| { | |
| "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": 7.662746774640641, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 1.7165, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 7.780540722281656, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 1.7426, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 7.607197789278567, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.7798, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 7.130757987580523, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.7233, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 5.495085856118403, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 1.722, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 3.041800034099146, | |
| "learning_rate": 5e-06, | |
| "loss": 1.6939, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 2.542359589530777, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 1.6384, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 5.774459254580831, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.7588, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 6.257658308713539, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 1.6071, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 1.0506329113924051, | |
| "grad_norm": 6.008023976019131, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 1.6135, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.1518987341772151, | |
| "grad_norm": 3.669999677591489, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 1.5655, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 1.2531645569620253, | |
| "grad_norm": 2.753120639647023, | |
| "learning_rate": 1e-05, | |
| "loss": 1.5644, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 1.3544303797468356, | |
| "grad_norm": 2.5691380431456663, | |
| "learning_rate": 9.997762161417517e-06, | |
| "loss": 1.5743, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 1.4556962025316456, | |
| "grad_norm": 2.306400993306503, | |
| "learning_rate": 9.991050648838676e-06, | |
| "loss": 1.5471, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.5569620253164556, | |
| "grad_norm": 1.8699341630503707, | |
| "learning_rate": 9.979871469976197e-06, | |
| "loss": 1.5378, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 1.6582278481012658, | |
| "grad_norm": 1.8464716260953586, | |
| "learning_rate": 9.964234631709188e-06, | |
| "loss": 1.4935, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 1.759493670886076, | |
| "grad_norm": 1.9681283515447843, | |
| "learning_rate": 9.944154131125643e-06, | |
| "loss": 1.5307, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 1.8607594936708862, | |
| "grad_norm": 1.375532665824249, | |
| "learning_rate": 9.91964794299315e-06, | |
| "loss": 1.4384, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 1.9620253164556962, | |
| "grad_norm": 1.2400340447874798, | |
| "learning_rate": 9.890738003669029e-06, | |
| "loss": 1.4461, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 2.1012658227848102, | |
| "grad_norm": 1.3683455968129878, | |
| "learning_rate": 9.857450191464337e-06, | |
| "loss": 1.4202, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.2025316455696204, | |
| "grad_norm": 1.5450660435455545, | |
| "learning_rate": 9.819814303479268e-06, | |
| "loss": 1.4057, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 2.3037974683544302, | |
| "grad_norm": 1.421865126741691, | |
| "learning_rate": 9.777864028930705e-06, | |
| "loss": 1.3565, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 2.4050632911392404, | |
| "grad_norm": 1.1091220370043315, | |
| "learning_rate": 9.731636918995821e-06, | |
| "loss": 1.4285, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 2.5063291139240507, | |
| "grad_norm": 1.3362714346219098, | |
| "learning_rate": 9.681174353198687e-06, | |
| "loss": 1.3864, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 2.607594936708861, | |
| "grad_norm": 1.1836385685657953, | |
| "learning_rate": 9.626521502369984e-06, | |
| "loss": 1.4496, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.708860759493671, | |
| "grad_norm": 1.1833726695686608, | |
| "learning_rate": 9.567727288213005e-06, | |
| "loss": 1.3678, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 2.810126582278481, | |
| "grad_norm": 1.2442525903549861, | |
| "learning_rate": 9.504844339512096e-06, | |
| "loss": 1.4286, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 2.911392405063291, | |
| "grad_norm": 0.8921509735901482, | |
| "learning_rate": 9.437928945022772e-06, | |
| "loss": 1.4368, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 3.050632911392405, | |
| "grad_norm": 0.975058407873496, | |
| "learning_rate": 9.36704100308565e-06, | |
| "loss": 1.3964, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 3.151898734177215, | |
| "grad_norm": 1.0405046837289915, | |
| "learning_rate": 9.292243968009332e-06, | |
| "loss": 1.3709, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 3.2531645569620253, | |
| "grad_norm": 1.002508170436684, | |
| "learning_rate": 9.213604793270196e-06, | |
| "loss": 1.3603, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 3.3544303797468356, | |
| "grad_norm": 0.9770942979916054, | |
| "learning_rate": 9.131193871579975e-06, | |
| "loss": 1.3237, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 3.4556962025316453, | |
| "grad_norm": 0.9448598440796758, | |
| "learning_rate": 9.045084971874738e-06, | |
| "loss": 1.3099, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 3.5569620253164556, | |
| "grad_norm": 1.0243961491413869, | |
| "learning_rate": 8.955355173281709e-06, | |
| "loss": 1.3294, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 3.6582278481012658, | |
| "grad_norm": 1.0498287783111133, | |
| "learning_rate": 8.862084796122998e-06, | |
| "loss": 1.318, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 3.759493670886076, | |
| "grad_norm": 1.0218749159310017, | |
| "learning_rate": 8.765357330018056e-06, | |
| "loss": 1.2685, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 3.8607594936708862, | |
| "grad_norm": 0.9212713824325545, | |
| "learning_rate": 8.665259359149132e-06, | |
| "loss": 1.3514, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 3.962025316455696, | |
| "grad_norm": 0.8182570482584121, | |
| "learning_rate": 8.561880484756726e-06, | |
| "loss": 1.2959, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 4.10126582278481, | |
| "grad_norm": 0.9147193534940985, | |
| "learning_rate": 8.455313244934324e-06, | |
| "loss": 1.253, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 4.2025316455696204, | |
| "grad_norm": 0.96152387197972, | |
| "learning_rate": 8.345653031794292e-06, | |
| "loss": 1.314, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 4.30379746835443, | |
| "grad_norm": 0.6819883742162923, | |
| "learning_rate": 8.232998006078998e-06, | |
| "loss": 1.2201, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 4.405063291139241, | |
| "grad_norm": 0.8517906017071996, | |
| "learning_rate": 8.117449009293668e-06, | |
| "loss": 1.2754, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 4.506329113924051, | |
| "grad_norm": 0.8006510836405661, | |
| "learning_rate": 7.99910947343957e-06, | |
| "loss": 1.252, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 4.6075949367088604, | |
| "grad_norm": 0.7555890892764205, | |
| "learning_rate": 7.87808532842837e-06, | |
| "loss": 1.2666, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 4.708860759493671, | |
| "grad_norm": 0.7706830701550539, | |
| "learning_rate": 7.754484907260513e-06, | |
| "loss": 1.2233, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 4.810126582278481, | |
| "grad_norm": 0.6829800645271492, | |
| "learning_rate": 7.628418849052523e-06, | |
| "loss": 1.207, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 4.911392405063291, | |
| "grad_norm": 0.7895120131715684, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 1.2885, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 5.050632911392405, | |
| "grad_norm": 0.6612273743116371, | |
| "learning_rate": 7.369343312364994e-06, | |
| "loss": 1.1948, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 5.151898734177215, | |
| "grad_norm": 0.7056214697506767, | |
| "learning_rate": 7.236565741578163e-06, | |
| "loss": 1.2146, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 5.253164556962025, | |
| "grad_norm": 0.7801071489761209, | |
| "learning_rate": 7.101786141547829e-06, | |
| "loss": 1.1638, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 5.3544303797468356, | |
| "grad_norm": 0.6935082057169631, | |
| "learning_rate": 6.965125158269619e-06, | |
| "loss": 1.1918, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 5.455696202531645, | |
| "grad_norm": 0.7832626229660001, | |
| "learning_rate": 6.8267051218319766e-06, | |
| "loss": 1.1404, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 5.556962025316456, | |
| "grad_norm": 0.7128517215818128, | |
| "learning_rate": 6.686649936914151e-06, | |
| "loss": 1.1713, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 5.658227848101266, | |
| "grad_norm": 0.6669492372849923, | |
| "learning_rate": 6.545084971874738e-06, | |
| "loss": 1.1722, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 5.759493670886076, | |
| "grad_norm": 0.6998005902041823, | |
| "learning_rate": 6.402136946530014e-06, | |
| "loss": 1.1979, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 5.860759493670886, | |
| "grad_norm": 0.6804960587652885, | |
| "learning_rate": 6.257933818722544e-06, | |
| "loss": 1.2215, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 5.962025316455696, | |
| "grad_norm": 0.69570116237831, | |
| "learning_rate": 6.112604669781572e-06, | |
| "loss": 1.2289, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 6.10126582278481, | |
| "grad_norm": 0.6730746695442814, | |
| "learning_rate": 5.9662795889777666e-06, | |
| "loss": 1.1524, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 6.2025316455696204, | |
| "grad_norm": 0.5660521980897932, | |
| "learning_rate": 5.819089557075689e-06, | |
| "loss": 1.1378, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 6.30379746835443, | |
| "grad_norm": 0.5824150839029619, | |
| "learning_rate": 5.671166329088278e-06, | |
| "loss": 1.165, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 6.405063291139241, | |
| "grad_norm": 0.622521853438727, | |
| "learning_rate": 5.522642316338268e-06, | |
| "loss": 1.1215, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 6.506329113924051, | |
| "grad_norm": 0.6254680553315882, | |
| "learning_rate": 5.373650467932122e-06, | |
| "loss": 1.1403, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 6.6075949367088604, | |
| "grad_norm": 0.5887398962697034, | |
| "learning_rate": 5.224324151752575e-06, | |
| "loss": 1.143, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 6.708860759493671, | |
| "grad_norm": 0.5620967468980758, | |
| "learning_rate": 5.074797035076319e-06, | |
| "loss": 1.0853, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 6.810126582278481, | |
| "grad_norm": 0.5699880251358381, | |
| "learning_rate": 4.9252029649236835e-06, | |
| "loss": 1.1419, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 6.911392405063291, | |
| "grad_norm": 0.57469212766138, | |
| "learning_rate": 4.775675848247427e-06, | |
| "loss": 1.0354, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 7.050632911392405, | |
| "grad_norm": 0.6139033127376718, | |
| "learning_rate": 4.626349532067879e-06, | |
| "loss": 1.1216, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 7.151898734177215, | |
| "grad_norm": 0.6036384494469975, | |
| "learning_rate": 4.477357683661734e-06, | |
| "loss": 1.0732, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 7.253164556962025, | |
| "grad_norm": 0.5613205780072209, | |
| "learning_rate": 4.3288336709117246e-06, | |
| "loss": 1.0591, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 7.3544303797468356, | |
| "grad_norm": 0.6595402890869546, | |
| "learning_rate": 4.180910442924312e-06, | |
| "loss": 1.0313, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 7.455696202531645, | |
| "grad_norm": 0.623560168571677, | |
| "learning_rate": 4.033720411022235e-06, | |
| "loss": 1.0663, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 7.556962025316456, | |
| "grad_norm": 0.607807775148035, | |
| "learning_rate": 3.887395330218429e-06, | |
| "loss": 1.0936, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 7.658227848101266, | |
| "grad_norm": 0.5754652743966044, | |
| "learning_rate": 3.7420661812774577e-06, | |
| "loss": 1.1259, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 7.759493670886076, | |
| "grad_norm": 0.5903980443441297, | |
| "learning_rate": 3.5978630534699873e-06, | |
| "loss": 1.1048, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 7.860759493670886, | |
| "grad_norm": 0.6196916120126528, | |
| "learning_rate": 3.4549150281252635e-06, | |
| "loss": 0.9815, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 7.962025316455696, | |
| "grad_norm": 0.6232283295788235, | |
| "learning_rate": 3.3133500630858507e-06, | |
| "loss": 1.0788, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 8.10126582278481, | |
| "grad_norm": 0.5987601770502657, | |
| "learning_rate": 3.173294878168025e-06, | |
| "loss": 0.9989, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 8.20253164556962, | |
| "grad_norm": 0.5604402144565509, | |
| "learning_rate": 3.0348748417303826e-06, | |
| "loss": 1.0041, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 8.30379746835443, | |
| "grad_norm": 0.5451666489977828, | |
| "learning_rate": 2.8982138584521734e-06, | |
| "loss": 1.0447, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 8.405063291139241, | |
| "grad_norm": 0.6344906778308087, | |
| "learning_rate": 2.7634342584218364e-06, | |
| "loss": 0.9988, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 8.50632911392405, | |
| "grad_norm": 0.5872351561833212, | |
| "learning_rate": 2.6306566876350072e-06, | |
| "loss": 1.0034, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 8.60759493670886, | |
| "grad_norm": 0.610141177277554, | |
| "learning_rate": 2.5000000000000015e-06, | |
| "loss": 1.0155, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 8.708860759493671, | |
| "grad_norm": 0.5823822940344505, | |
| "learning_rate": 2.371581150947476e-06, | |
| "loss": 1.0434, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 8.810126582278482, | |
| "grad_norm": 0.5841088794808743, | |
| "learning_rate": 2.245515092739488e-06, | |
| "loss": 1.0029, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 8.91139240506329, | |
| "grad_norm": 0.5339095226049001, | |
| "learning_rate": 2.1219146715716332e-06, | |
| "loss": 1.0095, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 9.050632911392405, | |
| "grad_norm": 0.5952832378445538, | |
| "learning_rate": 2.0008905265604316e-06, | |
| "loss": 0.9776, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 9.151898734177216, | |
| "grad_norm": 0.6543984849776463, | |
| "learning_rate": 1.8825509907063328e-06, | |
| "loss": 0.9483, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 9.253164556962025, | |
| "grad_norm": 0.5743239227937693, | |
| "learning_rate": 1.7670019939210025e-06, | |
| "loss": 0.9806, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 9.354430379746836, | |
| "grad_norm": 0.6601365553789452, | |
| "learning_rate": 1.6543469682057105e-06, | |
| "loss": 1.0314, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 9.455696202531646, | |
| "grad_norm": 0.6289586904940253, | |
| "learning_rate": 1.544686755065677e-06, | |
| "loss": 0.9512, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 9.556962025316455, | |
| "grad_norm": 0.614688522043007, | |
| "learning_rate": 1.438119515243277e-06, | |
| "loss": 0.9952, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 9.658227848101266, | |
| "grad_norm": 0.6108071548241799, | |
| "learning_rate": 1.3347406408508695e-06, | |
| "loss": 0.9396, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 9.759493670886076, | |
| "grad_norm": 0.6174148171968499, | |
| "learning_rate": 1.234642669981946e-06, | |
| "loss": 1.003, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 9.860759493670885, | |
| "grad_norm": 0.591800071365107, | |
| "learning_rate": 1.137915203877003e-06, | |
| "loss": 0.9779, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 9.962025316455696, | |
| "grad_norm": 0.5838573058052152, | |
| "learning_rate": 1.044644826718295e-06, | |
| "loss": 0.9907, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 10.10126582278481, | |
| "grad_norm": 0.5847615671433455, | |
| "learning_rate": 9.549150281252633e-07, | |
| "loss": 0.9369, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 10.20253164556962, | |
| "grad_norm": 0.6181551405935158, | |
| "learning_rate": 8.688061284200266e-07, | |
| "loss": 0.9118, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 10.30379746835443, | |
| "grad_norm": 0.5590355923937118, | |
| "learning_rate": 7.863952067298042e-07, | |
| "loss": 0.9465, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 10.405063291139241, | |
| "grad_norm": 0.590111346006904, | |
| "learning_rate": 7.077560319906696e-07, | |
| "loss": 0.9791, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 10.50632911392405, | |
| "grad_norm": 0.6124262283897487, | |
| "learning_rate": 6.329589969143518e-07, | |
| "loss": 0.9809, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 10.60759493670886, | |
| "grad_norm": 0.5838862008978108, | |
| "learning_rate": 5.620710549772295e-07, | |
| "loss": 0.8811, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 10.708860759493671, | |
| "grad_norm": 0.5755253550990348, | |
| "learning_rate": 4.951556604879049e-07, | |
| "loss": 0.9719, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 10.810126582278482, | |
| "grad_norm": 0.5777346624629859, | |
| "learning_rate": 4.322727117869951e-07, | |
| "loss": 1.0179, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 10.91139240506329, | |
| "grad_norm": 0.5579610761750595, | |
| "learning_rate": 3.734784976300165e-07, | |
| "loss": 0.935, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 11.050632911392405, | |
| "grad_norm": 0.600758752593525, | |
| "learning_rate": 3.18825646801314e-07, | |
| "loss": 0.9716, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 11.151898734177216, | |
| "grad_norm": 0.5763985708427721, | |
| "learning_rate": 2.6836308100417874e-07, | |
| "loss": 0.8525, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 11.253164556962025, | |
| "grad_norm": 0.5449406755955025, | |
| "learning_rate": 2.2213597106929608e-07, | |
| "loss": 0.9794, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 11.354430379746836, | |
| "grad_norm": 0.5418503074951475, | |
| "learning_rate": 1.801856965207338e-07, | |
| "loss": 0.9571, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 11.455696202531646, | |
| "grad_norm": 0.539582864821111, | |
| "learning_rate": 1.4254980853566248e-07, | |
| "loss": 0.9822, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 11.556962025316455, | |
| "grad_norm": 0.5355082589962255, | |
| "learning_rate": 1.0926199633097156e-07, | |
| "loss": 0.9958, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 11.658227848101266, | |
| "grad_norm": 0.5296272611774203, | |
| "learning_rate": 8.035205700685167e-08, | |
| "loss": 1.0107, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 11.759493670886076, | |
| "grad_norm": 0.5295685077197022, | |
| "learning_rate": 5.584586887435739e-08, | |
| "loss": 0.913, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 11.860759493670885, | |
| "grad_norm": 0.5323039521138976, | |
| "learning_rate": 3.576536829081323e-08, | |
| "loss": 0.8924, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 11.962025316455696, | |
| "grad_norm": 0.5116161587106628, | |
| "learning_rate": 2.012853002380466e-08, | |
| "loss": 0.9003, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 12.10126582278481, | |
| "grad_norm": 0.5109272463836937, | |
| "learning_rate": 8.949351161324227e-09, | |
| "loss": 0.9439, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 12.20253164556962, | |
| "grad_norm": 0.5127969897549903, | |
| "learning_rate": 2.237838582483387e-09, | |
| "loss": 0.9753, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "grad_norm": 0.5209306010094172, | |
| "learning_rate": 0.0, | |
| "loss": 0.9483, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "step": 117, | |
| "total_flos": 1.689220282402734e+17, | |
| "train_loss": 1.2013191203785758, | |
| "train_runtime": 3789.5587, | |
| "train_samples_per_second": 1.084, | |
| "train_steps_per_second": 0.031 | |
| } | |
| ], | |
| "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": 1.689220282402734e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |