| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 42423, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5e-05, | |
| "loss": 1.3894, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7218, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.879294111966783e-05, | |
| "loss": 0.7089, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.758588223933563e-05, | |
| "loss": 0.6974, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.637882335900346e-05, | |
| "loss": 0.68, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.517176447867127e-05, | |
| "loss": 0.6789, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.396470559833909e-05, | |
| "loss": 0.6668, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.275764671800691e-05, | |
| "loss": 0.6663, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.155058783767473e-05, | |
| "loss": 0.6634, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.034352895734254e-05, | |
| "loss": 0.6529, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.913647007701035e-05, | |
| "loss": 0.6555, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.792941119667818e-05, | |
| "loss": 0.6464, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.6722352316346e-05, | |
| "loss": 0.6397, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.55152934360138e-05, | |
| "loss": 0.6439, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.430823455568163e-05, | |
| "loss": 0.6393, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.310117567534945e-05, | |
| "loss": 0.6352, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.189411679501726e-05, | |
| "loss": 0.6304, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.068705791468509e-05, | |
| "loss": 0.6344, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.94799990343529e-05, | |
| "loss": 0.6336, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.827294015402071e-05, | |
| "loss": 0.623, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.706588127368853e-05, | |
| "loss": 0.6186, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.585882239335635e-05, | |
| "loss": 0.6288, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.465176351302417e-05, | |
| "loss": 0.6156, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.344470463269198e-05, | |
| "loss": 0.6113, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.223764575235981e-05, | |
| "loss": 0.6182, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 7.103058687202762e-05, | |
| "loss": 0.6233, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.982352799169543e-05, | |
| "loss": 0.6042, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.861646911136326e-05, | |
| "loss": 0.6043, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4832, | |
| "eval_loss": 0.5925313234329224, | |
| "eval_runtime": 306.7871, | |
| "eval_samples_per_second": 25.659, | |
| "eval_steps_per_second": 1.604, | |
| "step": 14141 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.740941023103108e-05, | |
| "loss": 0.5816, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.620235135069889e-05, | |
| "loss": 0.5907, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.499529247036672e-05, | |
| "loss": 0.5814, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.378823359003453e-05, | |
| "loss": 0.5949, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.258117470970234e-05, | |
| "loss": 0.5838, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.137411582937016e-05, | |
| "loss": 0.5823, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.016705694903798e-05, | |
| "loss": 0.5872, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.895999806870579e-05, | |
| "loss": 0.5779, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.775293918837361e-05, | |
| "loss": 0.5744, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.654588030804143e-05, | |
| "loss": 0.5833, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.533882142770924e-05, | |
| "loss": 0.5855, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.413176254737706e-05, | |
| "loss": 0.5836, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.292470366704488e-05, | |
| "loss": 0.5793, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.17176447867127e-05, | |
| "loss": 0.57, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 5.051058590638052e-05, | |
| "loss": 0.5792, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.930352702604834e-05, | |
| "loss": 0.5577, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.809646814571615e-05, | |
| "loss": 0.5729, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.6889409265383964e-05, | |
| "loss": 0.5618, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.5682350385051784e-05, | |
| "loss": 0.5678, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.4475291504719604e-05, | |
| "loss": 0.5723, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.326823262438742e-05, | |
| "loss": 0.5616, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.206117374405524e-05, | |
| "loss": 0.5707, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.085411486372305e-05, | |
| "loss": 0.5626, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.964705598339087e-05, | |
| "loss": 0.5703, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.843999710305869e-05, | |
| "loss": 0.5683, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.7232938222726505e-05, | |
| "loss": 0.555, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 3.6025879342394325e-05, | |
| "loss": 0.5664, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.481882046206214e-05, | |
| "loss": 0.5647, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.5152, | |
| "eval_loss": 0.5659283995628357, | |
| "eval_runtime": 306.5061, | |
| "eval_samples_per_second": 25.683, | |
| "eval_steps_per_second": 1.605, | |
| "step": 28282 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.361176158172996e-05, | |
| "loss": 0.5582, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.240470270139778e-05, | |
| "loss": 0.5302, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.119764382106559e-05, | |
| "loss": 0.545, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.9990584940733412e-05, | |
| "loss": 0.5382, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 2.878352606040123e-05, | |
| "loss": 0.5406, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.7576467180069043e-05, | |
| "loss": 0.5418, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 2.6369408299736863e-05, | |
| "loss": 0.5378, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 2.516234941940468e-05, | |
| "loss": 0.5412, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.3955290539072496e-05, | |
| "loss": 0.538, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.2748231658740317e-05, | |
| "loss": 0.5227, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.154117277840813e-05, | |
| "loss": 0.5314, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.033411389807595e-05, | |
| "loss": 0.5342, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1.9127055017743767e-05, | |
| "loss": 0.5381, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1.7919996137411584e-05, | |
| "loss": 0.5349, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 1.67129372570794e-05, | |
| "loss": 0.5349, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1.5505878376747217e-05, | |
| "loss": 0.5409, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1.4298819496415036e-05, | |
| "loss": 0.5149, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 1.3091760616082854e-05, | |
| "loss": 0.5289, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 1.1884701735750671e-05, | |
| "loss": 0.5207, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1.0677642855418488e-05, | |
| "loss": 0.5263, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 9.470583975086306e-06, | |
| "loss": 0.5246, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 8.263525094754123e-06, | |
| "loss": 0.5242, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 7.056466214421941e-06, | |
| "loss": 0.5349, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 5.849407334089757e-06, | |
| "loss": 0.5321, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.642348453757574e-06, | |
| "loss": 0.5215, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.435289573425392e-06, | |
| "loss": 0.5194, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 2.228230693093209e-06, | |
| "loss": 0.5224, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 1.0211718127610264e-06, | |
| "loss": 0.5237, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.5259, | |
| "eval_loss": 0.5759526491165161, | |
| "eval_runtime": 306.403, | |
| "eval_samples_per_second": 25.692, | |
| "eval_steps_per_second": 1.606, | |
| "step": 42423 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 42423, | |
| "total_flos": 4.593284195077325e+16, | |
| "train_loss": 0.05477445060331273, | |
| "train_runtime": 3408.5838, | |
| "train_samples_per_second": 99.567, | |
| "train_steps_per_second": 12.446 | |
| } | |
| ], | |
| "max_steps": 42423, | |
| "num_train_epochs": 3, | |
| "total_flos": 4.593284195077325e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |