| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 12.30379746835443, | |
| "eval_steps": 500, | |
| "global_step": 117, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.10126582278481013, | |
| "grad_norm": 6.4778934096410525, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.8098, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 5.612543112297245, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.7386, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 6.548065429151422, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.8181, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 5.44658060636073, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.7662, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 4.47695425931703, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.7373, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 2.445670129461038, | |
| "learning_rate": 5e-06, | |
| "loss": 0.7401, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 2.038374135196906, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.7213, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 3.6883261924243524, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.721, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 4.021835092791024, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.7316, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 1.0506329113924051, | |
| "grad_norm": 4.068463125902547, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.6951, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.1518987341772151, | |
| "grad_norm": 3.6102787398880323, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.651, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 1.2531645569620253, | |
| "grad_norm": 3.2413589478833225, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6401, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 1.3544303797468356, | |
| "grad_norm": 2.1039547455659036, | |
| "learning_rate": 9.997762161417517e-06, | |
| "loss": 0.6248, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 1.4556962025316456, | |
| "grad_norm": 1.506882982514818, | |
| "learning_rate": 9.991050648838676e-06, | |
| "loss": 0.5989, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.5569620253164556, | |
| "grad_norm": 1.961032694492409, | |
| "learning_rate": 9.979871469976197e-06, | |
| "loss": 0.5927, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 1.6582278481012658, | |
| "grad_norm": 1.7657958401095302, | |
| "learning_rate": 9.964234631709188e-06, | |
| "loss": 0.5798, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 1.759493670886076, | |
| "grad_norm": 1.1942801790386994, | |
| "learning_rate": 9.944154131125643e-06, | |
| "loss": 0.585, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 1.8607594936708862, | |
| "grad_norm": 1.0246056561560333, | |
| "learning_rate": 9.91964794299315e-06, | |
| "loss": 0.5636, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 1.9620253164556962, | |
| "grad_norm": 0.9759937830939072, | |
| "learning_rate": 9.890738003669029e-06, | |
| "loss": 0.5737, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 2.1012658227848102, | |
| "grad_norm": 0.8943349696312995, | |
| "learning_rate": 9.857450191464337e-06, | |
| "loss": 0.5544, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.2025316455696204, | |
| "grad_norm": 0.7778562356282119, | |
| "learning_rate": 9.819814303479268e-06, | |
| "loss": 0.5232, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 2.3037974683544302, | |
| "grad_norm": 0.7770753804028017, | |
| "learning_rate": 9.777864028930705e-06, | |
| "loss": 0.5007, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 2.4050632911392404, | |
| "grad_norm": 0.7830267056270357, | |
| "learning_rate": 9.731636918995821e-06, | |
| "loss": 0.4998, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 2.5063291139240507, | |
| "grad_norm": 0.6824790889336223, | |
| "learning_rate": 9.681174353198687e-06, | |
| "loss": 0.4864, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 2.607594936708861, | |
| "grad_norm": 0.6878043014822572, | |
| "learning_rate": 9.626521502369984e-06, | |
| "loss": 0.5195, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.708860759493671, | |
| "grad_norm": 0.6574977351787944, | |
| "learning_rate": 9.567727288213005e-06, | |
| "loss": 0.4816, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 2.810126582278481, | |
| "grad_norm": 0.9181938521695006, | |
| "learning_rate": 9.504844339512096e-06, | |
| "loss": 0.4971, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 2.911392405063291, | |
| "grad_norm": 0.6803920137285357, | |
| "learning_rate": 9.437928945022772e-06, | |
| "loss": 0.4937, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 3.050632911392405, | |
| "grad_norm": 0.5633128549281556, | |
| "learning_rate": 9.36704100308565e-06, | |
| "loss": 0.4523, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 3.151898734177215, | |
| "grad_norm": 0.6900059027767877, | |
| "learning_rate": 9.292243968009332e-06, | |
| "loss": 0.4404, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 3.2531645569620253, | |
| "grad_norm": 0.8283075120408294, | |
| "learning_rate": 9.213604793270196e-06, | |
| "loss": 0.4144, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 3.3544303797468356, | |
| "grad_norm": 0.6496042868146896, | |
| "learning_rate": 9.131193871579975e-06, | |
| "loss": 0.4377, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 3.4556962025316453, | |
| "grad_norm": 0.5756846729919667, | |
| "learning_rate": 9.045084971874738e-06, | |
| "loss": 0.4469, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 3.5569620253164556, | |
| "grad_norm": 0.6484891084170515, | |
| "learning_rate": 8.955355173281709e-06, | |
| "loss": 0.4325, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 3.6582278481012658, | |
| "grad_norm": 0.6716988018193619, | |
| "learning_rate": 8.862084796122998e-06, | |
| "loss": 0.4386, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 3.759493670886076, | |
| "grad_norm": 0.6227031440124084, | |
| "learning_rate": 8.765357330018056e-06, | |
| "loss": 0.4439, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 3.8607594936708862, | |
| "grad_norm": 0.5639927614442395, | |
| "learning_rate": 8.665259359149132e-06, | |
| "loss": 0.3859, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 3.962025316455696, | |
| "grad_norm": 0.7647456540040337, | |
| "learning_rate": 8.561880484756726e-06, | |
| "loss": 0.4299, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 4.10126582278481, | |
| "grad_norm": 0.6321083416415383, | |
| "learning_rate": 8.455313244934324e-06, | |
| "loss": 0.3883, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 4.2025316455696204, | |
| "grad_norm": 0.5198699064539194, | |
| "learning_rate": 8.345653031794292e-06, | |
| "loss": 0.3643, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 4.30379746835443, | |
| "grad_norm": 0.5086955081315121, | |
| "learning_rate": 8.232998006078998e-06, | |
| "loss": 0.3618, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 4.405063291139241, | |
| "grad_norm": 0.6081695043728169, | |
| "learning_rate": 8.117449009293668e-06, | |
| "loss": 0.383, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 4.506329113924051, | |
| "grad_norm": 0.5601990782087208, | |
| "learning_rate": 7.99910947343957e-06, | |
| "loss": 0.382, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 4.6075949367088604, | |
| "grad_norm": 0.5815202049512344, | |
| "learning_rate": 7.87808532842837e-06, | |
| "loss": 0.3602, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 4.708860759493671, | |
| "grad_norm": 0.5100502056093675, | |
| "learning_rate": 7.754484907260513e-06, | |
| "loss": 0.3502, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 4.810126582278481, | |
| "grad_norm": 0.4979200942364688, | |
| "learning_rate": 7.628418849052523e-06, | |
| "loss": 0.3271, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 4.911392405063291, | |
| "grad_norm": 0.6431923972069008, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3496, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 5.050632911392405, | |
| "grad_norm": 0.48926523770720753, | |
| "learning_rate": 7.369343312364994e-06, | |
| "loss": 0.3482, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 5.151898734177215, | |
| "grad_norm": 0.5144390139948329, | |
| "learning_rate": 7.236565741578163e-06, | |
| "loss": 0.3056, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 5.253164556962025, | |
| "grad_norm": 0.5324310796726963, | |
| "learning_rate": 7.101786141547829e-06, | |
| "loss": 0.3014, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 5.3544303797468356, | |
| "grad_norm": 0.570662846466903, | |
| "learning_rate": 6.965125158269619e-06, | |
| "loss": 0.3117, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 5.455696202531645, | |
| "grad_norm": 0.5973485364504665, | |
| "learning_rate": 6.8267051218319766e-06, | |
| "loss": 0.3101, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 5.556962025316456, | |
| "grad_norm": 0.5388100158678228, | |
| "learning_rate": 6.686649936914151e-06, | |
| "loss": 0.2752, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 5.658227848101266, | |
| "grad_norm": 0.5427783482814825, | |
| "learning_rate": 6.545084971874738e-06, | |
| "loss": 0.2832, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 5.759493670886076, | |
| "grad_norm": 0.4462088474534927, | |
| "learning_rate": 6.402136946530014e-06, | |
| "loss": 0.2956, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 5.860759493670886, | |
| "grad_norm": 0.5359550899851249, | |
| "learning_rate": 6.257933818722544e-06, | |
| "loss": 0.3135, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 5.962025316455696, | |
| "grad_norm": 0.5865839344475425, | |
| "learning_rate": 6.112604669781572e-06, | |
| "loss": 0.3025, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 6.10126582278481, | |
| "grad_norm": 0.6248214693634591, | |
| "learning_rate": 5.9662795889777666e-06, | |
| "loss": 0.2691, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 6.2025316455696204, | |
| "grad_norm": 0.5236794111254583, | |
| "learning_rate": 5.819089557075689e-06, | |
| "loss": 0.2553, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 6.30379746835443, | |
| "grad_norm": 0.7579361632562012, | |
| "learning_rate": 5.671166329088278e-06, | |
| "loss": 0.2617, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 6.405063291139241, | |
| "grad_norm": 0.5840342453987523, | |
| "learning_rate": 5.522642316338268e-06, | |
| "loss": 0.267, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 6.506329113924051, | |
| "grad_norm": 0.46177196353515965, | |
| "learning_rate": 5.373650467932122e-06, | |
| "loss": 0.2485, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 6.6075949367088604, | |
| "grad_norm": 0.5597684956497142, | |
| "learning_rate": 5.224324151752575e-06, | |
| "loss": 0.2293, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 6.708860759493671, | |
| "grad_norm": 0.5620672859815132, | |
| "learning_rate": 5.074797035076319e-06, | |
| "loss": 0.2467, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 6.810126582278481, | |
| "grad_norm": 0.48474841455975654, | |
| "learning_rate": 4.9252029649236835e-06, | |
| "loss": 0.2381, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 6.911392405063291, | |
| "grad_norm": 0.7101817876580098, | |
| "learning_rate": 4.775675848247427e-06, | |
| "loss": 0.2353, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 7.050632911392405, | |
| "grad_norm": 0.5476517720017136, | |
| "learning_rate": 4.626349532067879e-06, | |
| "loss": 0.2402, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 7.151898734177215, | |
| "grad_norm": 0.5724014642743455, | |
| "learning_rate": 4.477357683661734e-06, | |
| "loss": 0.2301, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 7.253164556962025, | |
| "grad_norm": 0.5856649190293501, | |
| "learning_rate": 4.3288336709117246e-06, | |
| "loss": 0.2003, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 7.3544303797468356, | |
| "grad_norm": 0.4501764347655313, | |
| "learning_rate": 4.180910442924312e-06, | |
| "loss": 0.2082, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 7.455696202531645, | |
| "grad_norm": 0.5497087888059933, | |
| "learning_rate": 4.033720411022235e-06, | |
| "loss": 0.2161, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 7.556962025316456, | |
| "grad_norm": 0.786401153735926, | |
| "learning_rate": 3.887395330218429e-06, | |
| "loss": 0.211, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 7.658227848101266, | |
| "grad_norm": 0.44392822635264506, | |
| "learning_rate": 3.7420661812774577e-06, | |
| "loss": 0.1776, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 7.759493670886076, | |
| "grad_norm": 0.47423922231310417, | |
| "learning_rate": 3.5978630534699873e-06, | |
| "loss": 0.1737, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 7.860759493670886, | |
| "grad_norm": 0.4971979227785128, | |
| "learning_rate": 3.4549150281252635e-06, | |
| "loss": 0.2106, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 7.962025316455696, | |
| "grad_norm": 0.49614655483664794, | |
| "learning_rate": 3.3133500630858507e-06, | |
| "loss": 0.1959, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 8.10126582278481, | |
| "grad_norm": 0.6181584183705097, | |
| "learning_rate": 3.173294878168025e-06, | |
| "loss": 0.1511, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 8.20253164556962, | |
| "grad_norm": 0.46174252453585185, | |
| "learning_rate": 3.0348748417303826e-06, | |
| "loss": 0.1739, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 8.30379746835443, | |
| "grad_norm": 0.6988174535684211, | |
| "learning_rate": 2.8982138584521734e-06, | |
| "loss": 0.1803, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 8.405063291139241, | |
| "grad_norm": 0.5349028931087013, | |
| "learning_rate": 2.7634342584218364e-06, | |
| "loss": 0.1734, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 8.50632911392405, | |
| "grad_norm": 0.45207010276189186, | |
| "learning_rate": 2.6306566876350072e-06, | |
| "loss": 0.179, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 8.60759493670886, | |
| "grad_norm": 0.6209707760111869, | |
| "learning_rate": 2.5000000000000015e-06, | |
| "loss": 0.1762, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 8.708860759493671, | |
| "grad_norm": 0.46590378024363177, | |
| "learning_rate": 2.371581150947476e-06, | |
| "loss": 0.1523, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 8.810126582278482, | |
| "grad_norm": 0.48616281321676563, | |
| "learning_rate": 2.245515092739488e-06, | |
| "loss": 0.1659, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 8.91139240506329, | |
| "grad_norm": 0.4573433266948846, | |
| "learning_rate": 2.1219146715716332e-06, | |
| "loss": 0.1761, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 9.050632911392405, | |
| "grad_norm": 0.4068922915948588, | |
| "learning_rate": 2.0008905265604316e-06, | |
| "loss": 0.1807, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 9.151898734177216, | |
| "grad_norm": 0.45799070927164987, | |
| "learning_rate": 1.8825509907063328e-06, | |
| "loss": 0.1524, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 9.253164556962025, | |
| "grad_norm": 0.4236447469361975, | |
| "learning_rate": 1.7670019939210025e-06, | |
| "loss": 0.1235, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 9.354430379746836, | |
| "grad_norm": 0.468507042121909, | |
| "learning_rate": 1.6543469682057105e-06, | |
| "loss": 0.1337, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 9.455696202531646, | |
| "grad_norm": 0.5430414054629514, | |
| "learning_rate": 1.544686755065677e-06, | |
| "loss": 0.1614, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 9.556962025316455, | |
| "grad_norm": 0.5532412129252906, | |
| "learning_rate": 1.438119515243277e-06, | |
| "loss": 0.1495, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 9.658227848101266, | |
| "grad_norm": 0.4349456263435749, | |
| "learning_rate": 1.3347406408508695e-06, | |
| "loss": 0.1607, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 9.759493670886076, | |
| "grad_norm": 0.4213621612269145, | |
| "learning_rate": 1.234642669981946e-06, | |
| "loss": 0.15, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 9.860759493670885, | |
| "grad_norm": 0.4654265405081889, | |
| "learning_rate": 1.137915203877003e-06, | |
| "loss": 0.1698, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 9.962025316455696, | |
| "grad_norm": 0.44369239489583484, | |
| "learning_rate": 1.044644826718295e-06, | |
| "loss": 0.1198, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 10.10126582278481, | |
| "grad_norm": 0.5123031700512586, | |
| "learning_rate": 9.549150281252633e-07, | |
| "loss": 0.1248, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 10.20253164556962, | |
| "grad_norm": 0.4983041518219975, | |
| "learning_rate": 8.688061284200266e-07, | |
| "loss": 0.1242, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 10.30379746835443, | |
| "grad_norm": 0.4222769994509792, | |
| "learning_rate": 7.863952067298042e-07, | |
| "loss": 0.1405, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 10.405063291139241, | |
| "grad_norm": 0.3841005492919697, | |
| "learning_rate": 7.077560319906696e-07, | |
| "loss": 0.1445, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 10.50632911392405, | |
| "grad_norm": 0.4200137360534297, | |
| "learning_rate": 6.329589969143518e-07, | |
| "loss": 0.1254, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 10.60759493670886, | |
| "grad_norm": 0.5003494790644746, | |
| "learning_rate": 5.620710549772295e-07, | |
| "loss": 0.133, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 10.708860759493671, | |
| "grad_norm": 0.5011493021736739, | |
| "learning_rate": 4.951556604879049e-07, | |
| "loss": 0.1168, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 10.810126582278482, | |
| "grad_norm": 0.48884310021347865, | |
| "learning_rate": 4.322727117869951e-07, | |
| "loss": 0.1388, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 10.91139240506329, | |
| "grad_norm": 0.5226686695455263, | |
| "learning_rate": 3.734784976300165e-07, | |
| "loss": 0.1527, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 11.050632911392405, | |
| "grad_norm": 0.3948238578792808, | |
| "learning_rate": 3.18825646801314e-07, | |
| "loss": 0.1274, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 11.151898734177216, | |
| "grad_norm": 0.35269461111410555, | |
| "learning_rate": 2.6836308100417874e-07, | |
| "loss": 0.139, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 11.253164556962025, | |
| "grad_norm": 0.3548380183475208, | |
| "learning_rate": 2.2213597106929608e-07, | |
| "loss": 0.1157, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 11.354430379746836, | |
| "grad_norm": 0.34864448919335145, | |
| "learning_rate": 1.801856965207338e-07, | |
| "loss": 0.1349, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 11.455696202531646, | |
| "grad_norm": 0.386010554165558, | |
| "learning_rate": 1.4254980853566248e-07, | |
| "loss": 0.1417, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 11.556962025316455, | |
| "grad_norm": 0.38535039808598326, | |
| "learning_rate": 1.0926199633097156e-07, | |
| "loss": 0.1449, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 11.658227848101266, | |
| "grad_norm": 0.3858057110646703, | |
| "learning_rate": 8.035205700685167e-08, | |
| "loss": 0.1072, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 11.759493670886076, | |
| "grad_norm": 0.34442383851045083, | |
| "learning_rate": 5.584586887435739e-08, | |
| "loss": 0.1233, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 11.860759493670885, | |
| "grad_norm": 0.3762221447736398, | |
| "learning_rate": 3.576536829081323e-08, | |
| "loss": 0.1273, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 11.962025316455696, | |
| "grad_norm": 0.3487932502007235, | |
| "learning_rate": 2.012853002380466e-08, | |
| "loss": 0.1302, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 12.10126582278481, | |
| "grad_norm": 0.37349970189214104, | |
| "learning_rate": 8.949351161324227e-09, | |
| "loss": 0.1193, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 12.20253164556962, | |
| "grad_norm": 0.387963316902507, | |
| "learning_rate": 2.237838582483387e-09, | |
| "loss": 0.1353, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "grad_norm": 0.3748478725901639, | |
| "learning_rate": 0.0, | |
| "loss": 0.1385, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "step": 117, | |
| "total_flos": 1.3184315127811277e+17, | |
| "train_loss": 0.32744092360520977, | |
| "train_runtime": 5136.4339, | |
| "train_samples_per_second": 0.8, | |
| "train_steps_per_second": 0.023 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 117, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 13, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.3184315127811277e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |