| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9960159362549801, | |
| "global_step": 51000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9999511756893996e-05, | |
| "loss": 10.8875, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.975587844699633e-05, | |
| "loss": 5.458, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.951175689399266e-05, | |
| "loss": 4.5892, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.926763534098899e-05, | |
| "loss": 4.4037, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9023513787985316e-05, | |
| "loss": 4.2792, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.8779392234981645e-05, | |
| "loss": 4.201, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.853527068197797e-05, | |
| "loss": 4.1614, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.82911491289743e-05, | |
| "loss": 4.1349, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.804702757597063e-05, | |
| "loss": 4.0883, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.780290602296696e-05, | |
| "loss": 4.0732, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.755878446996329e-05, | |
| "loss": 4.048, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.7314662916959616e-05, | |
| "loss": 4.0258, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7070541363955944e-05, | |
| "loss": 4.0192, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.682641981095227e-05, | |
| "loss": 4.0183, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.65822982579486e-05, | |
| "loss": 3.9832, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.633817670494493e-05, | |
| "loss": 3.9937, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.609405515194126e-05, | |
| "loss": 3.9712, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.5849933598937586e-05, | |
| "loss": 3.9645, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.5605812045933915e-05, | |
| "loss": 3.951, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.536169049293024e-05, | |
| "loss": 3.9571, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.511756893992657e-05, | |
| "loss": 3.9625, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.48734473869229e-05, | |
| "loss": 3.9393, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.462932583391923e-05, | |
| "loss": 3.9326, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.438520428091556e-05, | |
| "loss": 3.9223, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.4141082727911886e-05, | |
| "loss": 3.9397, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.3896961174908214e-05, | |
| "loss": 3.9159, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.365283962190454e-05, | |
| "loss": 3.905, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.3408718068900864e-05, | |
| "loss": 3.8848, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.31645965158972e-05, | |
| "loss": 3.887, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.292047496289353e-05, | |
| "loss": 3.8847, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.267635340988985e-05, | |
| "loss": 3.8797, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.2432231856886185e-05, | |
| "loss": 3.8924, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.2188110303882513e-05, | |
| "loss": 3.9005, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.1943988750878835e-05, | |
| "loss": 3.858, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.169986719787517e-05, | |
| "loss": 3.8763, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.14557456448715e-05, | |
| "loss": 3.8808, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.121162409186782e-05, | |
| "loss": 3.8721, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.0967502538864156e-05, | |
| "loss": 3.8733, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.0723380985860484e-05, | |
| "loss": 3.8694, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.0479259432856806e-05, | |
| "loss": 3.8685, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.023513787985314e-05, | |
| "loss": 3.8451, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.999101632684947e-05, | |
| "loss": 3.8267, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.974689477384579e-05, | |
| "loss": 3.8492, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.950277322084212e-05, | |
| "loss": 3.8322, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.9258651667838455e-05, | |
| "loss": 3.8824, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.901453011483478e-05, | |
| "loss": 3.8454, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.8770408561831105e-05, | |
| "loss": 3.8463, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.852628700882744e-05, | |
| "loss": 3.8457, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.828216545582376e-05, | |
| "loss": 3.8437, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.803804390282009e-05, | |
| "loss": 3.8046, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.7793922349816426e-05, | |
| "loss": 3.8114, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 3.754980079681275e-05, | |
| "loss": 3.8322, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 3.7305679243809076e-05, | |
| "loss": 3.797, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 3.706155769080541e-05, | |
| "loss": 3.855, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.681743613780173e-05, | |
| "loss": 3.8149, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 3.657331458479806e-05, | |
| "loss": 3.8168, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.63291930317944e-05, | |
| "loss": 3.8025, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.6085071478790725e-05, | |
| "loss": 3.8156, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.584094992578705e-05, | |
| "loss": 3.8241, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.559682837278338e-05, | |
| "loss": 3.7813, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.535270681977971e-05, | |
| "loss": 3.7708, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.510858526677603e-05, | |
| "loss": 3.7432, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.486446371377236e-05, | |
| "loss": 3.7315, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.4620342160768696e-05, | |
| "loss": 3.7491, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.437622060776502e-05, | |
| "loss": 3.7062, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.4132099054761346e-05, | |
| "loss": 3.7109, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.388797750175768e-05, | |
| "loss": 3.7077, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.3643855948754e-05, | |
| "loss": 3.6941, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.339973439575033e-05, | |
| "loss": 3.7062, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.315561284274667e-05, | |
| "loss": 3.6663, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.291149128974299e-05, | |
| "loss": 3.6651, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.266736973673932e-05, | |
| "loss": 3.6516, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.242324818373565e-05, | |
| "loss": 3.6591, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.2179126630731974e-05, | |
| "loss": 3.648, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.19350050777283e-05, | |
| "loss": 3.6457, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.169088352472464e-05, | |
| "loss": 3.6374, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.144676197172096e-05, | |
| "loss": 3.6312, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.120264041871729e-05, | |
| "loss": 3.6071, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.0958518865713616e-05, | |
| "loss": 3.5633, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.0714397312709945e-05, | |
| "loss": 3.5886, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.0470275759706273e-05, | |
| "loss": 3.5567, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.02261542067026e-05, | |
| "loss": 3.5192, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.9982032653698933e-05, | |
| "loss": 3.5032, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.973791110069526e-05, | |
| "loss": 3.4983, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.9493789547691587e-05, | |
| "loss": 3.485, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.924966799468792e-05, | |
| "loss": 3.4727, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.9005546441684244e-05, | |
| "loss": 3.44, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.8761424888680572e-05, | |
| "loss": 3.4306, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.8517303335676904e-05, | |
| "loss": 3.3684, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.827318178267323e-05, | |
| "loss": 3.354, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.8029060229669558e-05, | |
| "loss": 3.3174, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.778493867666589e-05, | |
| "loss": 3.2533, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.7540817123662215e-05, | |
| "loss": 3.2496, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.7296695570658543e-05, | |
| "loss": 3.2356, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.7052574017654875e-05, | |
| "loss": 3.2401, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.68084524646512e-05, | |
| "loss": 3.2198, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.656433091164753e-05, | |
| "loss": 3.174, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.6320209358643854e-05, | |
| "loss": 3.1457, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.6076087805640186e-05, | |
| "loss": 3.1575, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.5831966252636514e-05, | |
| "loss": 3.1276, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.558784469963284e-05, | |
| "loss": 3.0936, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.534372314662917e-05, | |
| "loss": 3.0638, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.50996015936255e-05, | |
| "loss": 3.0411, | |
| "step": 51000 | |
| } | |
| ], | |
| "max_steps": 102408, | |
| "num_train_epochs": 2, | |
| "total_flos": 2.469751052832e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |