| { | |
| "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": 6.093127927352865, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 1.0465, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 6.376749800604005, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 1.0595, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 6.134204231185576, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.0228, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 5.763454575379169, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.025, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 4.525487104312608, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.9958, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 2.7756526405966326, | |
| "learning_rate": 5e-06, | |
| "loss": 1.0136, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 2.488803007844571, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.9429, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 3.604953739785236, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.9477, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 3.8615522369171145, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.929, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 1.0506329113924051, | |
| "grad_norm": 3.849506320837983, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.9278, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.1518987341772151, | |
| "grad_norm": 3.5155044153587047, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.9158, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 1.2531645569620253, | |
| "grad_norm": 3.024242127694516, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8998, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 1.3544303797468356, | |
| "grad_norm": 2.294327896312, | |
| "learning_rate": 9.997762161417517e-06, | |
| "loss": 0.8402, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 1.4556962025316456, | |
| "grad_norm": 1.5960845186141464, | |
| "learning_rate": 9.991050648838676e-06, | |
| "loss": 0.8374, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.5569620253164556, | |
| "grad_norm": 1.4247689467150386, | |
| "learning_rate": 9.979871469976197e-06, | |
| "loss": 0.8112, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 1.6582278481012658, | |
| "grad_norm": 1.4239117892656694, | |
| "learning_rate": 9.964234631709188e-06, | |
| "loss": 0.816, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 1.759493670886076, | |
| "grad_norm": 1.1016961138715842, | |
| "learning_rate": 9.944154131125643e-06, | |
| "loss": 0.7541, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 1.8607594936708862, | |
| "grad_norm": 1.1350900162618598, | |
| "learning_rate": 9.91964794299315e-06, | |
| "loss": 0.8165, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 1.9620253164556962, | |
| "grad_norm": 1.1030702268983177, | |
| "learning_rate": 9.890738003669029e-06, | |
| "loss": 0.771, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 2.1012658227848102, | |
| "grad_norm": 1.2111126189062653, | |
| "learning_rate": 9.857450191464337e-06, | |
| "loss": 0.7178, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.2025316455696204, | |
| "grad_norm": 0.8944694533129416, | |
| "learning_rate": 9.819814303479268e-06, | |
| "loss": 0.7269, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 2.3037974683544302, | |
| "grad_norm": 0.9964788031302798, | |
| "learning_rate": 9.777864028930705e-06, | |
| "loss": 0.719, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 2.4050632911392404, | |
| "grad_norm": 0.8659417316261437, | |
| "learning_rate": 9.731636918995821e-06, | |
| "loss": 0.726, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 2.5063291139240507, | |
| "grad_norm": 0.9510298302530387, | |
| "learning_rate": 9.681174353198687e-06, | |
| "loss": 0.6787, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 2.607594936708861, | |
| "grad_norm": 0.7701042268025953, | |
| "learning_rate": 9.626521502369984e-06, | |
| "loss": 0.7382, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.708860759493671, | |
| "grad_norm": 0.9006674291637412, | |
| "learning_rate": 9.567727288213005e-06, | |
| "loss": 0.7065, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 2.810126582278481, | |
| "grad_norm": 0.9658095455512565, | |
| "learning_rate": 9.504844339512096e-06, | |
| "loss": 0.7043, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 2.911392405063291, | |
| "grad_norm": 0.8224675251003287, | |
| "learning_rate": 9.437928945022772e-06, | |
| "loss": 0.7208, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 3.050632911392405, | |
| "grad_norm": 1.010660898884819, | |
| "learning_rate": 9.36704100308565e-06, | |
| "loss": 0.6622, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 3.151898734177215, | |
| "grad_norm": 0.7475055771401168, | |
| "learning_rate": 9.292243968009332e-06, | |
| "loss": 0.6449, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 3.2531645569620253, | |
| "grad_norm": 0.728192858237667, | |
| "learning_rate": 9.213604793270196e-06, | |
| "loss": 0.6118, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 3.3544303797468356, | |
| "grad_norm": 0.7441891768849659, | |
| "learning_rate": 9.131193871579975e-06, | |
| "loss": 0.633, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 3.4556962025316453, | |
| "grad_norm": 0.6683330245620003, | |
| "learning_rate": 9.045084971874738e-06, | |
| "loss": 0.6422, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 3.5569620253164556, | |
| "grad_norm": 0.7015787987390913, | |
| "learning_rate": 8.955355173281709e-06, | |
| "loss": 0.6166, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 3.6582278481012658, | |
| "grad_norm": 0.7334621925363748, | |
| "learning_rate": 8.862084796122998e-06, | |
| "loss": 0.5979, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 3.759493670886076, | |
| "grad_norm": 0.6696926119741805, | |
| "learning_rate": 8.765357330018056e-06, | |
| "loss": 0.5899, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 3.8607594936708862, | |
| "grad_norm": 0.7855802125841747, | |
| "learning_rate": 8.665259359149132e-06, | |
| "loss": 0.6135, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 3.962025316455696, | |
| "grad_norm": 0.723294929138997, | |
| "learning_rate": 8.561880484756726e-06, | |
| "loss": 0.6351, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 4.10126582278481, | |
| "grad_norm": 0.7145068913864236, | |
| "learning_rate": 8.455313244934324e-06, | |
| "loss": 0.5904, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 4.2025316455696204, | |
| "grad_norm": 0.6555548301129988, | |
| "learning_rate": 8.345653031794292e-06, | |
| "loss": 0.5544, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 4.30379746835443, | |
| "grad_norm": 0.6519065973717273, | |
| "learning_rate": 8.232998006078998e-06, | |
| "loss": 0.58, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 4.405063291139241, | |
| "grad_norm": 0.634867043327632, | |
| "learning_rate": 8.117449009293668e-06, | |
| "loss": 0.4976, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 4.506329113924051, | |
| "grad_norm": 0.6126180718881084, | |
| "learning_rate": 7.99910947343957e-06, | |
| "loss": 0.4825, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 4.6075949367088604, | |
| "grad_norm": 0.6383735094624874, | |
| "learning_rate": 7.87808532842837e-06, | |
| "loss": 0.5524, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 4.708860759493671, | |
| "grad_norm": 0.6751704459512173, | |
| "learning_rate": 7.754484907260513e-06, | |
| "loss": 0.5276, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 4.810126582278481, | |
| "grad_norm": 0.6250362549179618, | |
| "learning_rate": 7.628418849052523e-06, | |
| "loss": 0.5188, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 4.911392405063291, | |
| "grad_norm": 0.6388687654056359, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.5732, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 5.050632911392405, | |
| "grad_norm": 0.643588325559921, | |
| "learning_rate": 7.369343312364994e-06, | |
| "loss": 0.5188, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 5.151898734177215, | |
| "grad_norm": 0.6544614320573806, | |
| "learning_rate": 7.236565741578163e-06, | |
| "loss": 0.4689, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 5.253164556962025, | |
| "grad_norm": 0.6193706127526757, | |
| "learning_rate": 7.101786141547829e-06, | |
| "loss": 0.4673, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 5.3544303797468356, | |
| "grad_norm": 0.6997314537621265, | |
| "learning_rate": 6.965125158269619e-06, | |
| "loss": 0.4431, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 5.455696202531645, | |
| "grad_norm": 0.6100449445272762, | |
| "learning_rate": 6.8267051218319766e-06, | |
| "loss": 0.4591, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 5.556962025316456, | |
| "grad_norm": 0.6436653081504855, | |
| "learning_rate": 6.686649936914151e-06, | |
| "loss": 0.5004, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 5.658227848101266, | |
| "grad_norm": 0.6960171144760199, | |
| "learning_rate": 6.545084971874738e-06, | |
| "loss": 0.4941, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 5.759493670886076, | |
| "grad_norm": 0.6986240890121468, | |
| "learning_rate": 6.402136946530014e-06, | |
| "loss": 0.4854, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 5.860759493670886, | |
| "grad_norm": 0.5953494033916403, | |
| "learning_rate": 6.257933818722544e-06, | |
| "loss": 0.4659, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 5.962025316455696, | |
| "grad_norm": 0.563769645700779, | |
| "learning_rate": 6.112604669781572e-06, | |
| "loss": 0.4791, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 6.10126582278481, | |
| "grad_norm": 0.6824803711165172, | |
| "learning_rate": 5.9662795889777666e-06, | |
| "loss": 0.3796, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 6.2025316455696204, | |
| "grad_norm": 0.5777404281034876, | |
| "learning_rate": 5.819089557075689e-06, | |
| "loss": 0.4297, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 6.30379746835443, | |
| "grad_norm": 0.8101764986930826, | |
| "learning_rate": 5.671166329088278e-06, | |
| "loss": 0.4154, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 6.405063291139241, | |
| "grad_norm": 0.6656993867530263, | |
| "learning_rate": 5.522642316338268e-06, | |
| "loss": 0.4374, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 6.506329113924051, | |
| "grad_norm": 0.6485857692094394, | |
| "learning_rate": 5.373650467932122e-06, | |
| "loss": 0.4026, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 6.6075949367088604, | |
| "grad_norm": 0.6638069216266845, | |
| "learning_rate": 5.224324151752575e-06, | |
| "loss": 0.392, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 6.708860759493671, | |
| "grad_norm": 0.5542938382680552, | |
| "learning_rate": 5.074797035076319e-06, | |
| "loss": 0.4205, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 6.810126582278481, | |
| "grad_norm": 0.5757651496141476, | |
| "learning_rate": 4.9252029649236835e-06, | |
| "loss": 0.4254, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 6.911392405063291, | |
| "grad_norm": 0.6186876933480754, | |
| "learning_rate": 4.775675848247427e-06, | |
| "loss": 0.3953, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 7.050632911392405, | |
| "grad_norm": 0.5636596152497861, | |
| "learning_rate": 4.626349532067879e-06, | |
| "loss": 0.386, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 7.151898734177215, | |
| "grad_norm": 0.6279576207714367, | |
| "learning_rate": 4.477357683661734e-06, | |
| "loss": 0.3778, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 7.253164556962025, | |
| "grad_norm": 0.510638676310019, | |
| "learning_rate": 4.3288336709117246e-06, | |
| "loss": 0.3687, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 7.3544303797468356, | |
| "grad_norm": 0.6295761327837356, | |
| "learning_rate": 4.180910442924312e-06, | |
| "loss": 0.367, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 7.455696202531645, | |
| "grad_norm": 0.6037769901395187, | |
| "learning_rate": 4.033720411022235e-06, | |
| "loss": 0.3505, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 7.556962025316456, | |
| "grad_norm": 0.5657764599634766, | |
| "learning_rate": 3.887395330218429e-06, | |
| "loss": 0.3455, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 7.658227848101266, | |
| "grad_norm": 0.6711486184759682, | |
| "learning_rate": 3.7420661812774577e-06, | |
| "loss": 0.3271, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 7.759493670886076, | |
| "grad_norm": 0.5595752566668041, | |
| "learning_rate": 3.5978630534699873e-06, | |
| "loss": 0.3611, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 7.860759493670886, | |
| "grad_norm": 0.5384341752905978, | |
| "learning_rate": 3.4549150281252635e-06, | |
| "loss": 0.3608, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 7.962025316455696, | |
| "grad_norm": 0.5691066973615871, | |
| "learning_rate": 3.3133500630858507e-06, | |
| "loss": 0.3467, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 8.10126582278481, | |
| "grad_norm": 0.5186698594500165, | |
| "learning_rate": 3.173294878168025e-06, | |
| "loss": 0.2888, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 8.20253164556962, | |
| "grad_norm": 0.5534197087640107, | |
| "learning_rate": 3.0348748417303826e-06, | |
| "loss": 0.3213, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 8.30379746835443, | |
| "grad_norm": 0.49709389364299633, | |
| "learning_rate": 2.8982138584521734e-06, | |
| "loss": 0.3021, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 8.405063291139241, | |
| "grad_norm": 0.6010934672082207, | |
| "learning_rate": 2.7634342584218364e-06, | |
| "loss": 0.314, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 8.50632911392405, | |
| "grad_norm": 0.5395680222621423, | |
| "learning_rate": 2.6306566876350072e-06, | |
| "loss": 0.3516, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 8.60759493670886, | |
| "grad_norm": 0.5185614060562257, | |
| "learning_rate": 2.5000000000000015e-06, | |
| "loss": 0.3206, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 8.708860759493671, | |
| "grad_norm": 0.49967476814912687, | |
| "learning_rate": 2.371581150947476e-06, | |
| "loss": 0.2944, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 8.810126582278482, | |
| "grad_norm": 0.5061530939422232, | |
| "learning_rate": 2.245515092739488e-06, | |
| "loss": 0.341, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 8.91139240506329, | |
| "grad_norm": 0.4909648345098446, | |
| "learning_rate": 2.1219146715716332e-06, | |
| "loss": 0.3064, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 9.050632911392405, | |
| "grad_norm": 0.5652583213655021, | |
| "learning_rate": 2.0008905265604316e-06, | |
| "loss": 0.2845, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 9.151898734177216, | |
| "grad_norm": 0.5317096330076252, | |
| "learning_rate": 1.8825509907063328e-06, | |
| "loss": 0.301, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 9.253164556962025, | |
| "grad_norm": 0.5473277995837466, | |
| "learning_rate": 1.7670019939210025e-06, | |
| "loss": 0.3262, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 9.354430379746836, | |
| "grad_norm": 0.5403409229947518, | |
| "learning_rate": 1.6543469682057105e-06, | |
| "loss": 0.2709, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 9.455696202531646, | |
| "grad_norm": 0.5536659315371194, | |
| "learning_rate": 1.544686755065677e-06, | |
| "loss": 0.2877, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 9.556962025316455, | |
| "grad_norm": 0.5385579167471305, | |
| "learning_rate": 1.438119515243277e-06, | |
| "loss": 0.2852, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 9.658227848101266, | |
| "grad_norm": 0.6147947834715497, | |
| "learning_rate": 1.3347406408508695e-06, | |
| "loss": 0.2937, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 9.759493670886076, | |
| "grad_norm": 0.6365599397873863, | |
| "learning_rate": 1.234642669981946e-06, | |
| "loss": 0.2254, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 9.860759493670885, | |
| "grad_norm": 0.5308077417679637, | |
| "learning_rate": 1.137915203877003e-06, | |
| "loss": 0.2844, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 9.962025316455696, | |
| "grad_norm": 0.49730719245253746, | |
| "learning_rate": 1.044644826718295e-06, | |
| "loss": 0.2863, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 10.10126582278481, | |
| "grad_norm": 0.5112130720046859, | |
| "learning_rate": 9.549150281252633e-07, | |
| "loss": 0.2497, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 10.20253164556962, | |
| "grad_norm": 0.49510148651875857, | |
| "learning_rate": 8.688061284200266e-07, | |
| "loss": 0.291, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 10.30379746835443, | |
| "grad_norm": 0.4614716490489081, | |
| "learning_rate": 7.863952067298042e-07, | |
| "loss": 0.2758, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 10.405063291139241, | |
| "grad_norm": 0.476649888745995, | |
| "learning_rate": 7.077560319906696e-07, | |
| "loss": 0.2354, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 10.50632911392405, | |
| "grad_norm": 0.5243721821647354, | |
| "learning_rate": 6.329589969143518e-07, | |
| "loss": 0.2915, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 10.60759493670886, | |
| "grad_norm": 0.4794666138385684, | |
| "learning_rate": 5.620710549772295e-07, | |
| "loss": 0.3087, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 10.708860759493671, | |
| "grad_norm": 0.4522704956058331, | |
| "learning_rate": 4.951556604879049e-07, | |
| "loss": 0.2708, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 10.810126582278482, | |
| "grad_norm": 0.4492683183416952, | |
| "learning_rate": 4.322727117869951e-07, | |
| "loss": 0.2441, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 10.91139240506329, | |
| "grad_norm": 0.445972108320586, | |
| "learning_rate": 3.734784976300165e-07, | |
| "loss": 0.255, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 11.050632911392405, | |
| "grad_norm": 0.4526618529426575, | |
| "learning_rate": 3.18825646801314e-07, | |
| "loss": 0.261, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 11.151898734177216, | |
| "grad_norm": 0.47197301700611427, | |
| "learning_rate": 2.6836308100417874e-07, | |
| "loss": 0.2432, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 11.253164556962025, | |
| "grad_norm": 0.4661672173811326, | |
| "learning_rate": 2.2213597106929608e-07, | |
| "loss": 0.2638, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 11.354430379746836, | |
| "grad_norm": 0.491824675531964, | |
| "learning_rate": 1.801856965207338e-07, | |
| "loss": 0.2171, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 11.455696202531646, | |
| "grad_norm": 0.46408120594682206, | |
| "learning_rate": 1.4254980853566248e-07, | |
| "loss": 0.2675, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 11.556962025316455, | |
| "grad_norm": 0.4289647062372751, | |
| "learning_rate": 1.0926199633097156e-07, | |
| "loss": 0.2768, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 11.658227848101266, | |
| "grad_norm": 0.469074565919927, | |
| "learning_rate": 8.035205700685167e-08, | |
| "loss": 0.2427, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 11.759493670886076, | |
| "grad_norm": 0.4311892072050466, | |
| "learning_rate": 5.584586887435739e-08, | |
| "loss": 0.2825, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 11.860759493670885, | |
| "grad_norm": 0.4359995544489271, | |
| "learning_rate": 3.576536829081323e-08, | |
| "loss": 0.2302, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 11.962025316455696, | |
| "grad_norm": 0.44634914594857533, | |
| "learning_rate": 2.012853002380466e-08, | |
| "loss": 0.2679, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 12.10126582278481, | |
| "grad_norm": 0.4250683605787237, | |
| "learning_rate": 8.949351161324227e-09, | |
| "loss": 0.2718, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 12.20253164556962, | |
| "grad_norm": 0.4717022769055403, | |
| "learning_rate": 2.237838582483387e-09, | |
| "loss": 0.2831, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "grad_norm": 0.409401020402968, | |
| "learning_rate": 0.0, | |
| "loss": 0.2209, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "step": 117, | |
| "total_flos": 1.473220092006236e+17, | |
| "train_loss": 0.49950172771246004, | |
| "train_runtime": 5489.261, | |
| "train_samples_per_second": 0.748, | |
| "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": 1.473220092006236e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |