| { | |
| "best_metric": 1.514884114265442, | |
| "best_model_checkpoint": "./pippa-sharegpt-13b-qlora/checkpoint-350", | |
| "epoch": 2.364864864864865, | |
| "eval_steps": 50, | |
| "global_step": 350, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-05, | |
| "loss": 2.0773, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4e-05, | |
| "loss": 2.0879, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6e-05, | |
| "loss": 2.0647, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8e-05, | |
| "loss": 1.9962, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001, | |
| "loss": 1.9106, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00012, | |
| "loss": 1.8651, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00014, | |
| "loss": 1.8169, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00016, | |
| "loss": 1.8535, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00018, | |
| "loss": 1.8226, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 1.7774, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00019999738007780348, | |
| "loss": 1.7297, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00019998952044849376, | |
| "loss": 1.7607, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00019997642152390314, | |
| "loss": 1.7323, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00019995808399039496, | |
| "loss": 1.6993, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00019993450880882735, | |
| "loss": 1.7053, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00019990569721450326, | |
| "loss": 1.7207, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00019987165071710527, | |
| "loss": 1.667, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00019983237110061697, | |
| "loss": 1.7375, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0001997878604232291, | |
| "loss": 1.725, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00019973812101723188, | |
| "loss": 1.658, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0001996831554888928, | |
| "loss": 1.6804, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00019962296671832003, | |
| "loss": 1.7273, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00019955755785931145, | |
| "loss": 1.6712, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00019948693233918952, | |
| "loss": 1.6261, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0001994110938586216, | |
| "loss": 1.673, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00019933004639142605, | |
| "loss": 1.6796, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00019924379418436404, | |
| "loss": 1.6701, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000199152341756917, | |
| "loss": 1.6619, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00019905569390104986, | |
| "loss": 1.6282, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00019895385568095982, | |
| "loss": 1.7152, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00019884683243281116, | |
| "loss": 1.6431, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00019873462976445553, | |
| "loss": 1.7584, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00019861725355513823, | |
| "loss": 1.6342, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00019849470995518992, | |
| "loss": 1.652, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00019836700538570457, | |
| "loss": 1.6472, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0001982341465382029, | |
| "loss": 1.6961, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00019809614037428176, | |
| "loss": 1.6556, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00019795299412524945, | |
| "loss": 1.6386, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00019780471529174664, | |
| "loss": 1.6692, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00019765131164335345, | |
| "loss": 1.5955, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00019749279121818235, | |
| "loss": 1.6292, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000197329162322457, | |
| "loss": 1.648, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00019716043353007693, | |
| "loss": 1.7011, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00019698661368216817, | |
| "loss": 1.6319, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00019680771188662044, | |
| "loss": 1.6143, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00019662373751760934, | |
| "loss": 1.6526, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0001964347002151056, | |
| "loss": 1.5946, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00019624060988436966, | |
| "loss": 1.6474, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00019604147669543282, | |
| "loss": 1.6756, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0001958373110825644, | |
| "loss": 1.6447, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "eval_loss": 1.6321444511413574, | |
| "eval_runtime": 311.9756, | |
| "eval_samples_per_second": 4.795, | |
| "eval_steps_per_second": 1.199, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00019562812374372473, | |
| "loss": 1.6902, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00019541392564000488, | |
| "loss": 1.6152, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0001951947279950522, | |
| "loss": 1.631, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00019497054229448223, | |
| "loss": 1.6445, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00019474138028527675, | |
| "loss": 1.6278, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0001945072539751685, | |
| "loss": 1.6202, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00019426817563201177, | |
| "loss": 1.6196, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00019402415778313977, | |
| "loss": 1.6062, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00019377521321470805, | |
| "loss": 1.5883, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00019352135497102463, | |
| "loss": 1.6769, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00019326259635386644, | |
| "loss": 1.6134, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0001929989509217824, | |
| "loss": 1.5884, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00019273043248938288, | |
| "loss": 1.6232, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0001924570551266159, | |
| "loss": 1.6303, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00019217883315802991, | |
| "loss": 1.6585, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00019189578116202307, | |
| "loss": 1.6264, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00019160791397007957, | |
| "loss": 1.5944, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00019131524666599233, | |
| "loss": 1.6371, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00019101779458507263, | |
| "loss": 1.592, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00019071557331334669, | |
| "loss": 1.6369, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00019040859868673887, | |
| "loss": 1.6355, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0001900968867902419, | |
| "loss": 1.6435, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00018978045395707418, | |
| "loss": 1.6074, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00018945931676782373, | |
| "loss": 1.5909, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0001891334920495795, | |
| "loss": 1.6151, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0001888029968750498, | |
| "loss": 1.5976, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0001884678485616675, | |
| "loss": 1.5716, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00018812806467068268, | |
| "loss": 1.5879, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00018778366300624245, | |
| "loss": 1.5664, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00018743466161445823, | |
| "loss": 1.6096, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00018708107878245977, | |
| "loss": 1.603, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00018672293303743738, | |
| "loss": 1.6123, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00018636024314567067, | |
| "loss": 1.6197, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00018599302811154572, | |
| "loss": 1.5905, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00018562130717655878, | |
| "loss": 1.6175, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00018524509981830852, | |
| "loss": 1.5753, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00018486442574947511, | |
| "loss": 1.6248, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00018447930491678733, | |
| "loss": 1.5616, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00018408975749997759, | |
| "loss": 1.596, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00018369580391072433, | |
| "loss": 1.5591, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00018329746479158265, | |
| "loss": 1.6116, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00018289476101490256, | |
| "loss": 1.5883, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00018248771368173524, | |
| "loss": 1.6456, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00018207634412072764, | |
| "loss": 1.5663, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0001816606738870046, | |
| "loss": 1.6058, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00018124072476103956, | |
| "loss": 1.5659, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00018081651874751327, | |
| "loss": 1.5531, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00018038807807416068, | |
| "loss": 1.5899, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00017995542519060647, | |
| "loss": 1.5176, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00017951858276718844, | |
| "loss": 1.6243, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "eval_loss": 1.570174217224121, | |
| "eval_runtime": 311.9281, | |
| "eval_samples_per_second": 4.796, | |
| "eval_steps_per_second": 1.199, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00017907757369376985, | |
| "loss": 1.5544, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00017863242107853995, | |
| "loss": 1.5498, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.000178183148246803, | |
| "loss": 1.5598, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0001777297787397563, | |
| "loss": 1.526, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00017727233631325664, | |
| "loss": 1.5929, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00017681084493657525, | |
| "loss": 1.5623, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0001763453287911422, | |
| "loss": 1.6259, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0001758758122692791, | |
| "loss": 1.5729, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00017540231997292114, | |
| "loss": 1.5769, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00017492487671232784, | |
| "loss": 1.542, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00017444350750478316, | |
| "loss": 1.5598, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00017395823757328444, | |
| "loss": 1.5982, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00017346909234522109, | |
| "loss": 1.54, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00017297609745104184, | |
| "loss": 1.5536, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.000172479278722912, | |
| "loss": 1.5671, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0001719786621933599, | |
| "loss": 1.5779, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00017147427409391265, | |
| "loss": 1.5637, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00017096614085372185, | |
| "loss": 1.5705, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0001704542890981785, | |
| "loss": 1.5621, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00016993874564751822, | |
| "loss": 1.56, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00016941953751541553, | |
| "loss": 1.5931, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00016889669190756868, | |
| "loss": 1.5508, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00016837023622027388, | |
| "loss": 1.528, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00016784019803899, | |
| "loss": 1.528, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00016730660513689292, | |
| "loss": 1.5825, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0001667694854734204, | |
| "loss": 1.5595, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00016622886719280705, | |
| "loss": 1.5411, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0001656847786226095, | |
| "loss": 1.5616, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00016513724827222227, | |
| "loss": 1.5769, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00016458630483138356, | |
| "loss": 1.5407, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0001640319771686725, | |
| "loss": 1.5193, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00016347429432999602, | |
| "loss": 1.4997, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.00016291328553706704, | |
| "loss": 1.5718, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.00016234898018587337, | |
| "loss": 1.5238, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0001617814078451373, | |
| "loss": 1.565, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0001612105982547663, | |
| "loss": 1.5667, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0001606365813242947, | |
| "loss": 1.5599, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00016005938713131642, | |
| "loss": 1.5277, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.00015947904591990907, | |
| "loss": 1.4543, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.00015889558809904902, | |
| "loss": 1.5824, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0001583090442410182, | |
| "loss": 1.561, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00015771944507980207, | |
| "loss": 1.5249, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00015712682150947923, | |
| "loss": 1.5219, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00015653120458260263, | |
| "loss": 1.4947, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00015593262550857233, | |
| "loss": 1.5432, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00015533111565200044, | |
| "loss": 1.5583, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00015472670653106745, | |
| "loss": 1.56, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0001541194298158708, | |
| "loss": 1.5059, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0001535093173267654, | |
| "loss": 1.5834, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00015289640103269625, | |
| "loss": 1.527, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "eval_loss": 1.5405874252319336, | |
| "eval_runtime": 311.8121, | |
| "eval_samples_per_second": 4.798, | |
| "eval_steps_per_second": 1.199, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0001522807130495235, | |
| "loss": 1.4796, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00015166228563833934, | |
| "loss": 1.466, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00015104115120377783, | |
| "loss": 1.469, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00015041734229231688, | |
| "loss": 1.5355, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00014979089159057265, | |
| "loss": 1.5019, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00014916183192358718, | |
| "loss": 1.5015, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00014853019625310813, | |
| "loss": 1.5418, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00014789601767586173, | |
| "loss": 1.5265, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00014725932942181872, | |
| "loss": 1.5189, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00014662016485245274, | |
| "loss": 1.4945, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00014597855745899274, | |
| "loss": 1.4742, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00014533454086066772, | |
| "loss": 1.5178, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0001446881488029453, | |
| "loss": 1.5304, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00014403941515576344, | |
| "loss": 1.5476, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00014338837391175582, | |
| "loss": 1.4911, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00014273505918447054, | |
| "loss": 1.4941, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00014207950520658274, | |
| "loss": 1.5374, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00014142174632810072, | |
| "loss": 1.4581, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00014076181701456623, | |
| "loss": 1.5601, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0001400997518452484, | |
| "loss": 1.4741, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00013943558551133186, | |
| "loss": 1.5178, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00013876935281409907, | |
| "loss": 1.5491, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0001381010886631066, | |
| "loss": 1.4954, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00013743082807435615, | |
| "loss": 1.4939, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00013675860616845954, | |
| "loss": 1.5415, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00013608445816879866, | |
| "loss": 1.5166, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00013540841939967962, | |
| "loss": 1.4965, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00013473052528448201, | |
| "loss": 1.4909, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00013405081134380264, | |
| "loss": 1.5002, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00013336931319359426, | |
| "loss": 1.4465, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001326860665432995, | |
| "loss": 1.5397, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013200110719397968, | |
| "loss": 1.4682, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013131447103643885, | |
| "loss": 1.4864, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013062619404934317, | |
| "loss": 1.4745, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00012993631229733582, | |
| "loss": 1.4745, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00012924486192914705, | |
| "loss": 1.4671, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0001285518791757002, | |
| "loss": 1.4875, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00012785740034821329, | |
| "loss": 1.5469, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.0001271614618362962, | |
| "loss": 1.5117, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00012646410010604397, | |
| "loss": 1.4958, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.00012576535169812615, | |
| "loss": 1.5226, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.00012506525322587207, | |
| "loss": 1.5602, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0001243638413733522, | |
| "loss": 1.4868, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.0001236611528934562, | |
| "loss": 1.4798, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.00012295722460596697, | |
| "loss": 1.5437, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.00012225209339563145, | |
| "loss": 1.4851, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00012154579621022777, | |
| "loss": 1.4799, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.00012083837005862946, | |
| "loss": 1.4948, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.00012012985200886602, | |
| "loss": 1.4851, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.00011942027918618074, | |
| "loss": 1.4873, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "eval_loss": 1.5275379419326782, | |
| "eval_runtime": 311.7793, | |
| "eval_samples_per_second": 4.798, | |
| "eval_steps_per_second": 1.2, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00011870968877108546, | |
| "loss": 1.4698, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.0001179981179974121, | |
| "loss": 1.5118, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.00011728560415036201, | |
| "loss": 1.4641, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.00011657218456455206, | |
| "loss": 1.4629, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.00011585789662205835, | |
| "loss": 1.4944, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.00011514277775045768, | |
| "loss": 1.4857, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.00011442686542086609, | |
| "loss": 1.4841, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00011371019714597562, | |
| "loss": 1.4868, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00011299281047808877, | |
| "loss": 1.4687, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.00011227474300715055, | |
| "loss": 1.4597, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.00011155603235877912, | |
| "loss": 1.5115, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.00011083671619229408, | |
| "loss": 1.525, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.00011011683219874323, | |
| "loss": 1.5116, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.00010939641809892767, | |
| "loss": 1.4964, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0001086755116414252, | |
| "loss": 1.4583, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.00010795415060061243, | |
| "loss": 1.4918, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.00010723237277468538, | |
| "loss": 1.5259, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.00010651021598367906, | |
| "loss": 1.5294, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00010578771806748546, | |
| "loss": 1.4972, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.00010506491688387127, | |
| "loss": 1.462, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.00010434185030649372, | |
| "loss": 1.4782, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.00010361855622291637, | |
| "loss": 1.4435, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00010289507253262358, | |
| "loss": 1.4993, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00010217143714503508, | |
| "loss": 1.5007, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00010144768797751905, | |
| "loss": 1.4938, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.00010072386295340572, | |
| "loss": 1.4986, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4512, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.927613704659429e-05, | |
| "loss": 1.5191, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.855231202248097e-05, | |
| "loss": 1.5092, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.782856285496495e-05, | |
| "loss": 1.4555, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.710492746737643e-05, | |
| "loss": 1.4859, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.638144377708367e-05, | |
| "loss": 1.5297, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.565814969350629e-05, | |
| "loss": 1.4883, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 9.493508311612874e-05, | |
| "loss": 1.4751, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.421228193251452e-05, | |
| "loss": 1.4805, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.348978401632101e-05, | |
| "loss": 1.4906, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.276762722531463e-05, | |
| "loss": 1.4829, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.204584939938762e-05, | |
| "loss": 1.4818, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.132448835857483e-05, | |
| "loss": 1.5235, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.060358190107234e-05, | |
| "loss": 1.4753, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 8.98831678012568e-05, | |
| "loss": 1.4732, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 8.916328380770595e-05, | |
| "loss": 1.4651, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 8.844396764122093e-05, | |
| "loss": 1.4852, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.772525699284946e-05, | |
| "loss": 1.4747, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.700718952191124e-05, | |
| "loss": 1.5208, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.628980285402439e-05, | |
| "loss": 1.4901, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.557313457913394e-05, | |
| "loss": 1.4801, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.485722224954237e-05, | |
| "loss": 1.4781, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.414210337794166e-05, | |
| "loss": 1.4976, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.342781543544798e-05, | |
| "loss": 1.5005, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "eval_loss": 1.5195521116256714, | |
| "eval_runtime": 311.8576, | |
| "eval_samples_per_second": 4.797, | |
| "eval_steps_per_second": 1.199, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.271439584963802e-05, | |
| "loss": 1.48, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.200188200258791e-05, | |
| "loss": 1.5102, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.129031122891459e-05, | |
| "loss": 1.4419, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.057972081381927e-05, | |
| "loss": 1.4944, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 7.987014799113397e-05, | |
| "loss": 1.4637, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 7.916162994137056e-05, | |
| "loss": 1.5022, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 7.845420378977223e-05, | |
| "loss": 1.4504, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 7.774790660436858e-05, | |
| "loss": 1.466, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.704277539403304e-05, | |
| "loss": 1.4501, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.633884710654383e-05, | |
| "loss": 1.5231, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.56361586266478e-05, | |
| "loss": 1.498, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.493474677412794e-05, | |
| "loss": 1.5227, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 7.423464830187386e-05, | |
| "loss": 1.5031, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 7.353589989395604e-05, | |
| "loss": 1.479, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 7.283853816370386e-05, | |
| "loss": 1.4836, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.214259965178674e-05, | |
| "loss": 1.4626, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.14481208242998e-05, | |
| "loss": 1.4506, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.075513807085299e-05, | |
| "loss": 1.4606, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.006368770266421e-05, | |
| "loss": 1.4762, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.937380595065685e-05, | |
| "loss": 1.5297, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.868552896356117e-05, | |
| "loss": 1.4686, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.799889280602031e-05, | |
| "loss": 1.4208, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.731393345670051e-05, | |
| "loss": 1.4916, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.663068680640574e-05, | |
| "loss": 1.4945, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 6.59491886561974e-05, | |
| "loss": 1.4625, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 6.526947471551798e-05, | |
| "loss": 1.4791, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 6.45915806003204e-05, | |
| "loss": 1.4837, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.391554183120138e-05, | |
| "loss": 1.4536, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.324139383154049e-05, | |
| "loss": 1.4307, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.25691719256439e-05, | |
| "loss": 1.4663, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.189891133689342e-05, | |
| "loss": 1.4668, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.123064718590099e-05, | |
| "loss": 1.4562, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.0564414488668165e-05, | |
| "loss": 1.527, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 5.9900248154751616e-05, | |
| "loss": 1.5169, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.923818298543379e-05, | |
| "loss": 1.4942, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.857825367189931e-05, | |
| "loss": 1.4828, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.7920494793417326e-05, | |
| "loss": 1.5331, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.7264940815529485e-05, | |
| "loss": 1.4857, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.6611626088244194e-05, | |
| "loss": 1.4651, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 5.596058484423656e-05, | |
| "loss": 1.4988, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.531185119705474e-05, | |
| "loss": 1.496, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.46654591393323e-05, | |
| "loss": 1.4849, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.402144254100725e-05, | |
| "loss": 1.5382, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.337983514754723e-05, | |
| "loss": 1.4595, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.274067057818131e-05, | |
| "loss": 1.5216, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.2103982324138244e-05, | |
| "loss": 1.472, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 5.146980374689192e-05, | |
| "loss": 1.4603, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 5.083816807641284e-05, | |
| "loss": 1.431, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 5.0209108409427384e-05, | |
| "loss": 1.4643, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.958265770768316e-05, | |
| "loss": 1.4054, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "eval_loss": 1.5153393745422363, | |
| "eval_runtime": 312.7093, | |
| "eval_samples_per_second": 4.784, | |
| "eval_steps_per_second": 1.196, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.895884879622216e-05, | |
| "loss": 1.4389, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.833771436166069e-05, | |
| "loss": 1.4052, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.7719286950476525e-05, | |
| "loss": 1.4469, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.710359896730379e-05, | |
| "loss": 1.4418, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.649068267323465e-05, | |
| "loss": 1.4759, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.5880570184129215e-05, | |
| "loss": 1.4329, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.5273293468932584e-05, | |
| "loss": 1.4015, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.466888434799958e-05, | |
| "loss": 1.4387, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.406737449142769e-05, | |
| "loss": 1.4713, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.34687954173974e-05, | |
| "loss": 1.4665, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.287317849052075e-05, | |
| "loss": 1.4308, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.2280554920197936e-05, | |
| "loss": 1.4307, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.169095575898181e-05, | |
| "loss": 1.4356, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.1104411900951015e-05, | |
| "loss": 1.4521, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.052095408009096e-05, | |
| "loss": 1.4454, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.994061286868361e-05, | |
| "loss": 1.4114, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.9363418675705334e-05, | |
| "loss": 1.4255, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 3.878940174523371e-05, | |
| "loss": 1.4438, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.821859215486274e-05, | |
| "loss": 1.4493, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.7651019814126654e-05, | |
| "loss": 1.4245, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.7086714462933006e-05, | |
| "loss": 1.4372, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.652570567000402e-05, | |
| "loss": 1.4101, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.5968022831327505e-05, | |
| "loss": 1.4433, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.541369516861648e-05, | |
| "loss": 1.4999, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.4862751727777797e-05, | |
| "loss": 1.4423, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.431522137739049e-05, | |
| "loss": 1.3616, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.377113280719295e-05, | |
| "loss": 1.4282, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.323051452657961e-05, | |
| "loss": 1.4705, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.269339486310711e-05, | |
| "loss": 1.4255, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.215980196101002e-05, | |
| "loss": 1.4869, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 3.162976377972614e-05, | |
| "loss": 1.4884, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 3.110330809243134e-05, | |
| "loss": 1.4677, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 3.058046248458446e-05, | |
| "loss": 1.3935, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 3.0061254352481804e-05, | |
| "loss": 1.4546, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 2.954571090182149e-05, | |
| "loss": 1.4612, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.9033859146278197e-05, | |
| "loss": 1.4626, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.852572590608735e-05, | |
| "loss": 1.4729, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.8021337806640135e-05, | |
| "loss": 1.4555, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 2.7520721277088024e-05, | |
| "loss": 1.4631, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.702390254895819e-05, | |
| "loss": 1.4732, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.6530907654778958e-05, | |
| "loss": 1.419, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 2.6041762426715566e-05, | |
| "loss": 1.4219, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.5556492495216867e-05, | |
| "loss": 1.4772, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.5075123287672175e-05, | |
| "loss": 1.4066, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.4597680027078873e-05, | |
| "loss": 1.4213, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.4124187730720917e-05, | |
| "loss": 1.4578, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.3654671208857826e-05, | |
| "loss": 1.4273, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.3189155063424782e-05, | |
| "loss": 1.4381, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.2727663686743385e-05, | |
| "loss": 1.4327, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.2270221260243673e-05, | |
| "loss": 1.4145, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "eval_loss": 1.514884114265442, | |
| "eval_runtime": 312.6299, | |
| "eval_samples_per_second": 4.785, | |
| "eval_steps_per_second": 1.196, | |
| "step": 350 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 444, | |
| "num_train_epochs": 3, | |
| "save_steps": 50, | |
| "total_flos": 2.7717336710066995e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |