| { | |
| "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": 1.3569283485412598, | |
| "learning_rate": 6.25e-05, | |
| "loss": 6.1959, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.8884565830230713, | |
| "learning_rate": 0.000125, | |
| "loss": 3.6302, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.9340901374816895, | |
| "learning_rate": 0.0001875, | |
| "loss": 3.3395, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.026479721069336, | |
| "learning_rate": 0.00025, | |
| "loss": 3.1647, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.9679443836212158, | |
| "learning_rate": 0.0003125, | |
| "loss": 3.0568, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.43760837510541645, | |
| "eval_loss": 3.0170083045959473, | |
| "eval_runtime": 3.3233, | |
| "eval_samples_per_second": 1351.351, | |
| "eval_steps_per_second": 10.832, | |
| "step": 2928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.8825171589851379, | |
| "learning_rate": 0.000375, | |
| "loss": 2.9449, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.9031845927238464, | |
| "learning_rate": 0.00043750000000000006, | |
| "loss": 2.8664, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.8770526051521301, | |
| "learning_rate": 0.0005, | |
| "loss": 2.8193, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.8624604940414429, | |
| "learning_rate": 0.0005625000000000001, | |
| "loss": 2.7658, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 0.8200684189796448, | |
| "learning_rate": 0.000625, | |
| "loss": 2.7358, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.8737301230430603, | |
| "learning_rate": 0.0006875, | |
| "loss": 2.7079, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.4585023064501707, | |
| "eval_loss": 2.7878565788269043, | |
| "eval_runtime": 3.4124, | |
| "eval_samples_per_second": 1316.077, | |
| "eval_steps_per_second": 10.55, | |
| "step": 5856 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.7897689342498779, | |
| "learning_rate": 0.00075, | |
| "loss": 2.6723, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.8149366974830627, | |
| "learning_rate": 0.0008125, | |
| "loss": 2.6196, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 0.684084951877594, | |
| "learning_rate": 0.0008750000000000001, | |
| "loss": 2.6148, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.6549258828163147, | |
| "learning_rate": 0.0009375, | |
| "loss": 2.6107, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 0.670656681060791, | |
| "learning_rate": 0.001, | |
| "loss": 2.5927, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.6147797703742981, | |
| "learning_rate": 0.0010625, | |
| "loss": 2.5856, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.47057018674268924, | |
| "eval_loss": 2.6923789978027344, | |
| "eval_runtime": 3.3929, | |
| "eval_samples_per_second": 1323.657, | |
| "eval_steps_per_second": 10.61, | |
| "step": 8784 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 0.6165093779563904, | |
| "learning_rate": 0.0011250000000000001, | |
| "loss": 2.5405, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 0.5891627669334412, | |
| "learning_rate": 0.0011875, | |
| "loss": 2.513, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "grad_norm": 0.5683926939964294, | |
| "learning_rate": 0.00125, | |
| "loss": 2.5239, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "grad_norm": 0.49277979135513306, | |
| "learning_rate": 0.0013125, | |
| "loss": 2.5101, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.5084903836250305, | |
| "learning_rate": 0.001375, | |
| "loss": 2.5176, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "grad_norm": 0.4873659014701843, | |
| "learning_rate": 0.0014375000000000002, | |
| "loss": 2.5108, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.4762087604780865, | |
| "eval_loss": 2.6403005123138428, | |
| "eval_runtime": 3.4307, | |
| "eval_samples_per_second": 1309.069, | |
| "eval_steps_per_second": 10.494, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 0.47938045859336853, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4624, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 0.4903472065925598, | |
| "learning_rate": 0.0015625, | |
| "loss": 2.4398, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "grad_norm": 0.43617236614227295, | |
| "learning_rate": 0.001625, | |
| "loss": 2.4683, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.40969905257225037, | |
| "learning_rate": 0.0016875, | |
| "loss": 2.4612, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 0.367087185382843, | |
| "learning_rate": 0.0017500000000000003, | |
| "loss": 2.4706, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 0.37903687357902527, | |
| "learning_rate": 0.0018124999999999999, | |
| "loss": 2.4576, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.47962591850367403, | |
| "eval_loss": 2.607229709625244, | |
| "eval_runtime": 3.4265, | |
| "eval_samples_per_second": 1310.668, | |
| "eval_steps_per_second": 10.506, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "grad_norm": 0.38673949241638184, | |
| "learning_rate": 0.001875, | |
| "loss": 2.4, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "grad_norm": 0.38303470611572266, | |
| "learning_rate": 0.0019375000000000002, | |
| "loss": 2.3994, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "grad_norm": 0.3417862057685852, | |
| "learning_rate": 0.002, | |
| "loss": 2.4168, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 0.34288668632507324, | |
| "learning_rate": 0.0020625, | |
| "loss": 2.4167, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "grad_norm": 0.36294203996658325, | |
| "learning_rate": 0.002125, | |
| "loss": 2.4319, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "grad_norm": 0.3338480293750763, | |
| "learning_rate": 0.0021874999999999998, | |
| "loss": 2.423, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.48246449153775617, | |
| "eval_loss": 2.5827646255493164, | |
| "eval_runtime": 3.3721, | |
| "eval_samples_per_second": 1331.828, | |
| "eval_steps_per_second": 10.676, | |
| "step": 17568 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "grad_norm": 0.34943464398384094, | |
| "learning_rate": 0.0022500000000000003, | |
| "loss": 2.3545, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "grad_norm": 0.33251243829727173, | |
| "learning_rate": 0.0023125000000000003, | |
| "loss": 2.371, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "grad_norm": 0.36209550499916077, | |
| "learning_rate": 0.002375, | |
| "loss": 2.3805, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "grad_norm": 0.36579540371894836, | |
| "learning_rate": 0.0024375, | |
| "loss": 2.3914, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "grad_norm": 0.3118489980697632, | |
| "learning_rate": 0.0025, | |
| "loss": 2.3921, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.48544858746364117, | |
| "eval_loss": 2.566972017288208, | |
| "eval_runtime": 3.3834, | |
| "eval_samples_per_second": 1327.359, | |
| "eval_steps_per_second": 10.64, | |
| "step": 20496 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.33852943778038025, | |
| "learning_rate": 0.0025625, | |
| "loss": 2.3805, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "grad_norm": 0.3410607874393463, | |
| "learning_rate": 0.002625, | |
| "loss": 2.3178, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "grad_norm": 0.3233848810195923, | |
| "learning_rate": 0.0026875000000000002, | |
| "loss": 2.3326, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "grad_norm": 0.3061616122722626, | |
| "learning_rate": 0.00275, | |
| "loss": 2.3472, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "grad_norm": 0.3195803463459015, | |
| "learning_rate": 0.0028125, | |
| "loss": 2.3588, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 0.37057217955589294, | |
| "learning_rate": 0.0028750000000000004, | |
| "loss": 2.3602, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.4870756385912683, | |
| "eval_loss": 2.553264856338501, | |
| "eval_runtime": 3.4098, | |
| "eval_samples_per_second": 1317.103, | |
| "eval_steps_per_second": 10.558, | |
| "step": 23424 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "grad_norm": 0.31338265538215637, | |
| "learning_rate": 0.0029375, | |
| "loss": 2.3546, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "grad_norm": 0.3508692979812622, | |
| "learning_rate": 0.003, | |
| "loss": 2.2832, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "grad_norm": 0.3429717421531677, | |
| "learning_rate": 0.002715909090909091, | |
| "loss": 2.3064, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "grad_norm": 0.3300691545009613, | |
| "learning_rate": 0.0024318181818181817, | |
| "loss": 2.3012, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "grad_norm": 0.31728777289390564, | |
| "learning_rate": 0.002147727272727273, | |
| "loss": 2.2953, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "grad_norm": 0.29462775588035583, | |
| "learning_rate": 0.0018636363636363638, | |
| "loss": 2.2861, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.49417470110825323, | |
| "eval_loss": 2.500990867614746, | |
| "eval_runtime": 3.4136, | |
| "eval_samples_per_second": 1315.633, | |
| "eval_steps_per_second": 10.546, | |
| "step": 26352 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "grad_norm": 0.29570701718330383, | |
| "learning_rate": 0.0015795454545454546, | |
| "loss": 2.2492, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "grad_norm": 0.3393871486186981, | |
| "learning_rate": 0.0012954545454545456, | |
| "loss": 2.1726, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "grad_norm": 0.32914644479751587, | |
| "learning_rate": 0.0010113636363636364, | |
| "loss": 2.1655, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "grad_norm": 0.2837465703487396, | |
| "learning_rate": 0.0007272727272727273, | |
| "loss": 2.1501, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "grad_norm": 0.3447483479976654, | |
| "learning_rate": 0.0004431818181818182, | |
| "loss": 2.1465, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "grad_norm": 0.30058619379997253, | |
| "learning_rate": 0.0001590909090909091, | |
| "loss": 2.1381, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.49974139004167567, | |
| "eval_loss": 2.4754340648651123, | |
| "eval_runtime": 3.3618, | |
| "eval_samples_per_second": 1335.893, | |
| "eval_steps_per_second": 10.709, | |
| "step": 29280 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 29280, | |
| "total_flos": 2273237316403200.0, | |
| "train_loss": 2.566941033035028, | |
| "train_runtime": 742.3412, | |
| "train_samples_per_second": 631.044, | |
| "train_steps_per_second": 39.443 | |
| } | |
| ], | |
| "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 | |
| } | |