| { | |
| "best_metric": 0.635071337223053, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 0.10085728693898134, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010085728693898135, | |
| "grad_norm": 1.311654806137085, | |
| "learning_rate": 1.0018000000000001e-05, | |
| "loss": 0.9276, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0010085728693898135, | |
| "eval_loss": 1.104080319404602, | |
| "eval_runtime": 47.2809, | |
| "eval_samples_per_second": 8.841, | |
| "eval_steps_per_second": 2.221, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.002017145738779627, | |
| "grad_norm": 1.4768002033233643, | |
| "learning_rate": 2.0036000000000003e-05, | |
| "loss": 1.1293, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0030257186081694403, | |
| "grad_norm": 1.863478422164917, | |
| "learning_rate": 3.0054e-05, | |
| "loss": 1.095, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.004034291477559254, | |
| "grad_norm": 1.4832903146743774, | |
| "learning_rate": 4.0072000000000005e-05, | |
| "loss": 1.0675, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.005042864346949067, | |
| "grad_norm": 1.516528606414795, | |
| "learning_rate": 5.009e-05, | |
| "loss": 1.1571, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.006051437216338881, | |
| "grad_norm": 1.0894883871078491, | |
| "learning_rate": 6.0108e-05, | |
| "loss": 0.8794, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0070600100857286935, | |
| "grad_norm": 1.1469849348068237, | |
| "learning_rate": 7.0126e-05, | |
| "loss": 0.7944, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.008068582955118508, | |
| "grad_norm": 0.9738489389419556, | |
| "learning_rate": 8.014400000000001e-05, | |
| "loss": 0.7899, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.009077155824508321, | |
| "grad_norm": 1.1924914121627808, | |
| "learning_rate": 9.016200000000001e-05, | |
| "loss": 0.8915, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.010085728693898134, | |
| "grad_norm": 0.986112117767334, | |
| "learning_rate": 0.00010018, | |
| "loss": 0.735, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.011094301563287948, | |
| "grad_norm": 0.9808599948883057, | |
| "learning_rate": 9.965273684210526e-05, | |
| "loss": 0.7869, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.012102874432677761, | |
| "grad_norm": 0.8853185176849365, | |
| "learning_rate": 9.912547368421053e-05, | |
| "loss": 0.7262, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.013111447302067574, | |
| "grad_norm": 0.7452360987663269, | |
| "learning_rate": 9.859821052631579e-05, | |
| "loss": 0.679, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.014120020171457387, | |
| "grad_norm": 0.7864087224006653, | |
| "learning_rate": 9.807094736842106e-05, | |
| "loss": 0.6494, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.015128593040847202, | |
| "grad_norm": 0.7922771573066711, | |
| "learning_rate": 9.754368421052633e-05, | |
| "loss": 0.6502, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.016137165910237016, | |
| "grad_norm": 0.7772244215011597, | |
| "learning_rate": 9.701642105263158e-05, | |
| "loss": 0.5793, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.01714573877962683, | |
| "grad_norm": 0.7786775827407837, | |
| "learning_rate": 9.648915789473685e-05, | |
| "loss": 0.5646, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.018154311649016642, | |
| "grad_norm": 0.9989523887634277, | |
| "learning_rate": 9.596189473684211e-05, | |
| "loss": 0.8072, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.019162884518406455, | |
| "grad_norm": 0.8197798728942871, | |
| "learning_rate": 9.543463157894737e-05, | |
| "loss": 0.549, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.020171457387796268, | |
| "grad_norm": 0.727142333984375, | |
| "learning_rate": 9.490736842105264e-05, | |
| "loss": 0.6405, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02118003025718608, | |
| "grad_norm": 0.7259921431541443, | |
| "learning_rate": 9.43801052631579e-05, | |
| "loss": 0.6329, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.022188603126575897, | |
| "grad_norm": 0.7431148886680603, | |
| "learning_rate": 9.385284210526316e-05, | |
| "loss": 0.522, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.02319717599596571, | |
| "grad_norm": 0.8169230818748474, | |
| "learning_rate": 9.332557894736843e-05, | |
| "loss": 0.7403, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.024205748865355523, | |
| "grad_norm": 0.9084967970848083, | |
| "learning_rate": 9.279831578947369e-05, | |
| "loss": 0.6538, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.025214321734745335, | |
| "grad_norm": 0.8007498383522034, | |
| "learning_rate": 9.227105263157896e-05, | |
| "loss": 0.6519, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.026222894604135148, | |
| "grad_norm": 0.7854844927787781, | |
| "learning_rate": 9.174378947368421e-05, | |
| "loss": 0.6472, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.02723146747352496, | |
| "grad_norm": 0.8193365335464478, | |
| "learning_rate": 9.121652631578948e-05, | |
| "loss": 0.5204, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.028240040342914774, | |
| "grad_norm": 0.8219320178031921, | |
| "learning_rate": 9.068926315789475e-05, | |
| "loss": 0.6833, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.02924861321230459, | |
| "grad_norm": 0.8295514583587646, | |
| "learning_rate": 9.016200000000001e-05, | |
| "loss": 0.6379, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.030257186081694403, | |
| "grad_norm": 0.9452921748161316, | |
| "learning_rate": 8.963473684210526e-05, | |
| "loss": 0.7919, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.031265758951084216, | |
| "grad_norm": 0.893158495426178, | |
| "learning_rate": 8.910747368421053e-05, | |
| "loss": 0.8002, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.03227433182047403, | |
| "grad_norm": 0.7392338514328003, | |
| "learning_rate": 8.858021052631579e-05, | |
| "loss": 0.6503, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.03328290468986384, | |
| "grad_norm": 0.8364588618278503, | |
| "learning_rate": 8.805294736842106e-05, | |
| "loss": 0.7389, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.03429147755925366, | |
| "grad_norm": 0.8333988189697266, | |
| "learning_rate": 8.752568421052633e-05, | |
| "loss": 0.718, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.03530005042864347, | |
| "grad_norm": 0.8555812835693359, | |
| "learning_rate": 8.699842105263159e-05, | |
| "loss": 0.8115, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.036308623298033284, | |
| "grad_norm": 0.909781277179718, | |
| "learning_rate": 8.647115789473686e-05, | |
| "loss": 0.7834, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.03731719616742309, | |
| "grad_norm": 0.8471423387527466, | |
| "learning_rate": 8.594389473684211e-05, | |
| "loss": 0.7363, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.03832576903681291, | |
| "grad_norm": 0.7747591137886047, | |
| "learning_rate": 8.541663157894737e-05, | |
| "loss": 0.7223, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.039334341906202726, | |
| "grad_norm": 0.7503566145896912, | |
| "learning_rate": 8.488936842105264e-05, | |
| "loss": 0.6464, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.040342914775592535, | |
| "grad_norm": 0.7485514283180237, | |
| "learning_rate": 8.43621052631579e-05, | |
| "loss": 0.624, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04135148764498235, | |
| "grad_norm": 0.8754268884658813, | |
| "learning_rate": 8.383484210526316e-05, | |
| "loss": 0.7319, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.04236006051437216, | |
| "grad_norm": 0.9374569058418274, | |
| "learning_rate": 8.330757894736843e-05, | |
| "loss": 0.8252, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.04336863338376198, | |
| "grad_norm": 1.0073909759521484, | |
| "learning_rate": 8.278031578947369e-05, | |
| "loss": 0.8029, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.044377206253151794, | |
| "grad_norm": 0.9302268624305725, | |
| "learning_rate": 8.225305263157896e-05, | |
| "loss": 0.7655, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0453857791225416, | |
| "grad_norm": 0.9419411420822144, | |
| "learning_rate": 8.172578947368422e-05, | |
| "loss": 0.7218, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.04639435199193142, | |
| "grad_norm": 0.936607301235199, | |
| "learning_rate": 8.119852631578947e-05, | |
| "loss": 0.7549, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.04740292486132123, | |
| "grad_norm": 0.8890344500541687, | |
| "learning_rate": 8.067126315789474e-05, | |
| "loss": 0.7809, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.048411497730711045, | |
| "grad_norm": 1.0612900257110596, | |
| "learning_rate": 8.014400000000001e-05, | |
| "loss": 0.7987, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.049420070600100854, | |
| "grad_norm": 0.9946021437644958, | |
| "learning_rate": 7.961673684210527e-05, | |
| "loss": 0.8006, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.05042864346949067, | |
| "grad_norm": 1.1615980863571167, | |
| "learning_rate": 7.908947368421054e-05, | |
| "loss": 0.8517, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05042864346949067, | |
| "eval_loss": 0.6663289666175842, | |
| "eval_runtime": 47.2665, | |
| "eval_samples_per_second": 8.843, | |
| "eval_steps_per_second": 2.221, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05143721633888049, | |
| "grad_norm": 0.8349344730377197, | |
| "learning_rate": 7.856221052631579e-05, | |
| "loss": 0.4587, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.052445789208270296, | |
| "grad_norm": 0.6953268647193909, | |
| "learning_rate": 7.803494736842106e-05, | |
| "loss": 0.666, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.05345436207766011, | |
| "grad_norm": 0.7224613428115845, | |
| "learning_rate": 7.750768421052632e-05, | |
| "loss": 0.7536, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.05446293494704992, | |
| "grad_norm": 0.6757810115814209, | |
| "learning_rate": 7.698042105263157e-05, | |
| "loss": 0.6169, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.05547150781643974, | |
| "grad_norm": 0.7635074257850647, | |
| "learning_rate": 7.645315789473686e-05, | |
| "loss": 0.6307, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.05648008068582955, | |
| "grad_norm": 0.6975710988044739, | |
| "learning_rate": 7.592589473684211e-05, | |
| "loss": 0.6233, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.057488653555219364, | |
| "grad_norm": 0.9252186417579651, | |
| "learning_rate": 7.539863157894737e-05, | |
| "loss": 0.6551, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.05849722642460918, | |
| "grad_norm": 0.8248884677886963, | |
| "learning_rate": 7.487136842105264e-05, | |
| "loss": 0.5789, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.05950579929399899, | |
| "grad_norm": 0.6074532866477966, | |
| "learning_rate": 7.43441052631579e-05, | |
| "loss": 0.5004, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.060514372163388806, | |
| "grad_norm": 0.6752936244010925, | |
| "learning_rate": 7.381684210526315e-05, | |
| "loss": 0.4884, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.061522945032778616, | |
| "grad_norm": 0.7720556259155273, | |
| "learning_rate": 7.328957894736844e-05, | |
| "loss": 0.7758, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.06253151790216843, | |
| "grad_norm": 0.699116051197052, | |
| "learning_rate": 7.276231578947369e-05, | |
| "loss": 0.5716, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.06354009077155824, | |
| "grad_norm": 0.7762227058410645, | |
| "learning_rate": 7.223505263157895e-05, | |
| "loss": 0.6935, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.06454866364094806, | |
| "grad_norm": 0.6913344264030457, | |
| "learning_rate": 7.170778947368422e-05, | |
| "loss": 0.6068, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.06555723651033787, | |
| "grad_norm": 0.7144531011581421, | |
| "learning_rate": 7.118052631578947e-05, | |
| "loss": 0.6431, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.06656580937972768, | |
| "grad_norm": 0.6667282581329346, | |
| "learning_rate": 7.065326315789474e-05, | |
| "loss": 0.5421, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.06757438224911749, | |
| "grad_norm": 0.6969847083091736, | |
| "learning_rate": 7.0126e-05, | |
| "loss": 0.6785, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.06858295511850732, | |
| "grad_norm": 0.6973515748977661, | |
| "learning_rate": 6.959873684210527e-05, | |
| "loss": 0.5455, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.06959152798789713, | |
| "grad_norm": 0.7713499069213867, | |
| "learning_rate": 6.907147368421054e-05, | |
| "loss": 0.7372, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.07060010085728693, | |
| "grad_norm": 0.5957585573196411, | |
| "learning_rate": 6.85442105263158e-05, | |
| "loss": 0.4954, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.07160867372667676, | |
| "grad_norm": 0.7025538086891174, | |
| "learning_rate": 6.801694736842105e-05, | |
| "loss": 0.5496, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.07261724659606657, | |
| "grad_norm": 0.7217448353767395, | |
| "learning_rate": 6.748968421052632e-05, | |
| "loss": 0.5567, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.07362581946545638, | |
| "grad_norm": 0.7021524906158447, | |
| "learning_rate": 6.696242105263158e-05, | |
| "loss": 0.5291, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.07463439233484619, | |
| "grad_norm": 0.7426285743713379, | |
| "learning_rate": 6.643515789473685e-05, | |
| "loss": 0.6815, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.07564296520423601, | |
| "grad_norm": 0.7391223311424255, | |
| "learning_rate": 6.590789473684212e-05, | |
| "loss": 0.6689, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07665153807362582, | |
| "grad_norm": 0.6848605871200562, | |
| "learning_rate": 6.538063157894737e-05, | |
| "loss": 0.5782, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.07766011094301563, | |
| "grad_norm": 0.711951732635498, | |
| "learning_rate": 6.485336842105264e-05, | |
| "loss": 0.5801, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.07866868381240545, | |
| "grad_norm": 0.5852221846580505, | |
| "learning_rate": 6.43261052631579e-05, | |
| "loss": 0.4995, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.07967725668179526, | |
| "grad_norm": 0.9040377736091614, | |
| "learning_rate": 6.379884210526315e-05, | |
| "loss": 0.5318, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.08068582955118507, | |
| "grad_norm": 0.756641149520874, | |
| "learning_rate": 6.327157894736842e-05, | |
| "loss": 0.6616, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.08169440242057488, | |
| "grad_norm": 0.7158970832824707, | |
| "learning_rate": 6.274431578947368e-05, | |
| "loss": 0.6113, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0827029752899647, | |
| "grad_norm": 0.6931173205375671, | |
| "learning_rate": 6.221705263157895e-05, | |
| "loss": 0.5882, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.08371154815935451, | |
| "grad_norm": 0.7512569427490234, | |
| "learning_rate": 6.168978947368422e-05, | |
| "loss": 0.6451, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.08472012102874432, | |
| "grad_norm": 0.7793223261833191, | |
| "learning_rate": 6.116252631578948e-05, | |
| "loss": 0.6186, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.08572869389813415, | |
| "grad_norm": 0.7280353307723999, | |
| "learning_rate": 6.063526315789474e-05, | |
| "loss": 0.6657, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.08673726676752395, | |
| "grad_norm": 0.7213184833526611, | |
| "learning_rate": 6.0108e-05, | |
| "loss": 0.6312, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.08774583963691376, | |
| "grad_norm": 0.7809893488883972, | |
| "learning_rate": 5.9580736842105264e-05, | |
| "loss": 0.6928, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.08875441250630359, | |
| "grad_norm": 0.8184586763381958, | |
| "learning_rate": 5.905347368421053e-05, | |
| "loss": 0.6551, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0897629853756934, | |
| "grad_norm": 0.8529835939407349, | |
| "learning_rate": 5.85262105263158e-05, | |
| "loss": 0.7481, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0907715582450832, | |
| "grad_norm": 0.8931265473365784, | |
| "learning_rate": 5.799894736842106e-05, | |
| "loss": 0.7928, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.09178013111447302, | |
| "grad_norm": 0.7958138585090637, | |
| "learning_rate": 5.747168421052632e-05, | |
| "loss": 0.6151, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.09278870398386284, | |
| "grad_norm": 0.7931579351425171, | |
| "learning_rate": 5.694442105263158e-05, | |
| "loss": 0.7141, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.09379727685325265, | |
| "grad_norm": 0.8281200528144836, | |
| "learning_rate": 5.641715789473684e-05, | |
| "loss": 0.6538, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.09480584972264246, | |
| "grad_norm": 0.8042008280754089, | |
| "learning_rate": 5.5889894736842104e-05, | |
| "loss": 0.6661, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.09581442259203228, | |
| "grad_norm": 0.8040047287940979, | |
| "learning_rate": 5.5362631578947374e-05, | |
| "loss": 0.5659, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.09682299546142209, | |
| "grad_norm": 1.044537901878357, | |
| "learning_rate": 5.483536842105264e-05, | |
| "loss": 0.835, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0978315683308119, | |
| "grad_norm": 0.9413779377937317, | |
| "learning_rate": 5.43081052631579e-05, | |
| "loss": 0.7605, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.09884014120020171, | |
| "grad_norm": 0.9656868577003479, | |
| "learning_rate": 5.378084210526316e-05, | |
| "loss": 0.7144, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.09984871406959153, | |
| "grad_norm": 1.0694280862808228, | |
| "learning_rate": 5.3253578947368426e-05, | |
| "loss": 0.8615, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.10085728693898134, | |
| "grad_norm": 1.2606571912765503, | |
| "learning_rate": 5.272631578947368e-05, | |
| "loss": 0.8549, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.10085728693898134, | |
| "eval_loss": 0.635071337223053, | |
| "eval_runtime": 47.0624, | |
| "eval_samples_per_second": 8.882, | |
| "eval_steps_per_second": 2.231, | |
| "step": 100 | |
| } | |
| ], | |
| "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": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8.648580773235917e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |