| { | |
| "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": 7.299582901894801, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 1.1424, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 7.8679570675843395, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 1.2252, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 7.861809879113161, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.2371, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 7.2497894877198785, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.2067, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 5.848324242885069, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 1.1472, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 3.051336996827792, | |
| "learning_rate": 5e-06, | |
| "loss": 1.0517, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 2.7201505310412677, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 1.0642, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 4.631799993416774, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.0026, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 5.196731942679453, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 1.081, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 1.0506329113924051, | |
| "grad_norm": 4.744784633633133, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.9863, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.1518987341772151, | |
| "grad_norm": 3.9508241416737055, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.9397, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 1.2531645569620253, | |
| "grad_norm": 3.8766980182332826, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9077, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 1.3544303797468356, | |
| "grad_norm": 3.2301587010318435, | |
| "learning_rate": 9.997762161417517e-06, | |
| "loss": 0.9354, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 1.4556962025316456, | |
| "grad_norm": 2.850492771697211, | |
| "learning_rate": 9.991050648838676e-06, | |
| "loss": 0.8813, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.5569620253164556, | |
| "grad_norm": 3.206856648907991, | |
| "learning_rate": 9.979871469976197e-06, | |
| "loss": 0.9118, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 1.6582278481012658, | |
| "grad_norm": 2.7183042960414796, | |
| "learning_rate": 9.964234631709188e-06, | |
| "loss": 0.8988, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 1.759493670886076, | |
| "grad_norm": 2.436146563577647, | |
| "learning_rate": 9.944154131125643e-06, | |
| "loss": 0.9022, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 1.8607594936708862, | |
| "grad_norm": 2.0863250145434384, | |
| "learning_rate": 9.91964794299315e-06, | |
| "loss": 0.8154, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 1.9620253164556962, | |
| "grad_norm": 1.6292601012290788, | |
| "learning_rate": 9.890738003669029e-06, | |
| "loss": 0.825, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 2.1012658227848102, | |
| "grad_norm": 1.5622399464238044, | |
| "learning_rate": 9.857450191464337e-06, | |
| "loss": 0.7521, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.2025316455696204, | |
| "grad_norm": 1.695359746259927, | |
| "learning_rate": 9.819814303479268e-06, | |
| "loss": 0.7023, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 2.3037974683544302, | |
| "grad_norm": 1.5596505541728882, | |
| "learning_rate": 9.777864028930705e-06, | |
| "loss": 0.7255, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 2.4050632911392404, | |
| "grad_norm": 1.3199344477203505, | |
| "learning_rate": 9.731636918995821e-06, | |
| "loss": 0.6752, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 2.5063291139240507, | |
| "grad_norm": 1.3766440053077489, | |
| "learning_rate": 9.681174353198687e-06, | |
| "loss": 0.6497, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 2.607594936708861, | |
| "grad_norm": 1.4118330177815521, | |
| "learning_rate": 9.626521502369984e-06, | |
| "loss": 0.6814, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.708860759493671, | |
| "grad_norm": 1.2453663764744523, | |
| "learning_rate": 9.567727288213005e-06, | |
| "loss": 0.6299, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 2.810126582278481, | |
| "grad_norm": 1.2863802389291816, | |
| "learning_rate": 9.504844339512096e-06, | |
| "loss": 0.6635, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 2.911392405063291, | |
| "grad_norm": 1.185392203547916, | |
| "learning_rate": 9.437928945022772e-06, | |
| "loss": 0.6322, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 3.050632911392405, | |
| "grad_norm": 1.1085471431194507, | |
| "learning_rate": 9.36704100308565e-06, | |
| "loss": 0.5715, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 3.151898734177215, | |
| "grad_norm": 1.112853327822833, | |
| "learning_rate": 9.292243968009332e-06, | |
| "loss": 0.4968, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 3.2531645569620253, | |
| "grad_norm": 1.1662068165325876, | |
| "learning_rate": 9.213604793270196e-06, | |
| "loss": 0.4938, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 3.3544303797468356, | |
| "grad_norm": 1.9593302066465312, | |
| "learning_rate": 9.131193871579975e-06, | |
| "loss": 0.481, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 3.4556962025316453, | |
| "grad_norm": 1.6118771600693187, | |
| "learning_rate": 9.045084971874738e-06, | |
| "loss": 0.4305, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 3.5569620253164556, | |
| "grad_norm": 1.3262608684836656, | |
| "learning_rate": 8.955355173281709e-06, | |
| "loss": 0.4748, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 3.6582278481012658, | |
| "grad_norm": 1.194898246839981, | |
| "learning_rate": 8.862084796122998e-06, | |
| "loss": 0.4714, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 3.759493670886076, | |
| "grad_norm": 1.0989605424227171, | |
| "learning_rate": 8.765357330018056e-06, | |
| "loss": 0.4253, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 3.8607594936708862, | |
| "grad_norm": 1.2169975324352948, | |
| "learning_rate": 8.665259359149132e-06, | |
| "loss": 0.4119, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 3.962025316455696, | |
| "grad_norm": 1.1189539106114386, | |
| "learning_rate": 8.561880484756726e-06, | |
| "loss": 0.4372, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 4.10126582278481, | |
| "grad_norm": 0.9741532780328732, | |
| "learning_rate": 8.455313244934324e-06, | |
| "loss": 0.398, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 4.2025316455696204, | |
| "grad_norm": 0.9294644404307557, | |
| "learning_rate": 8.345653031794292e-06, | |
| "loss": 0.3151, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 4.30379746835443, | |
| "grad_norm": 1.0269930506276628, | |
| "learning_rate": 8.232998006078998e-06, | |
| "loss": 0.3517, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 4.405063291139241, | |
| "grad_norm": 1.3197553011825012, | |
| "learning_rate": 8.117449009293668e-06, | |
| "loss": 0.257, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 4.506329113924051, | |
| "grad_norm": 1.1535360713757945, | |
| "learning_rate": 7.99910947343957e-06, | |
| "loss": 0.2681, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 4.6075949367088604, | |
| "grad_norm": 1.0776167028989367, | |
| "learning_rate": 7.87808532842837e-06, | |
| "loss": 0.255, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 4.708860759493671, | |
| "grad_norm": 1.0649943086868738, | |
| "learning_rate": 7.754484907260513e-06, | |
| "loss": 0.3157, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 4.810126582278481, | |
| "grad_norm": 1.0591733884921077, | |
| "learning_rate": 7.628418849052523e-06, | |
| "loss": 0.2667, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 4.911392405063291, | |
| "grad_norm": 0.957938045411849, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3087, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 5.050632911392405, | |
| "grad_norm": 0.9044058470012809, | |
| "learning_rate": 7.369343312364994e-06, | |
| "loss": 0.2961, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 5.151898734177215, | |
| "grad_norm": 0.8491076025156571, | |
| "learning_rate": 7.236565741578163e-06, | |
| "loss": 0.2599, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 5.253164556962025, | |
| "grad_norm": 0.7609167747692668, | |
| "learning_rate": 7.101786141547829e-06, | |
| "loss": 0.1466, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 5.3544303797468356, | |
| "grad_norm": 0.9112112919140543, | |
| "learning_rate": 6.965125158269619e-06, | |
| "loss": 0.1755, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 5.455696202531645, | |
| "grad_norm": 1.1380934970286636, | |
| "learning_rate": 6.8267051218319766e-06, | |
| "loss": 0.1634, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 5.556962025316456, | |
| "grad_norm": 1.2076749498070298, | |
| "learning_rate": 6.686649936914151e-06, | |
| "loss": 0.1867, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 5.658227848101266, | |
| "grad_norm": 0.9018661688039192, | |
| "learning_rate": 6.545084971874738e-06, | |
| "loss": 0.1853, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 5.759493670886076, | |
| "grad_norm": 0.8589154495631117, | |
| "learning_rate": 6.402136946530014e-06, | |
| "loss": 0.1902, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 5.860759493670886, | |
| "grad_norm": 1.0291028715137012, | |
| "learning_rate": 6.257933818722544e-06, | |
| "loss": 0.2115, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 5.962025316455696, | |
| "grad_norm": 0.860079721275676, | |
| "learning_rate": 6.112604669781572e-06, | |
| "loss": 0.1855, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 6.10126582278481, | |
| "grad_norm": 0.8486183875048875, | |
| "learning_rate": 5.9662795889777666e-06, | |
| "loss": 0.1389, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 6.2025316455696204, | |
| "grad_norm": 0.763310553078264, | |
| "learning_rate": 5.819089557075689e-06, | |
| "loss": 0.1731, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 6.30379746835443, | |
| "grad_norm": 0.7611212279623061, | |
| "learning_rate": 5.671166329088278e-06, | |
| "loss": 0.1378, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 6.405063291139241, | |
| "grad_norm": 0.8211245158263769, | |
| "learning_rate": 5.522642316338268e-06, | |
| "loss": 0.1367, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 6.506329113924051, | |
| "grad_norm": 0.8999235630581932, | |
| "learning_rate": 5.373650467932122e-06, | |
| "loss": 0.12, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 6.6075949367088604, | |
| "grad_norm": 0.9213162050705732, | |
| "learning_rate": 5.224324151752575e-06, | |
| "loss": 0.1165, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 6.708860759493671, | |
| "grad_norm": 0.9208893459257624, | |
| "learning_rate": 5.074797035076319e-06, | |
| "loss": 0.1356, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 6.810126582278481, | |
| "grad_norm": 0.7188693966627466, | |
| "learning_rate": 4.9252029649236835e-06, | |
| "loss": 0.106, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 6.911392405063291, | |
| "grad_norm": 0.7841053366878892, | |
| "learning_rate": 4.775675848247427e-06, | |
| "loss": 0.0994, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 7.050632911392405, | |
| "grad_norm": 0.8949112676931893, | |
| "learning_rate": 4.626349532067879e-06, | |
| "loss": 0.1163, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 7.151898734177215, | |
| "grad_norm": 0.8239923035043983, | |
| "learning_rate": 4.477357683661734e-06, | |
| "loss": 0.1141, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 7.253164556962025, | |
| "grad_norm": 0.5738134184558951, | |
| "learning_rate": 4.3288336709117246e-06, | |
| "loss": 0.0667, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 7.3544303797468356, | |
| "grad_norm": 0.6193520485355736, | |
| "learning_rate": 4.180910442924312e-06, | |
| "loss": 0.0766, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 7.455696202531645, | |
| "grad_norm": 0.6877788401578242, | |
| "learning_rate": 4.033720411022235e-06, | |
| "loss": 0.1088, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 7.556962025316456, | |
| "grad_norm": 0.6588845919798726, | |
| "learning_rate": 3.887395330218429e-06, | |
| "loss": 0.0681, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 7.658227848101266, | |
| "grad_norm": 0.8491650716522718, | |
| "learning_rate": 3.7420661812774577e-06, | |
| "loss": 0.1116, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 7.759493670886076, | |
| "grad_norm": 0.6741155869434873, | |
| "learning_rate": 3.5978630534699873e-06, | |
| "loss": 0.0882, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 7.860759493670886, | |
| "grad_norm": 0.7425592714206699, | |
| "learning_rate": 3.4549150281252635e-06, | |
| "loss": 0.0559, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 7.962025316455696, | |
| "grad_norm": 0.6083133508920224, | |
| "learning_rate": 3.3133500630858507e-06, | |
| "loss": 0.0862, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 8.10126582278481, | |
| "grad_norm": 0.5703359639834087, | |
| "learning_rate": 3.173294878168025e-06, | |
| "loss": 0.0583, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 8.20253164556962, | |
| "grad_norm": 0.5702178040009944, | |
| "learning_rate": 3.0348748417303826e-06, | |
| "loss": 0.0802, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 8.30379746835443, | |
| "grad_norm": 0.5106953644395711, | |
| "learning_rate": 2.8982138584521734e-06, | |
| "loss": 0.044, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 8.405063291139241, | |
| "grad_norm": 0.4703003855166727, | |
| "learning_rate": 2.7634342584218364e-06, | |
| "loss": 0.0643, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 8.50632911392405, | |
| "grad_norm": 0.4478970915658029, | |
| "learning_rate": 2.6306566876350072e-06, | |
| "loss": 0.0692, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 8.60759493670886, | |
| "grad_norm": 0.41091340284201244, | |
| "learning_rate": 2.5000000000000015e-06, | |
| "loss": 0.0766, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 8.708860759493671, | |
| "grad_norm": 0.4313055699867515, | |
| "learning_rate": 2.371581150947476e-06, | |
| "loss": 0.0566, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 8.810126582278482, | |
| "grad_norm": 0.45399385646128537, | |
| "learning_rate": 2.245515092739488e-06, | |
| "loss": 0.0407, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 8.91139240506329, | |
| "grad_norm": 0.6031405603152155, | |
| "learning_rate": 2.1219146715716332e-06, | |
| "loss": 0.0671, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 9.050632911392405, | |
| "grad_norm": 0.4689104191382645, | |
| "learning_rate": 2.0008905265604316e-06, | |
| "loss": 0.0535, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 9.151898734177216, | |
| "grad_norm": 0.3132411609146735, | |
| "learning_rate": 1.8825509907063328e-06, | |
| "loss": 0.0327, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 9.253164556962025, | |
| "grad_norm": 0.3712598223268114, | |
| "learning_rate": 1.7670019939210025e-06, | |
| "loss": 0.0537, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 9.354430379746836, | |
| "grad_norm": 0.28375561995282444, | |
| "learning_rate": 1.6543469682057105e-06, | |
| "loss": 0.0564, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 9.455696202531646, | |
| "grad_norm": 0.4547891843390434, | |
| "learning_rate": 1.544686755065677e-06, | |
| "loss": 0.0461, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 9.556962025316455, | |
| "grad_norm": 0.31317289225729794, | |
| "learning_rate": 1.438119515243277e-06, | |
| "loss": 0.0311, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 9.658227848101266, | |
| "grad_norm": 0.34801954744390123, | |
| "learning_rate": 1.3347406408508695e-06, | |
| "loss": 0.0489, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 9.759493670886076, | |
| "grad_norm": 0.34034099410579, | |
| "learning_rate": 1.234642669981946e-06, | |
| "loss": 0.0588, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 9.860759493670885, | |
| "grad_norm": 0.2972499793487943, | |
| "learning_rate": 1.137915203877003e-06, | |
| "loss": 0.0211, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 9.962025316455696, | |
| "grad_norm": 0.418233442162911, | |
| "learning_rate": 1.044644826718295e-06, | |
| "loss": 0.0608, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 10.10126582278481, | |
| "grad_norm": 0.2213168587589544, | |
| "learning_rate": 9.549150281252633e-07, | |
| "loss": 0.0193, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 10.20253164556962, | |
| "grad_norm": 0.34702635312362357, | |
| "learning_rate": 8.688061284200266e-07, | |
| "loss": 0.0814, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 10.30379746835443, | |
| "grad_norm": 0.28319706830030994, | |
| "learning_rate": 7.863952067298042e-07, | |
| "loss": 0.0531, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 10.405063291139241, | |
| "grad_norm": 0.24254469503409937, | |
| "learning_rate": 7.077560319906696e-07, | |
| "loss": 0.0179, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 10.50632911392405, | |
| "grad_norm": 0.27893085493438385, | |
| "learning_rate": 6.329589969143518e-07, | |
| "loss": 0.0318, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 10.60759493670886, | |
| "grad_norm": 0.3220464779706176, | |
| "learning_rate": 5.620710549772295e-07, | |
| "loss": 0.0439, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 10.708860759493671, | |
| "grad_norm": 0.29949437910567617, | |
| "learning_rate": 4.951556604879049e-07, | |
| "loss": 0.0356, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 10.810126582278482, | |
| "grad_norm": 0.28661555963363944, | |
| "learning_rate": 4.322727117869951e-07, | |
| "loss": 0.0417, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 10.91139240506329, | |
| "grad_norm": 0.2976812280846049, | |
| "learning_rate": 3.734784976300165e-07, | |
| "loss": 0.0304, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 11.050632911392405, | |
| "grad_norm": 0.2501451112825144, | |
| "learning_rate": 3.18825646801314e-07, | |
| "loss": 0.0407, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 11.151898734177216, | |
| "grad_norm": 0.27705144742246957, | |
| "learning_rate": 2.6836308100417874e-07, | |
| "loss": 0.0487, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 11.253164556962025, | |
| "grad_norm": 0.25313288132229145, | |
| "learning_rate": 2.2213597106929608e-07, | |
| "loss": 0.0364, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 11.354430379746836, | |
| "grad_norm": 0.24880892218795006, | |
| "learning_rate": 1.801856965207338e-07, | |
| "loss": 0.0484, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 11.455696202531646, | |
| "grad_norm": 0.23857099287292122, | |
| "learning_rate": 1.4254980853566248e-07, | |
| "loss": 0.0227, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 11.556962025316455, | |
| "grad_norm": 0.19516465037614425, | |
| "learning_rate": 1.0926199633097156e-07, | |
| "loss": 0.0114, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 11.658227848101266, | |
| "grad_norm": 0.28707453819899287, | |
| "learning_rate": 8.035205700685167e-08, | |
| "loss": 0.0532, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 11.759493670886076, | |
| "grad_norm": 0.2822180553967599, | |
| "learning_rate": 5.584586887435739e-08, | |
| "loss": 0.0516, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 11.860759493670885, | |
| "grad_norm": 0.2872468773457217, | |
| "learning_rate": 3.576536829081323e-08, | |
| "loss": 0.0397, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 11.962025316455696, | |
| "grad_norm": 0.2676970099024659, | |
| "learning_rate": 2.012853002380466e-08, | |
| "loss": 0.0482, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 12.10126582278481, | |
| "grad_norm": 0.24978778628921836, | |
| "learning_rate": 8.949351161324227e-09, | |
| "loss": 0.0316, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 12.20253164556962, | |
| "grad_norm": 0.22565121248393416, | |
| "learning_rate": 2.237838582483387e-09, | |
| "loss": 0.0379, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "grad_norm": 0.26025383330288776, | |
| "learning_rate": 0.0, | |
| "loss": 0.04, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 12.30379746835443, | |
| "step": 117, | |
| "total_flos": 3.924966136977818e+16, | |
| "train_loss": 0.33031460574358446, | |
| "train_runtime": 2250.4603, | |
| "train_samples_per_second": 1.825, | |
| "train_steps_per_second": 0.052 | |
| } | |
| ], | |
| "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": 3.924966136977818e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |