| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1e-05, | |
| "loss": 10.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2e-05, | |
| "loss": 9.5, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9998741276738753e-05, | |
| "loss": 9.6875, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9994965423831853e-05, | |
| "loss": 10.1875, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9988673391830082e-05, | |
| "loss": 8.625, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9979866764718846e-05, | |
| "loss": 9.375, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9968547759519426e-05, | |
| "loss": 9.0625, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9954719225730847e-05, | |
| "loss": 7.7812, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9938384644612542e-05, | |
| "loss": 8.25, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9919548128307954e-05, | |
| "loss": 7.0312, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.989821441880933e-05, | |
| "loss": 7.6875, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9874388886763944e-05, | |
| "loss": 8.25, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9848077530122083e-05, | |
| "loss": 7.9688, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9819286972627066e-05, | |
| "loss": 8.0, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.978802446214779e-05, | |
| "loss": 8.625, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9754297868854075e-05, | |
| "loss": 7.4688, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9718115683235418e-05, | |
| "loss": 7.6562, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9679487013963566e-05, | |
| "loss": 6.75, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9638421585599422e-05, | |
| "loss": 5.875, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9594929736144978e-05, | |
| "loss": 7.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9549022414440738e-05, | |
| "loss": 6.2812, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9500711177409456e-05, | |
| "loss": 6.6562, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9450008187146685e-05, | |
| "loss": 5.5625, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9396926207859085e-05, | |
| "loss": 6.5625, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9341478602651068e-05, | |
| "loss": 5.5625, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.9283679330160726e-05, | |
| "loss": 5.75, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.9223542941045817e-05, | |
| "loss": 5.875, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.9161084574320696e-05, | |
| "loss": 5.75, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9096319953545186e-05, | |
| "loss": 5.5625, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9029265382866216e-05, | |
| "loss": 4.8125, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.895993774291336e-05, | |
| "loss": 5.4062, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.8888354486549238e-05, | |
| "loss": 4.6875, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.881453363447582e-05, | |
| "loss": 3.6875, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.873849377069785e-05, | |
| "loss": 3.6094, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.866025403784439e-05, | |
| "loss": 4.4375, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8579834132349773e-05, | |
| "loss": 3.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8497254299495147e-05, | |
| "loss": 5.2812, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.8412535328311813e-05, | |
| "loss": 4.6562, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.8325698546347714e-05, | |
| "loss": 5.0312, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.8236765814298328e-05, | |
| "loss": 4.3125, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.814575952050336e-05, | |
| "loss": 4.1562, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8052702575310588e-05, | |
| "loss": 2.7344, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.7957618405308323e-05, | |
| "loss": 3.8906, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.7860530947427878e-05, | |
| "loss": 4.0938, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.776146464291757e-05, | |
| "loss": 3.625, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.766044443118978e-05, | |
| "loss": 4.3125, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.7557495743542586e-05, | |
| "loss": 3.9844, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.745264449675755e-05, | |
| "loss": 4.3438, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.734591708657533e-05, | |
| "loss": 4.4062, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.72373403810507e-05, | |
| "loss": 1.9531, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.7126941713788633e-05, | |
| "loss": 4.1562, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.7014748877063212e-05, | |
| "loss": 1.7891, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.6900790114821122e-05, | |
| "loss": 4.1562, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.6785094115571323e-05, | |
| "loss": 3.3125, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.666769000516292e-05, | |
| "loss": 1.5156, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6548607339452853e-05, | |
| "loss": 3.8594, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.6427876096865394e-05, | |
| "loss": 3.2656, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6305526670845225e-05, | |
| "loss": 3.0156, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.6181589862206053e-05, | |
| "loss": 3.8438, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.6056096871376667e-05, | |
| "loss": 3.1406, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.5929079290546408e-05, | |
| "loss": 3.1719, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.5800569095711983e-05, | |
| "loss": 1.3672, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.5670598638627707e-05, | |
| "loss": 1.2422, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.5539200638661106e-05, | |
| "loss": 2.3906, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5406408174555978e-05, | |
| "loss": 3.3438, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5272254676105026e-05, | |
| "loss": 2.6406, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5136773915734067e-05, | |
| "loss": 3.5781, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 2.5, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.4861967361004687e-05, | |
| "loss": 3.1562, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.472271074772683e-05, | |
| "loss": 2.5469, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4582265217274105e-05, | |
| "loss": 2.0469, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4440666126057743e-05, | |
| "loss": 1.7891, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.4297949120891718e-05, | |
| "loss": 0.8867, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.4154150130018867e-05, | |
| "loss": 2.7812, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4009305354066138e-05, | |
| "loss": 0.8828, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.3863451256931286e-05, | |
| "loss": 1.9297, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.3716624556603275e-05, | |
| "loss": 1.9062, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.356886221591872e-05, | |
| "loss": 0.8047, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.342020143325669e-05, | |
| "loss": 1.7656, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3270679633174219e-05, | |
| "loss": 1.4531, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3120334456984871e-05, | |
| "loss": 3.0156, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.296920375328275e-05, | |
| "loss": 2.2969, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.2817325568414299e-05, | |
| "loss": 1.4141, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.266473813690035e-05, | |
| "loss": 0.5703, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.2511479871810792e-05, | |
| "loss": 0.2754, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.2357589355094275e-05, | |
| "loss": 2.5938, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.2203105327865407e-05, | |
| "loss": 1.6953, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2048066680651908e-05, | |
| "loss": 0.3672, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1892512443604103e-05, | |
| "loss": 3.0938, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1736481776669307e-05, | |
| "loss": 0.6133, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.15800139597335e-05, | |
| "loss": 1.5234, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.1423148382732854e-05, | |
| "loss": 1.2891, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.1265924535737494e-05, | |
| "loss": 1.6875, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.1108381999010111e-05, | |
| "loss": 2.625, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0950560433041825e-05, | |
| "loss": 0.4844, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.0792499568567885e-05, | |
| "loss": 0.2383, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0634239196565646e-05, | |
| "loss": 2.7812, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0475819158237426e-05, | |
| "loss": 2.7969, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.031727933498068e-05, | |
| "loss": 0.1514, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.015865963834808e-05, | |
| "loss": 2.1719, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1e-05, | |
| "loss": 1.2656, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.841340361651921e-06, | |
| "loss": 2.1094, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.682720665019325e-06, | |
| "loss": 2.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.524180841762577e-06, | |
| "loss": 2.3594, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.365760803434356e-06, | |
| "loss": 0.957, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.207500431432115e-06, | |
| "loss": 0.1865, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.049439566958176e-06, | |
| "loss": 1.3047, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.89161800098989e-06, | |
| "loss": 2.2031, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.734075464262507e-06, | |
| "loss": 1.0312, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.576851617267151e-06, | |
| "loss": 2.25, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.419986040266502e-06, | |
| "loss": 0.9258, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.263518223330698e-06, | |
| "loss": 0.5547, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.107487556395902e-06, | |
| "loss": 0.9297, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.951933319348095e-06, | |
| "loss": 0.2178, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.796894672134594e-06, | |
| "loss": 0.4336, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.642410644905726e-06, | |
| "loss": 1.2109, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.488520128189209e-06, | |
| "loss": 1.125, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.335261863099652e-06, | |
| "loss": 0.8438, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.182674431585703e-06, | |
| "loss": 0.1201, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.0307962467172555e-06, | |
| "loss": 1.6016, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.87966554301513e-06, | |
| "loss": 0.1318, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.729320366825785e-06, | |
| "loss": 0.6875, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.579798566743314e-06, | |
| "loss": 0.7812, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.431137784081283e-06, | |
| "loss": 1.6641, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.283375443396726e-06, | |
| "loss": 0.1494, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.136548743068713e-06, | |
| "loss": 0.7891, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.990694645933866e-06, | |
| "loss": 0.7188, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.845849869981137e-06, | |
| "loss": 0.25, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.702050879108284e-06, | |
| "loss": 1.5625, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.559333873942259e-06, | |
| "loss": 0.6055, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.417734782725896e-06, | |
| "loss": 0.8633, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.277289252273175e-06, | |
| "loss": 0.1157, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.138032638995315e-06, | |
| "loss": 0.5664, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.000000000000003e-06, | |
| "loss": 2.25, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.863226084265939e-06, | |
| "loss": 1.625, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.727745323894976e-06, | |
| "loss": 0.2051, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.593591825444028e-06, | |
| "loss": 1.8906, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.460799361338898e-06, | |
| "loss": 1.8906, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.3294013613722944e-06, | |
| "loss": 0.7109, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.19943090428802e-06, | |
| "loss": 1.5, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.070920709453597e-06, | |
| "loss": 0.6836, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 3.943903128623336e-06, | |
| "loss": 1.7344, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.818410137793947e-06, | |
| "loss": 0.5742, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.6944733291547784e-06, | |
| "loss": 0.7109, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.5721239031346067e-06, | |
| "loss": 0.5352, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.4513926605471504e-06, | |
| "loss": 0.5078, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.3323099948370853e-06, | |
| "loss": 0.2041, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.2149058844286796e-06, | |
| "loss": 1.4375, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.099209885178882e-06, | |
| "loss": 0.1147, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.9852511229367862e-06, | |
| "loss": 0.5625, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.8730582862113743e-06, | |
| "loss": 0.3262, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.7626596189492983e-06, | |
| "loss": 0.5234, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.6540829134246683e-06, | |
| "loss": 0.6016, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.5473555032424534e-06, | |
| "loss": 0.4609, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4425042564574186e-06, | |
| "loss": 0.0835, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.339555568810221e-06, | |
| "loss": 1.3672, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.2385353570824308e-06, | |
| "loss": 1.8906, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.1394690525721275e-06, | |
| "loss": 1.6953, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.0423815946916783e-06, | |
| "loss": 1.7578, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.947297424689414e-06, | |
| "loss": 1.9688, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.854240479496643e-06, | |
| "loss": 1.2578, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.7632341857016733e-06, | |
| "loss": 0.5234, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.6743014536522872e-06, | |
| "loss": 1.8984, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.587464671688187e-06, | |
| "loss": 0.6484, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.5027457005048573e-06, | |
| "loss": 0.7383, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.4201658676502294e-06, | |
| "loss": 0.4297, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.339745962155613e-06, | |
| "loss": 0.2139, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.2615062293021508e-06, | |
| "loss": 0.5352, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.1854663655241804e-06, | |
| "loss": 0.832, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.1116455134507665e-06, | |
| "loss": 0.377, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.0400622570866426e-06, | |
| "loss": 0.5859, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.707346171337895e-07, | |
| "loss": 0.3574, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.036800464548157e-07, | |
| "loss": 0.3145, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.389154256793042e-07, | |
| "loss": 0.5664, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.764570589541876e-07, | |
| "loss": 0.8281, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.163206698392744e-07, | |
| "loss": 1.125, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.585213973489335e-07, | |
| "loss": 1.7188, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.030737921409169e-07, | |
| "loss": 0.6914, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.499918128533155e-07, | |
| "loss": 0.6328, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.992888225905467e-07, | |
| "loss": 0.0991, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.509775855592613e-07, | |
| "loss": 1.7812, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.0507026385502747e-07, | |
| "loss": 0.6211, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.615784144005796e-07, | |
| "loss": 0.0645, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.2051298603643754e-07, | |
| "loss": 1.5312, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.818843167645835e-07, | |
| "loss": 1.2266, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.4570213114592957e-07, | |
| "loss": 0.3691, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.119755378522137e-07, | |
| "loss": 0.1953, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8071302737293294e-07, | |
| "loss": 0.3418, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.519224698779198e-07, | |
| "loss": 0.418, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.2561111323605714e-07, | |
| "loss": 0.2676, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.0178558119067316e-07, | |
| "loss": 0.7695, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.04518716920466e-08, | |
| "loss": 1.6719, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.161535538745877e-08, | |
| "loss": 1.8594, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.528077426915412e-08, | |
| "loss": 0.2598, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.1452240480577265e-08, | |
| "loss": 1.5078, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.013323528115674e-08, | |
| "loss": 1.7891, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.1326608169920373e-08, | |
| "loss": 1.9688, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.034576168149175e-09, | |
| "loss": 0.0664, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.2587232612493172e-09, | |
| "loss": 0.5742, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.0525, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 200, | |
| "total_flos": 128857731072.0, | |
| "train_loss": 2.513626708984375, | |
| "train_runtime": 417.978, | |
| "train_samples_per_second": 0.957, | |
| "train_steps_per_second": 0.478 | |
| } | |
| ], | |
| "max_steps": 200, | |
| "num_train_epochs": 2, | |
| "total_flos": 128857731072.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |