| { | |
| "best_metric": 1.5667701959609985, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 0.9404388714733543, | |
| "eval_steps": 50, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.006269592476489028, | |
| "grad_norm": 0.1956695169210434, | |
| "learning_rate": 1.0100000000000002e-05, | |
| "loss": 1.4797, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.006269592476489028, | |
| "eval_loss": 2.014873743057251, | |
| "eval_runtime": 21.7647, | |
| "eval_samples_per_second": 98.738, | |
| "eval_steps_per_second": 3.124, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.012539184952978056, | |
| "grad_norm": 0.21951091289520264, | |
| "learning_rate": 2.0200000000000003e-05, | |
| "loss": 1.8897, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.018808777429467086, | |
| "grad_norm": 0.2182653844356537, | |
| "learning_rate": 3.0299999999999998e-05, | |
| "loss": 1.8755, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.025078369905956112, | |
| "grad_norm": 0.3972433805465698, | |
| "learning_rate": 4.0400000000000006e-05, | |
| "loss": 2.0184, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03134796238244514, | |
| "grad_norm": 0.7481153607368469, | |
| "learning_rate": 5.05e-05, | |
| "loss": 2.1242, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03761755485893417, | |
| "grad_norm": 0.6163584589958191, | |
| "learning_rate": 6.0599999999999996e-05, | |
| "loss": 2.254, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0438871473354232, | |
| "grad_norm": 0.17153511941432953, | |
| "learning_rate": 7.07e-05, | |
| "loss": 1.4166, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.050156739811912224, | |
| "grad_norm": 0.16584382951259613, | |
| "learning_rate": 8.080000000000001e-05, | |
| "loss": 1.7842, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.05642633228840126, | |
| "grad_norm": 0.18644414842128754, | |
| "learning_rate": 9.09e-05, | |
| "loss": 1.8403, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.06269592476489028, | |
| "grad_norm": 0.21256040036678314, | |
| "learning_rate": 0.000101, | |
| "loss": 1.9229, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06896551724137931, | |
| "grad_norm": 0.452114462852478, | |
| "learning_rate": 0.00010046842105263158, | |
| "loss": 1.9909, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.07523510971786834, | |
| "grad_norm": 0.3007873296737671, | |
| "learning_rate": 9.993684210526315e-05, | |
| "loss": 1.995, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.08150470219435736, | |
| "grad_norm": 0.22591519355773926, | |
| "learning_rate": 9.940526315789473e-05, | |
| "loss": 1.3466, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0877742946708464, | |
| "grad_norm": 0.22962772846221924, | |
| "learning_rate": 9.887368421052632e-05, | |
| "loss": 1.5942, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.09404388714733543, | |
| "grad_norm": 0.20066531002521515, | |
| "learning_rate": 9.83421052631579e-05, | |
| "loss": 1.759, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.10031347962382445, | |
| "grad_norm": 0.24357616901397705, | |
| "learning_rate": 9.781052631578948e-05, | |
| "loss": 1.7682, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.10658307210031348, | |
| "grad_norm": 0.42487001419067383, | |
| "learning_rate": 9.727894736842106e-05, | |
| "loss": 1.8977, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.11285266457680251, | |
| "grad_norm": 0.5519868731498718, | |
| "learning_rate": 9.674736842105263e-05, | |
| "loss": 1.8652, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.11912225705329153, | |
| "grad_norm": 0.1837894469499588, | |
| "learning_rate": 9.621578947368421e-05, | |
| "loss": 1.2621, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.12539184952978055, | |
| "grad_norm": 0.16041411459445953, | |
| "learning_rate": 9.568421052631578e-05, | |
| "loss": 1.4973, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.13166144200626959, | |
| "grad_norm": 0.1466640830039978, | |
| "learning_rate": 9.515263157894737e-05, | |
| "loss": 1.6563, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.13793103448275862, | |
| "grad_norm": 0.15411381423473358, | |
| "learning_rate": 9.462105263157895e-05, | |
| "loss": 1.8432, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.14420062695924765, | |
| "grad_norm": 0.20786531269550323, | |
| "learning_rate": 9.408947368421054e-05, | |
| "loss": 1.866, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.15047021943573669, | |
| "grad_norm": 0.27923616766929626, | |
| "learning_rate": 9.355789473684211e-05, | |
| "loss": 1.8369, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.15673981191222572, | |
| "grad_norm": 0.45824792981147766, | |
| "learning_rate": 9.302631578947369e-05, | |
| "loss": 1.9946, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.16300940438871472, | |
| "grad_norm": 0.11721207201480865, | |
| "learning_rate": 9.249473684210526e-05, | |
| "loss": 1.3217, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.16927899686520376, | |
| "grad_norm": 0.15526717901229858, | |
| "learning_rate": 9.196315789473685e-05, | |
| "loss": 1.7139, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.1755485893416928, | |
| "grad_norm": 0.15869063138961792, | |
| "learning_rate": 9.143157894736843e-05, | |
| "loss": 1.7312, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.18181818181818182, | |
| "grad_norm": 0.1636732816696167, | |
| "learning_rate": 9.09e-05, | |
| "loss": 1.7714, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.18808777429467086, | |
| "grad_norm": 0.20718851685523987, | |
| "learning_rate": 9.036842105263158e-05, | |
| "loss": 1.828, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.19435736677115986, | |
| "grad_norm": 0.30376043915748596, | |
| "learning_rate": 8.983684210526316e-05, | |
| "loss": 1.8697, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.2006269592476489, | |
| "grad_norm": 0.11045503616333008, | |
| "learning_rate": 8.930526315789474e-05, | |
| "loss": 1.3708, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.20689655172413793, | |
| "grad_norm": 0.14193548262119293, | |
| "learning_rate": 8.877368421052632e-05, | |
| "loss": 1.6054, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.21316614420062696, | |
| "grad_norm": 0.15867547690868378, | |
| "learning_rate": 8.82421052631579e-05, | |
| "loss": 1.6872, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.219435736677116, | |
| "grad_norm": 0.16795365512371063, | |
| "learning_rate": 8.771052631578948e-05, | |
| "loss": 1.8028, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.22570532915360503, | |
| "grad_norm": 0.19065167009830475, | |
| "learning_rate": 8.717894736842105e-05, | |
| "loss": 1.8083, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.23197492163009403, | |
| "grad_norm": 0.23978182673454285, | |
| "learning_rate": 8.664736842105263e-05, | |
| "loss": 1.8856, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.23824451410658307, | |
| "grad_norm": 0.13266095519065857, | |
| "learning_rate": 8.61157894736842e-05, | |
| "loss": 1.239, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.2445141065830721, | |
| "grad_norm": 0.1396464705467224, | |
| "learning_rate": 8.55842105263158e-05, | |
| "loss": 1.528, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.2507836990595611, | |
| "grad_norm": 0.15594753623008728, | |
| "learning_rate": 8.505263157894737e-05, | |
| "loss": 1.6795, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.25705329153605017, | |
| "grad_norm": 0.16817249357700348, | |
| "learning_rate": 8.452105263157896e-05, | |
| "loss": 1.7164, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.26332288401253917, | |
| "grad_norm": 0.18988773226737976, | |
| "learning_rate": 8.398947368421053e-05, | |
| "loss": 1.7163, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.26959247648902823, | |
| "grad_norm": 0.20697522163391113, | |
| "learning_rate": 8.345789473684211e-05, | |
| "loss": 1.7748, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.27586206896551724, | |
| "grad_norm": 0.14124108850955963, | |
| "learning_rate": 8.292631578947368e-05, | |
| "loss": 1.3559, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.28213166144200624, | |
| "grad_norm": 0.10386940836906433, | |
| "learning_rate": 8.239473684210526e-05, | |
| "loss": 1.4066, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.2884012539184953, | |
| "grad_norm": 0.15532410144805908, | |
| "learning_rate": 8.186315789473683e-05, | |
| "loss": 1.6759, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2946708463949843, | |
| "grad_norm": 0.16163372993469238, | |
| "learning_rate": 8.133157894736842e-05, | |
| "loss": 1.6765, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.30094043887147337, | |
| "grad_norm": 0.18687112629413605, | |
| "learning_rate": 8.080000000000001e-05, | |
| "loss": 1.7736, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.3072100313479624, | |
| "grad_norm": 0.19891808927059174, | |
| "learning_rate": 8.026842105263159e-05, | |
| "loss": 1.7138, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.31347962382445144, | |
| "grad_norm": 0.546370267868042, | |
| "learning_rate": 7.973684210526316e-05, | |
| "loss": 1.8809, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.31347962382445144, | |
| "eval_loss": 1.6940747499465942, | |
| "eval_runtime": 22.5071, | |
| "eval_samples_per_second": 95.481, | |
| "eval_steps_per_second": 3.021, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.31974921630094044, | |
| "grad_norm": 0.11734400689601898, | |
| "learning_rate": 7.920526315789474e-05, | |
| "loss": 1.4185, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.32601880877742945, | |
| "grad_norm": 0.24825617671012878, | |
| "learning_rate": 7.867368421052631e-05, | |
| "loss": 1.6477, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.3322884012539185, | |
| "grad_norm": 0.2786071002483368, | |
| "learning_rate": 7.814210526315789e-05, | |
| "loss": 1.6508, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.3385579937304075, | |
| "grad_norm": 0.22523944079875946, | |
| "learning_rate": 7.761052631578946e-05, | |
| "loss": 1.7142, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.3448275862068966, | |
| "grad_norm": 0.20191609859466553, | |
| "learning_rate": 7.707894736842105e-05, | |
| "loss": 1.7316, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.3510971786833856, | |
| "grad_norm": 0.2791290283203125, | |
| "learning_rate": 7.654736842105264e-05, | |
| "loss": 1.754, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.3573667711598746, | |
| "grad_norm": 0.10366496443748474, | |
| "learning_rate": 7.601578947368422e-05, | |
| "loss": 1.2027, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.36363636363636365, | |
| "grad_norm": 0.15324437618255615, | |
| "learning_rate": 7.548421052631579e-05, | |
| "loss": 1.6548, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.36990595611285265, | |
| "grad_norm": 0.1608223021030426, | |
| "learning_rate": 7.495263157894737e-05, | |
| "loss": 1.6525, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.3761755485893417, | |
| "grad_norm": 0.20208290219306946, | |
| "learning_rate": 7.442105263157894e-05, | |
| "loss": 1.6865, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3824451410658307, | |
| "grad_norm": 0.23002368211746216, | |
| "learning_rate": 7.388947368421053e-05, | |
| "loss": 1.7319, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.3887147335423197, | |
| "grad_norm": 0.25719794631004333, | |
| "learning_rate": 7.335789473684211e-05, | |
| "loss": 1.7345, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.3949843260188088, | |
| "grad_norm": 0.11874490976333618, | |
| "learning_rate": 7.282631578947368e-05, | |
| "loss": 1.1791, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.4012539184952978, | |
| "grad_norm": 0.13194750249385834, | |
| "learning_rate": 7.229473684210527e-05, | |
| "loss": 1.5006, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.40752351097178685, | |
| "grad_norm": 0.16368307173252106, | |
| "learning_rate": 7.176315789473685e-05, | |
| "loss": 1.653, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.41379310344827586, | |
| "grad_norm": 0.18413026630878448, | |
| "learning_rate": 7.123157894736842e-05, | |
| "loss": 1.6872, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.4200626959247649, | |
| "grad_norm": 0.23217302560806274, | |
| "learning_rate": 7.07e-05, | |
| "loss": 1.7738, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.4263322884012539, | |
| "grad_norm": 0.21129940450191498, | |
| "learning_rate": 7.016842105263159e-05, | |
| "loss": 1.6499, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.43260188087774293, | |
| "grad_norm": 0.14536724984645844, | |
| "learning_rate": 6.963684210526316e-05, | |
| "loss": 1.0403, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.438871473354232, | |
| "grad_norm": 0.12869617342948914, | |
| "learning_rate": 6.910526315789474e-05, | |
| "loss": 1.4696, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.445141065830721, | |
| "grad_norm": 0.1599850058555603, | |
| "learning_rate": 6.857368421052631e-05, | |
| "loss": 1.6434, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.45141065830721006, | |
| "grad_norm": 0.17733348906040192, | |
| "learning_rate": 6.80421052631579e-05, | |
| "loss": 1.6627, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.45768025078369906, | |
| "grad_norm": 0.1968277245759964, | |
| "learning_rate": 6.751052631578948e-05, | |
| "loss": 1.641, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.46394984326018807, | |
| "grad_norm": 0.2661347985267639, | |
| "learning_rate": 6.697894736842105e-05, | |
| "loss": 1.6584, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.4702194357366771, | |
| "grad_norm": 0.574530303478241, | |
| "learning_rate": 6.644736842105264e-05, | |
| "loss": 1.8574, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.47648902821316613, | |
| "grad_norm": 0.11632688343524933, | |
| "learning_rate": 6.591578947368422e-05, | |
| "loss": 1.4023, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.4827586206896552, | |
| "grad_norm": 0.2113131582736969, | |
| "learning_rate": 6.538421052631579e-05, | |
| "loss": 1.6341, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.4890282131661442, | |
| "grad_norm": 0.2565387189388275, | |
| "learning_rate": 6.485263157894737e-05, | |
| "loss": 1.6345, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.4952978056426332, | |
| "grad_norm": 0.2454068958759308, | |
| "learning_rate": 6.432105263157894e-05, | |
| "loss": 1.6885, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.5015673981191222, | |
| "grad_norm": 0.25979533791542053, | |
| "learning_rate": 6.378947368421053e-05, | |
| "loss": 1.7208, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.5078369905956113, | |
| "grad_norm": 0.30401724576950073, | |
| "learning_rate": 6.32578947368421e-05, | |
| "loss": 1.7026, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.5141065830721003, | |
| "grad_norm": 0.11303433030843735, | |
| "learning_rate": 6.27263157894737e-05, | |
| "loss": 1.1272, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.5203761755485894, | |
| "grad_norm": 0.21448008716106415, | |
| "learning_rate": 6.219473684210527e-05, | |
| "loss": 1.6056, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.5266457680250783, | |
| "grad_norm": 0.18290068209171295, | |
| "learning_rate": 6.166315789473685e-05, | |
| "loss": 1.6274, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.5329153605015674, | |
| "grad_norm": 0.21106426417827606, | |
| "learning_rate": 6.113157894736842e-05, | |
| "loss": 1.6398, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.5391849529780565, | |
| "grad_norm": 0.24094125628471375, | |
| "learning_rate": 6.0599999999999996e-05, | |
| "loss": 1.7705, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.5454545454545454, | |
| "grad_norm": 0.2784807085990906, | |
| "learning_rate": 6.006842105263158e-05, | |
| "loss": 1.6553, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.5517241379310345, | |
| "grad_norm": 0.135740265250206, | |
| "learning_rate": 5.953684210526315e-05, | |
| "loss": 1.0888, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.5579937304075235, | |
| "grad_norm": 0.13163161277770996, | |
| "learning_rate": 5.900526315789474e-05, | |
| "loss": 1.5075, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.5642633228840125, | |
| "grad_norm": 0.1819342076778412, | |
| "learning_rate": 5.847368421052632e-05, | |
| "loss": 1.6406, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.5705329153605015, | |
| "grad_norm": 0.20158928632736206, | |
| "learning_rate": 5.79421052631579e-05, | |
| "loss": 1.6991, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.5768025078369906, | |
| "grad_norm": 0.218618705868721, | |
| "learning_rate": 5.7410526315789475e-05, | |
| "loss": 1.6872, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.5830721003134797, | |
| "grad_norm": 0.2431667596101761, | |
| "learning_rate": 5.687894736842105e-05, | |
| "loss": 1.6567, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.5893416927899686, | |
| "grad_norm": 0.15539388358592987, | |
| "learning_rate": 5.6347368421052625e-05, | |
| "loss": 1.1379, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.5956112852664577, | |
| "grad_norm": 0.14918771386146545, | |
| "learning_rate": 5.5815789473684214e-05, | |
| "loss": 1.4123, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.6018808777429467, | |
| "grad_norm": 0.19607971608638763, | |
| "learning_rate": 5.5284210526315796e-05, | |
| "loss": 1.6222, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.6081504702194357, | |
| "grad_norm": 0.21519030630588531, | |
| "learning_rate": 5.475263157894737e-05, | |
| "loss": 1.6053, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.6144200626959248, | |
| "grad_norm": 0.2148403525352478, | |
| "learning_rate": 5.422105263157895e-05, | |
| "loss": 1.7118, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.6206896551724138, | |
| "grad_norm": 0.24617306888103485, | |
| "learning_rate": 5.368947368421053e-05, | |
| "loss": 1.6654, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.6269592476489029, | |
| "grad_norm": 0.5618337392807007, | |
| "learning_rate": 5.3157894736842104e-05, | |
| "loss": 1.7828, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6269592476489029, | |
| "eval_loss": 1.6180227994918823, | |
| "eval_runtime": 21.2838, | |
| "eval_samples_per_second": 100.969, | |
| "eval_steps_per_second": 3.195, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6332288401253918, | |
| "grad_norm": 0.12550345063209534, | |
| "learning_rate": 5.262631578947368e-05, | |
| "loss": 1.1544, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.6394984326018809, | |
| "grad_norm": 0.20995737612247467, | |
| "learning_rate": 5.209473684210527e-05, | |
| "loss": 1.5736, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.64576802507837, | |
| "grad_norm": 0.25661516189575195, | |
| "learning_rate": 5.1563157894736844e-05, | |
| "loss": 1.6052, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.6520376175548589, | |
| "grad_norm": 0.2354477196931839, | |
| "learning_rate": 5.1031578947368426e-05, | |
| "loss": 1.6452, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.658307210031348, | |
| "grad_norm": 0.2458198219537735, | |
| "learning_rate": 5.05e-05, | |
| "loss": 1.7015, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.664576802507837, | |
| "grad_norm": 0.33329248428344727, | |
| "learning_rate": 4.9968421052631576e-05, | |
| "loss": 1.5996, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.670846394984326, | |
| "grad_norm": 0.1337110549211502, | |
| "learning_rate": 4.943684210526316e-05, | |
| "loss": 1.3261, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.677115987460815, | |
| "grad_norm": 0.17378275096416473, | |
| "learning_rate": 4.890526315789474e-05, | |
| "loss": 1.5064, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6833855799373041, | |
| "grad_norm": 0.20934435725212097, | |
| "learning_rate": 4.8373684210526316e-05, | |
| "loss": 1.6143, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.6896551724137931, | |
| "grad_norm": 0.24600297212600708, | |
| "learning_rate": 4.784210526315789e-05, | |
| "loss": 1.6473, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.6959247648902821, | |
| "grad_norm": 0.2352122813463211, | |
| "learning_rate": 4.731052631578947e-05, | |
| "loss": 1.59, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.7021943573667712, | |
| "grad_norm": 0.3226903975009918, | |
| "learning_rate": 4.6778947368421055e-05, | |
| "loss": 1.5846, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.7084639498432602, | |
| "grad_norm": 0.14687702059745789, | |
| "learning_rate": 4.624736842105263e-05, | |
| "loss": 1.1129, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.7147335423197492, | |
| "grad_norm": 0.15256668627262115, | |
| "learning_rate": 4.571578947368421e-05, | |
| "loss": 1.4691, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.7210031347962382, | |
| "grad_norm": 0.20408159494400024, | |
| "learning_rate": 4.518421052631579e-05, | |
| "loss": 1.6611, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 0.22884103655815125, | |
| "learning_rate": 4.465263157894737e-05, | |
| "loss": 1.6256, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.7335423197492164, | |
| "grad_norm": 0.23959754407405853, | |
| "learning_rate": 4.412105263157895e-05, | |
| "loss": 1.6332, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.7398119122257053, | |
| "grad_norm": 0.2993139624595642, | |
| "learning_rate": 4.358947368421053e-05, | |
| "loss": 1.6246, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.7460815047021944, | |
| "grad_norm": 0.19114616513252258, | |
| "learning_rate": 4.30578947368421e-05, | |
| "loss": 1.0553, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.7523510971786834, | |
| "grad_norm": 0.1459835320711136, | |
| "learning_rate": 4.2526315789473685e-05, | |
| "loss": 1.432, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.7586206896551724, | |
| "grad_norm": 0.19810743629932404, | |
| "learning_rate": 4.199473684210527e-05, | |
| "loss": 1.5597, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.7648902821316614, | |
| "grad_norm": 0.22155524790287018, | |
| "learning_rate": 4.146315789473684e-05, | |
| "loss": 1.5973, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.7711598746081505, | |
| "grad_norm": 0.252210795879364, | |
| "learning_rate": 4.093157894736842e-05, | |
| "loss": 1.7093, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.7774294670846394, | |
| "grad_norm": 0.2699624300003052, | |
| "learning_rate": 4.0400000000000006e-05, | |
| "loss": 1.6218, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.7836990595611285, | |
| "grad_norm": 0.5296167731285095, | |
| "learning_rate": 3.986842105263158e-05, | |
| "loss": 1.6447, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.7899686520376176, | |
| "grad_norm": 0.1305689811706543, | |
| "learning_rate": 3.933684210526316e-05, | |
| "loss": 1.2415, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.7962382445141066, | |
| "grad_norm": 0.19951651990413666, | |
| "learning_rate": 3.880526315789473e-05, | |
| "loss": 1.5846, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.8025078369905956, | |
| "grad_norm": 0.24844390153884888, | |
| "learning_rate": 3.827368421052632e-05, | |
| "loss": 1.5013, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.8087774294670846, | |
| "grad_norm": 0.26770254969596863, | |
| "learning_rate": 3.7742105263157896e-05, | |
| "loss": 1.6404, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.8150470219435737, | |
| "grad_norm": 0.268388956785202, | |
| "learning_rate": 3.721052631578947e-05, | |
| "loss": 1.6831, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.8213166144200627, | |
| "grad_norm": 0.3877102732658386, | |
| "learning_rate": 3.6678947368421054e-05, | |
| "loss": 1.5713, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.8275862068965517, | |
| "grad_norm": 0.14182406663894653, | |
| "learning_rate": 3.6147368421052636e-05, | |
| "loss": 1.1147, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.8338557993730408, | |
| "grad_norm": 0.1862500160932541, | |
| "learning_rate": 3.561578947368421e-05, | |
| "loss": 1.4998, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.8401253918495298, | |
| "grad_norm": 0.22701646387577057, | |
| "learning_rate": 3.508421052631579e-05, | |
| "loss": 1.5137, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.8463949843260188, | |
| "grad_norm": 0.2560668885707855, | |
| "learning_rate": 3.455263157894737e-05, | |
| "loss": 1.6129, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.8526645768025078, | |
| "grad_norm": 0.2551365792751312, | |
| "learning_rate": 3.402105263157895e-05, | |
| "loss": 1.6454, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.8589341692789969, | |
| "grad_norm": 0.331463485956192, | |
| "learning_rate": 3.3489473684210526e-05, | |
| "loss": 1.5863, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.8652037617554859, | |
| "grad_norm": 0.17028988897800446, | |
| "learning_rate": 3.295789473684211e-05, | |
| "loss": 1.1377, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.8714733542319749, | |
| "grad_norm": 0.1787910908460617, | |
| "learning_rate": 3.242631578947368e-05, | |
| "loss": 1.4605, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.877742946708464, | |
| "grad_norm": 0.21535782516002655, | |
| "learning_rate": 3.1894736842105265e-05, | |
| "loss": 1.5651, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.8840125391849529, | |
| "grad_norm": 0.2516005337238312, | |
| "learning_rate": 3.136315789473685e-05, | |
| "loss": 1.6457, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.890282131661442, | |
| "grad_norm": 0.26242345571517944, | |
| "learning_rate": 3.083157894736842e-05, | |
| "loss": 1.6603, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.896551724137931, | |
| "grad_norm": 0.31754815578460693, | |
| "learning_rate": 3.0299999999999998e-05, | |
| "loss": 1.6406, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.9028213166144201, | |
| "grad_norm": 0.19499145448207855, | |
| "learning_rate": 2.9768421052631577e-05, | |
| "loss": 1.0787, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.9090909090909091, | |
| "grad_norm": 0.16461507976055145, | |
| "learning_rate": 2.923684210526316e-05, | |
| "loss": 1.3539, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.9153605015673981, | |
| "grad_norm": 0.231450155377388, | |
| "learning_rate": 2.8705263157894737e-05, | |
| "loss": 1.6046, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.9216300940438872, | |
| "grad_norm": 0.25709661841392517, | |
| "learning_rate": 2.8173684210526313e-05, | |
| "loss": 1.5514, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.9278996865203761, | |
| "grad_norm": 0.26337385177612305, | |
| "learning_rate": 2.7642105263157898e-05, | |
| "loss": 1.6737, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.9341692789968652, | |
| "grad_norm": 0.29333168268203735, | |
| "learning_rate": 2.7110526315789473e-05, | |
| "loss": 1.5403, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.9404388714733543, | |
| "grad_norm": 0.6478084325790405, | |
| "learning_rate": 2.6578947368421052e-05, | |
| "loss": 1.7032, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9404388714733543, | |
| "eval_loss": 1.5667701959609985, | |
| "eval_runtime": 22.4342, | |
| "eval_samples_per_second": 95.791, | |
| "eval_steps_per_second": 3.031, | |
| "step": 150 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "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": 1.6208921818772275e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |