| { | |
| "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": 1.187337040901184, | |
| "learning_rate": 6.25e-05, | |
| "loss": 6.2023, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 1.0917108058929443, | |
| "learning_rate": 0.000125, | |
| "loss": 3.6263, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.0571826696395874, | |
| "learning_rate": 0.0001875, | |
| "loss": 3.3348, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.9771557450294495, | |
| "learning_rate": 0.00025, | |
| "loss": 3.1685, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.9500265717506409, | |
| "learning_rate": 0.0003125, | |
| "loss": 3.0457, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4381255950220651, | |
| "eval_loss": 3.016178607940674, | |
| "eval_runtime": 3.353, | |
| "eval_samples_per_second": 1339.396, | |
| "eval_steps_per_second": 10.737, | |
| "step": 2928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.9907193779945374, | |
| "learning_rate": 0.000375, | |
| "loss": 2.9546, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.9292730093002319, | |
| "learning_rate": 0.00043750000000000006, | |
| "loss": 2.8611, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.8982778787612915, | |
| "learning_rate": 0.0005, | |
| "loss": 2.8115, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.899147093296051, | |
| "learning_rate": 0.0005625000000000001, | |
| "loss": 2.7697, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 0.889224648475647, | |
| "learning_rate": 0.000625, | |
| "loss": 2.7455, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.8336099982261658, | |
| "learning_rate": 0.0006875, | |
| "loss": 2.7215, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.4596735027360057, | |
| "eval_loss": 2.7846643924713135, | |
| "eval_runtime": 3.384, | |
| "eval_samples_per_second": 1327.111, | |
| "eval_steps_per_second": 10.638, | |
| "step": 5856 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.8637300729751587, | |
| "learning_rate": 0.00075, | |
| "loss": 2.6694, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.7789953351020813, | |
| "learning_rate": 0.0008125, | |
| "loss": 2.6306, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 0.7371299862861633, | |
| "learning_rate": 0.0008750000000000001, | |
| "loss": 2.6088, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.6360913515090942, | |
| "learning_rate": 0.0009375, | |
| "loss": 2.6008, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 0.6919972896575928, | |
| "learning_rate": 0.001, | |
| "loss": 2.5956, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.653573751449585, | |
| "learning_rate": 0.0010625, | |
| "loss": 2.5825, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.47033349288252796, | |
| "eval_loss": 2.6924686431884766, | |
| "eval_runtime": 3.3904, | |
| "eval_samples_per_second": 1324.616, | |
| "eval_steps_per_second": 10.618, | |
| "step": 8784 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 0.621619701385498, | |
| "learning_rate": 0.0011250000000000001, | |
| "loss": 2.5387, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 0.6471651792526245, | |
| "learning_rate": 0.0011875, | |
| "loss": 2.5135, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "grad_norm": 0.5801098942756653, | |
| "learning_rate": 0.00125, | |
| "loss": 2.5157, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "grad_norm": 0.5154350399971008, | |
| "learning_rate": 0.0013125, | |
| "loss": 2.5178, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.4794241189956665, | |
| "learning_rate": 0.001375, | |
| "loss": 2.5135, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "grad_norm": 0.4597519040107727, | |
| "learning_rate": 0.0014375000000000002, | |
| "loss": 2.5059, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.47582479043826936, | |
| "eval_loss": 2.6392264366149902, | |
| "eval_runtime": 3.4286, | |
| "eval_samples_per_second": 1309.863, | |
| "eval_steps_per_second": 10.5, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 0.4645042419433594, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4568, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 0.46030470728874207, | |
| "learning_rate": 0.0015625, | |
| "loss": 2.4486, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "grad_norm": 0.44470083713531494, | |
| "learning_rate": 0.001625, | |
| "loss": 2.4657, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.4096473157405853, | |
| "learning_rate": 0.0016875, | |
| "loss": 2.4596, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 0.3900606334209442, | |
| "learning_rate": 0.0017500000000000003, | |
| "loss": 2.4585, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 0.40039312839508057, | |
| "learning_rate": 0.0018124999999999999, | |
| "loss": 2.4667, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.480080020057613, | |
| "eval_loss": 2.6032490730285645, | |
| "eval_runtime": 3.3841, | |
| "eval_samples_per_second": 1327.1, | |
| "eval_steps_per_second": 10.638, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "grad_norm": 0.3943752348423004, | |
| "learning_rate": 0.001875, | |
| "loss": 2.4054, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "grad_norm": 0.4194723069667816, | |
| "learning_rate": 0.0019375000000000002, | |
| "loss": 2.4006, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "grad_norm": 0.3874049484729767, | |
| "learning_rate": 0.002, | |
| "loss": 2.4215, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 0.4187428057193756, | |
| "learning_rate": 0.0020625, | |
| "loss": 2.409, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "grad_norm": 0.3771882951259613, | |
| "learning_rate": 0.002125, | |
| "loss": 2.4208, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "grad_norm": 0.34806564450263977, | |
| "learning_rate": 0.0021874999999999998, | |
| "loss": 2.4261, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.4827099518371827, | |
| "eval_loss": 2.585068941116333, | |
| "eval_runtime": 3.4455, | |
| "eval_samples_per_second": 1303.458, | |
| "eval_steps_per_second": 10.449, | |
| "step": 17568 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "grad_norm": 0.38157713413238525, | |
| "learning_rate": 0.0022500000000000003, | |
| "loss": 2.3536, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "grad_norm": 0.35192397236824036, | |
| "learning_rate": 0.0023125000000000003, | |
| "loss": 2.3635, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "grad_norm": 0.36528274416923523, | |
| "learning_rate": 0.002375, | |
| "loss": 2.3799, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "grad_norm": 0.3123536705970764, | |
| "learning_rate": 0.0024375, | |
| "loss": 2.3808, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "grad_norm": 0.32904407382011414, | |
| "learning_rate": 0.0025, | |
| "loss": 2.3888, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.484838443290781, | |
| "eval_loss": 2.566373348236084, | |
| "eval_runtime": 3.3915, | |
| "eval_samples_per_second": 1324.208, | |
| "eval_steps_per_second": 10.615, | |
| "step": 20496 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.316089928150177, | |
| "learning_rate": 0.0025625, | |
| "loss": 2.397, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "grad_norm": 0.36722826957702637, | |
| "learning_rate": 0.002625, | |
| "loss": 2.3088, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "grad_norm": 0.3584025800228119, | |
| "learning_rate": 0.0026875000000000002, | |
| "loss": 2.3392, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "grad_norm": 0.39147356152534485, | |
| "learning_rate": 0.00275, | |
| "loss": 2.3481, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "grad_norm": 0.30665042996406555, | |
| "learning_rate": 0.0028125, | |
| "loss": 2.3524, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 0.34735408425331116, | |
| "learning_rate": 0.0028750000000000004, | |
| "loss": 2.3623, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.4860008731373508, | |
| "eval_loss": 2.5573480129241943, | |
| "eval_runtime": 3.38, | |
| "eval_samples_per_second": 1328.716, | |
| "eval_steps_per_second": 10.651, | |
| "step": 23424 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "grad_norm": 0.3066130578517914, | |
| "learning_rate": 0.0029375, | |
| "loss": 2.359, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "grad_norm": 0.4078517556190491, | |
| "learning_rate": 0.003, | |
| "loss": 2.285, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "grad_norm": 0.2971705198287964, | |
| "learning_rate": 0.002715909090909091, | |
| "loss": 2.3045, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "grad_norm": 0.30169880390167236, | |
| "learning_rate": 0.0024318181818181817, | |
| "loss": 2.3081, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "grad_norm": 0.30579808354377747, | |
| "learning_rate": 0.002147727272727273, | |
| "loss": 2.2906, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "grad_norm": 0.3257807791233063, | |
| "learning_rate": 0.0018636363636363638, | |
| "loss": 2.2903, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.4940852834277479, | |
| "eval_loss": 2.502375841140747, | |
| "eval_runtime": 3.3782, | |
| "eval_samples_per_second": 1329.388, | |
| "eval_steps_per_second": 10.656, | |
| "step": 26352 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "grad_norm": 0.27855226397514343, | |
| "learning_rate": 0.0015795454545454546, | |
| "loss": 2.2424, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "grad_norm": 0.2956109941005707, | |
| "learning_rate": 0.0012954545454545456, | |
| "loss": 2.1635, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "grad_norm": 0.3226345479488373, | |
| "learning_rate": 0.0010113636363636364, | |
| "loss": 2.1661, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "grad_norm": 0.3127693831920624, | |
| "learning_rate": 0.0007272727272727273, | |
| "loss": 2.1605, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "grad_norm": 0.3304063379764557, | |
| "learning_rate": 0.0004431818181818182, | |
| "loss": 2.1557, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "grad_norm": 0.3265271782875061, | |
| "learning_rate": 0.0001590909090909091, | |
| "loss": 2.135, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.4996186598919624, | |
| "eval_loss": 2.4745402336120605, | |
| "eval_runtime": 3.376, | |
| "eval_samples_per_second": 1330.26, | |
| "eval_steps_per_second": 10.663, | |
| "step": 29280 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 29280, | |
| "total_flos": 2273237316403200.0, | |
| "train_loss": 2.566726820065024, | |
| "train_runtime": 751.1564, | |
| "train_samples_per_second": 623.638, | |
| "train_steps_per_second": 38.98 | |
| } | |
| ], | |
| "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 | |
| } | |