4.2147368421052626e-06, | |
| "loss": 0.0, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.10409924487594391, | |
| "grad_norm": 0.0008180232835002244, | |
| "learning_rate": 3.6878947368421047e-06, | |
| "loss": 0.0, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.104638619201726, | |
| "grad_norm": 0.0005253468989394605, | |
| "learning_rate": 3.1610526315789474e-06, | |
| "loss": 0.0, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10517799352750809, | |
| "grad_norm": 0.007774029858410358, | |
| "learning_rate": 2.634210526315789e-06, | |
| "loss": 0.0001, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.10571736785329018, | |
| "grad_norm": 0.001689290627837181, | |
| "learning_rate": 2.1073684210526313e-06, | |
| "loss": 0.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.10625674217907227, | |
| "grad_norm": 0.005365308839827776, | |
| "learning_rate": 1.5805263157894737e-06, | |
| "loss": 0.0001, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.10679611650485436, | |
| "grad_norm": 0.0035137736704200506, | |
| "learning_rate": 1.0536842105263156e-06, | |
| "loss": 0.0, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.10733549083063647, | |
| "grad_norm": 0.0472775362432003, | |
| "learning_rate": 5.268421052631578e-07, | |
| "loss": 0.0005, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.10787486515641856, | |
| "grad_norm": 0.029273375868797302, | |
| "learning_rate": 0.0, | |
| "loss": 0.0003, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.10787486515641856, | |
| "eval_loss": 0.000882925814948976, | |
| "eval_runtime": 43.7697, | |
| "eval_samples_per_second": 17.843, | |
| "eval_steps_per_second": 4.478, | |
| "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": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 7.580280733197926e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |