| { | |
| "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": 1.091705322265625, | |
| "learning_rate": 6.25e-05, | |
| "loss": 6.1952, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.9577378034591675, | |
| "learning_rate": 0.000125, | |
| "loss": 3.6423, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.2023097276687622, | |
| "learning_rate": 0.0001875, | |
| "loss": 3.3344, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.0954350233078003, | |
| "learning_rate": 0.00025, | |
| "loss": 3.1657, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.900829553604126, | |
| "learning_rate": 0.0003125, | |
| "loss": 3.0551, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4362285375650689, | |
| "eval_loss": 3.0203442573547363, | |
| "eval_runtime": 3.2937, | |
| "eval_samples_per_second": 1363.516, | |
| "eval_steps_per_second": 10.93, | |
| "step": 2928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.9360588192939758, | |
| "learning_rate": 0.000375, | |
| "loss": 2.9474, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.8689121007919312, | |
| "learning_rate": 0.00043750000000000006, | |
| "loss": 2.8718, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.8661333918571472, | |
| "learning_rate": 0.0005, | |
| "loss": 2.815, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.8203218579292297, | |
| "learning_rate": 0.0005625000000000001, | |
| "loss": 2.7786, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 0.8360357880592346, | |
| "learning_rate": 0.000625, | |
| "loss": 2.743, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.7822497487068176, | |
| "learning_rate": 0.0006875, | |
| "loss": 2.7076, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.45947012134505233, | |
| "eval_loss": 2.7878334522247314, | |
| "eval_runtime": 3.3426, | |
| "eval_samples_per_second": 1343.555, | |
| "eval_steps_per_second": 10.77, | |
| "step": 5856 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.7363527417182922, | |
| "learning_rate": 0.00075, | |
| "loss": 2.681, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.7337239384651184, | |
| "learning_rate": 0.0008125, | |
| "loss": 2.6194, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 0.7000368237495422, | |
| "learning_rate": 0.0008750000000000001, | |
| "loss": 2.6334, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.7184774279594421, | |
| "learning_rate": 0.0009375, | |
| "loss": 2.604, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 0.6264050602912903, | |
| "learning_rate": 0.001, | |
| "loss": 2.6012, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.6527601480484009, | |
| "learning_rate": 0.0010625, | |
| "loss": 2.5796, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.47054914728845265, | |
| "eval_loss": 2.6938107013702393, | |
| "eval_runtime": 3.3469, | |
| "eval_samples_per_second": 1341.821, | |
| "eval_steps_per_second": 10.756, | |
| "step": 8784 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 0.651335597038269, | |
| "learning_rate": 0.0011250000000000001, | |
| "loss": 2.5461, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 0.6495278477668762, | |
| "learning_rate": 0.0011875, | |
| "loss": 2.5131, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "grad_norm": 0.5576895475387573, | |
| "learning_rate": 0.00125, | |
| "loss": 2.5184, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "grad_norm": 0.5400159955024719, | |
| "learning_rate": 0.0013125, | |
| "loss": 2.5233, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.516352117061615, | |
| "learning_rate": 0.001375, | |
| "loss": 2.5253, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "grad_norm": 0.48890888690948486, | |
| "learning_rate": 0.0014375000000000002, | |
| "loss": 2.4976, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.47683643752947713, | |
| "eval_loss": 2.6370015144348145, | |
| "eval_runtime": 3.3506, | |
| "eval_samples_per_second": 1340.337, | |
| "eval_steps_per_second": 10.744, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 0.49809128046035767, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4639, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 0.453357994556427, | |
| "learning_rate": 0.0015625, | |
| "loss": 2.4492, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "grad_norm": 0.454194039106369, | |
| "learning_rate": 0.001625, | |
| "loss": 2.4562, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.4470754563808441, | |
| "learning_rate": 0.0016875, | |
| "loss": 2.455, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 0.3900794982910156, | |
| "learning_rate": 0.0017500000000000003, | |
| "loss": 2.464, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 0.3795139789581299, | |
| "learning_rate": 0.0018124999999999999, | |
| "loss": 2.4691, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.47998008264998937, | |
| "eval_loss": 2.614906072616577, | |
| "eval_runtime": 3.405, | |
| "eval_samples_per_second": 1318.952, | |
| "eval_steps_per_second": 10.573, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "grad_norm": 0.40433719754219055, | |
| "learning_rate": 0.001875, | |
| "loss": 2.4061, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "grad_norm": 0.3789573907852173, | |
| "learning_rate": 0.0019375000000000002, | |
| "loss": 2.4053, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "grad_norm": 0.4039715826511383, | |
| "learning_rate": 0.002, | |
| "loss": 2.4111, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 0.39668989181518555, | |
| "learning_rate": 0.0020625, | |
| "loss": 2.4235, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "grad_norm": 0.35224515199661255, | |
| "learning_rate": 0.002125, | |
| "loss": 2.4232, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "grad_norm": 0.360273152589798, | |
| "learning_rate": 0.0021874999999999998, | |
| "loss": 2.421, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.4827081985493296, | |
| "eval_loss": 2.5835256576538086, | |
| "eval_runtime": 3.4064, | |
| "eval_samples_per_second": 1318.419, | |
| "eval_steps_per_second": 10.568, | |
| "step": 17568 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "grad_norm": 0.34586748480796814, | |
| "learning_rate": 0.0022500000000000003, | |
| "loss": 2.3503, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "grad_norm": 0.4033253490924835, | |
| "learning_rate": 0.0023125000000000003, | |
| "loss": 2.3652, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "grad_norm": 0.3413168489933014, | |
| "learning_rate": 0.002375, | |
| "loss": 2.3791, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "grad_norm": 0.33606722950935364, | |
| "learning_rate": 0.0024375, | |
| "loss": 2.3848, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "grad_norm": 0.31669604778289795, | |
| "learning_rate": 0.0025, | |
| "loss": 2.3918, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.4860113928644691, | |
| "eval_loss": 2.5660340785980225, | |
| "eval_runtime": 3.4098, | |
| "eval_samples_per_second": 1317.099, | |
| "eval_steps_per_second": 10.558, | |
| "step": 20496 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.29715919494628906, | |
| "learning_rate": 0.0025625, | |
| "loss": 2.3943, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "grad_norm": 0.44059836864471436, | |
| "learning_rate": 0.002625, | |
| "loss": 2.3054, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "grad_norm": 0.33299922943115234, | |
| "learning_rate": 0.0026875000000000002, | |
| "loss": 2.3385, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "grad_norm": 0.3407984673976898, | |
| "learning_rate": 0.00275, | |
| "loss": 2.3518, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "grad_norm": 0.32114025950431824, | |
| "learning_rate": 0.0028125, | |
| "loss": 2.3566, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 0.3123725652694702, | |
| "learning_rate": 0.0028750000000000004, | |
| "loss": 2.3654, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.4854293012972577, | |
| "eval_loss": 2.5585596561431885, | |
| "eval_runtime": 3.3383, | |
| "eval_samples_per_second": 1345.276, | |
| "eval_steps_per_second": 10.784, | |
| "step": 23424 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "grad_norm": 0.3411466181278229, | |
| "learning_rate": 0.0029375, | |
| "loss": 2.3512, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "grad_norm": 0.35489559173583984, | |
| "learning_rate": 0.003, | |
| "loss": 2.2925, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "grad_norm": 0.32344290614128113, | |
| "learning_rate": 0.002715909090909091, | |
| "loss": 2.3065, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "grad_norm": 0.3349243104457855, | |
| "learning_rate": 0.0024318181818181817, | |
| "loss": 2.3006, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "grad_norm": 0.32749468088150024, | |
| "learning_rate": 0.002147727272727273, | |
| "loss": 2.2903, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "grad_norm": 0.29682689905166626, | |
| "learning_rate": 0.0018636363636363638, | |
| "loss": 2.2908, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.4947567926754647, | |
| "eval_loss": 2.505155324935913, | |
| "eval_runtime": 3.3638, | |
| "eval_samples_per_second": 1335.116, | |
| "eval_steps_per_second": 10.702, | |
| "step": 26352 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "grad_norm": 0.35732176899909973, | |
| "learning_rate": 0.0015795454545454546, | |
| "loss": 2.2388, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "grad_norm": 0.3471290171146393, | |
| "learning_rate": 0.0012954545454545456, | |
| "loss": 2.1577, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "grad_norm": 0.31934648752212524, | |
| "learning_rate": 0.0010113636363636364, | |
| "loss": 2.1666, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "grad_norm": 0.3358185887336731, | |
| "learning_rate": 0.0007272727272727273, | |
| "loss": 2.1574, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "grad_norm": 0.3271017074584961, | |
| "learning_rate": 0.0004431818181818182, | |
| "loss": 2.1576, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "grad_norm": 0.333063006401062, | |
| "learning_rate": 0.0001590909090909091, | |
| "loss": 2.1465, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.4999026925241559, | |
| "eval_loss": 2.4751529693603516, | |
| "eval_runtime": 3.4017, | |
| "eval_samples_per_second": 1320.234, | |
| "eval_steps_per_second": 10.583, | |
| "step": 29280 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 29280, | |
| "total_flos": 2273237316403200.0, | |
| "train_loss": 2.5679440357646004, | |
| "train_runtime": 768.0298, | |
| "train_samples_per_second": 609.937, | |
| "train_steps_per_second": 38.124 | |
| } | |
| ], | |
| "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 | |
| } | |