| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.774538386783285, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9912109375000002e-05, | |
| "loss": 1.9656, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.98193359375e-05, | |
| "loss": 1.8445, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.97216796875e-05, | |
| "loss": 1.59, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9624023437500002e-05, | |
| "loss": 1.6275, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.9526367187500002e-05, | |
| "loss": 1.473, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.9428710937500003e-05, | |
| "loss": 1.3701, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.93310546875e-05, | |
| "loss": 1.3224, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.92333984375e-05, | |
| "loss": 1.1423, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.91357421875e-05, | |
| "loss": 1.1652, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.9038085937500002e-05, | |
| "loss": 1.1422, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.8940429687500002e-05, | |
| "loss": 1.002, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8842773437500003e-05, | |
| "loss": 1.0779, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.87451171875e-05, | |
| "loss": 0.9887, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.86474609375e-05, | |
| "loss": 0.9543, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.85498046875e-05, | |
| "loss": 0.9371, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 1.8452148437500002e-05, | |
| "loss": 0.8701, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.8354492187500003e-05, | |
| "loss": 0.875, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.82568359375e-05, | |
| "loss": 0.7843, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 1.81591796875e-05, | |
| "loss": 0.7945, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.80615234375e-05, | |
| "loss": 0.799, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.79638671875e-05, | |
| "loss": 0.7623, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 1.7866210937500002e-05, | |
| "loss": 0.7263, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 1.7768554687500003e-05, | |
| "loss": 0.7779, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 1.76708984375e-05, | |
| "loss": 0.695, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.75732421875e-05, | |
| "loss": 0.7344, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.74755859375e-05, | |
| "loss": 0.695, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 1.7377929687500002e-05, | |
| "loss": 0.6504, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 1.7280273437500002e-05, | |
| "loss": 0.6447, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 1.7182617187500003e-05, | |
| "loss": 0.6931, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 1.70849609375e-05, | |
| "loss": 0.6256, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 1.69873046875e-05, | |
| "loss": 0.6132, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 1.68896484375e-05, | |
| "loss": 0.6001, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 1.6791992187500002e-05, | |
| "loss": 0.6176, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 1.6694335937500002e-05, | |
| "loss": 0.5709, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 1.65966796875e-05, | |
| "loss": 0.564, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 1.64990234375e-05, | |
| "loss": 0.5969, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 1.64013671875e-05, | |
| "loss": 0.5484, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 1.63037109375e-05, | |
| "loss": 0.5667, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.6206054687500002e-05, | |
| "loss": 0.5442, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 1.6108398437500003e-05, | |
| "loss": 0.4857, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 1.60107421875e-05, | |
| "loss": 0.5225, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 1.59130859375e-05, | |
| "loss": 0.5457, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 1.58154296875e-05, | |
| "loss": 0.5315, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 1.57177734375e-05, | |
| "loss": 0.5345, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.5620117187500002e-05, | |
| "loss": 0.5169, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 1.5522460937500003e-05, | |
| "loss": 0.5115, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 1.54248046875e-05, | |
| "loss": 0.4913, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 1.53271484375e-05, | |
| "loss": 0.4868, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 1.5229492187500001e-05, | |
| "loss": 0.5226, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 1.5131835937500002e-05, | |
| "loss": 0.4061, | |
| "step": 1000 | |
| } | |
| ], | |
| "max_steps": 4096, | |
| "num_train_epochs": 32, | |
| "total_flos": 1.2995638935552e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |