| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 20, | |
| "global_step": 776, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0, | |
| "eval_accuracy": 0.9369414101290964, | |
| "eval_f1": 0.18064516129032257, | |
| "eval_loss": 0.32557418942451477, | |
| "eval_precision": 0.14285714285714285, | |
| "eval_recall": 0.24561403508771928, | |
| "eval_runtime": 85.4903, | |
| "eval_samples_per_second": 5.322, | |
| "eval_steps_per_second": 0.175, | |
| "step": 0 | |
| }, | |
| { | |
| "epoch": 0.001288659793814433, | |
| "grad_norm": 4.25856876373291, | |
| "learning_rate": 2.564102564102564e-07, | |
| "loss": 0.5948, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.002577319587628866, | |
| "grad_norm": 3.965942144393921, | |
| "learning_rate": 5.128205128205128e-07, | |
| "loss": 0.5193, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.003865979381443299, | |
| "grad_norm": 4.168001174926758, | |
| "learning_rate": 7.692307692307694e-07, | |
| "loss": 0.5634, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.005154639175257732, | |
| "grad_norm": 3.510260820388794, | |
| "learning_rate": 1.0256410256410257e-06, | |
| "loss": 0.5141, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006443298969072165, | |
| "grad_norm": 3.4605071544647217, | |
| "learning_rate": 1.282051282051282e-06, | |
| "loss": 0.4554, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.007731958762886598, | |
| "grad_norm": 3.9275991916656494, | |
| "learning_rate": 1.5384615384615387e-06, | |
| "loss": 0.5348, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00902061855670103, | |
| "grad_norm": 3.4560351371765137, | |
| "learning_rate": 1.794871794871795e-06, | |
| "loss": 0.4784, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.010309278350515464, | |
| "grad_norm": 3.93674373626709, | |
| "learning_rate": 2.0512820512820513e-06, | |
| "loss": 0.5567, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.011597938144329897, | |
| "grad_norm": 3.893137216567993, | |
| "learning_rate": 2.307692307692308e-06, | |
| "loss": 0.5175, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.01288659793814433, | |
| "grad_norm": 3.8920130729675293, | |
| "learning_rate": 2.564102564102564e-06, | |
| "loss": 0.5187, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.014175257731958763, | |
| "grad_norm": 4.420604228973389, | |
| "learning_rate": 2.8205128205128207e-06, | |
| "loss": 0.5764, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.015463917525773196, | |
| "grad_norm": 3.916328191757202, | |
| "learning_rate": 3.0769230769230774e-06, | |
| "loss": 0.5103, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.01675257731958763, | |
| "grad_norm": 3.8103713989257812, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.5271, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.01804123711340206, | |
| "grad_norm": 3.2343883514404297, | |
| "learning_rate": 3.58974358974359e-06, | |
| "loss": 0.4981, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.019329896907216496, | |
| "grad_norm": 3.556180953979492, | |
| "learning_rate": 3.846153846153847e-06, | |
| "loss": 0.472, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.020618556701030927, | |
| "grad_norm": 3.4665050506591797, | |
| "learning_rate": 4.102564102564103e-06, | |
| "loss": 0.463, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02190721649484536, | |
| "grad_norm": 3.6362946033477783, | |
| "learning_rate": 4.358974358974359e-06, | |
| "loss": 0.5137, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.023195876288659795, | |
| "grad_norm": 3.6967854499816895, | |
| "learning_rate": 4.615384615384616e-06, | |
| "loss": 0.5288, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.024484536082474227, | |
| "grad_norm": 3.482665538787842, | |
| "learning_rate": 4.871794871794872e-06, | |
| "loss": 0.4946, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.02577319587628866, | |
| "grad_norm": 3.539522647857666, | |
| "learning_rate": 5.128205128205128e-06, | |
| "loss": 0.4529, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02577319587628866, | |
| "eval_accuracy": 0.9483614697120158, | |
| "eval_f1": 0.16129032258064516, | |
| "eval_loss": 0.28827786445617676, | |
| "eval_precision": 0.14925373134328357, | |
| "eval_recall": 0.17543859649122806, | |
| "eval_runtime": 85.2267, | |
| "eval_samples_per_second": 5.339, | |
| "eval_steps_per_second": 0.176, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.027061855670103094, | |
| "grad_norm": 3.701735496520996, | |
| "learning_rate": 5.384615384615385e-06, | |
| "loss": 0.457, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.028350515463917526, | |
| "grad_norm": 3.351079225540161, | |
| "learning_rate": 5.641025641025641e-06, | |
| "loss": 0.4086, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.029639175257731958, | |
| "grad_norm": 3.3507163524627686, | |
| "learning_rate": 5.897435897435898e-06, | |
| "loss": 0.3953, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.030927835051546393, | |
| "grad_norm": 4.022622108459473, | |
| "learning_rate": 6.153846153846155e-06, | |
| "loss": 0.4388, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03221649484536082, | |
| "grad_norm": 3.652367353439331, | |
| "learning_rate": 6.410256410256412e-06, | |
| "loss": 0.415, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03350515463917526, | |
| "grad_norm": 3.2059590816497803, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.3946, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.03479381443298969, | |
| "grad_norm": 3.6049065589904785, | |
| "learning_rate": 6.923076923076923e-06, | |
| "loss": 0.3836, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03608247422680412, | |
| "grad_norm": 3.0256381034851074, | |
| "learning_rate": 7.17948717948718e-06, | |
| "loss": 0.3502, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.037371134020618556, | |
| "grad_norm": 2.608344793319702, | |
| "learning_rate": 7.435897435897437e-06, | |
| "loss": 0.322, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.03865979381443299, | |
| "grad_norm": 2.5998220443725586, | |
| "learning_rate": 7.692307692307694e-06, | |
| "loss": 0.3078, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03994845360824742, | |
| "grad_norm": 3.184544801712036, | |
| "learning_rate": 7.948717948717949e-06, | |
| "loss": 0.3189, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.041237113402061855, | |
| "grad_norm": 2.9103171825408936, | |
| "learning_rate": 8.205128205128205e-06, | |
| "loss": 0.3387, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.04252577319587629, | |
| "grad_norm": 2.9772467613220215, | |
| "learning_rate": 8.461538461538462e-06, | |
| "loss": 0.3093, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.04381443298969072, | |
| "grad_norm": 2.77077054977417, | |
| "learning_rate": 8.717948717948719e-06, | |
| "loss": 0.2961, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.045103092783505154, | |
| "grad_norm": 2.3303472995758057, | |
| "learning_rate": 8.974358974358976e-06, | |
| "loss": 0.2714, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04639175257731959, | |
| "grad_norm": 2.529663324356079, | |
| "learning_rate": 9.230769230769232e-06, | |
| "loss": 0.2592, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.04768041237113402, | |
| "grad_norm": 2.3244917392730713, | |
| "learning_rate": 9.487179487179487e-06, | |
| "loss": 0.2439, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.04896907216494845, | |
| "grad_norm": 2.4089925289154053, | |
| "learning_rate": 9.743589743589744e-06, | |
| "loss": 0.237, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.05025773195876289, | |
| "grad_norm": 1.9498683214187622, | |
| "learning_rate": 1e-05, | |
| "loss": 0.202, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.05154639175257732, | |
| "grad_norm": 1.967588186264038, | |
| "learning_rate": 1.0256410256410256e-05, | |
| "loss": 0.2483, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05154639175257732, | |
| "eval_accuracy": 0.9672293942403177, | |
| "eval_f1": 0.08333333333333333, | |
| "eval_loss": 0.14610709249973297, | |
| "eval_precision": 0.2, | |
| "eval_recall": 0.05263157894736842, | |
| "eval_runtime": 84.9582, | |
| "eval_samples_per_second": 5.356, | |
| "eval_steps_per_second": 0.177, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05283505154639175, | |
| "grad_norm": 1.6313542127609253, | |
| "learning_rate": 1.0512820512820514e-05, | |
| "loss": 0.2022, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.05412371134020619, | |
| "grad_norm": 1.4857121706008911, | |
| "learning_rate": 1.076923076923077e-05, | |
| "loss": 0.1911, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.055412371134020616, | |
| "grad_norm": 1.229911208152771, | |
| "learning_rate": 1.1025641025641028e-05, | |
| "loss": 0.2323, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05670103092783505, | |
| "grad_norm": 1.344585657119751, | |
| "learning_rate": 1.1282051282051283e-05, | |
| "loss": 0.1473, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.05798969072164949, | |
| "grad_norm": 0.909356951713562, | |
| "learning_rate": 1.1538461538461538e-05, | |
| "loss": 0.1828, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.059278350515463915, | |
| "grad_norm": 0.9276474118232727, | |
| "learning_rate": 1.1794871794871796e-05, | |
| "loss": 0.1573, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.06056701030927835, | |
| "grad_norm": 0.984527051448822, | |
| "learning_rate": 1.2051282051282051e-05, | |
| "loss": 0.2004, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.061855670103092786, | |
| "grad_norm": 0.7970030307769775, | |
| "learning_rate": 1.230769230769231e-05, | |
| "loss": 0.1276, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.06314432989690721, | |
| "grad_norm": 1.0922330617904663, | |
| "learning_rate": 1.2564102564102565e-05, | |
| "loss": 0.168, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06443298969072164, | |
| "grad_norm": 1.187216877937317, | |
| "learning_rate": 1.2820512820512823e-05, | |
| "loss": 0.2007, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06572164948453608, | |
| "grad_norm": 1.1624428033828735, | |
| "learning_rate": 1.3076923076923078e-05, | |
| "loss": 0.1665, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06701030927835051, | |
| "grad_norm": 1.346639633178711, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.2013, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.06829896907216494, | |
| "grad_norm": 0.9596057534217834, | |
| "learning_rate": 1.3589743589743592e-05, | |
| "loss": 0.1412, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.06958762886597938, | |
| "grad_norm": 1.3882497549057007, | |
| "learning_rate": 1.3846153846153847e-05, | |
| "loss": 0.191, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.07087628865979381, | |
| "grad_norm": 0.7435745000839233, | |
| "learning_rate": 1.4102564102564105e-05, | |
| "loss": 0.1115, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.07216494845360824, | |
| "grad_norm": 0.8442493081092834, | |
| "learning_rate": 1.435897435897436e-05, | |
| "loss": 0.1239, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.07345360824742268, | |
| "grad_norm": 1.1027814149856567, | |
| "learning_rate": 1.4615384615384615e-05, | |
| "loss": 0.1605, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.07474226804123711, | |
| "grad_norm": 0.9518107771873474, | |
| "learning_rate": 1.4871794871794874e-05, | |
| "loss": 0.1253, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.07603092783505154, | |
| "grad_norm": 0.8097633719444275, | |
| "learning_rate": 1.5128205128205129e-05, | |
| "loss": 0.1258, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.07731958762886598, | |
| "grad_norm": 0.9735177159309387, | |
| "learning_rate": 1.5384615384615387e-05, | |
| "loss": 0.1622, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07731958762886598, | |
| "eval_accuracy": 0.9687189672293942, | |
| "eval_f1": 0.18181818181818182, | |
| "eval_loss": 0.10798302292823792, | |
| "eval_precision": 0.35, | |
| "eval_recall": 0.12280701754385964, | |
| "eval_runtime": 84.7661, | |
| "eval_samples_per_second": 5.368, | |
| "eval_steps_per_second": 0.177, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07860824742268041, | |
| "grad_norm": 0.9637501835823059, | |
| "learning_rate": 1.5641025641025644e-05, | |
| "loss": 0.1333, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.07989690721649484, | |
| "grad_norm": 0.6378239393234253, | |
| "learning_rate": 1.5897435897435897e-05, | |
| "loss": 0.076, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.08118556701030928, | |
| "grad_norm": 0.7773278951644897, | |
| "learning_rate": 1.6153846153846154e-05, | |
| "loss": 0.0937, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.08247422680412371, | |
| "grad_norm": 0.5744358897209167, | |
| "learning_rate": 1.641025641025641e-05, | |
| "loss": 0.089, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.08376288659793814, | |
| "grad_norm": 0.6605228781700134, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 0.058, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.08505154639175258, | |
| "grad_norm": 0.68232661485672, | |
| "learning_rate": 1.6923076923076924e-05, | |
| "loss": 0.1027, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.08634020618556701, | |
| "grad_norm": 0.7368768453598022, | |
| "learning_rate": 1.717948717948718e-05, | |
| "loss": 0.1222, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.08762886597938144, | |
| "grad_norm": 1.2759743928909302, | |
| "learning_rate": 1.7435897435897438e-05, | |
| "loss": 0.1637, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.08891752577319588, | |
| "grad_norm": 1.383555293083191, | |
| "learning_rate": 1.7692307692307694e-05, | |
| "loss": 0.1774, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.09020618556701031, | |
| "grad_norm": 1.048829436302185, | |
| "learning_rate": 1.794871794871795e-05, | |
| "loss": 0.0908, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09149484536082474, | |
| "grad_norm": 1.0387591123580933, | |
| "learning_rate": 1.8205128205128208e-05, | |
| "loss": 0.1263, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.09278350515463918, | |
| "grad_norm": 0.9198775291442871, | |
| "learning_rate": 1.8461538461538465e-05, | |
| "loss": 0.1042, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.09407216494845361, | |
| "grad_norm": 0.8574474453926086, | |
| "learning_rate": 1.8717948717948718e-05, | |
| "loss": 0.0845, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.09536082474226804, | |
| "grad_norm": 0.9596143364906311, | |
| "learning_rate": 1.8974358974358975e-05, | |
| "loss": 0.1468, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.09664948453608248, | |
| "grad_norm": 0.9079321026802063, | |
| "learning_rate": 1.923076923076923e-05, | |
| "loss": 0.0647, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0979381443298969, | |
| "grad_norm": 1.2310247421264648, | |
| "learning_rate": 1.9487179487179488e-05, | |
| "loss": 0.1315, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.09922680412371133, | |
| "grad_norm": 0.9216123223304749, | |
| "learning_rate": 1.9743589743589745e-05, | |
| "loss": 0.0794, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.10051546391752578, | |
| "grad_norm": 0.8806676864624023, | |
| "learning_rate": 2e-05, | |
| "loss": 0.0733, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.1018041237113402, | |
| "grad_norm": 1.0531110763549805, | |
| "learning_rate": 1.999989871195906e-05, | |
| "loss": 0.045, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.10309278350515463, | |
| "grad_norm": 1.5389978885650635, | |
| "learning_rate": 1.9999594849888083e-05, | |
| "loss": 0.1243, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.10309278350515463, | |
| "eval_accuracy": 0.9697120158887785, | |
| "eval_f1": 0.3838383838383838, | |
| "eval_loss": 0.08788777142763138, | |
| "eval_precision": 0.4523809523809524, | |
| "eval_recall": 0.3333333333333333, | |
| "eval_runtime": 84.9491, | |
| "eval_samples_per_second": 5.356, | |
| "eval_steps_per_second": 0.177, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.10438144329896908, | |
| "grad_norm": 1.1171326637268066, | |
| "learning_rate": 1.9999088419942598e-05, | |
| "loss": 0.081, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.1056701030927835, | |
| "grad_norm": 1.4132601022720337, | |
| "learning_rate": 1.999837943238166e-05, | |
| "loss": 0.0699, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.10695876288659793, | |
| "grad_norm": 2.1418771743774414, | |
| "learning_rate": 1.999746790156766e-05, | |
| "loss": 0.1157, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.10824742268041238, | |
| "grad_norm": 2.2251861095428467, | |
| "learning_rate": 1.9996353845966033e-05, | |
| "loss": 0.1371, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.1095360824742268, | |
| "grad_norm": 1.8240890502929688, | |
| "learning_rate": 1.999503728814488e-05, | |
| "loss": 0.0972, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.11082474226804123, | |
| "grad_norm": 2.1466102600097656, | |
| "learning_rate": 1.9993518254774517e-05, | |
| "loss": 0.0827, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.11211340206185567, | |
| "grad_norm": 1.60856032371521, | |
| "learning_rate": 1.999179677662692e-05, | |
| "loss": 0.0902, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.1134020618556701, | |
| "grad_norm": 1.7380954027175903, | |
| "learning_rate": 1.998987288857513e-05, | |
| "loss": 0.1057, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.11469072164948453, | |
| "grad_norm": 0.9695951342582703, | |
| "learning_rate": 1.9987746629592506e-05, | |
| "loss": 0.0485, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.11597938144329897, | |
| "grad_norm": 2.6375732421875, | |
| "learning_rate": 1.9985418042751975e-05, | |
| "loss": 0.0962, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.1172680412371134, | |
| "grad_norm": 1.3601967096328735, | |
| "learning_rate": 1.9982887175225136e-05, | |
| "loss": 0.0402, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.11855670103092783, | |
| "grad_norm": 1.6553303003311157, | |
| "learning_rate": 1.998015407828131e-05, | |
| "loss": 0.0576, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.11984536082474227, | |
| "grad_norm": 1.6685421466827393, | |
| "learning_rate": 1.9977218807286507e-05, | |
| "loss": 0.0666, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.1211340206185567, | |
| "grad_norm": 3.0571939945220947, | |
| "learning_rate": 1.9974081421702296e-05, | |
| "loss": 0.0754, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.12242268041237113, | |
| "grad_norm": 1.3410394191741943, | |
| "learning_rate": 1.99707419850846e-05, | |
| "loss": 0.0545, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.12371134020618557, | |
| "grad_norm": 1.2060542106628418, | |
| "learning_rate": 1.9967200565082426e-05, | |
| "loss": 0.0423, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 2.734278678894043, | |
| "learning_rate": 1.9963457233436468e-05, | |
| "loss": 0.0951, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.12628865979381443, | |
| "grad_norm": 1.210222601890564, | |
| "learning_rate": 1.9959512065977673e-05, | |
| "loss": 0.0476, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.12757731958762886, | |
| "grad_norm": 2.3914737701416016, | |
| "learning_rate": 1.9955365142625694e-05, | |
| "loss": 0.0498, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.12886597938144329, | |
| "grad_norm": 2.5380992889404297, | |
| "learning_rate": 1.9951016547387286e-05, | |
| "loss": 0.0678, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12886597938144329, | |
| "eval_accuracy": 0.9692154915590864, | |
| "eval_f1": 0.5753424657534246, | |
| "eval_loss": 0.06998522579669952, | |
| "eval_precision": 0.47191011235955055, | |
| "eval_recall": 0.7368421052631579, | |
| "eval_runtime": 84.7785, | |
| "eval_samples_per_second": 5.367, | |
| "eval_steps_per_second": 0.177, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13015463917525774, | |
| "grad_norm": 3.035583019256592, | |
| "learning_rate": 1.994646636835458e-05, | |
| "loss": 0.0746, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.13144329896907217, | |
| "grad_norm": 2.226928472518921, | |
| "learning_rate": 1.9941714697703333e-05, | |
| "loss": 0.0631, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.1327319587628866, | |
| "grad_norm": 0.8567290306091309, | |
| "learning_rate": 1.9936761631691007e-05, | |
| "loss": 0.0268, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.13402061855670103, | |
| "grad_norm": 3.362657308578491, | |
| "learning_rate": 1.993160727065489e-05, | |
| "loss": 0.0924, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.13530927835051546, | |
| "grad_norm": 0.8925597667694092, | |
| "learning_rate": 1.992625171901e-05, | |
| "loss": 0.0307, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.13659793814432988, | |
| "grad_norm": 3.0269739627838135, | |
| "learning_rate": 1.9920695085247012e-05, | |
| "loss": 0.0435, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.13788659793814434, | |
| "grad_norm": 1.2455264329910278, | |
| "learning_rate": 1.991493748193002e-05, | |
| "loss": 0.0365, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.13917525773195877, | |
| "grad_norm": 1.912764549255371, | |
| "learning_rate": 1.9908979025694312e-05, | |
| "loss": 0.0432, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.1404639175257732, | |
| "grad_norm": 0.9947022199630737, | |
| "learning_rate": 1.9902819837243954e-05, | |
| "loss": 0.0173, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.14175257731958762, | |
| "grad_norm": 1.1930323839187622, | |
| "learning_rate": 1.989646004134937e-05, | |
| "loss": 0.0339, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.14304123711340205, | |
| "grad_norm": 2.846757173538208, | |
| "learning_rate": 1.9889899766844817e-05, | |
| "loss": 0.0705, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.14432989690721648, | |
| "grad_norm": 2.564868927001953, | |
| "learning_rate": 1.9883139146625763e-05, | |
| "loss": 0.0359, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.14561855670103094, | |
| "grad_norm": 1.804880976676941, | |
| "learning_rate": 1.9876178317646203e-05, | |
| "loss": 0.0266, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.14690721649484537, | |
| "grad_norm": 1.020308256149292, | |
| "learning_rate": 1.9869017420915888e-05, | |
| "loss": 0.019, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.1481958762886598, | |
| "grad_norm": 1.854977011680603, | |
| "learning_rate": 1.9861656601497452e-05, | |
| "loss": 0.0279, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.14948453608247422, | |
| "grad_norm": 2.949525833129883, | |
| "learning_rate": 1.9854096008503495e-05, | |
| "loss": 0.0426, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.15077319587628865, | |
| "grad_norm": 3.297407627105713, | |
| "learning_rate": 1.9846335795093547e-05, | |
| "loss": 0.0677, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.15206185567010308, | |
| "grad_norm": 0.6557813286781311, | |
| "learning_rate": 1.9838376118470965e-05, | |
| "loss": 0.0143, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.15335051546391754, | |
| "grad_norm": 2.7906975746154785, | |
| "learning_rate": 1.9830217139879768e-05, | |
| "loss": 0.0333, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.15463917525773196, | |
| "grad_norm": 1.2422326803207397, | |
| "learning_rate": 1.9821859024601345e-05, | |
| "loss": 0.0301, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.15463917525773196, | |
| "eval_accuracy": 0.9836146971201589, | |
| "eval_f1": 0.7591240875912408, | |
| "eval_loss": 0.047399766743183136, | |
| "eval_precision": 0.65, | |
| "eval_recall": 0.9122807017543859, | |
| "eval_runtime": 83.3915, | |
| "eval_samples_per_second": 5.456, | |
| "eval_steps_per_second": 0.18, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.1559278350515464, | |
| "grad_norm": 1.9494062662124634, | |
| "learning_rate": 1.981330194195112e-05, | |
| "loss": 0.0154, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.15721649484536082, | |
| "grad_norm": 3.311889410018921, | |
| "learning_rate": 1.9804546065275116e-05, | |
| "loss": 0.0667, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.15850515463917525, | |
| "grad_norm": 2.091780424118042, | |
| "learning_rate": 1.9795591571946454e-05, | |
| "loss": 0.0305, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.15979381443298968, | |
| "grad_norm": 2.4576990604400635, | |
| "learning_rate": 1.978643864336176e-05, | |
| "loss": 0.0198, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.16108247422680413, | |
| "grad_norm": 2.812678575515747, | |
| "learning_rate": 1.9777087464937464e-05, | |
| "loss": 0.0375, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.16237113402061856, | |
| "grad_norm": 2.1656692028045654, | |
| "learning_rate": 1.9767538226106078e-05, | |
| "loss": 0.028, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.163659793814433, | |
| "grad_norm": 2.1985504627227783, | |
| "learning_rate": 1.9757791120312344e-05, | |
| "loss": 0.0267, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.16494845360824742, | |
| "grad_norm": 2.4062767028808594, | |
| "learning_rate": 1.9747846345009306e-05, | |
| "loss": 0.0434, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.16623711340206185, | |
| "grad_norm": 4.364503860473633, | |
| "learning_rate": 1.9737704101654335e-05, | |
| "loss": 0.0664, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.16752577319587628, | |
| "grad_norm": 0.7257900238037109, | |
| "learning_rate": 1.9727364595705012e-05, | |
| "loss": 0.0113, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.16881443298969073, | |
| "grad_norm": 2.333080530166626, | |
| "learning_rate": 1.9716828036615006e-05, | |
| "loss": 0.0246, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.17010309278350516, | |
| "grad_norm": 2.3138315677642822, | |
| "learning_rate": 1.9706094637829797e-05, | |
| "loss": 0.0499, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.1713917525773196, | |
| "grad_norm": 1.7137653827667236, | |
| "learning_rate": 1.9695164616782378e-05, | |
| "loss": 0.0217, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.17268041237113402, | |
| "grad_norm": 2.961329698562622, | |
| "learning_rate": 1.9684038194888827e-05, | |
| "loss": 0.046, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.17396907216494845, | |
| "grad_norm": 2.9298017024993896, | |
| "learning_rate": 1.9672715597543845e-05, | |
| "loss": 0.0239, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.17525773195876287, | |
| "grad_norm": 1.0312743186950684, | |
| "learning_rate": 1.9661197054116165e-05, | |
| "loss": 0.0111, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.17654639175257733, | |
| "grad_norm": 2.157925844192505, | |
| "learning_rate": 1.964948279794393e-05, | |
| "loss": 0.0308, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.17783505154639176, | |
| "grad_norm": 0.35301733016967773, | |
| "learning_rate": 1.963757306632996e-05, | |
| "loss": 0.0062, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.1791237113402062, | |
| "grad_norm": 0.8291146755218506, | |
| "learning_rate": 1.962546810053692e-05, | |
| "loss": 0.0121, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.18041237113402062, | |
| "grad_norm": 0.6177694201469421, | |
| "learning_rate": 1.9613168145782468e-05, | |
| "loss": 0.0105, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18041237113402062, | |
| "eval_accuracy": 0.9910625620655412, | |
| "eval_f1": 0.8421052631578947, | |
| "eval_loss": 0.03416102007031441, | |
| "eval_precision": 0.8421052631578947, | |
| "eval_recall": 0.8421052631578947, | |
| "eval_runtime": 85.1924, | |
| "eval_samples_per_second": 5.341, | |
| "eval_steps_per_second": 0.176, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18170103092783504, | |
| "grad_norm": 2.37103533744812, | |
| "learning_rate": 1.960067345123427e-05, | |
| "loss": 0.0273, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.18298969072164947, | |
| "grad_norm": 3.434865713119507, | |
| "learning_rate": 1.958798427000495e-05, | |
| "loss": 0.0304, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.18427835051546393, | |
| "grad_norm": 0.6154772639274597, | |
| "learning_rate": 1.9575100859146974e-05, | |
| "loss": 0.0135, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.18556701030927836, | |
| "grad_norm": 2.038938522338867, | |
| "learning_rate": 1.956202347964743e-05, | |
| "loss": 0.0248, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.18685567010309279, | |
| "grad_norm": 1.0157862901687622, | |
| "learning_rate": 1.954875239642274e-05, | |
| "loss": 0.0081, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.18814432989690721, | |
| "grad_norm": 1.953661561012268, | |
| "learning_rate": 1.9535287878313315e-05, | |
| "loss": 0.0169, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.18943298969072164, | |
| "grad_norm": 1.0837459564208984, | |
| "learning_rate": 1.952163019807809e-05, | |
| "loss": 0.0082, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.19072164948453607, | |
| "grad_norm": 3.747811794281006, | |
| "learning_rate": 1.9507779632388997e-05, | |
| "loss": 0.0215, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.19201030927835053, | |
| "grad_norm": 2.3342642784118652, | |
| "learning_rate": 1.9493736461825366e-05, | |
| "loss": 0.0138, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.19329896907216496, | |
| "grad_norm": 2.4509527683258057, | |
| "learning_rate": 1.947950097086825e-05, | |
| "loss": 0.0391, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.19458762886597938, | |
| "grad_norm": 1.6051100492477417, | |
| "learning_rate": 1.946507344789464e-05, | |
| "loss": 0.009, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.1958762886597938, | |
| "grad_norm": 0.9797040820121765, | |
| "learning_rate": 1.945045418517165e-05, | |
| "loss": 0.0115, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.19716494845360824, | |
| "grad_norm": 2.5942206382751465, | |
| "learning_rate": 1.9435643478850573e-05, | |
| "loss": 0.0237, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.19845360824742267, | |
| "grad_norm": 1.6278647184371948, | |
| "learning_rate": 1.9420641628960897e-05, | |
| "loss": 0.0139, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.19974226804123713, | |
| "grad_norm": 0.7787767648696899, | |
| "learning_rate": 1.9405448939404215e-05, | |
| "loss": 0.0091, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.20103092783505155, | |
| "grad_norm": 0.6835376024246216, | |
| "learning_rate": 1.9390065717948084e-05, | |
| "loss": 0.005, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.20231958762886598, | |
| "grad_norm": 0.6300978660583496, | |
| "learning_rate": 1.9374492276219776e-05, | |
| "loss": 0.0065, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.2036082474226804, | |
| "grad_norm": 1.7205617427825928, | |
| "learning_rate": 1.9358728929699966e-05, | |
| "loss": 0.0111, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.20489690721649484, | |
| "grad_norm": 1.923244595527649, | |
| "learning_rate": 1.9342775997716357e-05, | |
| "loss": 0.0065, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.20618556701030927, | |
| "grad_norm": 4.994507789611816, | |
| "learning_rate": 1.9326633803437197e-05, | |
| "loss": 0.041, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.20618556701030927, | |
| "eval_accuracy": 0.9925521350546177, | |
| "eval_f1": 0.8672566371681416, | |
| "eval_loss": 0.033311955630779266, | |
| "eval_precision": 0.875, | |
| "eval_recall": 0.8596491228070176, | |
| "eval_runtime": 83.558, | |
| "eval_samples_per_second": 5.445, | |
| "eval_steps_per_second": 0.18, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.20747422680412372, | |
| "grad_norm": 4.8891921043396, | |
| "learning_rate": 1.9310302673864724e-05, | |
| "loss": 0.0596, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.20876288659793815, | |
| "grad_norm": 2.8088436126708984, | |
| "learning_rate": 1.929378293982857e-05, | |
| "loss": 0.0281, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.21005154639175258, | |
| "grad_norm": 1.619964838027954, | |
| "learning_rate": 1.9277074935979034e-05, | |
| "loss": 0.0068, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.211340206185567, | |
| "grad_norm": 0.3896208703517914, | |
| "learning_rate": 1.926017900078031e-05, | |
| "loss": 0.0024, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.21262886597938144, | |
| "grad_norm": 6.722455024719238, | |
| "learning_rate": 1.924309547650363e-05, | |
| "loss": 0.0353, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.21391752577319587, | |
| "grad_norm": 1.7863891124725342, | |
| "learning_rate": 1.922582470922034e-05, | |
| "loss": 0.0063, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.21520618556701032, | |
| "grad_norm": 3.866225481033325, | |
| "learning_rate": 1.9208367048794878e-05, | |
| "loss": 0.0091, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.21649484536082475, | |
| "grad_norm": 3.38619327545166, | |
| "learning_rate": 1.9190722848877683e-05, | |
| "loss": 0.0152, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.21778350515463918, | |
| "grad_norm": 4.416774749755859, | |
| "learning_rate": 1.9172892466898047e-05, | |
| "loss": 0.0559, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.2190721649484536, | |
| "grad_norm": 1.8320109844207764, | |
| "learning_rate": 1.9154876264056863e-05, | |
| "loss": 0.0093, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.22036082474226804, | |
| "grad_norm": 1.0940265655517578, | |
| "learning_rate": 1.9136674605319304e-05, | |
| "loss": 0.0036, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.22164948453608246, | |
| "grad_norm": 0.4540210962295532, | |
| "learning_rate": 1.911828785940745e-05, | |
| "loss": 0.0025, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.22293814432989692, | |
| "grad_norm": 3.493422508239746, | |
| "learning_rate": 1.9099716398792788e-05, | |
| "loss": 0.035, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.22422680412371135, | |
| "grad_norm": 2.856956958770752, | |
| "learning_rate": 1.908096059968869e-05, | |
| "loss": 0.014, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.22551546391752578, | |
| "grad_norm": 4.453478813171387, | |
| "learning_rate": 1.906202084204279e-05, | |
| "loss": 0.0255, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.2268041237113402, | |
| "grad_norm": 1.2888829708099365, | |
| "learning_rate": 1.904289750952928e-05, | |
| "loss": 0.008, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.22809278350515463, | |
| "grad_norm": 2.4916348457336426, | |
| "learning_rate": 1.9023590989541126e-05, | |
| "loss": 0.0109, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.22938144329896906, | |
| "grad_norm": 6.163361549377441, | |
| "learning_rate": 1.900410167318226e-05, | |
| "loss": 0.0609, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.23067010309278352, | |
| "grad_norm": 2.3427462577819824, | |
| "learning_rate": 1.8984429955259607e-05, | |
| "loss": 0.0297, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.23195876288659795, | |
| "grad_norm": 5.827775001525879, | |
| "learning_rate": 1.8964576234275123e-05, | |
| "loss": 0.0291, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.23195876288659795, | |
| "eval_accuracy": 0.9930486593843099, | |
| "eval_f1": 0.8852459016393442, | |
| "eval_loss": 0.026750091463327408, | |
| "eval_precision": 0.8307692307692308, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.6749, | |
| "eval_samples_per_second": 5.438, | |
| "eval_steps_per_second": 0.179, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.23324742268041238, | |
| "grad_norm": 4.49289608001709, | |
| "learning_rate": 1.894454091241771e-05, | |
| "loss": 0.0457, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.2345360824742268, | |
| "grad_norm": 6.678655624389648, | |
| "learning_rate": 1.8924324395555066e-05, | |
| "loss": 0.0432, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.23582474226804123, | |
| "grad_norm": 4.152223587036133, | |
| "learning_rate": 1.8903927093225474e-05, | |
| "loss": 0.0314, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.23711340206185566, | |
| "grad_norm": 2.2259037494659424, | |
| "learning_rate": 1.8883349418629487e-05, | |
| "loss": 0.0524, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.23840206185567012, | |
| "grad_norm": 4.7238287925720215, | |
| "learning_rate": 1.8862591788621572e-05, | |
| "loss": 0.0404, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.23969072164948454, | |
| "grad_norm": 2.322946310043335, | |
| "learning_rate": 1.8841654623701673e-05, | |
| "loss": 0.0117, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.24097938144329897, | |
| "grad_norm": 1.2298915386199951, | |
| "learning_rate": 1.8820538348006666e-05, | |
| "loss": 0.0084, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.2422680412371134, | |
| "grad_norm": 0.8999142646789551, | |
| "learning_rate": 1.8799243389301796e-05, | |
| "loss": 0.0044, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.24355670103092783, | |
| "grad_norm": 3.322568416595459, | |
| "learning_rate": 1.877777017897199e-05, | |
| "loss": 0.0408, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.24484536082474226, | |
| "grad_norm": 2.4227094650268555, | |
| "learning_rate": 1.8756119152013134e-05, | |
| "loss": 0.0187, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.24613402061855671, | |
| "grad_norm": 4.87601375579834, | |
| "learning_rate": 1.873429074702324e-05, | |
| "loss": 0.0173, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.24742268041237114, | |
| "grad_norm": 1.4209668636322021, | |
| "learning_rate": 1.8712285406193585e-05, | |
| "loss": 0.0068, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.24871134020618557, | |
| "grad_norm": 1.8267419338226318, | |
| "learning_rate": 1.8690103575299754e-05, | |
| "loss": 0.0278, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 1.07807457447052, | |
| "learning_rate": 1.866774570369257e-05, | |
| "loss": 0.008, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.25128865979381443, | |
| "grad_norm": 1.8945574760437012, | |
| "learning_rate": 1.8645212244289047e-05, | |
| "loss": 0.0227, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.25257731958762886, | |
| "grad_norm": 1.3063504695892334, | |
| "learning_rate": 1.8622503653563173e-05, | |
| "loss": 0.0213, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.2538659793814433, | |
| "grad_norm": 3.3378217220306396, | |
| "learning_rate": 1.8599620391536682e-05, | |
| "loss": 0.0144, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.2551546391752577, | |
| "grad_norm": 6.446587085723877, | |
| "learning_rate": 1.8576562921769727e-05, | |
| "loss": 0.0255, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.25644329896907214, | |
| "grad_norm": 3.609628200531006, | |
| "learning_rate": 1.8553331711351502e-05, | |
| "loss": 0.04, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.25773195876288657, | |
| "grad_norm": 6.313230991363525, | |
| "learning_rate": 1.8529927230890757e-05, | |
| "loss": 0.0366, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.25773195876288657, | |
| "eval_accuracy": 0.9915590863952334, | |
| "eval_f1": 0.864, | |
| "eval_loss": 0.026168417185544968, | |
| "eval_precision": 0.7941176470588235, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.078, | |
| "eval_samples_per_second": 5.477, | |
| "eval_steps_per_second": 0.181, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.25902061855670105, | |
| "grad_norm": 2.778740406036377, | |
| "learning_rate": 1.85063499545063e-05, | |
| "loss": 0.0176, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.2603092783505155, | |
| "grad_norm": 2.0980327129364014, | |
| "learning_rate": 1.8482600359817344e-05, | |
| "loss": 0.0269, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.2615979381443299, | |
| "grad_norm": 3.3311469554901123, | |
| "learning_rate": 1.8458678927933884e-05, | |
| "loss": 0.0108, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.26288659793814434, | |
| "grad_norm": 2.4817209243774414, | |
| "learning_rate": 1.843458614344691e-05, | |
| "loss": 0.0164, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.26417525773195877, | |
| "grad_norm": 0.9327460527420044, | |
| "learning_rate": 1.8410322494418606e-05, | |
| "loss": 0.0046, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.2654639175257732, | |
| "grad_norm": 1.553088903427124, | |
| "learning_rate": 1.8385888472372474e-05, | |
| "loss": 0.0335, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.2667525773195876, | |
| "grad_norm": 0.8573108315467834, | |
| "learning_rate": 1.8361284572283356e-05, | |
| "loss": 0.0064, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.26804123711340205, | |
| "grad_norm": 1.9521257877349854, | |
| "learning_rate": 1.833651129256742e-05, | |
| "loss": 0.0189, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.2693298969072165, | |
| "grad_norm": 0.3993712365627289, | |
| "learning_rate": 1.831156913507206e-05, | |
| "loss": 0.0037, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.2706185567010309, | |
| "grad_norm": 1.7626994848251343, | |
| "learning_rate": 1.828645860506573e-05, | |
| "loss": 0.0472, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.27190721649484534, | |
| "grad_norm": 0.2588231861591339, | |
| "learning_rate": 1.826118021122771e-05, | |
| "loss": 0.0028, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.27319587628865977, | |
| "grad_norm": 2.898406982421875, | |
| "learning_rate": 1.8235734465637794e-05, | |
| "loss": 0.0172, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.27448453608247425, | |
| "grad_norm": 1.3976062536239624, | |
| "learning_rate": 1.821012188376593e-05, | |
| "loss": 0.0238, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.2757731958762887, | |
| "grad_norm": 1.2627792358398438, | |
| "learning_rate": 1.8184342984461766e-05, | |
| "loss": 0.0079, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.2770618556701031, | |
| "grad_norm": 1.2073824405670166, | |
| "learning_rate": 1.8158398289944145e-05, | |
| "loss": 0.0052, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.27835051546391754, | |
| "grad_norm": 1.785037636756897, | |
| "learning_rate": 1.8132288325790518e-05, | |
| "loss": 0.0113, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.27963917525773196, | |
| "grad_norm": 0.8981249928474426, | |
| "learning_rate": 1.8106013620926312e-05, | |
| "loss": 0.0245, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.2809278350515464, | |
| "grad_norm": 2.7547290325164795, | |
| "learning_rate": 1.8079574707614202e-05, | |
| "loss": 0.0119, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.2822164948453608, | |
| "grad_norm": 0.7960152626037598, | |
| "learning_rate": 1.8052972121443337e-05, | |
| "loss": 0.0068, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.28350515463917525, | |
| "grad_norm": 1.9430488348007202, | |
| "learning_rate": 1.802620640131848e-05, | |
| "loss": 0.0133, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.28350515463917525, | |
| "eval_accuracy": 0.9920556107249255, | |
| "eval_f1": 0.8688524590163934, | |
| "eval_loss": 0.020570827648043633, | |
| "eval_precision": 0.8153846153846154, | |
| "eval_recall": 0.9298245614035088, | |
| "eval_runtime": 86.1412, | |
| "eval_samples_per_second": 5.282, | |
| "eval_steps_per_second": 0.174, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.2847938144329897, | |
| "grad_norm": 2.146362066268921, | |
| "learning_rate": 1.799927808944911e-05, | |
| "loss": 0.0153, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.2860824742268041, | |
| "grad_norm": 2.4061279296875, | |
| "learning_rate": 1.797218773133841e-05, | |
| "loss": 0.0176, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.28737113402061853, | |
| "grad_norm": 4.6427903175354, | |
| "learning_rate": 1.7944935875772244e-05, | |
| "loss": 0.0226, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.28865979381443296, | |
| "grad_norm": 4.384968280792236, | |
| "learning_rate": 1.7917523074808024e-05, | |
| "loss": 0.0263, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.28994845360824745, | |
| "grad_norm": 1.4952014684677124, | |
| "learning_rate": 1.7889949883763532e-05, | |
| "loss": 0.0229, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.2912371134020619, | |
| "grad_norm": 2.8541419506073, | |
| "learning_rate": 1.786221686120567e-05, | |
| "loss": 0.0323, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.2925257731958763, | |
| "grad_norm": 0.25288426876068115, | |
| "learning_rate": 1.7834324568939137e-05, | |
| "loss": 0.0025, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.29381443298969073, | |
| "grad_norm": 2.647047281265259, | |
| "learning_rate": 1.7806273571995066e-05, | |
| "loss": 0.0205, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.29510309278350516, | |
| "grad_norm": 2.2785160541534424, | |
| "learning_rate": 1.7778064438619562e-05, | |
| "loss": 0.0153, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.2963917525773196, | |
| "grad_norm": 1.089971661567688, | |
| "learning_rate": 1.7749697740262197e-05, | |
| "loss": 0.0054, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.297680412371134, | |
| "grad_norm": 1.279832363128662, | |
| "learning_rate": 1.772117405156443e-05, | |
| "loss": 0.0118, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.29896907216494845, | |
| "grad_norm": 0.7529995441436768, | |
| "learning_rate": 1.769249395034797e-05, | |
| "loss": 0.0036, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.3002577319587629, | |
| "grad_norm": 1.4826620817184448, | |
| "learning_rate": 1.7663658017603073e-05, | |
| "loss": 0.0158, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.3015463917525773, | |
| "grad_norm": 2.0365002155303955, | |
| "learning_rate": 1.7634666837476765e-05, | |
| "loss": 0.0369, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.30283505154639173, | |
| "grad_norm": 1.6364721059799194, | |
| "learning_rate": 1.7605520997261014e-05, | |
| "loss": 0.0187, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.30412371134020616, | |
| "grad_norm": 1.508259654045105, | |
| "learning_rate": 1.757622108738083e-05, | |
| "loss": 0.0117, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.30541237113402064, | |
| "grad_norm": 1.2020483016967773, | |
| "learning_rate": 1.754676770138231e-05, | |
| "loss": 0.0233, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.30670103092783507, | |
| "grad_norm": 5.561709403991699, | |
| "learning_rate": 1.7517161435920606e-05, | |
| "loss": 0.075, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.3079896907216495, | |
| "grad_norm": 2.0912508964538574, | |
| "learning_rate": 1.7487402890747843e-05, | |
| "loss": 0.0203, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.30927835051546393, | |
| "grad_norm": 1.5323127508163452, | |
| "learning_rate": 1.7457492668700967e-05, | |
| "loss": 0.0075, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.30927835051546393, | |
| "eval_accuracy": 0.9955312810327706, | |
| "eval_f1": 0.918918918918919, | |
| "eval_loss": 0.018838409334421158, | |
| "eval_precision": 0.9444444444444444, | |
| "eval_recall": 0.8947368421052632, | |
| "eval_runtime": 86.7188, | |
| "eval_samples_per_second": 5.247, | |
| "eval_steps_per_second": 0.173, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.31056701030927836, | |
| "grad_norm": 1.113881230354309, | |
| "learning_rate": 1.7427431375689544e-05, | |
| "loss": 0.0142, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.3118556701030928, | |
| "grad_norm": 0.5411668419837952, | |
| "learning_rate": 1.7397219620683465e-05, | |
| "loss": 0.0029, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.3131443298969072, | |
| "grad_norm": 1.3847646713256836, | |
| "learning_rate": 1.7366858015700626e-05, | |
| "loss": 0.0172, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.31443298969072164, | |
| "grad_norm": 1.7475923299789429, | |
| "learning_rate": 1.7336347175794523e-05, | |
| "loss": 0.0168, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.31572164948453607, | |
| "grad_norm": 1.2648131847381592, | |
| "learning_rate": 1.73056877190418e-05, | |
| "loss": 0.0089, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.3170103092783505, | |
| "grad_norm": 1.3249045610427856, | |
| "learning_rate": 1.7274880266529716e-05, | |
| "loss": 0.0096, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.31829896907216493, | |
| "grad_norm": 1.3530569076538086, | |
| "learning_rate": 1.7243925442343578e-05, | |
| "loss": 0.0307, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.31958762886597936, | |
| "grad_norm": 1.448201060295105, | |
| "learning_rate": 1.721282387355408e-05, | |
| "loss": 0.0113, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.32087628865979384, | |
| "grad_norm": 1.4428458213806152, | |
| "learning_rate": 1.718157619020462e-05, | |
| "loss": 0.009, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.32216494845360827, | |
| "grad_norm": 1.675363540649414, | |
| "learning_rate": 1.715018302529852e-05, | |
| "loss": 0.0118, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.3234536082474227, | |
| "grad_norm": 1.7131081819534302, | |
| "learning_rate": 1.711864501478622e-05, | |
| "loss": 0.0142, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.3247422680412371, | |
| "grad_norm": 2.2574496269226074, | |
| "learning_rate": 1.7086962797552376e-05, | |
| "loss": 0.0139, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.32603092783505155, | |
| "grad_norm": 0.2616090476512909, | |
| "learning_rate": 1.7055137015402935e-05, | |
| "loss": 0.0042, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.327319587628866, | |
| "grad_norm": 1.805964708328247, | |
| "learning_rate": 1.7023168313052118e-05, | |
| "loss": 0.0243, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.3286082474226804, | |
| "grad_norm": 1.5689877271652222, | |
| "learning_rate": 1.6991057338109376e-05, | |
| "loss": 0.0071, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.32989690721649484, | |
| "grad_norm": 2.198000907897949, | |
| "learning_rate": 1.6958804741066254e-05, | |
| "loss": 0.0282, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.33118556701030927, | |
| "grad_norm": 1.7233505249023438, | |
| "learning_rate": 1.6926411175283227e-05, | |
| "loss": 0.0131, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.3324742268041237, | |
| "grad_norm": 0.9903095960617065, | |
| "learning_rate": 1.689387729697646e-05, | |
| "loss": 0.0049, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.3337628865979381, | |
| "grad_norm": 1.461641550064087, | |
| "learning_rate": 1.686120376520451e-05, | |
| "loss": 0.0106, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.33505154639175255, | |
| "grad_norm": 0.4015713930130005, | |
| "learning_rate": 1.6828391241854983e-05, | |
| "loss": 0.0036, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.33505154639175255, | |
| "eval_accuracy": 0.9945382323733863, | |
| "eval_f1": 0.9026548672566371, | |
| "eval_loss": 0.01681123673915863, | |
| "eval_precision": 0.9107142857142857, | |
| "eval_recall": 0.8947368421052632, | |
| "eval_runtime": 85.6487, | |
| "eval_samples_per_second": 5.312, | |
| "eval_steps_per_second": 0.175, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.33634020618556704, | |
| "grad_norm": 0.6473265886306763, | |
| "learning_rate": 1.6795440391631122e-05, | |
| "loss": 0.0056, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.33762886597938147, | |
| "grad_norm": 2.4610116481781006, | |
| "learning_rate": 1.6762351882038342e-05, | |
| "loss": 0.0137, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.3389175257731959, | |
| "grad_norm": 2.7106072902679443, | |
| "learning_rate": 1.6729126383370696e-05, | |
| "loss": 0.0343, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.3402061855670103, | |
| "grad_norm": 0.5585967898368835, | |
| "learning_rate": 1.669576456869733e-05, | |
| "loss": 0.0038, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.34149484536082475, | |
| "grad_norm": 0.7374432682991028, | |
| "learning_rate": 1.666226711384881e-05, | |
| "loss": 0.0031, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.3427835051546392, | |
| "grad_norm": 0.5863426923751831, | |
| "learning_rate": 1.6628634697403447e-05, | |
| "loss": 0.0028, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.3440721649484536, | |
| "grad_norm": 2.8820924758911133, | |
| "learning_rate": 1.6594868000673562e-05, | |
| "loss": 0.0219, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.34536082474226804, | |
| "grad_norm": 0.26523590087890625, | |
| "learning_rate": 1.6560967707691663e-05, | |
| "loss": 0.0032, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.34664948453608246, | |
| "grad_norm": 0.7281240224838257, | |
| "learning_rate": 1.6526934505196605e-05, | |
| "loss": 0.0056, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.3479381443298969, | |
| "grad_norm": 3.0071451663970947, | |
| "learning_rate": 1.649276908261967e-05, | |
| "loss": 0.0519, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3492268041237113, | |
| "grad_norm": 1.322235107421875, | |
| "learning_rate": 1.64584721320706e-05, | |
| "loss": 0.0094, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.35051546391752575, | |
| "grad_norm": 3.1698532104492188, | |
| "learning_rate": 1.642404434832358e-05, | |
| "loss": 0.0398, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.35180412371134023, | |
| "grad_norm": 1.9093055725097656, | |
| "learning_rate": 1.6389486428803173e-05, | |
| "loss": 0.0072, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.35309278350515466, | |
| "grad_norm": 2.027344226837158, | |
| "learning_rate": 1.635479907357016e-05, | |
| "loss": 0.0398, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.3543814432989691, | |
| "grad_norm": 2.1538121700286865, | |
| "learning_rate": 1.63199829853074e-05, | |
| "loss": 0.0248, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.3556701030927835, | |
| "grad_norm": 3.6502480506896973, | |
| "learning_rate": 1.6285038869305565e-05, | |
| "loss": 0.0144, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.35695876288659795, | |
| "grad_norm": 0.3177280128002167, | |
| "learning_rate": 1.624996743344887e-05, | |
| "loss": 0.0023, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.3582474226804124, | |
| "grad_norm": 1.3766614198684692, | |
| "learning_rate": 1.621476938820071e-05, | |
| "loss": 0.0098, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.3595360824742268, | |
| "grad_norm": 4.156036376953125, | |
| "learning_rate": 1.6179445446589308e-05, | |
| "loss": 0.0161, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.36082474226804123, | |
| "grad_norm": 2.045440435409546, | |
| "learning_rate": 1.6143996324193227e-05, | |
| "loss": 0.0081, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.36082474226804123, | |
| "eval_accuracy": 0.9960278053624627, | |
| "eval_f1": 0.9310344827586207, | |
| "eval_loss": 0.01818544790148735, | |
| "eval_precision": 0.9152542372881356, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 84.8439, | |
| "eval_samples_per_second": 5.363, | |
| "eval_steps_per_second": 0.177, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.36211340206185566, | |
| "grad_norm": 0.883822500705719, | |
| "learning_rate": 1.6108422739126896e-05, | |
| "loss": 0.0056, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.3634020618556701, | |
| "grad_norm": 6.268988132476807, | |
| "learning_rate": 1.6072725412026066e-05, | |
| "loss": 0.0257, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.3646907216494845, | |
| "grad_norm": 1.6119505167007446, | |
| "learning_rate": 1.6036905066033207e-05, | |
| "loss": 0.0025, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.36597938144329895, | |
| "grad_norm": 0.13591672480106354, | |
| "learning_rate": 1.6000962426782844e-05, | |
| "loss": 0.0009, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.36726804123711343, | |
| "grad_norm": 0.9825417995452881, | |
| "learning_rate": 1.596489822238689e-05, | |
| "loss": 0.009, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.36855670103092786, | |
| "grad_norm": 0.9686275720596313, | |
| "learning_rate": 1.592871318341986e-05, | |
| "loss": 0.0039, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.3698453608247423, | |
| "grad_norm": 3.8466947078704834, | |
| "learning_rate": 1.5892408042904098e-05, | |
| "loss": 0.012, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.3711340206185567, | |
| "grad_norm": 0.6667149066925049, | |
| "learning_rate": 1.585598353629492e-05, | |
| "loss": 0.0047, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.37242268041237114, | |
| "grad_norm": 0.26038888096809387, | |
| "learning_rate": 1.58194404014657e-05, | |
| "loss": 0.0013, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.37371134020618557, | |
| "grad_norm": 4.503498554229736, | |
| "learning_rate": 1.5782779378692957e-05, | |
| "loss": 0.0276, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 3.434586763381958, | |
| "learning_rate": 1.5746001210641316e-05, | |
| "loss": 0.0202, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.37628865979381443, | |
| "grad_norm": 0.5368062853813171, | |
| "learning_rate": 1.57091066423485e-05, | |
| "loss": 0.0027, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.37757731958762886, | |
| "grad_norm": 1.4007879495620728, | |
| "learning_rate": 1.5672096421210217e-05, | |
| "loss": 0.0129, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.3788659793814433, | |
| "grad_norm": 0.07279600203037262, | |
| "learning_rate": 1.5634971296965027e-05, | |
| "loss": 0.0007, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.3801546391752577, | |
| "grad_norm": 0.13010476529598236, | |
| "learning_rate": 1.5597732021679153e-05, | |
| "loss": 0.0008, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.38144329896907214, | |
| "grad_norm": 2.1944241523742676, | |
| "learning_rate": 1.5560379349731234e-05, | |
| "loss": 0.0167, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.38273195876288657, | |
| "grad_norm": 3.2190327644348145, | |
| "learning_rate": 1.552291403779707e-05, | |
| "loss": 0.0221, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.38402061855670105, | |
| "grad_norm": 4.059307098388672, | |
| "learning_rate": 1.5485336844834274e-05, | |
| "loss": 0.0166, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.3853092783505155, | |
| "grad_norm": 0.1012328714132309, | |
| "learning_rate": 1.544764853206689e-05, | |
| "loss": 0.0005, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.3865979381443299, | |
| "grad_norm": 1.7777224779129028, | |
| "learning_rate": 1.5409849862969994e-05, | |
| "loss": 0.0155, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.3865979381443299, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.014531011693179607, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 85.8668, | |
| "eval_samples_per_second": 5.299, | |
| "eval_steps_per_second": 0.175, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.38788659793814434, | |
| "grad_norm": 1.1837537288665771, | |
| "learning_rate": 1.5371941603254215e-05, | |
| "loss": 0.0047, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.38917525773195877, | |
| "grad_norm": 0.9767211675643921, | |
| "learning_rate": 1.5333924520850227e-05, | |
| "loss": 0.0026, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.3904639175257732, | |
| "grad_norm": 0.041874054819345474, | |
| "learning_rate": 1.529579938589319e-05, | |
| "loss": 0.0002, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.3917525773195876, | |
| "grad_norm": 1.1249877214431763, | |
| "learning_rate": 1.5257566970707147e-05, | |
| "loss": 0.0044, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.39304123711340205, | |
| "grad_norm": 0.562960684299469, | |
| "learning_rate": 1.5219228049789388e-05, | |
| "loss": 0.0029, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.3943298969072165, | |
| "grad_norm": 0.6778842806816101, | |
| "learning_rate": 1.5180783399794749e-05, | |
| "loss": 0.0022, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.3956185567010309, | |
| "grad_norm": 1.7381926774978638, | |
| "learning_rate": 1.514223379951989e-05, | |
| "loss": 0.0355, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.39690721649484534, | |
| "grad_norm": 3.0561463832855225, | |
| "learning_rate": 1.5103580029887504e-05, | |
| "loss": 0.0144, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.39819587628865977, | |
| "grad_norm": 0.7643241882324219, | |
| "learning_rate": 1.5064822873930516e-05, | |
| "loss": 0.0024, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.39948453608247425, | |
| "grad_norm": 0.043119773268699646, | |
| "learning_rate": 1.5025963116776203e-05, | |
| "loss": 0.0003, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.4007731958762887, | |
| "grad_norm": 1.9049817323684692, | |
| "learning_rate": 1.49870015456303e-05, | |
| "loss": 0.01, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.4020618556701031, | |
| "grad_norm": 3.531217336654663, | |
| "learning_rate": 1.4947938949761054e-05, | |
| "loss": 0.013, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.40335051546391754, | |
| "grad_norm": 0.1271563023328781, | |
| "learning_rate": 1.490877612048322e-05, | |
| "loss": 0.0006, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.40463917525773196, | |
| "grad_norm": 0.8592988848686218, | |
| "learning_rate": 1.4869513851142051e-05, | |
| "loss": 0.0032, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.4059278350515464, | |
| "grad_norm": 0.9091683030128479, | |
| "learning_rate": 1.483015293709722e-05, | |
| "loss": 0.0039, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.4072164948453608, | |
| "grad_norm": 1.7787505388259888, | |
| "learning_rate": 1.4790694175706698e-05, | |
| "loss": 0.0162, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.40850515463917525, | |
| "grad_norm": 0.26541659235954285, | |
| "learning_rate": 1.4751138366310612e-05, | |
| "loss": 0.0012, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.4097938144329897, | |
| "grad_norm": 0.29303649067878723, | |
| "learning_rate": 1.4711486310215053e-05, | |
| "loss": 0.0028, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.4110824742268041, | |
| "grad_norm": 0.35369065403938293, | |
| "learning_rate": 1.4671738810675838e-05, | |
| "loss": 0.0021, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.41237113402061853, | |
| "grad_norm": 2.0567262172698975, | |
| "learning_rate": 1.4631896672882235e-05, | |
| "loss": 0.0075, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.41237113402061853, | |
| "eval_accuracy": 0.9975173783515392, | |
| "eval_f1": 0.9557522123893806, | |
| "eval_loss": 0.016517719253897667, | |
| "eval_precision": 0.9642857142857143, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.1704, | |
| "eval_samples_per_second": 5.471, | |
| "eval_steps_per_second": 0.18, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.41365979381443296, | |
| "grad_norm": 3.277583599090576, | |
| "learning_rate": 1.4591960703940662e-05, | |
| "loss": 0.0208, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.41494845360824745, | |
| "grad_norm": 3.1945223808288574, | |
| "learning_rate": 1.4551931712858334e-05, | |
| "loss": 0.0115, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.4162371134020619, | |
| "grad_norm": 3.821542501449585, | |
| "learning_rate": 1.4511810510526869e-05, | |
| "loss": 0.0471, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.4175257731958763, | |
| "grad_norm": 7.283623218536377, | |
| "learning_rate": 1.4471597909705858e-05, | |
| "loss": 0.0285, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.41881443298969073, | |
| "grad_norm": 1.544988751411438, | |
| "learning_rate": 1.4431294725006415e-05, | |
| "loss": 0.0028, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.42010309278350516, | |
| "grad_norm": 5.9445881843566895, | |
| "learning_rate": 1.4390901772874668e-05, | |
| "loss": 0.0244, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.4213917525773196, | |
| "grad_norm": 0.5280015468597412, | |
| "learning_rate": 1.435041987157521e-05, | |
| "loss": 0.0039, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.422680412371134, | |
| "grad_norm": 4.258032321929932, | |
| "learning_rate": 1.4309849841174538e-05, | |
| "loss": 0.0251, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.42396907216494845, | |
| "grad_norm": 2.0573043823242188, | |
| "learning_rate": 1.4269192503524435e-05, | |
| "loss": 0.0133, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.4252577319587629, | |
| "grad_norm": 0.4584275782108307, | |
| "learning_rate": 1.422844868224531e-05, | |
| "loss": 0.003, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.4265463917525773, | |
| "grad_norm": 2.158039093017578, | |
| "learning_rate": 1.4187619202709538e-05, | |
| "loss": 0.0036, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.42783505154639173, | |
| "grad_norm": 1.538727045059204, | |
| "learning_rate": 1.4146704892024714e-05, | |
| "loss": 0.0054, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.42912371134020616, | |
| "grad_norm": 1.976838231086731, | |
| "learning_rate": 1.4105706579016916e-05, | |
| "loss": 0.0192, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.43041237113402064, | |
| "grad_norm": 2.7637200355529785, | |
| "learning_rate": 1.40646250942139e-05, | |
| "loss": 0.0654, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.43170103092783507, | |
| "grad_norm": 3.6735239028930664, | |
| "learning_rate": 1.4023461269828297e-05, | |
| "loss": 0.0346, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.4329896907216495, | |
| "grad_norm": 0.587160050868988, | |
| "learning_rate": 1.3982215939740726e-05, | |
| "loss": 0.0012, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.43427835051546393, | |
| "grad_norm": 1.3197327852249146, | |
| "learning_rate": 1.3940889939482925e-05, | |
| "loss": 0.0146, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.43556701030927836, | |
| "grad_norm": 1.715801477432251, | |
| "learning_rate": 1.3899484106220816e-05, | |
| "loss": 0.0075, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.4368556701030928, | |
| "grad_norm": 0.30534547567367554, | |
| "learning_rate": 1.3857999278737546e-05, | |
| "loss": 0.0025, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.4381443298969072, | |
| "grad_norm": 0.6033890843391418, | |
| "learning_rate": 1.3816436297416496e-05, | |
| "loss": 0.0033, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.4381443298969072, | |
| "eval_accuracy": 0.9975173783515392, | |
| "eval_f1": 0.9557522123893806, | |
| "eval_loss": 0.01393709797412157, | |
| "eval_precision": 0.9642857142857143, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 84.0685, | |
| "eval_samples_per_second": 5.412, | |
| "eval_steps_per_second": 0.178, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.43943298969072164, | |
| "grad_norm": 0.35391122102737427, | |
| "learning_rate": 1.3774796004224258e-05, | |
| "loss": 0.0016, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.44072164948453607, | |
| "grad_norm": 3.3299925327301025, | |
| "learning_rate": 1.3733079242693572e-05, | |
| "loss": 0.0217, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.4420103092783505, | |
| "grad_norm": 3.1309425830841064, | |
| "learning_rate": 1.3691286857906254e-05, | |
| "loss": 0.0181, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.44329896907216493, | |
| "grad_norm": 1.195021152496338, | |
| "learning_rate": 1.3649419696476057e-05, | |
| "loss": 0.0041, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.44458762886597936, | |
| "grad_norm": 1.9298248291015625, | |
| "learning_rate": 1.3607478606531533e-05, | |
| "loss": 0.0173, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.44587628865979384, | |
| "grad_norm": 0.45631176233291626, | |
| "learning_rate": 1.356546443769885e-05, | |
| "loss": 0.0053, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.44716494845360827, | |
| "grad_norm": 3.206501007080078, | |
| "learning_rate": 1.3523378041084576e-05, | |
| "loss": 0.0157, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.4484536082474227, | |
| "grad_norm": 1.0008442401885986, | |
| "learning_rate": 1.3481220269258449e-05, | |
| "loss": 0.027, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.4497422680412371, | |
| "grad_norm": 0.5664035081863403, | |
| "learning_rate": 1.3438991976236087e-05, | |
| "loss": 0.0033, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.45103092783505155, | |
| "grad_norm": 0.9563103914260864, | |
| "learning_rate": 1.3396694017461708e-05, | |
| "loss": 0.006, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.452319587628866, | |
| "grad_norm": 0.3030118942260742, | |
| "learning_rate": 1.3354327249790786e-05, | |
| "loss": 0.0016, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.4536082474226804, | |
| "grad_norm": 2.2578694820404053, | |
| "learning_rate": 1.3311892531472705e-05, | |
| "loss": 0.0119, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.45489690721649484, | |
| "grad_norm": 2.2603442668914795, | |
| "learning_rate": 1.3269390722133358e-05, | |
| "loss": 0.0083, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.45618556701030927, | |
| "grad_norm": 0.17129381000995636, | |
| "learning_rate": 1.3226822682757745e-05, | |
| "loss": 0.0013, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.4574742268041237, | |
| "grad_norm": 2.6453981399536133, | |
| "learning_rate": 1.3184189275672532e-05, | |
| "loss": 0.0235, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.4587628865979381, | |
| "grad_norm": 0.2814205586910248, | |
| "learning_rate": 1.3141491364528576e-05, | |
| "loss": 0.0019, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.46005154639175255, | |
| "grad_norm": 1.1380839347839355, | |
| "learning_rate": 1.3098729814283426e-05, | |
| "loss": 0.0071, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.46134020618556704, | |
| "grad_norm": 1.4015003442764282, | |
| "learning_rate": 1.3055905491183822e-05, | |
| "loss": 0.009, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.46262886597938147, | |
| "grad_norm": 2.173774480819702, | |
| "learning_rate": 1.3013019262748112e-05, | |
| "loss": 0.0447, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.4639175257731959, | |
| "grad_norm": 3.0909972190856934, | |
| "learning_rate": 1.2970071997748712e-05, | |
| "loss": 0.01, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.4639175257731959, | |
| "eval_accuracy": 0.997020854021847, | |
| "eval_f1": 0.9473684210526315, | |
| "eval_loss": 0.013612103648483753, | |
| "eval_precision": 0.9473684210526315, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.2253, | |
| "eval_samples_per_second": 5.467, | |
| "eval_steps_per_second": 0.18, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.4652061855670103, | |
| "grad_norm": 1.6566423177719116, | |
| "learning_rate": 1.2927064566194493e-05, | |
| "loss": 0.0186, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.46649484536082475, | |
| "grad_norm": 0.39793241024017334, | |
| "learning_rate": 1.288399783931315e-05, | |
| "loss": 0.0023, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.4677835051546392, | |
| "grad_norm": 1.0389119386672974, | |
| "learning_rate": 1.2840872689533562e-05, | |
| "loss": 0.0138, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.4690721649484536, | |
| "grad_norm": 1.4510776996612549, | |
| "learning_rate": 1.2797689990468113e-05, | |
| "loss": 0.0334, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.47036082474226804, | |
| "grad_norm": 0.7168273329734802, | |
| "learning_rate": 1.2754450616895006e-05, | |
| "loss": 0.0031, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.47164948453608246, | |
| "grad_norm": 0.7474784255027771, | |
| "learning_rate": 1.2711155444740529e-05, | |
| "loss": 0.0041, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.4729381443298969, | |
| "grad_norm": 1.4126038551330566, | |
| "learning_rate": 1.2667805351061314e-05, | |
| "loss": 0.0077, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.4742268041237113, | |
| "grad_norm": 3.0654406547546387, | |
| "learning_rate": 1.2624401214026574e-05, | |
| "loss": 0.0206, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.47551546391752575, | |
| "grad_norm": 2.1934995651245117, | |
| "learning_rate": 1.2580943912900309e-05, | |
| "loss": 0.0232, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.47680412371134023, | |
| "grad_norm": 3.323518753051758, | |
| "learning_rate": 1.2537434328023501e-05, | |
| "loss": 0.0349, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.47809278350515466, | |
| "grad_norm": 0.4509038031101227, | |
| "learning_rate": 1.2493873340796271e-05, | |
| "loss": 0.0029, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.4793814432989691, | |
| "grad_norm": 1.1871294975280762, | |
| "learning_rate": 1.2450261833660033e-05, | |
| "loss": 0.0152, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.4806701030927835, | |
| "grad_norm": 0.5248959064483643, | |
| "learning_rate": 1.2406600690079608e-05, | |
| "loss": 0.0045, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.48195876288659795, | |
| "grad_norm": 3.3892788887023926, | |
| "learning_rate": 1.2362890794525342e-05, | |
| "loss": 0.0354, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.4832474226804124, | |
| "grad_norm": 0.7496523857116699, | |
| "learning_rate": 1.2319133032455164e-05, | |
| "loss": 0.0044, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.4845360824742268, | |
| "grad_norm": 0.684665858745575, | |
| "learning_rate": 1.2275328290296677e-05, | |
| "loss": 0.0035, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.48582474226804123, | |
| "grad_norm": 0.21576450765132904, | |
| "learning_rate": 1.2231477455429185e-05, | |
| "loss": 0.0021, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.48711340206185566, | |
| "grad_norm": 0.6430164575576782, | |
| "learning_rate": 1.2187581416165721e-05, | |
| "loss": 0.0049, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.4884020618556701, | |
| "grad_norm": 2.2819581031799316, | |
| "learning_rate": 1.2143641061735048e-05, | |
| "loss": 0.025, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.4896907216494845, | |
| "grad_norm": 0.38613247871398926, | |
| "learning_rate": 1.2099657282263651e-05, | |
| "loss": 0.0018, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.4896907216494845, | |
| "eval_accuracy": 0.997020854021847, | |
| "eval_f1": 0.9473684210526315, | |
| "eval_loss": 0.014573891647160053, | |
| "eval_precision": 0.9473684210526315, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 85.034, | |
| "eval_samples_per_second": 5.351, | |
| "eval_steps_per_second": 0.176, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.49097938144329895, | |
| "grad_norm": 0.1862691044807434, | |
| "learning_rate": 1.2055630968757696e-05, | |
| "loss": 0.0012, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.49226804123711343, | |
| "grad_norm": 1.6418074369430542, | |
| "learning_rate": 1.2011563013084996e-05, | |
| "loss": 0.0084, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.49355670103092786, | |
| "grad_norm": 1.3631302118301392, | |
| "learning_rate": 1.1967454307956933e-05, | |
| "loss": 0.0178, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.4948453608247423, | |
| "grad_norm": 1.1505086421966553, | |
| "learning_rate": 1.1923305746910372e-05, | |
| "loss": 0.0043, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.4961340206185567, | |
| "grad_norm": 1.7858433723449707, | |
| "learning_rate": 1.1879118224289563e-05, | |
| "loss": 0.008, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.49742268041237114, | |
| "grad_norm": 0.8051453232765198, | |
| "learning_rate": 1.1834892635228024e-05, | |
| "loss": 0.0115, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.49871134020618557, | |
| "grad_norm": 1.9402683973312378, | |
| "learning_rate": 1.1790629875630412e-05, | |
| "loss": 0.0128, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.4412771463394165, | |
| "learning_rate": 1.1746330842154371e-05, | |
| "loss": 0.0038, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.5012886597938144, | |
| "grad_norm": 1.3783646821975708, | |
| "learning_rate": 1.1701996432192363e-05, | |
| "loss": 0.011, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.5025773195876289, | |
| "grad_norm": 0.6311134099960327, | |
| "learning_rate": 1.1657627543853491e-05, | |
| "loss": 0.0049, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5038659793814433, | |
| "grad_norm": 1.1378337144851685, | |
| "learning_rate": 1.1613225075945316e-05, | |
| "loss": 0.0069, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.5051546391752577, | |
| "grad_norm": 0.2164492905139923, | |
| "learning_rate": 1.156878992795563e-05, | |
| "loss": 0.002, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.5064432989690721, | |
| "grad_norm": 2.2421934604644775, | |
| "learning_rate": 1.1524323000034256e-05, | |
| "loss": 0.0117, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.5077319587628866, | |
| "grad_norm": 0.8325400948524475, | |
| "learning_rate": 1.1479825192974791e-05, | |
| "loss": 0.0081, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.509020618556701, | |
| "grad_norm": 2.1096835136413574, | |
| "learning_rate": 1.1435297408196382e-05, | |
| "loss": 0.0131, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.5103092783505154, | |
| "grad_norm": 2.2667651176452637, | |
| "learning_rate": 1.1390740547725443e-05, | |
| "loss": 0.0116, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.5115979381443299, | |
| "grad_norm": 2.102983236312866, | |
| "learning_rate": 1.13461555141774e-05, | |
| "loss": 0.0294, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.5128865979381443, | |
| "grad_norm": 1.3082245588302612, | |
| "learning_rate": 1.1301543210738383e-05, | |
| "loss": 0.0073, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.5141752577319587, | |
| "grad_norm": 3.114409923553467, | |
| "learning_rate": 1.1256904541146966e-05, | |
| "loss": 0.0221, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.5154639175257731, | |
| "grad_norm": 0.09964155405759811, | |
| "learning_rate": 1.1212240409675825e-05, | |
| "loss": 0.0006, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5154639175257731, | |
| "eval_accuracy": 0.9975173783515392, | |
| "eval_f1": 0.954954954954955, | |
| "eval_loss": 0.013772569596767426, | |
| "eval_precision": 0.9814814814814815, | |
| "eval_recall": 0.9298245614035088, | |
| "eval_runtime": 83.0778, | |
| "eval_samples_per_second": 5.477, | |
| "eval_steps_per_second": 0.181, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5167525773195877, | |
| "grad_norm": 1.4406157732009888, | |
| "learning_rate": 1.1167551721113435e-05, | |
| "loss": 0.0054, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.5180412371134021, | |
| "grad_norm": 0.4032902121543884, | |
| "learning_rate": 1.1122839380745738e-05, | |
| "loss": 0.003, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.5193298969072165, | |
| "grad_norm": 0.26270851492881775, | |
| "learning_rate": 1.1078104294337806e-05, | |
| "loss": 0.0016, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.520618556701031, | |
| "grad_norm": 0.2894816994667053, | |
| "learning_rate": 1.1033347368115494e-05, | |
| "loss": 0.0009, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.5219072164948454, | |
| "grad_norm": 0.30011385679244995, | |
| "learning_rate": 1.0988569508747075e-05, | |
| "loss": 0.0016, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.5231958762886598, | |
| "grad_norm": 1.4098050594329834, | |
| "learning_rate": 1.0943771623324884e-05, | |
| "loss": 0.0304, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.5244845360824743, | |
| "grad_norm": 0.08646271377801895, | |
| "learning_rate": 1.0898954619346924e-05, | |
| "loss": 0.0009, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.5257731958762887, | |
| "grad_norm": 0.11550068110227585, | |
| "learning_rate": 1.085411940469851e-05, | |
| "loss": 0.0012, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.5270618556701031, | |
| "grad_norm": 0.7148799896240234, | |
| "learning_rate": 1.0809266887633849e-05, | |
| "loss": 0.0017, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.5283505154639175, | |
| "grad_norm": 0.25288528203964233, | |
| "learning_rate": 1.0764397976757658e-05, | |
| "loss": 0.0027, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.529639175257732, | |
| "grad_norm": 2.467442274093628, | |
| "learning_rate": 1.0719513581006751e-05, | |
| "loss": 0.0226, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.5309278350515464, | |
| "grad_norm": 0.13029828667640686, | |
| "learning_rate": 1.0674614609631634e-05, | |
| "loss": 0.0005, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.5322164948453608, | |
| "grad_norm": 1.6410340070724487, | |
| "learning_rate": 1.062970197217808e-05, | |
| "loss": 0.016, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.5335051546391752, | |
| "grad_norm": 1.967556118965149, | |
| "learning_rate": 1.0584776578468698e-05, | |
| "loss": 0.0079, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.5347938144329897, | |
| "grad_norm": 0.3348277509212494, | |
| "learning_rate": 1.0539839338584509e-05, | |
| "loss": 0.0022, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.5360824742268041, | |
| "grad_norm": 1.085183024406433, | |
| "learning_rate": 1.0494891162846515e-05, | |
| "loss": 0.0024, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.5373711340206185, | |
| "grad_norm": 0.23125191032886505, | |
| "learning_rate": 1.0449932961797249e-05, | |
| "loss": 0.0014, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.538659793814433, | |
| "grad_norm": 4.222436904907227, | |
| "learning_rate": 1.040496564618233e-05, | |
| "loss": 0.0356, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.5399484536082474, | |
| "grad_norm": 2.120659112930298, | |
| "learning_rate": 1.0359990126932022e-05, | |
| "loss": 0.0152, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.5412371134020618, | |
| "grad_norm": 0.6565424203872681, | |
| "learning_rate": 1.0315007315142772e-05, | |
| "loss": 0.003, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.5412371134020618, | |
| "eval_accuracy": 0.9965243296921549, | |
| "eval_f1": 0.9391304347826087, | |
| "eval_loss": 0.013511905446648598, | |
| "eval_precision": 0.9310344827586207, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 85.5974, | |
| "eval_samples_per_second": 5.316, | |
| "eval_steps_per_second": 0.175, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.5425257731958762, | |
| "grad_norm": 2.273611545562744, | |
| "learning_rate": 1.0270018122058753e-05, | |
| "loss": 0.013, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.5438144329896907, | |
| "grad_norm": 0.7861140370368958, | |
| "learning_rate": 1.0225023459053416e-05, | |
| "loss": 0.0029, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.5451030927835051, | |
| "grad_norm": 2.014556407928467, | |
| "learning_rate": 1.018002423761101e-05, | |
| "loss": 0.0503, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.5463917525773195, | |
| "grad_norm": 0.169442817568779, | |
| "learning_rate": 1.0135021369308138e-05, | |
| "loss": 0.001, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.5476804123711341, | |
| "grad_norm": 0.9103741645812988, | |
| "learning_rate": 1.0090015765795265e-05, | |
| "loss": 0.0037, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.5489690721649485, | |
| "grad_norm": 1.2520557641983032, | |
| "learning_rate": 1.004500833877828e-05, | |
| "loss": 0.0356, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.5502577319587629, | |
| "grad_norm": 2.632268190383911, | |
| "learning_rate": 1e-05, | |
| "loss": 0.0104, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.5515463917525774, | |
| "grad_norm": 2.597231388092041, | |
| "learning_rate": 9.954991661221724e-06, | |
| "loss": 0.0117, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.5528350515463918, | |
| "grad_norm": 1.045973300933838, | |
| "learning_rate": 9.909984234204738e-06, | |
| "loss": 0.0026, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.5541237113402062, | |
| "grad_norm": 0.3830992877483368, | |
| "learning_rate": 9.864978630691865e-06, | |
| "loss": 0.0025, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.5554123711340206, | |
| "grad_norm": 0.4651045799255371, | |
| "learning_rate": 9.819975762388993e-06, | |
| "loss": 0.0026, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.5567010309278351, | |
| "grad_norm": 1.1568673849105835, | |
| "learning_rate": 9.774976540946589e-06, | |
| "loss": 0.0222, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.5579896907216495, | |
| "grad_norm": 1.9485609531402588, | |
| "learning_rate": 9.729981877941249e-06, | |
| "loss": 0.0284, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.5592783505154639, | |
| "grad_norm": 1.6155321598052979, | |
| "learning_rate": 9.684992684857232e-06, | |
| "loss": 0.0056, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.5605670103092784, | |
| "grad_norm": 1.5538249015808105, | |
| "learning_rate": 9.640009873067981e-06, | |
| "loss": 0.0197, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.5618556701030928, | |
| "grad_norm": 0.2794590890407562, | |
| "learning_rate": 9.595034353817673e-06, | |
| "loss": 0.001, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.5631443298969072, | |
| "grad_norm": 1.1001949310302734, | |
| "learning_rate": 9.550067038202756e-06, | |
| "loss": 0.0031, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.5644329896907216, | |
| "grad_norm": 1.667214035987854, | |
| "learning_rate": 9.505108837153489e-06, | |
| "loss": 0.0169, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.5657216494845361, | |
| "grad_norm": 0.4470076560974121, | |
| "learning_rate": 9.460160661415496e-06, | |
| "loss": 0.0026, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.5670103092783505, | |
| "grad_norm": 0.4521142542362213, | |
| "learning_rate": 9.415223421531308e-06, | |
| "loss": 0.0035, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.5670103092783505, | |
| "eval_accuracy": 0.9965243296921549, | |
| "eval_f1": 0.9357798165137615, | |
| "eval_loss": 0.014128734357655048, | |
| "eval_precision": 0.9807692307692307, | |
| "eval_recall": 0.8947368421052632, | |
| "eval_runtime": 87.0117, | |
| "eval_samples_per_second": 5.229, | |
| "eval_steps_per_second": 0.172, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.5682989690721649, | |
| "grad_norm": 0.24049897491931915, | |
| "learning_rate": 9.370298027821924e-06, | |
| "loss": 0.0011, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.5695876288659794, | |
| "grad_norm": 1.0193203687667847, | |
| "learning_rate": 9.325385390368367e-06, | |
| "loss": 0.0107, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.5708762886597938, | |
| "grad_norm": 0.09646586328744888, | |
| "learning_rate": 9.280486418993254e-06, | |
| "loss": 0.0009, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.5721649484536082, | |
| "grad_norm": 4.3729047775268555, | |
| "learning_rate": 9.23560202324235e-06, | |
| "loss": 0.0321, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.5734536082474226, | |
| "grad_norm": 0.5603130459785461, | |
| "learning_rate": 9.190733112366158e-06, | |
| "loss": 0.0033, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.5747422680412371, | |
| "grad_norm": 1.4428563117980957, | |
| "learning_rate": 9.145880595301495e-06, | |
| "loss": 0.0119, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.5760309278350515, | |
| "grad_norm": 1.241681694984436, | |
| "learning_rate": 9.101045380653076e-06, | |
| "loss": 0.0046, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.5773195876288659, | |
| "grad_norm": 1.7557916641235352, | |
| "learning_rate": 9.056228376675118e-06, | |
| "loss": 0.0186, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.5786082474226805, | |
| "grad_norm": 0.7375532984733582, | |
| "learning_rate": 9.011430491252924e-06, | |
| "loss": 0.0027, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.5798969072164949, | |
| "grad_norm": 1.8209916353225708, | |
| "learning_rate": 8.966652631884506e-06, | |
| "loss": 0.0122, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.5811855670103093, | |
| "grad_norm": 0.12816821038722992, | |
| "learning_rate": 8.921895705662194e-06, | |
| "loss": 0.0008, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.5824742268041238, | |
| "grad_norm": 0.6932976245880127, | |
| "learning_rate": 8.877160619254264e-06, | |
| "loss": 0.0026, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.5837628865979382, | |
| "grad_norm": 0.6997652053833008, | |
| "learning_rate": 8.832448278886567e-06, | |
| "loss": 0.0058, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.5850515463917526, | |
| "grad_norm": 0.24016019701957703, | |
| "learning_rate": 8.787759590324177e-06, | |
| "loss": 0.0011, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.586340206185567, | |
| "grad_norm": 0.17693662643432617, | |
| "learning_rate": 8.743095458853034e-06, | |
| "loss": 0.0016, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.5876288659793815, | |
| "grad_norm": 0.9566256403923035, | |
| "learning_rate": 8.698456789261617e-06, | |
| "loss": 0.0055, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.5889175257731959, | |
| "grad_norm": 0.15974543988704681, | |
| "learning_rate": 8.653844485822603e-06, | |
| "loss": 0.0008, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.5902061855670103, | |
| "grad_norm": 1.2823374271392822, | |
| "learning_rate": 8.609259452274559e-06, | |
| "loss": 0.0029, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.5914948453608248, | |
| "grad_norm": 3.422497510910034, | |
| "learning_rate": 8.56470259180362e-06, | |
| "loss": 0.0382, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.5927835051546392, | |
| "grad_norm": 0.6285749077796936, | |
| "learning_rate": 8.52017480702521e-06, | |
| "loss": 0.0024, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5927835051546392, | |
| "eval_accuracy": 0.9965243296921549, | |
| "eval_f1": 0.9357798165137615, | |
| "eval_loss": 0.01483297348022461, | |
| "eval_precision": 0.9807692307692307, | |
| "eval_recall": 0.8947368421052632, | |
| "eval_runtime": 85.4893, | |
| "eval_samples_per_second": 5.322, | |
| "eval_steps_per_second": 0.175, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5940721649484536, | |
| "grad_norm": 2.1351022720336914, | |
| "learning_rate": 8.475676999965747e-06, | |
| "loss": 0.0109, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.595360824742268, | |
| "grad_norm": 0.22357627749443054, | |
| "learning_rate": 8.431210072044371e-06, | |
| "loss": 0.0009, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.5966494845360825, | |
| "grad_norm": 1.897233247756958, | |
| "learning_rate": 8.386774924054686e-06, | |
| "loss": 0.0064, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.5979381443298969, | |
| "grad_norm": 1.4000173807144165, | |
| "learning_rate": 8.342372456146512e-06, | |
| "loss": 0.008, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.5992268041237113, | |
| "grad_norm": 1.9625613689422607, | |
| "learning_rate": 8.29800356780764e-06, | |
| "loss": 0.0096, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.6005154639175257, | |
| "grad_norm": 0.664243221282959, | |
| "learning_rate": 8.253669157845632e-06, | |
| "loss": 0.0038, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.6018041237113402, | |
| "grad_norm": 2.411156177520752, | |
| "learning_rate": 8.20937012436959e-06, | |
| "loss": 0.0132, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.6030927835051546, | |
| "grad_norm": 0.44838207960128784, | |
| "learning_rate": 8.165107364771979e-06, | |
| "loss": 0.0014, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.604381443298969, | |
| "grad_norm": 1.7571401596069336, | |
| "learning_rate": 8.12088177571044e-06, | |
| "loss": 0.012, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.6056701030927835, | |
| "grad_norm": 0.4941975176334381, | |
| "learning_rate": 8.076694253089632e-06, | |
| "loss": 0.0026, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.6069587628865979, | |
| "grad_norm": 0.18339809775352478, | |
| "learning_rate": 8.032545692043068e-06, | |
| "loss": 0.0006, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.6082474226804123, | |
| "grad_norm": 0.23761658370494843, | |
| "learning_rate": 7.988436986915005e-06, | |
| "loss": 0.0016, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.6095360824742269, | |
| "grad_norm": 1.723071575164795, | |
| "learning_rate": 7.944369031242307e-06, | |
| "loss": 0.0367, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.6108247422680413, | |
| "grad_norm": 0.24809041619300842, | |
| "learning_rate": 7.900342717736354e-06, | |
| "loss": 0.0023, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.6121134020618557, | |
| "grad_norm": 0.11259906738996506, | |
| "learning_rate": 7.856358938264953e-06, | |
| "loss": 0.0005, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.6134020618556701, | |
| "grad_norm": 3.289266586303711, | |
| "learning_rate": 7.812418583834282e-06, | |
| "loss": 0.0119, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.6146907216494846, | |
| "grad_norm": 0.9513089656829834, | |
| "learning_rate": 7.768522544570818e-06, | |
| "loss": 0.0035, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.615979381443299, | |
| "grad_norm": 1.184780478477478, | |
| "learning_rate": 7.724671709703328e-06, | |
| "loss": 0.0115, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.6172680412371134, | |
| "grad_norm": 0.5122628211975098, | |
| "learning_rate": 7.680866967544841e-06, | |
| "loss": 0.0015, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.6185567010309279, | |
| "grad_norm": 2.2791271209716797, | |
| "learning_rate": 7.637109205474665e-06, | |
| "loss": 0.0203, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.6185567010309279, | |
| "eval_accuracy": 0.997020854021847, | |
| "eval_f1": 0.9473684210526315, | |
| "eval_loss": 0.013630267232656479, | |
| "eval_precision": 0.9473684210526315, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 85.2359, | |
| "eval_samples_per_second": 5.338, | |
| "eval_steps_per_second": 0.176, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.6198453608247423, | |
| "grad_norm": 4.049417495727539, | |
| "learning_rate": 7.5933993099203936e-06, | |
| "loss": 0.0196, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.6211340206185567, | |
| "grad_norm": 1.3563647270202637, | |
| "learning_rate": 7.5497381663399716e-06, | |
| "loss": 0.0149, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.6224226804123711, | |
| "grad_norm": 0.8095941543579102, | |
| "learning_rate": 7.506126659203733e-06, | |
| "loss": 0.0055, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.6237113402061856, | |
| "grad_norm": 1.3925622701644897, | |
| "learning_rate": 7.462565671976504e-06, | |
| "loss": 0.0051, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 0.6791156530380249, | |
| "learning_rate": 7.419056087099695e-06, | |
| "loss": 0.005, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.6262886597938144, | |
| "grad_norm": 0.4225018322467804, | |
| "learning_rate": 7.375598785973429e-06, | |
| "loss": 0.0021, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.6275773195876289, | |
| "grad_norm": 1.5622189044952393, | |
| "learning_rate": 7.332194648938689e-06, | |
| "loss": 0.0087, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.6288659793814433, | |
| "grad_norm": 1.0303385257720947, | |
| "learning_rate": 7.288844555259471e-06, | |
| "loss": 0.0041, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.6301546391752577, | |
| "grad_norm": 0.8188164234161377, | |
| "learning_rate": 7.245549383104993e-06, | |
| "loss": 0.003, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.6314432989690721, | |
| "grad_norm": 0.49154236912727356, | |
| "learning_rate": 7.202310009531886e-06, | |
| "loss": 0.0033, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.6327319587628866, | |
| "grad_norm": 2.7728259563446045, | |
| "learning_rate": 7.159127310466441e-06, | |
| "loss": 0.0136, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.634020618556701, | |
| "grad_norm": 0.9476536512374878, | |
| "learning_rate": 7.116002160686851e-06, | |
| "loss": 0.0069, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.6353092783505154, | |
| "grad_norm": 1.1846680641174316, | |
| "learning_rate": 7.072935433805508e-06, | |
| "loss": 0.0097, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.6365979381443299, | |
| "grad_norm": 2.1318306922912598, | |
| "learning_rate": 7.0299280022512875e-06, | |
| "loss": 0.0088, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.6378865979381443, | |
| "grad_norm": 0.6336405277252197, | |
| "learning_rate": 6.986980737251889e-06, | |
| "loss": 0.0021, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.6391752577319587, | |
| "grad_norm": 0.39115601778030396, | |
| "learning_rate": 6.944094508816182e-06, | |
| "loss": 0.0047, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.6404639175257731, | |
| "grad_norm": 0.08056578785181046, | |
| "learning_rate": 6.9012701857165755e-06, | |
| "loss": 0.0003, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.6417525773195877, | |
| "grad_norm": 1.444043517112732, | |
| "learning_rate": 6.858508635471428e-06, | |
| "loss": 0.0032, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.6430412371134021, | |
| "grad_norm": 1.2117236852645874, | |
| "learning_rate": 6.815810724327469e-06, | |
| "loss": 0.0046, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.6443298969072165, | |
| "grad_norm": 3.499089241027832, | |
| "learning_rate": 6.773177317242257e-06, | |
| "loss": 0.0293, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.6443298969072165, | |
| "eval_accuracy": 0.997020854021847, | |
| "eval_f1": 0.9454545454545454, | |
| "eval_loss": 0.016382448375225067, | |
| "eval_precision": 0.9811320754716981, | |
| "eval_recall": 0.9122807017543859, | |
| "eval_runtime": 85.2826, | |
| "eval_samples_per_second": 5.335, | |
| "eval_steps_per_second": 0.176, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.645618556701031, | |
| "grad_norm": 0.2069253921508789, | |
| "learning_rate": 6.730609277866644e-06, | |
| "loss": 0.0008, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.6469072164948454, | |
| "grad_norm": 0.7460423111915588, | |
| "learning_rate": 6.688107468527297e-06, | |
| "loss": 0.0043, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.6481958762886598, | |
| "grad_norm": 0.1024048775434494, | |
| "learning_rate": 6.645672750209216e-06, | |
| "loss": 0.0008, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.6494845360824743, | |
| "grad_norm": 3.0588488578796387, | |
| "learning_rate": 6.603305982538295e-06, | |
| "loss": 0.0153, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.6507731958762887, | |
| "grad_norm": 0.7051161527633667, | |
| "learning_rate": 6.561008023763915e-06, | |
| "loss": 0.0014, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.6520618556701031, | |
| "grad_norm": 1.1912554502487183, | |
| "learning_rate": 6.518779730741555e-06, | |
| "loss": 0.0085, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.6533505154639175, | |
| "grad_norm": 0.12031827121973038, | |
| "learning_rate": 6.476621958915426e-06, | |
| "loss": 0.0008, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.654639175257732, | |
| "grad_norm": 0.19412291049957275, | |
| "learning_rate": 6.434535562301153e-06, | |
| "loss": 0.0005, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.6559278350515464, | |
| "grad_norm": 1.5401062965393066, | |
| "learning_rate": 6.392521393468471e-06, | |
| "loss": 0.003, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.6572164948453608, | |
| "grad_norm": 1.881408929824829, | |
| "learning_rate": 6.350580303523947e-06, | |
| "loss": 0.0053, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.6585051546391752, | |
| "grad_norm": 1.8034359216690063, | |
| "learning_rate": 6.308713142093749e-06, | |
| "loss": 0.0255, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.6597938144329897, | |
| "grad_norm": 2.4420573711395264, | |
| "learning_rate": 6.266920757306429e-06, | |
| "loss": 0.0248, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.6610824742268041, | |
| "grad_norm": 2.5794460773468018, | |
| "learning_rate": 6.225203995775746e-06, | |
| "loss": 0.0085, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.6623711340206185, | |
| "grad_norm": 0.4044947922229767, | |
| "learning_rate": 6.183563702583506e-06, | |
| "loss": 0.0016, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.663659793814433, | |
| "grad_norm": 0.47102847695350647, | |
| "learning_rate": 6.1420007212624584e-06, | |
| "loss": 0.0011, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.6649484536082474, | |
| "grad_norm": 0.08942103385925293, | |
| "learning_rate": 6.100515893779188e-06, | |
| "loss": 0.0006, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.6662371134020618, | |
| "grad_norm": 1.772922158241272, | |
| "learning_rate": 6.05911006051708e-06, | |
| "loss": 0.0078, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.6675257731958762, | |
| "grad_norm": 1.0783969163894653, | |
| "learning_rate": 6.01778406025928e-06, | |
| "loss": 0.0041, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.6688144329896907, | |
| "grad_norm": 1.0612661838531494, | |
| "learning_rate": 5.976538730171708e-06, | |
| "loss": 0.0064, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.6701030927835051, | |
| "grad_norm": 3.4996323585510254, | |
| "learning_rate": 5.935374905786102e-06, | |
| "loss": 0.0078, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.6701030927835051, | |
| "eval_accuracy": 0.997020854021847, | |
| "eval_f1": 0.9473684210526315, | |
| "eval_loss": 0.014883686788380146, | |
| "eval_precision": 0.9473684210526315, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.6623, | |
| "eval_samples_per_second": 5.439, | |
| "eval_steps_per_second": 0.179, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.6713917525773195, | |
| "grad_norm": 4.155304431915283, | |
| "learning_rate": 5.89429342098309e-06, | |
| "loss": 0.0209, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.6726804123711341, | |
| "grad_norm": 1.2153571844100952, | |
| "learning_rate": 5.8532951079752895e-06, | |
| "loss": 0.0045, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.6739690721649485, | |
| "grad_norm": 0.23559901118278503, | |
| "learning_rate": 5.812380797290465e-06, | |
| "loss": 0.0011, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.6752577319587629, | |
| "grad_norm": 0.7906387448310852, | |
| "learning_rate": 5.771551317754691e-06, | |
| "loss": 0.0016, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.6765463917525774, | |
| "grad_norm": 0.6385716199874878, | |
| "learning_rate": 5.730807496475568e-06, | |
| "loss": 0.0028, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.6778350515463918, | |
| "grad_norm": 0.6003400087356567, | |
| "learning_rate": 5.690150158825462e-06, | |
| "loss": 0.0017, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.6791237113402062, | |
| "grad_norm": 0.0828804075717926, | |
| "learning_rate": 5.649580128424792e-06, | |
| "loss": 0.0013, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.6804123711340206, | |
| "grad_norm": 0.4008159637451172, | |
| "learning_rate": 5.609098227125334e-06, | |
| "loss": 0.0012, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.6817010309278351, | |
| "grad_norm": 1.0619193315505981, | |
| "learning_rate": 5.568705274993584e-06, | |
| "loss": 0.0026, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.6829896907216495, | |
| "grad_norm": 0.24862630665302277, | |
| "learning_rate": 5.528402090294142e-06, | |
| "loss": 0.0013, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.6842783505154639, | |
| "grad_norm": 2.5749034881591797, | |
| "learning_rate": 5.488189489473133e-06, | |
| "loss": 0.0279, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.6855670103092784, | |
| "grad_norm": 0.7257778644561768, | |
| "learning_rate": 5.448068287141663e-06, | |
| "loss": 0.0029, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.6868556701030928, | |
| "grad_norm": 0.047684453427791595, | |
| "learning_rate": 5.4080392960593355e-06, | |
| "loss": 0.0003, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.6881443298969072, | |
| "grad_norm": 2.056273937225342, | |
| "learning_rate": 5.368103327117768e-06, | |
| "loss": 0.017, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.6894329896907216, | |
| "grad_norm": 2.3359081745147705, | |
| "learning_rate": 5.328261189324166e-06, | |
| "loss": 0.0234, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.6907216494845361, | |
| "grad_norm": 0.4394116699695587, | |
| "learning_rate": 5.288513689784951e-06, | |
| "loss": 0.0013, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.6920103092783505, | |
| "grad_norm": 0.6693306565284729, | |
| "learning_rate": 5.2488616336893915e-06, | |
| "loss": 0.0016, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.6932989690721649, | |
| "grad_norm": 1.8861000537872314, | |
| "learning_rate": 5.209305824293307e-06, | |
| "loss": 0.0068, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.6945876288659794, | |
| "grad_norm": 1.8001142740249634, | |
| "learning_rate": 5.1698470629027845e-06, | |
| "loss": 0.0163, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.6958762886597938, | |
| "grad_norm": 1.7382943630218506, | |
| "learning_rate": 5.130486148857952e-06, | |
| "loss": 0.0291, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.6958762886597938, | |
| "eval_accuracy": 0.9975173783515392, | |
| "eval_f1": 0.9557522123893806, | |
| "eval_loss": 0.014678677543997765, | |
| "eval_precision": 0.9642857142857143, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.1931, | |
| "eval_samples_per_second": 5.469, | |
| "eval_steps_per_second": 0.18, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.6971649484536082, | |
| "grad_norm": 1.584878921508789, | |
| "learning_rate": 5.0912238795167845e-06, | |
| "loss": 0.0218, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.6984536082474226, | |
| "grad_norm": 0.42845746874809265, | |
| "learning_rate": 5.05206105023895e-06, | |
| "loss": 0.0022, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.6997422680412371, | |
| "grad_norm": 3.7735812664031982, | |
| "learning_rate": 5.012998454369701e-06, | |
| "loss": 0.0381, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.7010309278350515, | |
| "grad_norm": 1.2307202816009521, | |
| "learning_rate": 4.974036883223798e-06, | |
| "loss": 0.0065, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.7023195876288659, | |
| "grad_norm": 0.17536193132400513, | |
| "learning_rate": 4.935177126069485e-06, | |
| "loss": 0.0006, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.7036082474226805, | |
| "grad_norm": 1.0268815755844116, | |
| "learning_rate": 4.896419970112499e-06, | |
| "loss": 0.0072, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.7048969072164949, | |
| "grad_norm": 0.6802368760108948, | |
| "learning_rate": 4.857766200480115e-06, | |
| "loss": 0.0018, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.7061855670103093, | |
| "grad_norm": 0.3945528268814087, | |
| "learning_rate": 4.819216600205254e-06, | |
| "loss": 0.0019, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.7074742268041238, | |
| "grad_norm": 1.625831127166748, | |
| "learning_rate": 4.780771950210616e-06, | |
| "loss": 0.0087, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.7087628865979382, | |
| "grad_norm": 1.2541440725326538, | |
| "learning_rate": 4.742433029292856e-06, | |
| "loss": 0.0031, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.7100515463917526, | |
| "grad_norm": 0.1559123545885086, | |
| "learning_rate": 4.704200614106813e-06, | |
| "loss": 0.0011, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.711340206185567, | |
| "grad_norm": 1.7901103496551514, | |
| "learning_rate": 4.6660754791497755e-06, | |
| "loss": 0.008, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.7126288659793815, | |
| "grad_norm": 2.5184080600738525, | |
| "learning_rate": 4.628058396745787e-06, | |
| "loss": 0.0062, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.7139175257731959, | |
| "grad_norm": 1.7694346904754639, | |
| "learning_rate": 4.590150137030009e-06, | |
| "loss": 0.0067, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.7152061855670103, | |
| "grad_norm": 0.8861400485038757, | |
| "learning_rate": 4.552351467933115e-06, | |
| "loss": 0.0027, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.7164948453608248, | |
| "grad_norm": 1.867985725402832, | |
| "learning_rate": 4.514663155165731e-06, | |
| "loss": 0.0076, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.7177835051546392, | |
| "grad_norm": 0.6749151945114136, | |
| "learning_rate": 4.477085962202931e-06, | |
| "loss": 0.0026, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.7190721649484536, | |
| "grad_norm": 0.47943294048309326, | |
| "learning_rate": 4.439620650268771e-06, | |
| "loss": 0.0014, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.720360824742268, | |
| "grad_norm": 0.7041411399841309, | |
| "learning_rate": 4.402267978320854e-06, | |
| "loss": 0.0024, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.7216494845360825, | |
| "grad_norm": 2.199207067489624, | |
| "learning_rate": 4.365028703034976e-06, | |
| "loss": 0.0119, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.7216494845360825, | |
| "eval_accuracy": 0.997020854021847, | |
| "eval_f1": 0.9473684210526315, | |
| "eval_loss": 0.013599889352917671, | |
| "eval_precision": 0.9473684210526315, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 84.0133, | |
| "eval_samples_per_second": 5.416, | |
| "eval_steps_per_second": 0.179, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.7229381443298969, | |
| "grad_norm": 2.070563554763794, | |
| "learning_rate": 4.327903578789785e-06, | |
| "loss": 0.0297, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.7242268041237113, | |
| "grad_norm": 0.7736282348632812, | |
| "learning_rate": 4.290893357651502e-06, | |
| "loss": 0.0014, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.7255154639175257, | |
| "grad_norm": 1.5043028593063354, | |
| "learning_rate": 4.253998789358683e-06, | |
| "loss": 0.0121, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.7268041237113402, | |
| "grad_norm": 2.060772180557251, | |
| "learning_rate": 4.217220621307043e-06, | |
| "loss": 0.0076, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.7280927835051546, | |
| "grad_norm": 0.3604072034358978, | |
| "learning_rate": 4.180559598534297e-06, | |
| "loss": 0.0012, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.729381443298969, | |
| "grad_norm": 1.343294382095337, | |
| "learning_rate": 4.144016463705081e-06, | |
| "loss": 0.0043, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.7306701030927835, | |
| "grad_norm": 1.6220418214797974, | |
| "learning_rate": 4.107591957095903e-06, | |
| "loss": 0.0232, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.7319587628865979, | |
| "grad_norm": 0.1144634336233139, | |
| "learning_rate": 4.071286816580142e-06, | |
| "loss": 0.001, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.7332474226804123, | |
| "grad_norm": 0.7282357811927795, | |
| "learning_rate": 4.035101777613113e-06, | |
| "loss": 0.0018, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.7345360824742269, | |
| "grad_norm": 3.7139105796813965, | |
| "learning_rate": 3.999037573217157e-06, | |
| "loss": 0.0249, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.7358247422680413, | |
| "grad_norm": 2.923678159713745, | |
| "learning_rate": 3.963094933966797e-06, | |
| "loss": 0.0207, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.7371134020618557, | |
| "grad_norm": 0.20384177565574646, | |
| "learning_rate": 3.927274587973935e-06, | |
| "loss": 0.0006, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.7384020618556701, | |
| "grad_norm": 0.675167977809906, | |
| "learning_rate": 3.8915772608731066e-06, | |
| "loss": 0.0014, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.7396907216494846, | |
| "grad_norm": 3.719918727874756, | |
| "learning_rate": 3.856003675806777e-06, | |
| "loss": 0.0142, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.740979381443299, | |
| "grad_norm": 0.10527591407299042, | |
| "learning_rate": 3.820554553410693e-06, | |
| "loss": 0.0006, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.7422680412371134, | |
| "grad_norm": 2.821239948272705, | |
| "learning_rate": 3.78523061179929e-06, | |
| "loss": 0.0142, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.7435567010309279, | |
| "grad_norm": 0.972940981388092, | |
| "learning_rate": 3.7500325665511337e-06, | |
| "loss": 0.0039, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.7448453608247423, | |
| "grad_norm": 0.08761493116617203, | |
| "learning_rate": 3.7149611306944356e-06, | |
| "loss": 0.0008, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.7461340206185567, | |
| "grad_norm": 1.4573255777359009, | |
| "learning_rate": 3.680017014692604e-06, | |
| "loss": 0.0063, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.7474226804123711, | |
| "grad_norm": 0.7035017609596252, | |
| "learning_rate": 3.645200926429844e-06, | |
| "loss": 0.002, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.7474226804123711, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.0138359684497118, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.1039, | |
| "eval_samples_per_second": 5.475, | |
| "eval_steps_per_second": 0.18, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.7487113402061856, | |
| "grad_norm": 0.15962012112140656, | |
| "learning_rate": 3.610513571196832e-06, | |
| "loss": 0.0007, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.1321505457162857, | |
| "learning_rate": 3.5759556516764205e-06, | |
| "loss": 0.0008, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.7512886597938144, | |
| "grad_norm": 0.07236671447753906, | |
| "learning_rate": 3.541527867929403e-06, | |
| "loss": 0.0004, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.7525773195876289, | |
| "grad_norm": 0.5433388352394104, | |
| "learning_rate": 3.507230917380332e-06, | |
| "loss": 0.0017, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.7538659793814433, | |
| "grad_norm": 3.733372449874878, | |
| "learning_rate": 3.4730654948033957e-06, | |
| "loss": 0.0024, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.7551546391752577, | |
| "grad_norm": 4.992002010345459, | |
| "learning_rate": 3.4390322923083385e-06, | |
| "loss": 0.0182, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.7564432989690721, | |
| "grad_norm": 0.3195149302482605, | |
| "learning_rate": 3.4051319993264397e-06, | |
| "loss": 0.0009, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.7577319587628866, | |
| "grad_norm": 0.13421054184436798, | |
| "learning_rate": 3.3713653025965544e-06, | |
| "loss": 0.0008, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.759020618556701, | |
| "grad_norm": 0.12820366024971008, | |
| "learning_rate": 3.3377328861511927e-06, | |
| "loss": 0.0005, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.7603092783505154, | |
| "grad_norm": 0.8797692060470581, | |
| "learning_rate": 3.3042354313026702e-06, | |
| "loss": 0.004, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.7615979381443299, | |
| "grad_norm": 0.11013241112232208, | |
| "learning_rate": 3.2708736166293064e-06, | |
| "loss": 0.0004, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.7628865979381443, | |
| "grad_norm": 0.743179202079773, | |
| "learning_rate": 3.237648117961665e-06, | |
| "loss": 0.0013, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.7641752577319587, | |
| "grad_norm": 4.201583385467529, | |
| "learning_rate": 3.2045596083688814e-06, | |
| "loss": 0.0186, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.7654639175257731, | |
| "grad_norm": 1.279377818107605, | |
| "learning_rate": 3.1716087581450193e-06, | |
| "loss": 0.0031, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.7667525773195877, | |
| "grad_norm": 2.0569307804107666, | |
| "learning_rate": 3.1387962347954936e-06, | |
| "loss": 0.009, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.7680412371134021, | |
| "grad_norm": 0.5968018174171448, | |
| "learning_rate": 3.1061227030235442e-06, | |
| "loss": 0.0011, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.7693298969072165, | |
| "grad_norm": 2.1051347255706787, | |
| "learning_rate": 3.073588824716777e-06, | |
| "loss": 0.005, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.770618556701031, | |
| "grad_norm": 0.3999452292919159, | |
| "learning_rate": 3.041195258933749e-06, | |
| "loss": 0.0024, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.7719072164948454, | |
| "grad_norm": 1.3566821813583374, | |
| "learning_rate": 3.008942661890627e-06, | |
| "loss": 0.0027, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.7731958762886598, | |
| "grad_norm": 0.30999279022216797, | |
| "learning_rate": 2.976831686947884e-06, | |
| "loss": 0.0009, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.7731958762886598, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.013978274539113045, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.5538, | |
| "eval_samples_per_second": 5.446, | |
| "eval_steps_per_second": 0.18, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.7744845360824743, | |
| "grad_norm": 0.12384779006242752, | |
| "learning_rate": 2.944862984597068e-06, | |
| "loss": 0.0006, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.7757731958762887, | |
| "grad_norm": 0.06338214129209518, | |
| "learning_rate": 2.913037202447625e-06, | |
| "loss": 0.0003, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.7770618556701031, | |
| "grad_norm": 1.3061492443084717, | |
| "learning_rate": 2.8813549852137824e-06, | |
| "loss": 0.0066, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.7783505154639175, | |
| "grad_norm": 0.06470157206058502, | |
| "learning_rate": 2.8498169747014824e-06, | |
| "loss": 0.0005, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.779639175257732, | |
| "grad_norm": 1.6503907442092896, | |
| "learning_rate": 2.818423809795384e-06, | |
| "loss": 0.0081, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.7809278350515464, | |
| "grad_norm": 1.9376052618026733, | |
| "learning_rate": 2.787176126445923e-06, | |
| "loss": 0.0147, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.7822164948453608, | |
| "grad_norm": 2.6921439170837402, | |
| "learning_rate": 2.756074557656424e-06, | |
| "loss": 0.0037, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.7835051546391752, | |
| "grad_norm": 0.5326633453369141, | |
| "learning_rate": 2.725119733470284e-06, | |
| "loss": 0.0017, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.7847938144329897, | |
| "grad_norm": 0.4451991021633148, | |
| "learning_rate": 2.6943122809582e-06, | |
| "loss": 0.0017, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.7860824742268041, | |
| "grad_norm": 0.46482527256011963, | |
| "learning_rate": 2.663652824205476e-06, | |
| "loss": 0.0016, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.7873711340206185, | |
| "grad_norm": 0.07450009137392044, | |
| "learning_rate": 2.6331419842993746e-06, | |
| "loss": 0.0003, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.788659793814433, | |
| "grad_norm": 0.21978060901165009, | |
| "learning_rate": 2.6027803793165353e-06, | |
| "loss": 0.001, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.7899484536082474, | |
| "grad_norm": 0.566473662853241, | |
| "learning_rate": 2.5725686243104586e-06, | |
| "loss": 0.0007, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.7912371134020618, | |
| "grad_norm": 2.9466750621795654, | |
| "learning_rate": 2.5425073312990334e-06, | |
| "loss": 0.0208, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.7925257731958762, | |
| "grad_norm": 3.2940099239349365, | |
| "learning_rate": 2.5125971092521607e-06, | |
| "loss": 0.0521, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.7938144329896907, | |
| "grad_norm": 0.08639179915189743, | |
| "learning_rate": 2.4828385640793974e-06, | |
| "loss": 0.0003, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.7951030927835051, | |
| "grad_norm": 2.585515260696411, | |
| "learning_rate": 2.4532322986176926e-06, | |
| "loss": 0.0057, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.7963917525773195, | |
| "grad_norm": 0.4970383048057556, | |
| "learning_rate": 2.4237789126191715e-06, | |
| "loss": 0.0023, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.7976804123711341, | |
| "grad_norm": 0.30445781350135803, | |
| "learning_rate": 2.3944790027389888e-06, | |
| "loss": 0.0018, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.7989690721649485, | |
| "grad_norm": 0.4191214144229889, | |
| "learning_rate": 2.3653331625232367e-06, | |
| "loss": 0.0022, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.7989690721649485, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.013430897146463394, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 84.121, | |
| "eval_samples_per_second": 5.409, | |
| "eval_steps_per_second": 0.178, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.8002577319587629, | |
| "grad_norm": 0.36169302463531494, | |
| "learning_rate": 2.336341982396928e-06, | |
| "loss": 0.0006, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.8015463917525774, | |
| "grad_norm": 0.40979477763175964, | |
| "learning_rate": 2.307506049652031e-06, | |
| "loss": 0.0036, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.8028350515463918, | |
| "grad_norm": 4.417389392852783, | |
| "learning_rate": 2.2788259484355712e-06, | |
| "loss": 0.0303, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.8041237113402062, | |
| "grad_norm": 1.4798436164855957, | |
| "learning_rate": 2.250302259737803e-06, | |
| "loss": 0.006, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.8054123711340206, | |
| "grad_norm": 0.44748619198799133, | |
| "learning_rate": 2.2219355613804406e-06, | |
| "loss": 0.0018, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.8067010309278351, | |
| "grad_norm": 0.01932459883391857, | |
| "learning_rate": 2.1937264280049365e-06, | |
| "loss": 0.0001, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.8079896907216495, | |
| "grad_norm": 0.01475608441978693, | |
| "learning_rate": 2.1656754310608663e-06, | |
| "loss": 0.0001, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.8092783505154639, | |
| "grad_norm": 2.7333805561065674, | |
| "learning_rate": 2.137783138794335e-06, | |
| "loss": 0.0055, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.8105670103092784, | |
| "grad_norm": 2.6362922191619873, | |
| "learning_rate": 2.1100501162364707e-06, | |
| "loss": 0.0365, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.8118556701030928, | |
| "grad_norm": 0.2980014383792877, | |
| "learning_rate": 2.0824769251919775e-06, | |
| "loss": 0.0006, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.8131443298969072, | |
| "grad_norm": 0.7731969356536865, | |
| "learning_rate": 2.055064124227758e-06, | |
| "loss": 0.0016, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.8144329896907216, | |
| "grad_norm": 0.166972815990448, | |
| "learning_rate": 2.027812268661592e-06, | |
| "loss": 0.0006, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.8157216494845361, | |
| "grad_norm": 0.33119046688079834, | |
| "learning_rate": 2.0007219105508935e-06, | |
| "loss": 0.0015, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.8170103092783505, | |
| "grad_norm": 0.7209166884422302, | |
| "learning_rate": 1.9737935986815205e-06, | |
| "loss": 0.0023, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.8182989690721649, | |
| "grad_norm": 0.3935668170452118, | |
| "learning_rate": 1.9470278785566653e-06, | |
| "loss": 0.0015, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.8195876288659794, | |
| "grad_norm": 2.5623619556427, | |
| "learning_rate": 1.9204252923858003e-06, | |
| "loss": 0.0139, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.8208762886597938, | |
| "grad_norm": 1.0038329362869263, | |
| "learning_rate": 1.8939863790736923e-06, | |
| "loss": 0.0047, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.8221649484536082, | |
| "grad_norm": 1.09479820728302, | |
| "learning_rate": 1.8677116742094858e-06, | |
| "loss": 0.0043, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.8234536082474226, | |
| "grad_norm": 1.43467116355896, | |
| "learning_rate": 1.8416017100558592e-06, | |
| "loss": 0.0031, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.8247422680412371, | |
| "grad_norm": 1.7519943714141846, | |
| "learning_rate": 1.8156570155382357e-06, | |
| "loss": 0.0149, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.8247422680412371, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.013629155233502388, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.6673, | |
| "eval_samples_per_second": 5.438, | |
| "eval_steps_per_second": 0.179, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.8260309278350515, | |
| "grad_norm": 0.3471926748752594, | |
| "learning_rate": 1.7898781162340683e-06, | |
| "loss": 0.0013, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.8273195876288659, | |
| "grad_norm": 2.4162745475769043, | |
| "learning_rate": 1.764265534362205e-06, | |
| "loss": 0.0047, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.8286082474226805, | |
| "grad_norm": 1.6409410238265991, | |
| "learning_rate": 1.7388197887722914e-06, | |
| "loss": 0.0027, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.8298969072164949, | |
| "grad_norm": 0.545690655708313, | |
| "learning_rate": 1.7135413949342706e-06, | |
| "loss": 0.0025, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.8311855670103093, | |
| "grad_norm": 1.0303876399993896, | |
| "learning_rate": 1.6884308649279413e-06, | |
| "loss": 0.0023, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.8324742268041238, | |
| "grad_norm": 0.33830779790878296, | |
| "learning_rate": 1.6634887074325844e-06, | |
| "loss": 0.0037, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.8337628865979382, | |
| "grad_norm": 3.1400089263916016, | |
| "learning_rate": 1.6387154277166484e-06, | |
| "loss": 0.0166, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.8350515463917526, | |
| "grad_norm": 4.601196765899658, | |
| "learning_rate": 1.6141115276275298e-06, | |
| "loss": 0.0089, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.836340206185567, | |
| "grad_norm": 2.343628168106079, | |
| "learning_rate": 1.5896775055813973e-06, | |
| "loss": 0.0135, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.8376288659793815, | |
| "grad_norm": 0.42886826395988464, | |
| "learning_rate": 1.565413856553095e-06, | |
| "loss": 0.0009, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.8389175257731959, | |
| "grad_norm": 2.6612963676452637, | |
| "learning_rate": 1.5413210720661187e-06, | |
| "loss": 0.0107, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.8402061855670103, | |
| "grad_norm": 2.667137384414673, | |
| "learning_rate": 1.5173996401826563e-06, | |
| "loss": 0.0063, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.8414948453608248, | |
| "grad_norm": 0.3507553040981293, | |
| "learning_rate": 1.493650045493703e-06, | |
| "loss": 0.0005, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.8427835051546392, | |
| "grad_norm": 0.04581530764698982, | |
| "learning_rate": 1.470072769109242e-06, | |
| "loss": 0.0002, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.8440721649484536, | |
| "grad_norm": 1.1834570169448853, | |
| "learning_rate": 1.4466682886485007e-06, | |
| "loss": 0.024, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.845360824742268, | |
| "grad_norm": 1.0574672222137451, | |
| "learning_rate": 1.4234370782302742e-06, | |
| "loss": 0.0023, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.8466494845360825, | |
| "grad_norm": 2.036065101623535, | |
| "learning_rate": 1.4003796084633204e-06, | |
| "loss": 0.0025, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.8479381443298969, | |
| "grad_norm": 1.696574330329895, | |
| "learning_rate": 1.3774963464368295e-06, | |
| "loss": 0.0193, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.8492268041237113, | |
| "grad_norm": 1.5883897542953491, | |
| "learning_rate": 1.3547877557109546e-06, | |
| "loss": 0.0027, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.8505154639175257, | |
| "grad_norm": 4.670267581939697, | |
| "learning_rate": 1.3322542963074314e-06, | |
| "loss": 0.0397, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.8505154639175257, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.014034120365977287, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.9766, | |
| "eval_samples_per_second": 5.418, | |
| "eval_steps_per_second": 0.179, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.8518041237113402, | |
| "grad_norm": 1.0197376012802124, | |
| "learning_rate": 1.3098964247002499e-06, | |
| "loss": 0.0082, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.8530927835051546, | |
| "grad_norm": 1.4475185871124268, | |
| "learning_rate": 1.287714593806415e-06, | |
| "loss": 0.0407, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.854381443298969, | |
| "grad_norm": 1.8474105596542358, | |
| "learning_rate": 1.2657092529767644e-06, | |
| "loss": 0.0162, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.8556701030927835, | |
| "grad_norm": 1.8313465118408203, | |
| "learning_rate": 1.2438808479868715e-06, | |
| "loss": 0.0056, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.8569587628865979, | |
| "grad_norm": 0.1645200103521347, | |
| "learning_rate": 1.2222298210280103e-06, | |
| "loss": 0.0003, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.8582474226804123, | |
| "grad_norm": 0.8122232556343079, | |
| "learning_rate": 1.200756610698205e-06, | |
| "loss": 0.0013, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.8595360824742269, | |
| "grad_norm": 0.12824606895446777, | |
| "learning_rate": 1.1794616519933344e-06, | |
| "loss": 0.0006, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.8608247422680413, | |
| "grad_norm": 2.5258920192718506, | |
| "learning_rate": 1.1583453762983289e-06, | |
| "loss": 0.0158, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.8621134020618557, | |
| "grad_norm": 4.607170581817627, | |
| "learning_rate": 1.137408211378429e-06, | |
| "loss": 0.0285, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.8634020618556701, | |
| "grad_norm": 2.1147348880767822, | |
| "learning_rate": 1.1166505813705187e-06, | |
| "loss": 0.0086, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.8646907216494846, | |
| "grad_norm": 3.446589469909668, | |
| "learning_rate": 1.09607290677453e-06, | |
| "loss": 0.0197, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.865979381443299, | |
| "grad_norm": 1.567643404006958, | |
| "learning_rate": 1.0756756044449358e-06, | |
| "loss": 0.0097, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.8672680412371134, | |
| "grad_norm": 0.9470865726470947, | |
| "learning_rate": 1.0554590875822924e-06, | |
| "loss": 0.0016, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.8685567010309279, | |
| "grad_norm": 0.1424911469221115, | |
| "learning_rate": 1.035423765724879e-06, | |
| "loss": 0.0005, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.8698453608247423, | |
| "grad_norm": 0.6392320394515991, | |
| "learning_rate": 1.015570044740396e-06, | |
| "loss": 0.0032, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.8711340206185567, | |
| "grad_norm": 0.08668619394302368, | |
| "learning_rate": 9.958983268177425e-07, | |
| "loss": 0.0004, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.8724226804123711, | |
| "grad_norm": 1.2266596555709839, | |
| "learning_rate": 9.764090104588741e-07, | |
| "loss": 0.0025, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.8737113402061856, | |
| "grad_norm": 0.201644629240036, | |
| "learning_rate": 9.571024904707238e-07, | |
| "loss": 0.0014, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "grad_norm": 1.0185980796813965, | |
| "learning_rate": 9.379791579572118e-07, | |
| "loss": 0.0036, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.8762886597938144, | |
| "grad_norm": 1.772665023803711, | |
| "learning_rate": 9.190394003113123e-07, | |
| "loss": 0.0058, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.8762886597938144, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.013515458442270756, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 84.5145, | |
| "eval_samples_per_second": 5.384, | |
| "eval_steps_per_second": 0.177, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.8775773195876289, | |
| "grad_norm": 0.43585294485092163, | |
| "learning_rate": 9.002836012072169e-07, | |
| "loss": 0.001, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.8788659793814433, | |
| "grad_norm": 0.862573504447937, | |
| "learning_rate": 8.817121405925543e-07, | |
| "loss": 0.004, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.8801546391752577, | |
| "grad_norm": 1.9756450653076172, | |
| "learning_rate": 8.633253946806974e-07, | |
| "loss": 0.0131, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.8814432989690721, | |
| "grad_norm": 0.09257616102695465, | |
| "learning_rate": 8.451237359431397e-07, | |
| "loss": 0.0004, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.8827319587628866, | |
| "grad_norm": 0.9120633602142334, | |
| "learning_rate": 8.271075331019541e-07, | |
| "loss": 0.0085, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.884020618556701, | |
| "grad_norm": 0.1704445630311966, | |
| "learning_rate": 8.092771511223185e-07, | |
| "loss": 0.0012, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.8853092783505154, | |
| "grad_norm": 0.10662536323070526, | |
| "learning_rate": 7.916329512051235e-07, | |
| "loss": 0.0006, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.8865979381443299, | |
| "grad_norm": 2.8713998794555664, | |
| "learning_rate": 7.741752907796584e-07, | |
| "loss": 0.0159, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.8878865979381443, | |
| "grad_norm": 0.5300127863883972, | |
| "learning_rate": 7.56904523496369e-07, | |
| "loss": 0.0013, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.8891752577319587, | |
| "grad_norm": 1.1316462755203247, | |
| "learning_rate": 7.398209992196914e-07, | |
| "loss": 0.0019, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.8904639175257731, | |
| "grad_norm": 0.06225062906742096, | |
| "learning_rate": 7.229250640209661e-07, | |
| "loss": 0.0003, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.8917525773195877, | |
| "grad_norm": 0.21658547222614288, | |
| "learning_rate": 7.062170601714302e-07, | |
| "loss": 0.0009, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.8930412371134021, | |
| "grad_norm": 5.541469573974609, | |
| "learning_rate": 6.896973261352779e-07, | |
| "loss": 0.0294, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.8943298969072165, | |
| "grad_norm": 0.24933123588562012, | |
| "learning_rate": 6.73366196562808e-07, | |
| "loss": 0.0013, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.895618556701031, | |
| "grad_norm": 2.7014365196228027, | |
| "learning_rate": 6.572240022836451e-07, | |
| "loss": 0.0166, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.8969072164948454, | |
| "grad_norm": 1.3881912231445312, | |
| "learning_rate": 6.412710703000368e-07, | |
| "loss": 0.0037, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.8981958762886598, | |
| "grad_norm": 4.935601711273193, | |
| "learning_rate": 6.255077237802288e-07, | |
| "loss": 0.0208, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.8994845360824743, | |
| "grad_norm": 2.3995628356933594, | |
| "learning_rate": 6.099342820519183e-07, | |
| "loss": 0.0219, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.9007731958762887, | |
| "grad_norm": 0.5882774591445923, | |
| "learning_rate": 5.945510605957861e-07, | |
| "loss": 0.0023, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.9020618556701031, | |
| "grad_norm": 3.2632617950439453, | |
| "learning_rate": 5.79358371039106e-07, | |
| "loss": 0.0153, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.9020618556701031, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.01318343449383974, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 83.7071, | |
| "eval_samples_per_second": 5.436, | |
| "eval_steps_per_second": 0.179, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.9033505154639175, | |
| "grad_norm": 4.556031703948975, | |
| "learning_rate": 5.643565211494285e-07, | |
| "loss": 0.041, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.904639175257732, | |
| "grad_norm": 0.027816738933324814, | |
| "learning_rate": 5.495458148283505e-07, | |
| "loss": 0.0002, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.9059278350515464, | |
| "grad_norm": 0.04451802000403404, | |
| "learning_rate": 5.349265521053603e-07, | |
| "loss": 0.0002, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.9072164948453608, | |
| "grad_norm": 1.494473934173584, | |
| "learning_rate": 5.204990291317535e-07, | |
| "loss": 0.003, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.9085051546391752, | |
| "grad_norm": 2.967031955718994, | |
| "learning_rate": 5.062635381746362e-07, | |
| "loss": 0.0254, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.9097938144329897, | |
| "grad_norm": 0.6380704045295715, | |
| "learning_rate": 4.92220367611006e-07, | |
| "loss": 0.0036, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.9110824742268041, | |
| "grad_norm": 0.40568777918815613, | |
| "learning_rate": 4.783698019219118e-07, | |
| "loss": 0.0014, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.9123711340206185, | |
| "grad_norm": 3.13850736618042, | |
| "learning_rate": 4.647121216866857e-07, | |
| "loss": 0.0343, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.913659793814433, | |
| "grad_norm": 1.518782377243042, | |
| "learning_rate": 4.512476035772628e-07, | |
| "loss": 0.0098, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.9149484536082474, | |
| "grad_norm": 0.19677811861038208, | |
| "learning_rate": 4.3797652035257544e-07, | |
| "loss": 0.0016, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.9162371134020618, | |
| "grad_norm": 1.9592320919036865, | |
| "learning_rate": 4.248991408530279e-07, | |
| "loss": 0.0126, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.9175257731958762, | |
| "grad_norm": 0.1629909873008728, | |
| "learning_rate": 4.1201572999505e-07, | |
| "loss": 0.0004, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.9188144329896907, | |
| "grad_norm": 5.022075653076172, | |
| "learning_rate": 3.9932654876573164e-07, | |
| "loss": 0.0158, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.9201030927835051, | |
| "grad_norm": 0.3693143427371979, | |
| "learning_rate": 3.8683185421753313e-07, | |
| "loss": 0.0014, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.9213917525773195, | |
| "grad_norm": 1.231784462928772, | |
| "learning_rate": 3.74531899463082e-07, | |
| "loss": 0.0082, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.9226804123711341, | |
| "grad_norm": 0.04480728134512901, | |
| "learning_rate": 3.6242693367004365e-07, | |
| "loss": 0.0002, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.9239690721649485, | |
| "grad_norm": 0.14905039966106415, | |
| "learning_rate": 3.5051720205606877e-07, | |
| "loss": 0.0009, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.9252577319587629, | |
| "grad_norm": 0.11093373596668243, | |
| "learning_rate": 3.38802945883836e-07, | |
| "loss": 0.0008, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.9265463917525774, | |
| "grad_norm": 1.7350820302963257, | |
| "learning_rate": 3.2728440245615724e-07, | |
| "loss": 0.0041, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.9278350515463918, | |
| "grad_norm": 4.958645343780518, | |
| "learning_rate": 3.1596180511117235e-07, | |
| "loss": 0.0122, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.9278350515463918, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.01316594984382391, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 85.875, | |
| "eval_samples_per_second": 5.298, | |
| "eval_steps_per_second": 0.175, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.9291237113402062, | |
| "grad_norm": 3.578312873840332, | |
| "learning_rate": 3.048353832176221e-07, | |
| "loss": 0.0234, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.9304123711340206, | |
| "grad_norm": 0.2854086756706238, | |
| "learning_rate": 2.939053621702015e-07, | |
| "loss": 0.0011, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.9317010309278351, | |
| "grad_norm": 2.446420907974243, | |
| "learning_rate": 2.83171963384995e-07, | |
| "loss": 0.0121, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.9329896907216495, | |
| "grad_norm": 3.754582643508911, | |
| "learning_rate": 2.7263540429498747e-07, | |
| "loss": 0.0285, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.9342783505154639, | |
| "grad_norm": 1.2466968297958374, | |
| "learning_rate": 2.6229589834566807e-07, | |
| "loss": 0.0334, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.9355670103092784, | |
| "grad_norm": 4.552764415740967, | |
| "learning_rate": 2.5215365499069446e-07, | |
| "loss": 0.013, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.9368556701030928, | |
| "grad_norm": 0.8621684312820435, | |
| "learning_rate": 2.4220887968765873e-07, | |
| "loss": 0.0022, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.9381443298969072, | |
| "grad_norm": 3.7333528995513916, | |
| "learning_rate": 2.3246177389392388e-07, | |
| "loss": 0.0402, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.9394329896907216, | |
| "grad_norm": 0.3883923590183258, | |
| "learning_rate": 2.229125350625394e-07, | |
| "loss": 0.0012, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.9407216494845361, | |
| "grad_norm": 0.36370164155960083, | |
| "learning_rate": 2.1356135663824328e-07, | |
| "loss": 0.0017, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.9420103092783505, | |
| "grad_norm": 1.1851253509521484, | |
| "learning_rate": 2.0440842805354522e-07, | |
| "loss": 0.0178, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.9432989690721649, | |
| "grad_norm": 0.40920475125312805, | |
| "learning_rate": 1.9545393472488738e-07, | |
| "loss": 0.0024, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.9445876288659794, | |
| "grad_norm": 1.3488298654556274, | |
| "learning_rate": 1.866980580488842e-07, | |
| "loss": 0.0269, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.9458762886597938, | |
| "grad_norm": 1.7972975969314575, | |
| "learning_rate": 1.7814097539865626e-07, | |
| "loss": 0.0037, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.9471649484536082, | |
| "grad_norm": 0.23618119955062866, | |
| "learning_rate": 1.6978286012023225e-07, | |
| "loss": 0.0013, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.9484536082474226, | |
| "grad_norm": 0.15299421548843384, | |
| "learning_rate": 1.6162388152903498e-07, | |
| "loss": 0.0007, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.9497422680412371, | |
| "grad_norm": 0.8923954963684082, | |
| "learning_rate": 1.5366420490645738e-07, | |
| "loss": 0.0033, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.9510309278350515, | |
| "grad_norm": 0.06458217650651932, | |
| "learning_rate": 1.4590399149650769e-07, | |
| "loss": 0.0003, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.9523195876288659, | |
| "grad_norm": 2.2554473876953125, | |
| "learning_rate": 1.3834339850254952e-07, | |
| "loss": 0.0097, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.9536082474226805, | |
| "grad_norm": 1.226751685142517, | |
| "learning_rate": 1.309825790841146e-07, | |
| "loss": 0.0276, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.9536082474226805, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.013158504851162434, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 85.185, | |
| "eval_samples_per_second": 5.341, | |
| "eval_steps_per_second": 0.176, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.9548969072164949, | |
| "grad_norm": 0.9453576803207397, | |
| "learning_rate": 1.2382168235379742e-07, | |
| "loss": 0.001, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.9561855670103093, | |
| "grad_norm": 6.605914115905762, | |
| "learning_rate": 1.1686085337423991e-07, | |
| "loss": 0.0179, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.9574742268041238, | |
| "grad_norm": 3.2699971199035645, | |
| "learning_rate": 1.1010023315518592e-07, | |
| "loss": 0.0252, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.9587628865979382, | |
| "grad_norm": 2.245004415512085, | |
| "learning_rate": 1.0353995865063138e-07, | |
| "loss": 0.0321, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.9600515463917526, | |
| "grad_norm": 0.08078460395336151, | |
| "learning_rate": 9.718016275604759e-08, | |
| "loss": 0.0005, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.961340206185567, | |
| "grad_norm": 0.14461307227611542, | |
| "learning_rate": 9.10209743056889e-08, | |
| "loss": 0.0006, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.9626288659793815, | |
| "grad_norm": 2.124976396560669, | |
| "learning_rate": 8.506251806997934e-08, | |
| "loss": 0.0429, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.9639175257731959, | |
| "grad_norm": 0.512088418006897, | |
| "learning_rate": 7.930491475299229e-08, | |
| "loss": 0.0009, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.9652061855670103, | |
| "grad_norm": 0.8166017532348633, | |
| "learning_rate": 7.37482809900003e-08, | |
| "loss": 0.002, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.9664948453608248, | |
| "grad_norm": 0.9644703269004822, | |
| "learning_rate": 6.839272934511143e-08, | |
| "loss": 0.0045, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.9677835051546392, | |
| "grad_norm": 0.47802242636680603, | |
| "learning_rate": 6.323836830899321e-08, | |
| "loss": 0.0015, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.9690721649484536, | |
| "grad_norm": 4.951274871826172, | |
| "learning_rate": 5.828530229667228e-08, | |
| "loss": 0.0172, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.970360824742268, | |
| "grad_norm": 0.19146917760372162, | |
| "learning_rate": 5.353363164541825e-08, | |
| "loss": 0.0011, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.9716494845360825, | |
| "grad_norm": 1.4356184005737305, | |
| "learning_rate": 4.898345261271531e-08, | |
| "loss": 0.002, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.9729381443298969, | |
| "grad_norm": 2.761564254760742, | |
| "learning_rate": 4.463485737430606e-08, | |
| "loss": 0.0068, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.9742268041237113, | |
| "grad_norm": 3.234301805496216, | |
| "learning_rate": 4.0487934022328533e-08, | |
| "loss": 0.0142, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.9755154639175257, | |
| "grad_norm": 2.400604248046875, | |
| "learning_rate": 3.654276656353206e-08, | |
| "loss": 0.0098, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.9768041237113402, | |
| "grad_norm": 0.21227847039699554, | |
| "learning_rate": 3.27994349175742e-08, | |
| "loss": 0.0006, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.9780927835051546, | |
| "grad_norm": 1.0701520442962646, | |
| "learning_rate": 2.9258014915399813e-08, | |
| "loss": 0.0013, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.979381443298969, | |
| "grad_norm": 1.216143250465393, | |
| "learning_rate": 2.591857829770672e-08, | |
| "loss": 0.0042, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.979381443298969, | |
| "eval_accuracy": 0.9980139026812314, | |
| "eval_f1": 0.9642857142857143, | |
| "eval_loss": 0.013023993000388145, | |
| "eval_precision": 0.9818181818181818, | |
| "eval_recall": 0.9473684210526315, | |
| "eval_runtime": 86.4332, | |
| "eval_samples_per_second": 5.264, | |
| "eval_steps_per_second": 0.174, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.9806701030927835, | |
| "grad_norm": 3.9802327156066895, | |
| "learning_rate": 2.278119271349466e-08, | |
| "loss": 0.0287, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.9819587628865979, | |
| "grad_norm": 1.6140069961547852, | |
| "learning_rate": 1.984592171869082e-08, | |
| "loss": 0.0041, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.9832474226804123, | |
| "grad_norm": 0.06459134072065353, | |
| "learning_rate": 1.711282477486642e-08, | |
| "loss": 0.0003, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.9845360824742269, | |
| "grad_norm": 0.7752759456634521, | |
| "learning_rate": 1.4581957248026579e-08, | |
| "loss": 0.0037, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.9858247422680413, | |
| "grad_norm": 1.614241361618042, | |
| "learning_rate": 1.2253370407495636e-08, | |
| "loss": 0.0067, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.9871134020618557, | |
| "grad_norm": 0.4276859760284424, | |
| "learning_rate": 1.0127111424872437e-08, | |
| "loss": 0.0019, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.9884020618556701, | |
| "grad_norm": 3.409043788909912, | |
| "learning_rate": 8.203223373078883e-09, | |
| "loss": 0.016, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.9896907216494846, | |
| "grad_norm": 0.6996654868125916, | |
| "learning_rate": 6.481745225485059e-09, | |
| "loss": 0.0018, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.990979381443299, | |
| "grad_norm": 0.7112312912940979, | |
| "learning_rate": 4.962711855120983e-09, | |
| "loss": 0.0026, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.9922680412371134, | |
| "grad_norm": 4.281740665435791, | |
| "learning_rate": 3.6461540339682855e-09, | |
| "loss": 0.0217, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.9935567010309279, | |
| "grad_norm": 0.22678984701633453, | |
| "learning_rate": 2.532098432341812e-09, | |
| "loss": 0.0009, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.9948453608247423, | |
| "grad_norm": 2.139557123184204, | |
| "learning_rate": 1.6205676183411733e-09, | |
| "loss": 0.0065, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.9961340206185567, | |
| "grad_norm": 1.8708549737930298, | |
| "learning_rate": 9.115800574022171e-10, | |
| "loss": 0.0096, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.9974226804123711, | |
| "grad_norm": 1.9189296960830688, | |
| "learning_rate": 4.0515011191621933e-10, | |
| "loss": 0.0148, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.9987113402061856, | |
| "grad_norm": 1.2406359910964966, | |
| "learning_rate": 1.0128804094233779e-10, | |
| "loss": 0.0034, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 2.146103858947754, | |
| "learning_rate": 0.0, | |
| "loss": 0.0056, | |
| "step": 776 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 776, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.5857289592242176e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |