| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "eval_steps": 500, | |
| "global_step": 29280, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 1.1839450597763062, | |
| "learning_rate": 6.25e-05, | |
| "loss": 6.2404, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.9727169871330261, | |
| "learning_rate": 0.000125, | |
| "loss": 3.63, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.0313897132873535, | |
| "learning_rate": 0.0001875, | |
| "loss": 3.3461, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.0492364168167114, | |
| "learning_rate": 0.00025, | |
| "loss": 3.1763, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.9632183909416199, | |
| "learning_rate": 0.0003125, | |
| "loss": 3.0421, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.437592595514739, | |
| "eval_loss": 3.014521598815918, | |
| "eval_runtime": 3.2929, | |
| "eval_samples_per_second": 1363.845, | |
| "eval_steps_per_second": 10.933, | |
| "step": 2928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.8621485829353333, | |
| "learning_rate": 0.000375, | |
| "loss": 2.9614, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.8864262700080872, | |
| "learning_rate": 0.00043750000000000006, | |
| "loss": 2.8642, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.974511981010437, | |
| "learning_rate": 0.0005, | |
| "loss": 2.8239, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.843772828578949, | |
| "learning_rate": 0.0005625000000000001, | |
| "loss": 2.7889, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 0.8417291045188904, | |
| "learning_rate": 0.000625, | |
| "loss": 2.7435, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.7840189933776855, | |
| "learning_rate": 0.0006875, | |
| "loss": 2.7062, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.459005500063995, | |
| "eval_loss": 2.790151596069336, | |
| "eval_runtime": 3.3884, | |
| "eval_samples_per_second": 1325.389, | |
| "eval_steps_per_second": 10.624, | |
| "step": 5856 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.7942721843719482, | |
| "learning_rate": 0.00075, | |
| "loss": 2.6608, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.7657195925712585, | |
| "learning_rate": 0.0008125, | |
| "loss": 2.6336, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 0.6993751525878906, | |
| "learning_rate": 0.0008750000000000001, | |
| "loss": 2.6207, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.6740849614143372, | |
| "learning_rate": 0.0009375, | |
| "loss": 2.6097, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 0.6975589990615845, | |
| "learning_rate": 0.001, | |
| "loss": 2.5888, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.6254838705062866, | |
| "learning_rate": 0.0010625, | |
| "loss": 2.5829, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.46816818238401564, | |
| "eval_loss": 2.6946029663085938, | |
| "eval_runtime": 3.3593, | |
| "eval_samples_per_second": 1336.889, | |
| "eval_steps_per_second": 10.717, | |
| "step": 8784 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 0.6288382411003113, | |
| "learning_rate": 0.0011250000000000001, | |
| "loss": 2.5418, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 0.5383425354957581, | |
| "learning_rate": 0.0011875, | |
| "loss": 2.5137, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "grad_norm": 0.5171464681625366, | |
| "learning_rate": 0.00125, | |
| "loss": 2.5227, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "grad_norm": 0.5070509910583496, | |
| "learning_rate": 0.0013125, | |
| "loss": 2.5239, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.4720727801322937, | |
| "learning_rate": 0.001375, | |
| "loss": 2.5135, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "grad_norm": 0.5138970017433167, | |
| "learning_rate": 0.0014375000000000002, | |
| "loss": 2.5042, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.47502003131372106, | |
| "eval_loss": 2.644279956817627, | |
| "eval_runtime": 3.3434, | |
| "eval_samples_per_second": 1343.254, | |
| "eval_steps_per_second": 10.768, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 0.499744713306427, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4579, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 0.4518820345401764, | |
| "learning_rate": 0.0015625, | |
| "loss": 2.4445, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "grad_norm": 0.4289190173149109, | |
| "learning_rate": 0.001625, | |
| "loss": 2.462, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.44824743270874023, | |
| "learning_rate": 0.0016875, | |
| "loss": 2.4688, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 0.4198445677757263, | |
| "learning_rate": 0.0017500000000000003, | |
| "loss": 2.4673, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 0.3872630298137665, | |
| "learning_rate": 0.0018124999999999999, | |
| "loss": 2.4588, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.4792787675087708, | |
| "eval_loss": 2.6083338260650635, | |
| "eval_runtime": 3.3503, | |
| "eval_samples_per_second": 1340.489, | |
| "eval_steps_per_second": 10.745, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "grad_norm": 0.39676180481910706, | |
| "learning_rate": 0.001875, | |
| "loss": 2.4106, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "grad_norm": 0.4363040626049042, | |
| "learning_rate": 0.0019375000000000002, | |
| "loss": 2.3922, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "grad_norm": 0.41588276624679565, | |
| "learning_rate": 0.002, | |
| "loss": 2.4188, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 0.38543984293937683, | |
| "learning_rate": 0.0020625, | |
| "loss": 2.4218, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "grad_norm": 0.35857653617858887, | |
| "learning_rate": 0.002125, | |
| "loss": 2.4239, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "grad_norm": 0.331021249294281, | |
| "learning_rate": 0.0021874999999999998, | |
| "loss": 2.4252, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.48289229377389953, | |
| "eval_loss": 2.5867631435394287, | |
| "eval_runtime": 3.3509, | |
| "eval_samples_per_second": 1340.236, | |
| "eval_steps_per_second": 10.743, | |
| "step": 17568 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "grad_norm": 0.34858438372612, | |
| "learning_rate": 0.0022500000000000003, | |
| "loss": 2.3489, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "grad_norm": 0.35453295707702637, | |
| "learning_rate": 0.0023125000000000003, | |
| "loss": 2.3836, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "grad_norm": 0.3333473205566406, | |
| "learning_rate": 0.002375, | |
| "loss": 2.3823, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "grad_norm": 0.34180307388305664, | |
| "learning_rate": 0.0024375, | |
| "loss": 2.3823, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "grad_norm": 0.36889737844467163, | |
| "learning_rate": 0.0025, | |
| "loss": 2.3884, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.4853994954037559, | |
| "eval_loss": 2.5670363903045654, | |
| "eval_runtime": 3.3422, | |
| "eval_samples_per_second": 1343.741, | |
| "eval_steps_per_second": 10.771, | |
| "step": 20496 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.34776511788368225, | |
| "learning_rate": 0.0025625, | |
| "loss": 2.3873, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "grad_norm": 0.3375663161277771, | |
| "learning_rate": 0.002625, | |
| "loss": 2.3223, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "grad_norm": 0.35632824897766113, | |
| "learning_rate": 0.0026875000000000002, | |
| "loss": 2.3341, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "grad_norm": 0.34955427050590515, | |
| "learning_rate": 0.00275, | |
| "loss": 2.3471, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "grad_norm": 0.33797112107276917, | |
| "learning_rate": 0.0028125, | |
| "loss": 2.3604, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 0.33891069889068604, | |
| "learning_rate": 0.0028750000000000004, | |
| "loss": 2.3624, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.4855029393870856, | |
| "eval_loss": 2.5581905841827393, | |
| "eval_runtime": 3.415, | |
| "eval_samples_per_second": 1315.097, | |
| "eval_steps_per_second": 10.542, | |
| "step": 23424 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "grad_norm": 0.3400173783302307, | |
| "learning_rate": 0.0029375, | |
| "loss": 2.3493, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "grad_norm": 0.37537550926208496, | |
| "learning_rate": 0.003, | |
| "loss": 2.2908, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "grad_norm": 0.31209731101989746, | |
| "learning_rate": 0.002715909090909091, | |
| "loss": 2.3134, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "grad_norm": 0.30068129301071167, | |
| "learning_rate": 0.0024318181818181817, | |
| "loss": 2.3028, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "grad_norm": 0.30200621485710144, | |
| "learning_rate": 0.002147727272727273, | |
| "loss": 2.3009, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "grad_norm": 0.3131857216358185, | |
| "learning_rate": 0.0018636363636363638, | |
| "loss": 2.2859, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.494355289757117, | |
| "eval_loss": 2.501802921295166, | |
| "eval_runtime": 3.3486, | |
| "eval_samples_per_second": 1341.16, | |
| "eval_steps_per_second": 10.751, | |
| "step": 26352 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "grad_norm": 0.3251318633556366, | |
| "learning_rate": 0.0015795454545454546, | |
| "loss": 2.2396, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "grad_norm": 0.32136398553848267, | |
| "learning_rate": 0.0012954545454545456, | |
| "loss": 2.1712, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "grad_norm": 0.2987048625946045, | |
| "learning_rate": 0.0010113636363636364, | |
| "loss": 2.1659, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "grad_norm": 0.34757348895072937, | |
| "learning_rate": 0.0007272727272727273, | |
| "loss": 2.1607, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "grad_norm": 0.2865099012851715, | |
| "learning_rate": 0.0004431818181818182, | |
| "loss": 2.1531, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "grad_norm": 0.3215331733226776, | |
| "learning_rate": 0.0001590909090909091, | |
| "loss": 2.1433, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.49926975560920617, | |
| "eval_loss": 2.47407603263855, | |
| "eval_runtime": 3.3548, | |
| "eval_samples_per_second": 1338.689, | |
| "eval_steps_per_second": 10.731, | |
| "step": 29280 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 29280, | |
| "total_flos": 2273237316403200.0, | |
| "train_loss": 2.5694013522622363, | |
| "train_runtime": 748.563, | |
| "train_samples_per_second": 625.799, | |
| "train_steps_per_second": 39.115 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 29280, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 2000, | |
| "total_flos": 2273237316403200.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |