| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 260760, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9904126399754563e-05, | |
| "loss": 2.9959, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.980825279950913e-05, | |
| "loss": 2.0327, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.971237919926369e-05, | |
| "loss": 1.7134, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.961650559901826e-05, | |
| "loss": 1.6017, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.952063199877282e-05, | |
| "loss": 1.4858, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.942475839852738e-05, | |
| "loss": 1.3887, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.932888479828195e-05, | |
| "loss": 1.348, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.923301119803651e-05, | |
| "loss": 1.2965, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.913713759779107e-05, | |
| "loss": 1.2528, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.904126399754564e-05, | |
| "loss": 1.2444, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.89453903973002e-05, | |
| "loss": 1.2153, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.884951679705477e-05, | |
| "loss": 1.189, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.875364319680933e-05, | |
| "loss": 1.162, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.865776959656389e-05, | |
| "loss": 1.144, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.856189599631846e-05, | |
| "loss": 1.1463, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.846602239607302e-05, | |
| "loss": 1.1062, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.837014879582759e-05, | |
| "loss": 1.0981, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.827427519558215e-05, | |
| "loss": 1.0894, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.817840159533671e-05, | |
| "loss": 1.0681, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.808252799509128e-05, | |
| "loss": 1.0633, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.798665439484584e-05, | |
| "loss": 1.0484, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7890780794600406e-05, | |
| "loss": 1.0343, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.779490719435497e-05, | |
| "loss": 1.0362, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.769903359410953e-05, | |
| "loss": 1.0078, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7603159993864096e-05, | |
| "loss": 1.0099, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.750728639361866e-05, | |
| "loss": 0.987, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.741141279337322e-05, | |
| "loss": 0.9977, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7315539193127786e-05, | |
| "loss": 0.9912, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.721966559288235e-05, | |
| "loss": 0.9769, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.7123791992636915e-05, | |
| "loss": 0.9701, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.7027918392391476e-05, | |
| "loss": 0.9703, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.6932044792146037e-05, | |
| "loss": 0.9531, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6836171191900604e-05, | |
| "loss": 0.9525, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6740297591655165e-05, | |
| "loss": 0.9588, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.664442399140973e-05, | |
| "loss": 0.9511, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.6548550391164294e-05, | |
| "loss": 0.9375, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.6452676790918855e-05, | |
| "loss": 0.9315, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.635680319067342e-05, | |
| "loss": 0.9313, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6260929590427984e-05, | |
| "loss": 0.9272, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6165055990182545e-05, | |
| "loss": 0.929, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.606918238993711e-05, | |
| "loss": 0.9215, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.5973308789691674e-05, | |
| "loss": 0.9066, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.587743518944624e-05, | |
| "loss": 0.8864, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.57815615892008e-05, | |
| "loss": 0.8892, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5685687988955364e-05, | |
| "loss": 0.8988, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.558981438870993e-05, | |
| "loss": 0.902, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.549394078846449e-05, | |
| "loss": 0.8844, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5398067188219054e-05, | |
| "loss": 0.8954, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.530219358797362e-05, | |
| "loss": 0.8988, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.520631998772818e-05, | |
| "loss": 0.8754, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5110446387482743e-05, | |
| "loss": 0.8661, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.501457278723731e-05, | |
| "loss": 0.8818, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.491869918699187e-05, | |
| "loss": 0.8647, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.482282558674643e-05, | |
| "loss": 0.8867, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.4726951986501e-05, | |
| "loss": 0.8702, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.463107838625556e-05, | |
| "loss": 0.853, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.453520478601012e-05, | |
| "loss": 0.8685, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.443933118576469e-05, | |
| "loss": 0.8677, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.434345758551925e-05, | |
| "loss": 0.8651, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.424758398527381e-05, | |
| "loss": 0.8494, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.415171038502838e-05, | |
| "loss": 0.8409, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.405583678478294e-05, | |
| "loss": 0.8441, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.39599631845375e-05, | |
| "loss": 0.8382, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.386408958429207e-05, | |
| "loss": 0.8346, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.376821598404663e-05, | |
| "loss": 0.8507, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.36723423838012e-05, | |
| "loss": 0.8328, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.357646878355576e-05, | |
| "loss": 0.8281, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.348059518331032e-05, | |
| "loss": 0.8423, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.338472158306489e-05, | |
| "loss": 0.8161, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.328884798281945e-05, | |
| "loss": 0.8438, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.319297438257401e-05, | |
| "loss": 0.8278, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.309710078232858e-05, | |
| "loss": 0.8269, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.300122718208314e-05, | |
| "loss": 0.8266, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.290535358183771e-05, | |
| "loss": 0.8211, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.280947998159227e-05, | |
| "loss": 0.8376, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.271360638134683e-05, | |
| "loss": 0.8106, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.26177327811014e-05, | |
| "loss": 0.8044, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.252185918085596e-05, | |
| "loss": 0.8068, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.242598558061053e-05, | |
| "loss": 0.8109, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.233011198036509e-05, | |
| "loss": 0.8165, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.223423838011965e-05, | |
| "loss": 0.8064, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.213836477987422e-05, | |
| "loss": 0.7974, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.204249117962878e-05, | |
| "loss": 0.7813, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.194661757938334e-05, | |
| "loss": 0.8056, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.1850743979137906e-05, | |
| "loss": 0.7912, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.175487037889247e-05, | |
| "loss": 0.7834, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.1658996778647035e-05, | |
| "loss": 0.7965, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.1563123178401596e-05, | |
| "loss": 0.7965, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.146724957815616e-05, | |
| "loss": 0.7842, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.1371375977910725e-05, | |
| "loss": 0.7985, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.1275502377665286e-05, | |
| "loss": 0.796, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1179628777419854e-05, | |
| "loss": 0.7936, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1083755177174415e-05, | |
| "loss": 0.7953, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.0987881576928976e-05, | |
| "loss": 0.7797, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.0892007976683544e-05, | |
| "loss": 0.7824, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.0796134376438105e-05, | |
| "loss": 0.7743, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.0700260776192666e-05, | |
| "loss": 0.7841, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.0604387175947234e-05, | |
| "loss": 0.7778, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.0508513575701795e-05, | |
| "loss": 0.7811, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.041263997545636e-05, | |
| "loss": 0.7662, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.0316766375210924e-05, | |
| "loss": 0.7768, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.0220892774965485e-05, | |
| "loss": 0.7689, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.012501917472005e-05, | |
| "loss": 0.7785, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.0029145574474613e-05, | |
| "loss": 0.7688, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.993327197422918e-05, | |
| "loss": 0.7787, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.983739837398374e-05, | |
| "loss": 0.7649, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.97415247737383e-05, | |
| "loss": 0.7692, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.964565117349287e-05, | |
| "loss": 0.7682, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.954977757324743e-05, | |
| "loss": 0.7614, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.9453903973002e-05, | |
| "loss": 0.7726, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.935803037275656e-05, | |
| "loss": 0.7621, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.926215677251112e-05, | |
| "loss": 0.7552, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.916628317226569e-05, | |
| "loss": 0.7518, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.907040957202025e-05, | |
| "loss": 0.77, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.897453597177481e-05, | |
| "loss": 0.7494, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.887866237152938e-05, | |
| "loss": 0.7673, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.878278877128394e-05, | |
| "loss": 0.7541, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.868691517103851e-05, | |
| "loss": 0.7398, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.859104157079307e-05, | |
| "loss": 0.744, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.849516797054763e-05, | |
| "loss": 0.7443, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.83992943703022e-05, | |
| "loss": 0.7514, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.830342077005676e-05, | |
| "loss": 0.7395, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.820754716981133e-05, | |
| "loss": 0.7477, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.811167356956589e-05, | |
| "loss": 0.7485, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.801579996932045e-05, | |
| "loss": 0.74, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.791992636907502e-05, | |
| "loss": 0.7349, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.782405276882958e-05, | |
| "loss": 0.746, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.772817916858414e-05, | |
| "loss": 0.7446, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.763230556833871e-05, | |
| "loss": 0.7328, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.753643196809327e-05, | |
| "loss": 0.7449, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.7440558367847836e-05, | |
| "loss": 0.7202, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.73446847676024e-05, | |
| "loss": 0.7384, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.724881116735696e-05, | |
| "loss": 0.7249, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.7152937567111526e-05, | |
| "loss": 0.7286, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.7057063966866087e-05, | |
| "loss": 0.728, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.6961190366620654e-05, | |
| "loss": 0.7224, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.6865316766375215e-05, | |
| "loss": 0.7306, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.6769443166129776e-05, | |
| "loss": 0.7264, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.6673569565884344e-05, | |
| "loss": 0.7392, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.6577695965638905e-05, | |
| "loss": 0.7216, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.6481822365393466e-05, | |
| "loss": 0.7271, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.6385948765148034e-05, | |
| "loss": 0.7233, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.6290075164902595e-05, | |
| "loss": 0.7364, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.619420156465716e-05, | |
| "loss": 0.7186, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.6098327964411724e-05, | |
| "loss": 0.7246, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.6002454364166285e-05, | |
| "loss": 0.7228, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.590658076392085e-05, | |
| "loss": 0.7183, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.5810707163675414e-05, | |
| "loss": 0.7187, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.5714833563429975e-05, | |
| "loss": 0.7188, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.561895996318454e-05, | |
| "loss": 0.7234, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.5523086362939104e-05, | |
| "loss": 0.7102, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.5427212762693665e-05, | |
| "loss": 0.7201, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.533133916244823e-05, | |
| "loss": 0.7176, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.5235465562202793e-05, | |
| "loss": 0.7069, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.5139591961957355e-05, | |
| "loss": 0.7118, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.504371836171192e-05, | |
| "loss": 0.7147, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.494784476146648e-05, | |
| "loss": 0.7132, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.4851971161221044e-05, | |
| "loss": 0.7059, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.475609756097561e-05, | |
| "loss": 0.713, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.466022396073017e-05, | |
| "loss": 0.7154, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.4564350360484734e-05, | |
| "loss": 0.69, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.44684767602393e-05, | |
| "loss": 0.7152, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.437260315999386e-05, | |
| "loss": 0.6957, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.4276729559748424e-05, | |
| "loss": 0.7189, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.418085595950299e-05, | |
| "loss": 0.705, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.408498235925755e-05, | |
| "loss": 0.7036, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.398910875901212e-05, | |
| "loss": 0.7008, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.389323515876668e-05, | |
| "loss": 0.6923, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.379736155852124e-05, | |
| "loss": 0.7158, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.370148795827581e-05, | |
| "loss": 0.7009, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.360561435803037e-05, | |
| "loss": 0.7074, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.350974075778493e-05, | |
| "loss": 0.7037, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.34138671575395e-05, | |
| "loss": 0.6913, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.331799355729406e-05, | |
| "loss": 0.693, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.322211995704863e-05, | |
| "loss": 0.6385, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.312624635680319e-05, | |
| "loss": 0.6387, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.303037275655775e-05, | |
| "loss": 0.6385, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.293449915631232e-05, | |
| "loss": 0.6346, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.283862555606688e-05, | |
| "loss": 0.6373, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.274275195582145e-05, | |
| "loss": 0.6377, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.264687835557601e-05, | |
| "loss": 0.6451, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.255100475533057e-05, | |
| "loss": 0.6343, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.245513115508514e-05, | |
| "loss": 0.6436, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.23592575548397e-05, | |
| "loss": 0.6248, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.226338395459426e-05, | |
| "loss": 0.6393, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.216751035434883e-05, | |
| "loss": 0.6388, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.207163675410339e-05, | |
| "loss": 0.643, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.1975763153857956e-05, | |
| "loss": 0.6426, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.187988955361252e-05, | |
| "loss": 0.6269, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.178401595336708e-05, | |
| "loss": 0.6372, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.1688142353121646e-05, | |
| "loss": 0.6213, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.159226875287621e-05, | |
| "loss": 0.6396, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.1496395152630775e-05, | |
| "loss": 0.6388, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.1400521552385336e-05, | |
| "loss": 0.6325, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.13046479521399e-05, | |
| "loss": 0.6311, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.1208774351894465e-05, | |
| "loss": 0.635, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.1112900751649026e-05, | |
| "loss": 0.6311, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.1017027151403594e-05, | |
| "loss": 0.6406, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.0921153551158155e-05, | |
| "loss": 0.6436, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.0825279950912716e-05, | |
| "loss": 0.6294, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.0729406350667284e-05, | |
| "loss": 0.6307, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.0633532750421845e-05, | |
| "loss": 0.6325, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.0537659150176406e-05, | |
| "loss": 0.6343, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.0441785549930974e-05, | |
| "loss": 0.6296, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0345911949685535e-05, | |
| "loss": 0.6244, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0250038349440102e-05, | |
| "loss": 0.6342, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0154164749194663e-05, | |
| "loss": 0.6259, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.0058291148949224e-05, | |
| "loss": 0.6332, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.9962417548703792e-05, | |
| "loss": 0.626, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9866543948458353e-05, | |
| "loss": 0.6314, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.977067034821292e-05, | |
| "loss": 0.6276, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9674796747967482e-05, | |
| "loss": 0.6274, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9578923147722043e-05, | |
| "loss": 0.6231, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.948304954747661e-05, | |
| "loss": 0.6257, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9387175947231172e-05, | |
| "loss": 0.6379, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9291302346985733e-05, | |
| "loss": 0.6294, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.91954287467403e-05, | |
| "loss": 0.6297, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9099555146494862e-05, | |
| "loss": 0.6273, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.900368154624943e-05, | |
| "loss": 0.6223, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.890780794600399e-05, | |
| "loss": 0.6352, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.881193434575855e-05, | |
| "loss": 0.6209, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.871606074551312e-05, | |
| "loss": 0.627, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.862018714526768e-05, | |
| "loss": 0.63, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.8524313545022245e-05, | |
| "loss": 0.6264, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.842843994477681e-05, | |
| "loss": 0.6212, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.833256634453137e-05, | |
| "loss": 0.6137, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.8236692744285935e-05, | |
| "loss": 0.6187, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.81408191440405e-05, | |
| "loss": 0.6266, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.804494554379506e-05, | |
| "loss": 0.6157, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.7949071943549625e-05, | |
| "loss": 0.623, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.785319834330419e-05, | |
| "loss": 0.6202, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7757324743058753e-05, | |
| "loss": 0.6169, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.7661451142813314e-05, | |
| "loss": 0.6134, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.756557754256788e-05, | |
| "loss": 0.6189, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7469703942322443e-05, | |
| "loss": 0.6263, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7373830342077004e-05, | |
| "loss": 0.6331, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7277956741831572e-05, | |
| "loss": 0.6204, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7182083141586133e-05, | |
| "loss": 0.6187, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7086209541340694e-05, | |
| "loss": 0.6099, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.6990335941095262e-05, | |
| "loss": 0.6184, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.6894462340849823e-05, | |
| "loss": 0.6134, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.679858874060439e-05, | |
| "loss": 0.6064, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6702715140358952e-05, | |
| "loss": 0.6315, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6606841540113513e-05, | |
| "loss": 0.6245, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.651096793986808e-05, | |
| "loss": 0.6069, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.641509433962264e-05, | |
| "loss": 0.6224, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6319220739377203e-05, | |
| "loss": 0.6165, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.622334713913177e-05, | |
| "loss": 0.6124, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.612747353888633e-05, | |
| "loss": 0.6112, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.60315999386409e-05, | |
| "loss": 0.6093, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.593572633839546e-05, | |
| "loss": 0.6035, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.583985273815002e-05, | |
| "loss": 0.609, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.574397913790459e-05, | |
| "loss": 0.5955, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.564810553765915e-05, | |
| "loss": 0.6195, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5552231937413718e-05, | |
| "loss": 0.6143, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.545635833716828e-05, | |
| "loss": 0.6083, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.536048473692284e-05, | |
| "loss": 0.6209, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5264611136677408e-05, | |
| "loss": 0.6203, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.516873753643197e-05, | |
| "loss": 0.6001, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.507286393618653e-05, | |
| "loss": 0.6096, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.4976990335941098e-05, | |
| "loss": 0.6131, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.488111673569566e-05, | |
| "loss": 0.6104, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.4785243135450223e-05, | |
| "loss": 0.6044, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4689369535204788e-05, | |
| "loss": 0.5952, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4593495934959352e-05, | |
| "loss": 0.5971, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.4497622334713916e-05, | |
| "loss": 0.6026, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4401748734468477e-05, | |
| "loss": 0.5951, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4305875134223042e-05, | |
| "loss": 0.6101, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4210001533977606e-05, | |
| "loss": 0.599, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.411412793373217e-05, | |
| "loss": 0.6086, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4018254333486732e-05, | |
| "loss": 0.6069, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.3922380733241296e-05, | |
| "loss": 0.606, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.382650713299586e-05, | |
| "loss": 0.6106, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3730633532750425e-05, | |
| "loss": 0.5935, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.363475993250499e-05, | |
| "loss": 0.6018, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.353888633225955e-05, | |
| "loss": 0.6004, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.3443012732014115e-05, | |
| "loss": 0.5984, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.334713913176868e-05, | |
| "loss": 0.6003, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.325126553152324e-05, | |
| "loss": 0.6122, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3155391931277805e-05, | |
| "loss": 0.5897, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.305951833103237e-05, | |
| "loss": 0.604, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.296364473078693e-05, | |
| "loss": 0.5999, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2867771130541495e-05, | |
| "loss": 0.5959, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.277189753029606e-05, | |
| "loss": 0.6054, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.267602393005062e-05, | |
| "loss": 0.6028, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2580150329805184e-05, | |
| "loss": 0.6035, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.248427672955975e-05, | |
| "loss": 0.5916, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2388403129314313e-05, | |
| "loss": 0.6072, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2292529529068874e-05, | |
| "loss": 0.5871, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.219665592882344e-05, | |
| "loss": 0.601, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.2100782328578003e-05, | |
| "loss": 0.5954, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.2004908728332567e-05, | |
| "loss": 0.5971, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.190903512808713e-05, | |
| "loss": 0.5976, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1813161527841693e-05, | |
| "loss": 0.5951, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1717287927596257e-05, | |
| "loss": 0.5925, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1621414327350822e-05, | |
| "loss": 0.5913, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1525540727105386e-05, | |
| "loss": 0.6006, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1429667126859947e-05, | |
| "loss": 0.5942, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.133379352661451e-05, | |
| "loss": 0.5825, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1237919926369076e-05, | |
| "loss": 0.592, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.114204632612364e-05, | |
| "loss": 0.5848, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.10461727258782e-05, | |
| "loss": 0.594, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.0950299125632766e-05, | |
| "loss": 0.598, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.085442552538733e-05, | |
| "loss": 0.5973, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0758551925141895e-05, | |
| "loss": 0.5906, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.0662678324896456e-05, | |
| "loss": 0.5904, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.056680472465102e-05, | |
| "loss": 0.5919, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.0470931124405585e-05, | |
| "loss": 0.5979, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.037505752416015e-05, | |
| "loss": 0.5892, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.0279183923914713e-05, | |
| "loss": 0.5922, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0183310323669274e-05, | |
| "loss": 0.5945, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.008743672342384e-05, | |
| "loss": 0.5907, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.9991563123178403e-05, | |
| "loss": 0.5775, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9895689522932968e-05, | |
| "loss": 0.5845, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.979981592268753e-05, | |
| "loss": 0.5941, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9703942322442093e-05, | |
| "loss": 0.5806, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9608068722196658e-05, | |
| "loss": 0.5867, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.9512195121951222e-05, | |
| "loss": 0.5862, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9416321521705786e-05, | |
| "loss": 0.5726, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9320447921460347e-05, | |
| "loss": 0.5832, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9224574321214912e-05, | |
| "loss": 0.5858, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9128700720969476e-05, | |
| "loss": 0.5805, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.903282712072404e-05, | |
| "loss": 0.5764, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.89369535204786e-05, | |
| "loss": 0.6018, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8841079920233166e-05, | |
| "loss": 0.5776, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.874520631998773e-05, | |
| "loss": 0.595, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8649332719742295e-05, | |
| "loss": 0.5891, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8553459119496856e-05, | |
| "loss": 0.5952, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.845758551925142e-05, | |
| "loss": 0.5861, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.8361711919005985e-05, | |
| "loss": 0.5746, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.826583831876055e-05, | |
| "loss": 0.5905, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.816996471851511e-05, | |
| "loss": 0.5871, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.8074091118269675e-05, | |
| "loss": 0.5833, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.797821751802424e-05, | |
| "loss": 0.5864, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.78823439177788e-05, | |
| "loss": 0.5847, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.7786470317533364e-05, | |
| "loss": 0.5895, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.769059671728793e-05, | |
| "loss": 0.5862, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.759472311704249e-05, | |
| "loss": 0.5939, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.7498849516797054e-05, | |
| "loss": 0.5705, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.740297591655162e-05, | |
| "loss": 0.5706, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7307102316306183e-05, | |
| "loss": 0.5667, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7211228716060744e-05, | |
| "loss": 0.5772, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.711535511581531e-05, | |
| "loss": 0.5799, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.7019481515569873e-05, | |
| "loss": 0.5877, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.6923607915324437e-05, | |
| "loss": 0.5628, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6827734315079e-05, | |
| "loss": 0.5812, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6731860714833563e-05, | |
| "loss": 0.5675, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6635987114588127e-05, | |
| "loss": 0.5465, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.654011351434269e-05, | |
| "loss": 0.5017, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.6444239914097253e-05, | |
| "loss": 0.4992, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.6348366313851817e-05, | |
| "loss": 0.4933, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.625249271360638e-05, | |
| "loss": 0.525, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6156619113360946e-05, | |
| "loss": 0.5102, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.606074551311551e-05, | |
| "loss": 0.5177, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.596487191287007e-05, | |
| "loss": 0.5103, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.5868998312624636e-05, | |
| "loss": 0.5144, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.57731247123792e-05, | |
| "loss": 0.5163, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.5677251112133765e-05, | |
| "loss": 0.5015, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5581377511888326e-05, | |
| "loss": 0.5067, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.548550391164289e-05, | |
| "loss": 0.5135, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5389630311397454e-05, | |
| "loss": 0.5042, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.529375671115202e-05, | |
| "loss": 0.5123, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5197883110906583e-05, | |
| "loss": 0.5128, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5102009510661144e-05, | |
| "loss": 0.5003, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.5006135910415709e-05, | |
| "loss": 0.511, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4910262310170273e-05, | |
| "loss": 0.5044, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4814388709924838e-05, | |
| "loss": 0.5104, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4718515109679399e-05, | |
| "loss": 0.512, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4622641509433963e-05, | |
| "loss": 0.516, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4526767909188527e-05, | |
| "loss": 0.5064, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.443089430894309e-05, | |
| "loss": 0.5139, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.4335020708697653e-05, | |
| "loss": 0.5052, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4239147108452217e-05, | |
| "loss": 0.5083, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.414327350820678e-05, | |
| "loss": 0.4972, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.4047399907961344e-05, | |
| "loss": 0.5115, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.3951526307715909e-05, | |
| "loss": 0.5049, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.385565270747047e-05, | |
| "loss": 0.5188, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3759779107225034e-05, | |
| "loss": 0.5102, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3663905506979599e-05, | |
| "loss": 0.5034, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3568031906734163e-05, | |
| "loss": 0.5158, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3472158306488724e-05, | |
| "loss": 0.5021, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3376284706243289e-05, | |
| "loss": 0.5116, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3280411105997853e-05, | |
| "loss": 0.5078, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3184537505752417e-05, | |
| "loss": 0.5182, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3088663905506982e-05, | |
| "loss": 0.4994, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.2992790305261543e-05, | |
| "loss": 0.5111, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.2896916705016107e-05, | |
| "loss": 0.5058, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.2801043104770672e-05, | |
| "loss": 0.503, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2705169504525236e-05, | |
| "loss": 0.5092, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2609295904279797e-05, | |
| "loss": 0.5071, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2513422304034362e-05, | |
| "loss": 0.5145, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2417548703788926e-05, | |
| "loss": 0.5087, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2321675103543489e-05, | |
| "loss": 0.5097, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2225801503298053e-05, | |
| "loss": 0.5132, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2129927903052617e-05, | |
| "loss": 0.5063, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.203405430280718e-05, | |
| "loss": 0.5143, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.1938180702561743e-05, | |
| "loss": 0.5013, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1842307102316307e-05, | |
| "loss": 0.4953, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.174643350207087e-05, | |
| "loss": 0.5026, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.1650559901825433e-05, | |
| "loss": 0.5026, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.1554686301579997e-05, | |
| "loss": 0.5042, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.145881270133456e-05, | |
| "loss": 0.4996, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1362939101089124e-05, | |
| "loss": 0.5076, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1267065500843687e-05, | |
| "loss": 0.4997, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1171191900598251e-05, | |
| "loss": 0.5052, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.1075318300352816e-05, | |
| "loss": 0.5025, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.0979444700107379e-05, | |
| "loss": 0.5061, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0883571099861943e-05, | |
| "loss": 0.511, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0787697499616506e-05, | |
| "loss": 0.5031, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.069182389937107e-05, | |
| "loss": 0.5071, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0595950299125633e-05, | |
| "loss": 0.5003, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0500076698880197e-05, | |
| "loss": 0.4956, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.040420309863476e-05, | |
| "loss": 0.5051, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0308329498389324e-05, | |
| "loss": 0.5016, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0212455898143887e-05, | |
| "loss": 0.5055, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0116582297898452e-05, | |
| "loss": 0.5049, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.0020708697653016e-05, | |
| "loss": 0.5079, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.924835097407579e-06, | |
| "loss": 0.5163, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.828961497162143e-06, | |
| "loss": 0.5015, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.733087896916706e-06, | |
| "loss": 0.5029, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.63721429667127e-06, | |
| "loss": 0.5034, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.541340696425833e-06, | |
| "loss": 0.5001, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.445467096180396e-06, | |
| "loss": 0.5036, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.34959349593496e-06, | |
| "loss": 0.5091, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.253719895689523e-06, | |
| "loss": 0.505, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.157846295444086e-06, | |
| "loss": 0.4992, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.06197269519865e-06, | |
| "loss": 0.5052, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 8.966099094953214e-06, | |
| "loss": 0.4838, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.870225494707777e-06, | |
| "loss": 0.503, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.774351894462341e-06, | |
| "loss": 0.509, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.678478294216904e-06, | |
| "loss": 0.4943, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.582604693971469e-06, | |
| "loss": 0.5004, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.486731093726031e-06, | |
| "loss": 0.494, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.390857493480596e-06, | |
| "loss": 0.4979, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.294983893235158e-06, | |
| "loss": 0.506, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.199110292989723e-06, | |
| "loss": 0.5048, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.103236692744286e-06, | |
| "loss": 0.5033, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.00736309249885e-06, | |
| "loss": 0.4989, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.911489492253414e-06, | |
| "loss": 0.4928, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.815615892007977e-06, | |
| "loss": 0.4981, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.719742291762542e-06, | |
| "loss": 0.505, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.623868691517104e-06, | |
| "loss": 0.502, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.527995091271668e-06, | |
| "loss": 0.4872, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.432121491026231e-06, | |
| "loss": 0.4967, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.336247890780795e-06, | |
| "loss": 0.5011, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.240374290535358e-06, | |
| "loss": 0.5026, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.144500690289922e-06, | |
| "loss": 0.5013, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.048627090044485e-06, | |
| "loss": 0.4961, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 6.952753489799049e-06, | |
| "loss": 0.4968, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.856879889553614e-06, | |
| "loss": 0.4882, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.761006289308176e-06, | |
| "loss": 0.4921, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.665132689062741e-06, | |
| "loss": 0.4931, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.5692590888173035e-06, | |
| "loss": 0.4924, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.473385488571868e-06, | |
| "loss": 0.5035, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.377511888326431e-06, | |
| "loss": 0.4953, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.281638288080994e-06, | |
| "loss": 0.4973, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.185764687835558e-06, | |
| "loss": 0.4982, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.089891087590121e-06, | |
| "loss": 0.4866, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.994017487344685e-06, | |
| "loss": 0.5021, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.8981438870992485e-06, | |
| "loss": 0.4912, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.802270286853812e-06, | |
| "loss": 0.4966, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.706396686608376e-06, | |
| "loss": 0.4913, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.610523086362939e-06, | |
| "loss": 0.4875, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.514649486117503e-06, | |
| "loss": 0.4958, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.418775885872066e-06, | |
| "loss": 0.496, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.322902285626631e-06, | |
| "loss": 0.4889, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.227028685381194e-06, | |
| "loss": 0.4899, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.131155085135757e-06, | |
| "loss": 0.4972, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.035281484890321e-06, | |
| "loss": 0.5017, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.939407884644884e-06, | |
| "loss": 0.4922, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.843534284399448e-06, | |
| "loss": 0.5002, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.747660684154011e-06, | |
| "loss": 0.4961, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.651787083908575e-06, | |
| "loss": 0.4859, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.5559134836631385e-06, | |
| "loss": 0.4967, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.460039883417702e-06, | |
| "loss": 0.5026, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.364166283172266e-06, | |
| "loss": 0.4883, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.26829268292683e-06, | |
| "loss": 0.4866, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.172419082681394e-06, | |
| "loss": 0.4926, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.076545482435957e-06, | |
| "loss": 0.4991, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.980671882190521e-06, | |
| "loss": 0.4936, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.8847982819450834e-06, | |
| "loss": 0.4907, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.7889246816996474e-06, | |
| "loss": 0.4962, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.693051081454211e-06, | |
| "loss": 0.4909, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.5971774812087746e-06, | |
| "loss": 0.4886, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.501303880963338e-06, | |
| "loss": 0.489, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.4054302807179013e-06, | |
| "loss": 0.486, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.309556680472465e-06, | |
| "loss": 0.4891, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.2136830802270293e-06, | |
| "loss": 0.4843, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.1178094799815924e-06, | |
| "loss": 0.4943, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.021935879736156e-06, | |
| "loss": 0.4828, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.9260622794907195e-06, | |
| "loss": 0.4961, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.830188679245283e-06, | |
| "loss": 0.4975, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.7343150789998467e-06, | |
| "loss": 0.4924, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.6384414787544102e-06, | |
| "loss": 0.4941, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.542567878508974e-06, | |
| "loss": 0.4773, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.446694278263538e-06, | |
| "loss": 0.4858, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.350820678018101e-06, | |
| "loss": 0.4967, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.2549470777726645e-06, | |
| "loss": 0.4899, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.159073477527228e-06, | |
| "loss": 0.4959, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.0631998772817916e-06, | |
| "loss": 0.4907, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1.967326277036355e-06, | |
| "loss": 0.4921, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.871452676790919e-06, | |
| "loss": 0.4896, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.7755790765454826e-06, | |
| "loss": 0.4888, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.6797054763000461e-06, | |
| "loss": 0.4903, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.5838318760546097e-06, | |
| "loss": 0.4881, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.4879582758091733e-06, | |
| "loss": 0.4895, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.3920846755637368e-06, | |
| "loss": 0.4815, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.2962110753183004e-06, | |
| "loss": 0.4906, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.200337475072864e-06, | |
| "loss": 0.4936, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.1044638748274275e-06, | |
| "loss": 0.4753, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.0085902745819913e-06, | |
| "loss": 0.4883, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 9.127166743365548e-07, | |
| "loss": 0.4933, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.168430740911182e-07, | |
| "loss": 0.4831, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 7.209694738456819e-07, | |
| "loss": 0.4861, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.250958736002455e-07, | |
| "loss": 0.4806, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.29222273354809e-07, | |
| "loss": 0.4767, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.333486731093726e-07, | |
| "loss": 0.4845, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.3747507286393617e-07, | |
| "loss": 0.4924, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.416014726184998e-07, | |
| "loss": 0.4937, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.4572787237306336e-07, | |
| "loss": 0.4803, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.985427212762693e-08, | |
| "loss": 0.4802, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 260760, | |
| "total_flos": 2.990387392931758e+17, | |
| "train_loss": 0.11540543585032806, | |
| "train_runtime": 18089.1482, | |
| "train_samples_per_second": 144.153, | |
| "train_steps_per_second": 14.415 | |
| } | |
| ], | |
| "max_steps": 260760, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.990387392931758e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |