| { | |
| "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": 6.546009533031287, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 1.0179, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 6.460799362837899, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 1.0183, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 6.332065761714634, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.0191, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 6.212814677568616, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.0676, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 4.7287446661188035, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.9786, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 2.7777179200668574, | |
| "learning_rate": 5e-06, | |
| "loss": 0.9481, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 2.613951749021039, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.9575, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 3.8348663962376137, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.9209, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 4.35182857951712, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.9102, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 1.0506329113924051, | |
| "grad_norm": 4.169721379167235, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.9366, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.1518987341772151, | |
| "grad_norm": 3.680839452144861, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.9049, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 1.2531645569620253, | |
| "grad_norm": 3.0763292190467966, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8525, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 1.3544303797468356, | |
| "grad_norm": 2.2677743431179835, | |
| "learning_rate": 9.997762161417517e-06, | |
| "loss": 0.8504, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 1.4556962025316456, | |
| "grad_norm": 1.5595552228079432, | |
| "learning_rate": 9.991050648838676e-06, | |
| "loss": 0.8096, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.5569620253164556, | |
| "grad_norm": 1.536676547610973, | |
| "learning_rate": 9.979871469976197e-06, | |
| "loss": 0.7859, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 1.6582278481012658, | |
| "grad_norm": 1.4734589039716286, | |
| "learning_rate": 9.964234631709188e-06, | |
| "loss": 0.7917, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 1.759493670886076, | |
| "grad_norm": 1.1590212230198378, | |
| "learning_rate": 9.944154131125643e-06, | |
| "loss": 0.7685, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 1.8607594936708862, | |
| "grad_norm": 1.004405339326733, | |
| "learning_rate": 9.91964794299315e-06, | |
| "loss": 0.7392, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 1.9620253164556962, | |
| "grad_norm": 1.1234894312354784, | |
| "learning_rate": 9.890738003669029e-06, | |
| "loss": 0.7411, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 2.1012658227848102, | |
| "grad_norm": 0.8713502001355584, | |
| "learning_rate": 9.857450191464337e-06, | |
| "loss": 0.7268, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.2025316455696204, | |
| "grad_norm": 0.9414835222648549, | |
| "learning_rate": 9.819814303479268e-06, | |
| "loss": 0.6639, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 2.3037974683544302, | |
| "grad_norm": 0.9258413227375791, | |
| "learning_rate": 9.777864028930705e-06, | |
| "loss": 0.7213, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 2.4050632911392404, | |
| "grad_norm": 1.002092836679983, | |
| "learning_rate": 9.731636918995821e-06, | |
| "loss": 0.6793, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 2.5063291139240507, | |
| "grad_norm": 0.9045518131394497, | |
| "learning_rate": 9.681174353198687e-06, | |
| "loss": 0.6959, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 2.607594936708861, | |
| "grad_norm": 0.9769135172818063, | |
| "learning_rate": 9.626521502369984e-06, | |
| "loss": 0.6782, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.708860759493671, | |
| "grad_norm": 0.9160217984512808, | |
| "learning_rate": 9.567727288213005e-06, | |
| "loss": 0.6589, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 2.810126582278481, | |
| "grad_norm": 0.852604515984319, | |
| "learning_rate": 9.504844339512096e-06, | |
| "loss": 0.6789, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 2.911392405063291, | |
| "grad_norm": 1.0542347872826128, | |
| "learning_rate": 9.437928945022772e-06, | |
| "loss": 0.6657, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 3.050632911392405, | |
| "grad_norm": 0.8361647229145174, | |
| "learning_rate": 9.36704100308565e-06, | |
| "loss": 0.633, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 3.151898734177215, | |
| "grad_norm": 0.8401459237265612, | |
| "learning_rate": 9.292243968009332e-06, | |
| "loss": 0.627, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 3.2531645569620253, | |
| "grad_norm": 0.798406897352322, | |
| "learning_rate": 9.213604793270196e-06, | |
| "loss": 0.5847, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 3.3544303797468356, | |
| "grad_norm": 0.7785463168108498, | |
| "learning_rate": 9.131193871579975e-06, | |
| "loss": 0.6225, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 3.4556962025316453, | |
| "grad_norm": 0.811784071692141, | |
| "learning_rate": 9.045084971874738e-06, | |
| "loss": 0.5831, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 3.5569620253164556, | |
| "grad_norm": 0.7579955542500708, | |
| "learning_rate": 8.955355173281709e-06, | |
| "loss": 0.575, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 3.6582278481012658, | |
| "grad_norm": 0.8659809911621335, | |
| "learning_rate": 8.862084796122998e-06, | |
| "loss": 0.586, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 3.759493670886076, | |
| "grad_norm": 0.8425842114890063, | |
| "learning_rate": 8.765357330018056e-06, | |
| "loss": 0.6199, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 3.8607594936708862, | |
| "grad_norm": 0.9104748398758332, | |
| "learning_rate": 8.665259359149132e-06, | |
| "loss": 0.5611, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 3.962025316455696, | |
| "grad_norm": 0.8559298089159043, | |
| "learning_rate": 8.561880484756726e-06, | |
| "loss": 0.588, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 4.10126582278481, | |
| "grad_norm": 0.6428700019417081, | |
| "learning_rate": 8.455313244934324e-06, | |
| "loss": 0.5639, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 4.2025316455696204, | |
| "grad_norm": 0.6941203934435238, | |
| "learning_rate": 8.345653031794292e-06, | |
| "loss": 0.5407, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 4.30379746835443, | |
| "grad_norm": 0.693205219032939, | |
| "learning_rate": 8.232998006078998e-06, | |
| "loss": 0.5205, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 4.405063291139241, | |
| "grad_norm": 0.6263066276526804, | |
| "learning_rate": 8.117449009293668e-06, | |
| "loss": 0.5072, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 4.506329113924051, | |
| "grad_norm": 0.7014603226763829, | |
| "learning_rate": 7.99910947343957e-06, | |
| "loss": 0.4799, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 4.6075949367088604, | |
| "grad_norm": 0.6882083439072155, | |
| "learning_rate": 7.87808532842837e-06, | |
| "loss": 0.5162, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 4.708860759493671, | |
| "grad_norm": 0.62937439204103, | |
| "learning_rate": 7.754484907260513e-06, | |
| "loss": 0.5004, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 4.810126582278481, | |
| "grad_norm": 0.7149728999417626, | |
| "learning_rate": 7.628418849052523e-06, | |
| "loss": 0.4682, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 4.911392405063291, | |
| "grad_norm": 0.6832943836501206, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.5221, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 5.050632911392405, | |
| "grad_norm": 0.6822445566734908, | |
| "learning_rate": 7.369343312364994e-06, | |
| "loss": 0.4407, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 5.151898734177215, | |
| "grad_norm": 0.5791117912757918, | |
| "learning_rate": 7.236565741578163e-06, | |
| "loss": 0.4529, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 5.253164556962025, | |
| "grad_norm": 0.682442212729935, | |
| "learning_rate": 7.101786141547829e-06, | |
| "loss": 0.4536, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 5.3544303797468356, | |
| "grad_norm": 0.6658822753925351, | |
| "learning_rate": 6.965125158269619e-06, | |
| "loss": 0.4647, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 5.455696202531645, | |
| "grad_norm": 0.6916148619591792, | |
| "learning_rate": 6.8267051218319766e-06, | |
| "loss": 0.4266, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 5.556962025316456, | |
| "grad_norm": 0.6558155132689918, | |
| "learning_rate": 6.686649936914151e-06, | |
| "loss": 0.4344, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 5.658227848101266, | |
| "grad_norm": 0.5901666069165875, | |
| "learning_rate": 6.545084971874738e-06, | |
| "loss": 0.4983, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 5.759493670886076, | |
| "grad_norm": 0.6354510055047695, | |
| "learning_rate": 6.402136946530014e-06, | |
| "loss": 0.4208, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 5.860759493670886, | |
| "grad_norm": 0.6617892714492388, | |
| "learning_rate": 6.257933818722544e-06, | |
| "loss": 0.4858, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 5.962025316455696, | |
| "grad_norm": 0.5951723333271441, | |
| "learning_rate": 6.112604669781572e-06, | |
| "loss": 0.4202, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 6.10126582278481, | |
| "grad_norm": 0.7286362666813779, | |
| "learning_rate": 5.9662795889777666e-06, | |
| "loss": 0.4054, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 6.2025316455696204, | |
| "grad_norm": 0.5693058773864588, | |
| "learning_rate": 5.819089557075689e-06, | |
| "loss": 0.4125, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 6.30379746835443, | |
| "grad_norm": 0.6380984932603142, | |
| "learning_rate": 5.671166329088278e-06, | |
| "loss": 0.4284, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 6.405063291139241, | |
| "grad_norm": 0.6472799307403144, | |
| "learning_rate": 5.522642316338268e-06, | |
| "loss": 0.3876, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 6.506329113924051, | |
| "grad_norm": 0.6040003054463189, | |
| "learning_rate": 5.373650467932122e-06, | |
| "loss": 0.3913, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 6.6075949367088604, | |
| "grad_norm": 0.7335515508341466, | |
| "learning_rate": 5.224324151752575e-06, | |
| "loss": 0.3677, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 6.708860759493671, | |
| "grad_norm": 0.5490801269030343, | |
| "learning_rate": 5.074797035076319e-06, | |
| "loss": 0.3488, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 6.810126582278481, | |
| "grad_norm": 0.6086098788544544, | |
| "learning_rate": 4.9252029649236835e-06, | |
| "loss": 0.3614, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 6.911392405063291, | |
| "grad_norm": 0.5992575440705068, | |
| "learning_rate": 4.775675848247427e-06, | |
| "loss": 0.3642, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 7.050632911392405, | |
| "grad_norm": 0.5676686997653896, | |
| "learning_rate": 4.626349532067879e-06, | |
| "loss": 0.3453, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 7.151898734177215, | |
| "grad_norm": 0.5741894338625257, | |
| "learning_rate": 4.477357683661734e-06, | |
| "loss": 0.3557, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 7.253164556962025, | |
| "grad_norm": 0.5456905272285272, | |
| "learning_rate": 4.3288336709117246e-06, | |
| "loss": 0.3288, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 7.3544303797468356, | |
| "grad_norm": 0.6901700868402513, | |
| "learning_rate": 4.180910442924312e-06, | |
| "loss": 0.3227, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 7.455696202531645, | |
| "grad_norm": 0.6919666602864087, | |
| "learning_rate": 4.033720411022235e-06, | |
| "loss": 0.3652, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 7.556962025316456, | |
| "grad_norm": 0.6491412548699199, | |
| "learning_rate": 3.887395330218429e-06, | |
| "loss": 0.3718, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 7.658227848101266, | |
| "grad_norm": 0.6331333131436986, | |
| "learning_rate": 3.7420661812774577e-06, | |
| "loss": 0.284, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 7.759493670886076, | |
| "grad_norm": 0.5632802484819347, | |
| "learning_rate": 3.5978630534699873e-06, | |
| "loss": 0.3393, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 7.860759493670886, | |
| "grad_norm": 0.5294198963211196, | |
| "learning_rate": 3.4549150281252635e-06, | |
| "loss": 0.3094, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 7.962025316455696, | |
| "grad_norm": 0.6286411745612709, | |
| "learning_rate": 3.3133500630858507e-06, | |
| "loss": 0.3093, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 8.10126582278481, | |
| "grad_norm": 0.5522931278991092, | |
| "learning_rate": 3.173294878168025e-06, | |
| "loss": 0.3069, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 8.20253164556962, | |
| "grad_norm": 0.5392969070232384, | |
| "learning_rate": 3.0348748417303826e-06, | |
| "loss": 0.3053, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 8.30379746835443, | |
| "grad_norm": 0.5053894308026001, | |
| "learning_rate": 2.8982138584521734e-06, | |
| "loss": 0.2778, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 8.405063291139241, | |
| "grad_norm": 0.5630020125868113, | |
| "learning_rate": 2.7634342584218364e-06, | |
| "loss": 0.2992, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 8.50632911392405, | |
| "grad_norm": 0.551765484774629, | |
| "learning_rate": 2.6306566876350072e-06, | |
| "loss": 0.297, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 8.60759493670886, | |
| "grad_norm": 0.5575647289561099, | |
| "learning_rate": 2.5000000000000015e-06, | |
| "loss": 0.3169, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 8.708860759493671, | |
| "grad_norm": 0.5140632699613491, | |
| "learning_rate": 2.371581150947476e-06, | |
| "loss": 0.2725, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 8.810126582278482, | |
| "grad_norm": 0.47997163922167685, | |
| "learning_rate": 2.245515092739488e-06, | |
| "loss": 0.2936, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 8.91139240506329, | |
| "grad_norm": 0.4859192227954078, | |
| "learning_rate": 2.1219146715716332e-06, | |
| "loss": 0.2858, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 9.050632911392405, | |
| "grad_norm": 0.5411447217328182, | |
| "learning_rate": 2.0008905265604316e-06, | |
| "loss": 0.245, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 9.151898734177216, | |
| "grad_norm": 0.6184713042762273, | |
| "learning_rate": 1.8825509907063328e-06, | |
| "loss": 0.263, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 9.253164556962025, | |
| "grad_norm": 0.4670465816516778, | |
| "learning_rate": 1.7670019939210025e-06, | |
| "loss": 0.2263, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 9.354430379746836, | |
| "grad_norm": 0.6334298168655356, | |
| "learning_rate": 1.6543469682057105e-06, | |
| "loss": 0.2687, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 9.455696202531646, | |
| "grad_norm": 0.7002478830469511, | |
| "learning_rate": 1.544686755065677e-06, | |
| "loss": 0.2365, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 9.556962025316455, | |
| "grad_norm": 0.5880456768176067, | |
| "learning_rate": 1.438119515243277e-06, | |
| "loss": 0.2839, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 9.658227848101266, | |
| "grad_norm": 0.5032015419834202, | |
| "learning_rate": 1.3347406408508695e-06, | |
| "loss": 0.2935, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 9.759493670886076, | |
| "grad_norm": 0.6082684001275916, | |
| "learning_rate": 1.234642669981946e-06, | |
| "loss": 0.2612, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 9.860759493670885, | |
| "grad_norm": 0.6426627482872871, | |
| "learning_rate": 1.137915203877003e-06, | |
| "loss": 0.2828, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 9.962025316455696, | |
| "grad_norm": 0.6012394971886591, | |
| "learning_rate": 1.044644826718295e-06, | |
| "loss": 0.2608, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 10.10126582278481, | |
| "grad_norm": 0.5668410845892721, | |
| "learning_rate": 9.549150281252633e-07, | |
| "loss": 0.2621, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 10.20253164556962, | |
| "grad_norm": 0.5310326662066904, | |
| "learning_rate": 8.688061284200266e-07, | |
| "loss": 0.2955, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 10.30379746835443, | |
| "grad_norm": 0.44957691016662316, | |
| "learning_rate": 7.863952067298042e-07, | |
| "loss": 0.2385, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 10.405063291139241, | |
| "grad_norm": 0.5145974097097716, | |
| "learning_rate": 7.077560319906696e-07, | |
| "loss": 0.1838, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 10.50632911392405, | |
| "grad_norm": 0.7889381784170816, | |
| "learning_rate": 6.329589969143518e-07, | |
| "loss": 0.2734, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 10.60759493670886, | |
| "grad_norm": 0.6239124366991347, | |
| "learning_rate": 5.620710549772295e-07, | |
| "loss": 0.2139, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 10.708860759493671, | |
| "grad_norm": 0.5530707992253046, | |
| "learning_rate": 4.951556604879049e-07, | |
| "loss": 0.2523, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 10.810126582278482, | |
| "grad_norm": 0.5185905793224019, | |
| "learning_rate": 4.322727117869951e-07, | |
| "loss": 0.2675, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 10.91139240506329, | |
| "grad_norm": 0.45547899663391855, | |
| "learning_rate": 3.734784976300165e-07, | |
| "loss": 0.2326, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 11.050632911392405, | |
| "grad_norm": 0.45598050696439335, | |
| "learning_rate": 3.18825646801314e-07, | |
| "loss": 0.2589, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 11.151898734177216, | |
| "grad_norm": 0.5451557423128508, | |
| "learning_rate": 2.6836308100417874e-07, | |
| "loss": 0.2152, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 11.253164556962025, | |
| "grad_norm": 0.5336777529047616, | |
| "learning_rate": 2.2213597106929608e-07, | |
| "loss": 0.2209, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 11.354430379746836, | |
| "grad_norm": 0.548460153552006, | |
| "learning_rate": 1.801856965207338e-07, | |
| "loss": 0.2699, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 11.455696202531646, | |
| "grad_norm": 0.48957272210988534, | |
| "learning_rate": 1.4254980853566248e-07, | |
| "loss": 0.2582, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 11.556962025316455, | |
| "grad_norm": 0.441175274421175, | |
| "learning_rate": 1.0926199633097156e-07, | |
| "loss": 0.2489, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 11.658227848101266, | |
| "grad_norm": 0.46794935792199505, | |
| "learning_rate": 8.035205700685167e-08, | |
| "loss": 0.2069, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 11.759493670886076, | |
| "grad_norm": 0.40708726908026116, | |
| "learning_rate": 5.584586887435739e-08, | |
| "loss": 0.2486, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 11.860759493670885, | |
| "grad_norm": 0.44319163488452995, | |
| "learning_rate": 3.576536829081323e-08, | |
| "loss": 0.2424, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 11.962025316455696, | |
| "grad_norm": 0.4277006145736423, | |
| "learning_rate": 2.012853002380466e-08, | |
| "loss": 0.2637, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 12.10126582278481, | |
| "grad_norm": 0.42790757929601125, | |
| "learning_rate": 8.949351161324227e-09, | |
| "loss": 0.2358, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 12.20253164556962, | |
| "grad_norm": 0.4472063613414348, | |
| "learning_rate": 2.237838582483387e-09, | |
| "loss": 0.2522, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "grad_norm": 0.4154156187584164, | |
| "learning_rate": 0.0, | |
| "loss": 0.2439, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "step": 117, | |
| "total_flos": 1.3482231997661184e+17, | |
| "train_loss": 0.4764612058543751, | |
| "train_runtime": 5513.9458, | |
| "train_samples_per_second": 0.745, | |
| "train_steps_per_second": 0.021 | |
| } | |
| ], | |
| "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": 1.3482231997661184e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |