| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 261603, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.990443534668945e-05, | |
| "loss": 2.4122, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.98088706933789e-05, | |
| "loss": 2.0865, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.971330604006835e-05, | |
| "loss": 1.9392, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.96177413867578e-05, | |
| "loss": 1.8163, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.952217673344725e-05, | |
| "loss": 1.7553, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9426612080136694e-05, | |
| "loss": 1.6586, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.933104742682615e-05, | |
| "loss": 1.6214, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9235482773515594e-05, | |
| "loss": 1.58, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.913991812020505e-05, | |
| "loss": 1.5398, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9044353466894494e-05, | |
| "loss": 1.5098, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.894878881358394e-05, | |
| "loss": 1.4674, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.885322416027339e-05, | |
| "loss": 1.4478, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.875765950696284e-05, | |
| "loss": 1.4264, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.866209485365229e-05, | |
| "loss": 1.413, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.856653020034174e-05, | |
| "loss": 1.3918, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8470965547031185e-05, | |
| "loss": 1.3834, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.837540089372064e-05, | |
| "loss": 1.3515, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8279836240410085e-05, | |
| "loss": 1.3263, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.818427158709954e-05, | |
| "loss": 1.3411, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.808870693378899e-05, | |
| "loss": 1.3176, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7993142280478434e-05, | |
| "loss": 1.3117, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.789757762716789e-05, | |
| "loss": 1.305, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7802012973857334e-05, | |
| "loss": 1.2867, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7706448320546784e-05, | |
| "loss": 1.2586, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7610883667236234e-05, | |
| "loss": 1.2645, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.751531901392568e-05, | |
| "loss": 1.2608, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.741975436061513e-05, | |
| "loss": 1.2586, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.732418970730458e-05, | |
| "loss": 1.2374, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.722862505399403e-05, | |
| "loss": 1.2306, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.713306040068348e-05, | |
| "loss": 1.2437, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.7037495747372926e-05, | |
| "loss": 1.2214, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.694193109406238e-05, | |
| "loss": 1.2365, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6846366440751825e-05, | |
| "loss": 1.2184, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.675080178744128e-05, | |
| "loss": 1.2098, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6655237134130725e-05, | |
| "loss": 1.1901, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.6559672480820174e-05, | |
| "loss": 1.1875, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.6464107827509624e-05, | |
| "loss": 1.1871, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6368543174199074e-05, | |
| "loss": 1.1909, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6272978520888524e-05, | |
| "loss": 1.176, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6177413867577974e-05, | |
| "loss": 1.18, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.608184921426742e-05, | |
| "loss": 1.1598, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.598628456095687e-05, | |
| "loss": 1.1607, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.589071990764632e-05, | |
| "loss": 1.1591, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.579515525433577e-05, | |
| "loss": 1.1464, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.569959060102522e-05, | |
| "loss": 1.1605, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5604025947714666e-05, | |
| "loss": 1.1387, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.550846129440412e-05, | |
| "loss": 1.1553, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5412896641093565e-05, | |
| "loss": 1.1238, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5317331987783015e-05, | |
| "loss": 1.1291, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5221767334472465e-05, | |
| "loss": 1.1154, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5126202681161914e-05, | |
| "loss": 1.1228, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5030638027851364e-05, | |
| "loss": 1.1204, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4935073374540814e-05, | |
| "loss": 1.113, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4839508721230264e-05, | |
| "loss": 1.1165, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.4743944067919714e-05, | |
| "loss": 1.0987, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.464837941460916e-05, | |
| "loss": 1.1014, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.455281476129861e-05, | |
| "loss": 1.1041, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.4457250107988056e-05, | |
| "loss": 1.0836, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.436168545467751e-05, | |
| "loss": 1.1053, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.426612080136696e-05, | |
| "loss": 1.0941, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.4170556148056406e-05, | |
| "loss": 1.0864, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.407499149474586e-05, | |
| "loss": 1.1057, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.3979426841435305e-05, | |
| "loss": 1.0829, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.3883862188124755e-05, | |
| "loss": 1.077, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.3788297534814205e-05, | |
| "loss": 1.0654, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.3692732881503655e-05, | |
| "loss": 1.0889, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.3597168228193104e-05, | |
| "loss": 1.0631, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.3501603574882554e-05, | |
| "loss": 1.079, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.3406038921572004e-05, | |
| "loss": 1.0663, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.3310474268261454e-05, | |
| "loss": 1.0479, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.32149096149509e-05, | |
| "loss": 1.0503, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.311934496164035e-05, | |
| "loss": 1.0586, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.3023780308329796e-05, | |
| "loss": 1.053, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.2928215655019246e-05, | |
| "loss": 1.0698, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.2832651001708696e-05, | |
| "loss": 1.0644, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.2737086348398146e-05, | |
| "loss": 1.0363, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.2641521695087595e-05, | |
| "loss": 1.0343, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.2545957041777045e-05, | |
| "loss": 1.0459, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.2450392388466495e-05, | |
| "loss": 1.0376, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.2354827735155945e-05, | |
| "loss": 1.0394, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.2259263081845395e-05, | |
| "loss": 1.0396, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.2163698428534844e-05, | |
| "loss": 1.042, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.2068133775224294e-05, | |
| "loss": 1.0336, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.1972569121913744e-05, | |
| "loss": 1.0184, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.1877004468603194e-05, | |
| "loss": 1.0268, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.178143981529264e-05, | |
| "loss": 1.0142, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.168587516198209e-05, | |
| "loss": 1.0199, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.1590310508671536e-05, | |
| "loss": 1.0097, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.1494745855360986e-05, | |
| "loss": 1.0123, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.1399181202050436e-05, | |
| "loss": 1.009, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.1303616548739886e-05, | |
| "loss": 1.0122, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1208051895429335e-05, | |
| "loss": 0.9982, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1112487242118785e-05, | |
| "loss": 1.0064, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.1016922588808235e-05, | |
| "loss": 0.9934, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.0921357935497685e-05, | |
| "loss": 1.0044, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.082579328218713e-05, | |
| "loss": 1.004, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.0730228628876584e-05, | |
| "loss": 0.9874, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.0634663975566034e-05, | |
| "loss": 1.0173, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.053909932225548e-05, | |
| "loss": 1.0207, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.0443534668944934e-05, | |
| "loss": 1.0013, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.034797001563438e-05, | |
| "loss": 1.0093, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.025240536232383e-05, | |
| "loss": 0.9874, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.0156840709013276e-05, | |
| "loss": 1.0028, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.0061276055702726e-05, | |
| "loss": 0.9882, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.9965711402392176e-05, | |
| "loss": 1.001, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.9870146749081626e-05, | |
| "loss": 0.998, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.9774582095771076e-05, | |
| "loss": 0.9912, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9679017442460525e-05, | |
| "loss": 0.9872, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9583452789149975e-05, | |
| "loss": 0.9958, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.9487888135839425e-05, | |
| "loss": 0.9797, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.939232348252887e-05, | |
| "loss": 0.9761, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.9296758829218324e-05, | |
| "loss": 0.9781, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.920119417590777e-05, | |
| "loss": 0.9645, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.910562952259722e-05, | |
| "loss": 0.986, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.901006486928667e-05, | |
| "loss": 0.9766, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.891450021597612e-05, | |
| "loss": 0.945, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.881893556266557e-05, | |
| "loss": 0.9671, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.8723370909355016e-05, | |
| "loss": 0.9774, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.8627806256044466e-05, | |
| "loss": 0.9646, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.8532241602733916e-05, | |
| "loss": 0.959, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.8436676949423366e-05, | |
| "loss": 0.9623, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.8341112296112816e-05, | |
| "loss": 0.9637, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.8245547642802265e-05, | |
| "loss": 0.9704, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.814998298949171e-05, | |
| "loss": 0.9557, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.8054418336181165e-05, | |
| "loss": 0.9609, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.795885368287061e-05, | |
| "loss": 0.9465, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.7863289029560065e-05, | |
| "loss": 0.9526, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.776772437624951e-05, | |
| "loss": 0.96, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.767215972293896e-05, | |
| "loss": 0.9513, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.757659506962841e-05, | |
| "loss": 0.959, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.748103041631786e-05, | |
| "loss": 0.9484, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.738546576300731e-05, | |
| "loss": 0.9443, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.7289901109696756e-05, | |
| "loss": 0.9556, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.7194336456386206e-05, | |
| "loss": 0.9579, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.7098771803075656e-05, | |
| "loss": 0.9432, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.70032071497651e-05, | |
| "loss": 0.9415, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.6907642496454556e-05, | |
| "loss": 0.9305, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.6812077843144005e-05, | |
| "loss": 0.9464, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.671651318983345e-05, | |
| "loss": 0.9551, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.6620948536522905e-05, | |
| "loss": 0.9435, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.652538388321235e-05, | |
| "loss": 0.9429, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.6429819229901805e-05, | |
| "loss": 0.9314, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.633425457659125e-05, | |
| "loss": 0.9499, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.62386899232807e-05, | |
| "loss": 0.9335, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.614312526997015e-05, | |
| "loss": 0.9423, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.60475606166596e-05, | |
| "loss": 0.924, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.595199596334905e-05, | |
| "loss": 0.9386, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.5856431310038497e-05, | |
| "loss": 0.923, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.576086665672794e-05, | |
| "loss": 0.9288, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.5665302003417396e-05, | |
| "loss": 0.9335, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.556973735010684e-05, | |
| "loss": 0.9173, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.5474172696796296e-05, | |
| "loss": 0.9339, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.537860804348574e-05, | |
| "loss": 0.9105, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.528304339017519e-05, | |
| "loss": 0.9377, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.518747873686464e-05, | |
| "loss": 0.9166, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.509191408355409e-05, | |
| "loss": 0.9168, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.4996349430243545e-05, | |
| "loss": 0.911, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.490078477693299e-05, | |
| "loss": 0.9174, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.480522012362244e-05, | |
| "loss": 0.9292, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.470965547031189e-05, | |
| "loss": 0.9146, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.461409081700134e-05, | |
| "loss": 0.9122, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.451852616369079e-05, | |
| "loss": 0.9186, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.4422961510380237e-05, | |
| "loss": 0.919, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.432739685706968e-05, | |
| "loss": 0.9108, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4231832203759136e-05, | |
| "loss": 0.8935, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.413626755044858e-05, | |
| "loss": 0.903, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.4040702897138036e-05, | |
| "loss": 0.92, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.394513824382748e-05, | |
| "loss": 0.9104, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.384957359051693e-05, | |
| "loss": 0.9074, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.375400893720638e-05, | |
| "loss": 0.9063, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.365844428389583e-05, | |
| "loss": 0.905, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.356287963058528e-05, | |
| "loss": 0.8945, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.346731497727473e-05, | |
| "loss": 0.9077, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.337175032396417e-05, | |
| "loss": 0.8878, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.327618567065363e-05, | |
| "loss": 0.8388, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.318062101734308e-05, | |
| "loss": 0.8148, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.308505636403253e-05, | |
| "loss": 0.8024, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.298949171072198e-05, | |
| "loss": 0.8132, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.289392705741142e-05, | |
| "loss": 0.8028, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.2798362404100876e-05, | |
| "loss": 0.8078, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.270279775079032e-05, | |
| "loss": 0.8252, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.2607233097479776e-05, | |
| "loss": 0.7964, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.251166844416922e-05, | |
| "loss": 0.813, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.241610379085867e-05, | |
| "loss": 0.798, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.232053913754812e-05, | |
| "loss": 0.8003, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.222497448423757e-05, | |
| "loss": 0.8108, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.212940983092702e-05, | |
| "loss": 0.8256, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.203384517761647e-05, | |
| "loss": 0.8262, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.193828052430591e-05, | |
| "loss": 0.8144, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.184271587099537e-05, | |
| "loss": 0.8183, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.174715121768481e-05, | |
| "loss": 0.8124, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.165158656437427e-05, | |
| "loss": 0.8143, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.155602191106371e-05, | |
| "loss": 0.8122, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.146045725775316e-05, | |
| "loss": 0.815, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.136489260444261e-05, | |
| "loss": 0.8069, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.126932795113206e-05, | |
| "loss": 0.8088, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.117376329782151e-05, | |
| "loss": 0.8184, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.107819864451096e-05, | |
| "loss": 0.8036, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.098263399120041e-05, | |
| "loss": 0.8372, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.088706933788986e-05, | |
| "loss": 0.8116, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.079150468457931e-05, | |
| "loss": 0.8077, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.069594003126876e-05, | |
| "loss": 0.81, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.060037537795821e-05, | |
| "loss": 0.7957, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.0504810724647654e-05, | |
| "loss": 0.818, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0409246071337104e-05, | |
| "loss": 0.8097, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.031368141802655e-05, | |
| "loss": 0.8113, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0218116764716004e-05, | |
| "loss": 0.7916, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.012255211140545e-05, | |
| "loss": 0.8194, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.0026987458094903e-05, | |
| "loss": 0.8172, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.993142280478435e-05, | |
| "loss": 0.8141, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.98358581514738e-05, | |
| "loss": 0.8155, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9740293498163246e-05, | |
| "loss": 0.8197, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.96447288448527e-05, | |
| "loss": 0.8015, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9549164191542145e-05, | |
| "loss": 0.7978, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9453599538231595e-05, | |
| "loss": 0.8057, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9358034884921048e-05, | |
| "loss": 0.8102, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9262470231610495e-05, | |
| "loss": 0.7861, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9166905578299948e-05, | |
| "loss": 0.8137, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.9071340924989394e-05, | |
| "loss": 0.8057, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.8975776271678844e-05, | |
| "loss": 0.8239, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.888021161836829e-05, | |
| "loss": 0.7989, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8784646965057744e-05, | |
| "loss": 0.8048, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.868908231174719e-05, | |
| "loss": 0.8038, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8593517658436643e-05, | |
| "loss": 0.7906, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.849795300512609e-05, | |
| "loss": 0.8044, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.840238835181554e-05, | |
| "loss": 0.8114, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8306823698504986e-05, | |
| "loss": 0.7954, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.821125904519444e-05, | |
| "loss": 0.7984, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.8115694391883885e-05, | |
| "loss": 0.8192, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.8020129738573335e-05, | |
| "loss": 0.8027, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.792456508526278e-05, | |
| "loss": 0.8054, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7829000431952235e-05, | |
| "loss": 0.8059, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.773343577864168e-05, | |
| "loss": 0.8027, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.7637871125331134e-05, | |
| "loss": 0.8053, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7542306472020584e-05, | |
| "loss": 0.796, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.744674181871003e-05, | |
| "loss": 0.7983, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7351177165399484e-05, | |
| "loss": 0.7908, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.725561251208893e-05, | |
| "loss": 0.7987, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.716004785877838e-05, | |
| "loss": 0.7959, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.7064483205467826e-05, | |
| "loss": 0.8101, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.696891855215728e-05, | |
| "loss": 0.7921, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.6873353898846726e-05, | |
| "loss": 0.7842, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.677778924553618e-05, | |
| "loss": 0.7932, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6682224592225625e-05, | |
| "loss": 0.7845, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6586659938915075e-05, | |
| "loss": 0.7833, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.649109528560452e-05, | |
| "loss": 0.7894, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6395530632293975e-05, | |
| "loss": 0.7718, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.629996597898342e-05, | |
| "loss": 0.7878, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6204401325672874e-05, | |
| "loss": 0.7986, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6108836672362317e-05, | |
| "loss": 0.797, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.601327201905177e-05, | |
| "loss": 0.7918, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.5917707365741217e-05, | |
| "loss": 0.7852, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.582214271243067e-05, | |
| "loss": 0.7873, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.572657805912012e-05, | |
| "loss": 0.7999, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5631013405809566e-05, | |
| "loss": 0.7919, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.553544875249902e-05, | |
| "loss": 0.7941, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5439884099188466e-05, | |
| "loss": 0.7808, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.534431944587792e-05, | |
| "loss": 0.7806, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5248754792567365e-05, | |
| "loss": 0.78, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5153190139256815e-05, | |
| "loss": 0.7908, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.505762548594626e-05, | |
| "loss": 0.7819, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.496206083263571e-05, | |
| "loss": 0.7863, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.486649617932516e-05, | |
| "loss": 0.7927, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.477093152601461e-05, | |
| "loss": 0.7974, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.467536687270406e-05, | |
| "loss": 0.7836, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.457980221939351e-05, | |
| "loss": 0.7831, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.448423756608296e-05, | |
| "loss": 0.7866, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.438867291277241e-05, | |
| "loss": 0.787, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4293108259461857e-05, | |
| "loss": 0.7782, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4197543606151306e-05, | |
| "loss": 0.7848, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4101978952840756e-05, | |
| "loss": 0.7858, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4006414299530206e-05, | |
| "loss": 0.7909, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3910849646219656e-05, | |
| "loss": 0.7842, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3815284992909102e-05, | |
| "loss": 0.7716, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3719720339598552e-05, | |
| "loss": 0.7858, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3624155686288e-05, | |
| "loss": 0.7856, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.352859103297745e-05, | |
| "loss": 0.7797, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.34330263796669e-05, | |
| "loss": 0.7619, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.333746172635635e-05, | |
| "loss": 0.7902, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3241897073045797e-05, | |
| "loss": 0.7639, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3146332419735247e-05, | |
| "loss": 0.7784, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.3050767766424697e-05, | |
| "loss": 0.7805, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.295520311311415e-05, | |
| "loss": 0.7716, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2859638459803597e-05, | |
| "loss": 0.7781, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2764073806493046e-05, | |
| "loss": 0.7929, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.2668509153182496e-05, | |
| "loss": 0.7791, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2572944499871946e-05, | |
| "loss": 0.763, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2477379846561396e-05, | |
| "loss": 0.7859, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2381815193250842e-05, | |
| "loss": 0.7727, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2286250539940292e-05, | |
| "loss": 0.7804, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.2190685886629742e-05, | |
| "loss": 0.7923, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.209512123331919e-05, | |
| "loss": 0.7538, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.199955658000864e-05, | |
| "loss": 0.7815, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1903991926698088e-05, | |
| "loss": 0.7655, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1808427273387537e-05, | |
| "loss": 0.7767, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1712862620076987e-05, | |
| "loss": 0.7731, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1617297966766437e-05, | |
| "loss": 0.7742, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1521733313455887e-05, | |
| "loss": 0.755, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1426168660145333e-05, | |
| "loss": 0.7631, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1330604006834783e-05, | |
| "loss": 0.7636, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1235039353524233e-05, | |
| "loss": 0.7712, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1139474700213683e-05, | |
| "loss": 0.7581, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.1043910046903136e-05, | |
| "loss": 0.7654, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.0948345393592582e-05, | |
| "loss": 0.7581, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0852780740282032e-05, | |
| "loss": 0.7595, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0757216086971482e-05, | |
| "loss": 0.7693, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.066165143366093e-05, | |
| "loss": 0.7599, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.056608678035038e-05, | |
| "loss": 0.7563, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.0470522127039828e-05, | |
| "loss": 0.7644, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.0374957473729278e-05, | |
| "loss": 0.7669, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.0279392820418727e-05, | |
| "loss": 0.7559, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0183828167108177e-05, | |
| "loss": 0.7646, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0088263513797627e-05, | |
| "loss": 0.7602, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.9992698860487073e-05, | |
| "loss": 0.7594, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9897134207176523e-05, | |
| "loss": 0.7593, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9801569553865973e-05, | |
| "loss": 0.7578, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9706004900555423e-05, | |
| "loss": 0.7612, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9610440247244872e-05, | |
| "loss": 0.7553, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.951487559393432e-05, | |
| "loss": 0.7599, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.941931094062377e-05, | |
| "loss": 0.7387, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.932374628731322e-05, | |
| "loss": 0.7558, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9228181634002668e-05, | |
| "loss": 0.7482, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9132616980692118e-05, | |
| "loss": 0.7703, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.9037052327381568e-05, | |
| "loss": 0.7375, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.8941487674071018e-05, | |
| "loss": 0.7565, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8845923020760467e-05, | |
| "loss": 0.7591, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8750358367449917e-05, | |
| "loss": 0.7639, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8654793714139367e-05, | |
| "loss": 0.7493, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8559229060828813e-05, | |
| "loss": 0.7435, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8463664407518263e-05, | |
| "loss": 0.7492, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.8368099754207713e-05, | |
| "loss": 0.745, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.8272535100897163e-05, | |
| "loss": 0.7529, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8176970447586612e-05, | |
| "loss": 0.7599, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.808140579427606e-05, | |
| "loss": 0.7546, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.798584114096551e-05, | |
| "loss": 0.7485, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.789027648765496e-05, | |
| "loss": 0.7591, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.7794711834344408e-05, | |
| "loss": 0.7466, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7699147181033858e-05, | |
| "loss": 0.7399, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7603582527723304e-05, | |
| "loss": 0.7547, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.7508017874412754e-05, | |
| "loss": 0.7459, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7412453221102204e-05, | |
| "loss": 0.7503, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7316888567791657e-05, | |
| "loss": 0.7451, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7221323914481104e-05, | |
| "loss": 0.7369, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7125759261170553e-05, | |
| "loss": 0.7363, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.7030194607860003e-05, | |
| "loss": 0.7457, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.6934629954549453e-05, | |
| "loss": 0.7473, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6839065301238903e-05, | |
| "loss": 0.7508, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.674350064792835e-05, | |
| "loss": 0.7477, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.66479359946178e-05, | |
| "loss": 0.7332, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.655237134130725e-05, | |
| "loss": 0.6564, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.64568066879967e-05, | |
| "loss": 0.66, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.636124203468615e-05, | |
| "loss": 0.6528, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.6265677381375595e-05, | |
| "loss": 0.6626, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6170112728065044e-05, | |
| "loss": 0.6507, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.6074548074754494e-05, | |
| "loss": 0.6543, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.5978983421443944e-05, | |
| "loss": 0.6613, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.5883418768133394e-05, | |
| "loss": 0.6535, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.5787854114822844e-05, | |
| "loss": 0.6544, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.569228946151229e-05, | |
| "loss": 0.6553, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.559672480820174e-05, | |
| "loss": 0.6576, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.550116015489119e-05, | |
| "loss": 0.6497, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5405595501580643e-05, | |
| "loss": 0.6618, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.531003084827009e-05, | |
| "loss": 0.646, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5214466194959539e-05, | |
| "loss": 0.6451, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5118901541648989e-05, | |
| "loss": 0.6566, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.5023336888338439e-05, | |
| "loss": 0.6641, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.4927772235027887e-05, | |
| "loss": 0.6587, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4832207581717336e-05, | |
| "loss": 0.655, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4736642928406785e-05, | |
| "loss": 0.6515, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4641078275096234e-05, | |
| "loss": 0.6415, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4545513621785684e-05, | |
| "loss": 0.642, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4449948968475132e-05, | |
| "loss": 0.6513, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.4354384315164582e-05, | |
| "loss": 0.6538, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.4258819661854032e-05, | |
| "loss": 0.6543, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.416325500854348e-05, | |
| "loss": 0.6587, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.406769035523293e-05, | |
| "loss": 0.6544, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.3972125701922378e-05, | |
| "loss": 0.6556, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3876561048611828e-05, | |
| "loss": 0.6509, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3780996395301277e-05, | |
| "loss": 0.6514, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3685431741990725e-05, | |
| "loss": 0.6606, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3589867088680177e-05, | |
| "loss": 0.654, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3494302435369627e-05, | |
| "loss": 0.6494, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3398737782059077e-05, | |
| "loss": 0.6482, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3303173128748525e-05, | |
| "loss": 0.6615, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3207608475437974e-05, | |
| "loss": 0.6517, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3112043822127424e-05, | |
| "loss": 0.6415, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.3016479168816872e-05, | |
| "loss": 0.6388, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.2920914515506322e-05, | |
| "loss": 0.6472, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.282534986219577e-05, | |
| "loss": 0.6396, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.272978520888522e-05, | |
| "loss": 0.6549, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.263422055557467e-05, | |
| "loss": 0.6471, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2538655902264118e-05, | |
| "loss": 0.6502, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2443091248953568e-05, | |
| "loss": 0.6448, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2347526595643016e-05, | |
| "loss": 0.6471, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2251961942332465e-05, | |
| "loss": 0.639, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2156397289021915e-05, | |
| "loss": 0.6507, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.2060832635711365e-05, | |
| "loss": 0.6513, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.1965267982400815e-05, | |
| "loss": 0.642, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1869703329090263e-05, | |
| "loss": 0.6459, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1774138675779713e-05, | |
| "loss": 0.6431, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.1678574022469163e-05, | |
| "loss": 0.6497, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.158300936915861e-05, | |
| "loss": 0.6513, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.148744471584806e-05, | |
| "loss": 0.6426, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1391880062537509e-05, | |
| "loss": 0.6454, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.129631540922696e-05, | |
| "loss": 0.6461, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1200750755916408e-05, | |
| "loss": 0.6513, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1105186102605858e-05, | |
| "loss": 0.6426, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.1009621449295308e-05, | |
| "loss": 0.6445, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0914056795984756e-05, | |
| "loss": 0.6455, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0818492142674206e-05, | |
| "loss": 0.6621, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0722927489363654e-05, | |
| "loss": 0.6403, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0627362836053103e-05, | |
| "loss": 0.6339, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0531798182742553e-05, | |
| "loss": 0.6407, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0436233529432001e-05, | |
| "loss": 0.6573, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0340668876121453e-05, | |
| "loss": 0.6348, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0245104222810901e-05, | |
| "loss": 0.6425, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.014953956950035e-05, | |
| "loss": 0.6475, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.00539749161898e-05, | |
| "loss": 0.6416, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.958410262879249e-06, | |
| "loss": 0.6496, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.862845609568698e-06, | |
| "loss": 0.6585, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.767280956258146e-06, | |
| "loss": 0.636, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.671716302947596e-06, | |
| "loss": 0.6479, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.576151649637046e-06, | |
| "loss": 0.6557, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.480586996326494e-06, | |
| "loss": 0.6438, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.385022343015946e-06, | |
| "loss": 0.6371, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.289457689705394e-06, | |
| "loss": 0.635, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.193893036394843e-06, | |
| "loss": 0.6456, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.098328383084293e-06, | |
| "loss": 0.6419, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.002763729773741e-06, | |
| "loss": 0.6355, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.907199076463191e-06, | |
| "loss": 0.6464, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.81163442315264e-06, | |
| "loss": 0.6468, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.716069769842089e-06, | |
| "loss": 0.6324, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.620505116531539e-06, | |
| "loss": 0.6495, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.524940463220987e-06, | |
| "loss": 0.6412, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.429375809910438e-06, | |
| "loss": 0.6302, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.333811156599886e-06, | |
| "loss": 0.6464, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.238246503289336e-06, | |
| "loss": 0.6353, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.142681849978784e-06, | |
| "loss": 0.632, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.047117196668234e-06, | |
| "loss": 0.6438, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.951552543357684e-06, | |
| "loss": 0.6522, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.855987890047132e-06, | |
| "loss": 0.633, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.760423236736582e-06, | |
| "loss": 0.6465, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.664858583426032e-06, | |
| "loss": 0.6442, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.569293930115481e-06, | |
| "loss": 0.6342, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.47372927680493e-06, | |
| "loss": 0.6526, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.378164623494379e-06, | |
| "loss": 0.6327, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.282599970183829e-06, | |
| "loss": 0.6281, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.187035316873278e-06, | |
| "loss": 0.6431, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.091470663562727e-06, | |
| "loss": 0.6428, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 6.995906010252176e-06, | |
| "loss": 0.64, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.900341356941626e-06, | |
| "loss": 0.6345, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.804776703631075e-06, | |
| "loss": 0.6446, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.7092120503205236e-06, | |
| "loss": 0.6405, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.613647397009974e-06, | |
| "loss": 0.6269, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.518082743699423e-06, | |
| "loss": 0.6423, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.422518090388872e-06, | |
| "loss": 0.6197, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.326953437078322e-06, | |
| "loss": 0.6273, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.231388783767771e-06, | |
| "loss": 0.6245, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.13582413045722e-06, | |
| "loss": 0.6333, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 6.040259477146669e-06, | |
| "loss": 0.6312, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.944694823836118e-06, | |
| "loss": 0.653, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.849130170525568e-06, | |
| "loss": 0.6325, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.753565517215017e-06, | |
| "loss": 0.6387, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.658000863904466e-06, | |
| "loss": 0.6298, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.562436210593915e-06, | |
| "loss": 0.6368, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.466871557283365e-06, | |
| "loss": 0.6298, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.371306903972814e-06, | |
| "loss": 0.6233, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.275742250662264e-06, | |
| "loss": 0.6265, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.1801775973517126e-06, | |
| "loss": 0.6283, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.0846129440411615e-06, | |
| "loss": 0.6316, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.989048290730611e-06, | |
| "loss": 0.617, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.89348363742006e-06, | |
| "loss": 0.6298, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.79791898410951e-06, | |
| "loss": 0.6439, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.702354330798959e-06, | |
| "loss": 0.6437, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.606789677488408e-06, | |
| "loss": 0.6245, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.511225024177858e-06, | |
| "loss": 0.6295, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.415660370867307e-06, | |
| "loss": 0.6264, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.320095717556756e-06, | |
| "loss": 0.6121, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.224531064246205e-06, | |
| "loss": 0.6365, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.128966410935654e-06, | |
| "loss": 0.6243, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.033401757625104e-06, | |
| "loss": 0.6247, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.937837104314553e-06, | |
| "loss": 0.6258, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.842272451004003e-06, | |
| "loss": 0.6272, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.7467077976934518e-06, | |
| "loss": 0.6284, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.6511431443829007e-06, | |
| "loss": 0.6213, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.5555784910723505e-06, | |
| "loss": 0.6263, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.4600138377618e-06, | |
| "loss": 0.6336, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.364449184451249e-06, | |
| "loss": 0.6324, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.2688845311406977e-06, | |
| "loss": 0.6398, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.173319877830147e-06, | |
| "loss": 0.6373, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.0777552245195965e-06, | |
| "loss": 0.6203, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.982190571209046e-06, | |
| "loss": 0.6392, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.886625917898495e-06, | |
| "loss": 0.6196, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.7910612645879446e-06, | |
| "loss": 0.629, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.695496611277394e-06, | |
| "loss": 0.6216, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.599931957966843e-06, | |
| "loss": 0.6272, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.5043673046562922e-06, | |
| "loss": 0.6156, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.4088026513457416e-06, | |
| "loss": 0.6135, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.313237998035191e-06, | |
| "loss": 0.6118, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.2176733447246403e-06, | |
| "loss": 0.6151, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.1221086914140893e-06, | |
| "loss": 0.6281, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.0265440381035386e-06, | |
| "loss": 0.6241, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1.930979384792988e-06, | |
| "loss": 0.6334, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.8354147314824374e-06, | |
| "loss": 0.6143, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.7398500781718865e-06, | |
| "loss": 0.631, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.6442854248613357e-06, | |
| "loss": 0.622, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.548720771550785e-06, | |
| "loss": 0.6228, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.4531561182402342e-06, | |
| "loss": 0.6224, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.3575914649296835e-06, | |
| "loss": 0.6253, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.262026811619133e-06, | |
| "loss": 0.639, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.166462158308582e-06, | |
| "loss": 0.6161, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.0708975049980314e-06, | |
| "loss": 0.618, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 9.753328516874808e-07, | |
| "loss": 0.6155, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.7976819837693e-07, | |
| "loss": 0.6096, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 7.842035450663792e-07, | |
| "loss": 0.6226, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.886388917558285e-07, | |
| "loss": 0.618, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 5.930742384452778e-07, | |
| "loss": 0.6241, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.975095851347271e-07, | |
| "loss": 0.6303, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.0194493182417635e-07, | |
| "loss": 0.626, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.063802785136256e-07, | |
| "loss": 0.6324, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.1081562520307492e-07, | |
| "loss": 0.619, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.1525097189252418e-07, | |
| "loss": 0.609, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1.968631858197345e-08, | |
| "loss": 0.6266, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 261603, | |
| "total_flos": 2.70854049202176e+17, | |
| "train_loss": 0.8388812589635725, | |
| "train_runtime": 66410.1599, | |
| "train_samples_per_second": 39.392, | |
| "train_steps_per_second": 3.939 | |
| } | |
| ], | |
| "max_steps": 261603, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.70854049202176e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |