| { | |
| "best_metric": 0.7564328908920288, | |
| "best_model_checkpoint": "../../output/llama3-8b-sft/LangGPT/checkpoint-700", | |
| "epoch": 9.0, | |
| "eval_steps": 100, | |
| "global_step": 720, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 0.33964911103248596, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.1993, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.4269595146179199, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1546, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 0.39937543869018555, | |
| "learning_rate": 4.997482666353287e-05, | |
| "loss": 1.0663, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.3308526277542114, | |
| "learning_rate": 4.989935734988098e-05, | |
| "loss": 1.0087, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 0.2543019950389862, | |
| "learning_rate": 4.977374404419837e-05, | |
| "loss": 0.9476, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.2518838346004486, | |
| "learning_rate": 4.959823971496574e-05, | |
| "loss": 0.9267, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "grad_norm": 0.2807057201862335, | |
| "learning_rate": 4.937319780454559e-05, | |
| "loss": 0.8875, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.25611793994903564, | |
| "learning_rate": 4.909907151739633e-05, | |
| "loss": 0.8747, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.125, | |
| "grad_norm": 0.28418758511543274, | |
| "learning_rate": 4.877641290737884e-05, | |
| "loss": 0.8721, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 0.294350266456604, | |
| "learning_rate": 4.8405871765993433e-05, | |
| "loss": 0.8441, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "eval_loss": 0.8461617231369019, | |
| "eval_runtime": 158.2385, | |
| "eval_samples_per_second": 5.397, | |
| "eval_steps_per_second": 1.352, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.375, | |
| "grad_norm": 0.33300408720970154, | |
| "learning_rate": 4.7988194313786275e-05, | |
| "loss": 0.8252, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 0.29645124077796936, | |
| "learning_rate": 4.752422169756048e-05, | |
| "loss": 0.8232, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.625, | |
| "grad_norm": 0.3273533284664154, | |
| "learning_rate": 4.701488829641845e-05, | |
| "loss": 0.8061, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "grad_norm": 0.3246830105781555, | |
| "learning_rate": 4.6461219840046654e-05, | |
| "loss": 0.8023, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.875, | |
| "grad_norm": 0.3237927556037903, | |
| "learning_rate": 4.586433134303257e-05, | |
| "loss": 0.8131, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.3896329402923584, | |
| "learning_rate": 4.522542485937369e-05, | |
| "loss": 0.8096, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.125, | |
| "grad_norm": 0.4054020345211029, | |
| "learning_rate": 4.454578706170075e-05, | |
| "loss": 0.8003, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "grad_norm": 0.42092710733413696, | |
| "learning_rate": 4.382678665009028e-05, | |
| "loss": 0.7769, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.375, | |
| "grad_norm": 0.35680437088012695, | |
| "learning_rate": 4.306987159568479e-05, | |
| "loss": 0.78, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.379029780626297, | |
| "learning_rate": 4.227656622467162e-05, | |
| "loss": 0.7829, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "eval_loss": 0.7941725254058838, | |
| "eval_runtime": 158.4268, | |
| "eval_samples_per_second": 5.391, | |
| "eval_steps_per_second": 1.351, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.625, | |
| "grad_norm": 0.44896718859672546, | |
| "learning_rate": 4.144846814849282e-05, | |
| "loss": 0.7874, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 0.40545764565467834, | |
| "learning_rate": 4.058724504646834e-05, | |
| "loss": 0.7787, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.875, | |
| "grad_norm": 0.37240272760391235, | |
| "learning_rate": 3.969463130731183e-05, | |
| "loss": 0.7789, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.4221615493297577, | |
| "learning_rate": 3.8772424536302564e-05, | |
| "loss": 0.77, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 3.125, | |
| "grad_norm": 0.38211557269096375, | |
| "learning_rate": 3.782248193514766e-05, | |
| "loss": 0.7775, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "grad_norm": 0.44149255752563477, | |
| "learning_rate": 3.6846716561824965e-05, | |
| "loss": 0.7697, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 3.375, | |
| "grad_norm": 0.41183075308799744, | |
| "learning_rate": 3.5847093477938956e-05, | |
| "loss": 0.7688, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "grad_norm": 0.4264132082462311, | |
| "learning_rate": 3.4825625791348096e-05, | |
| "loss": 0.7503, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 3.625, | |
| "grad_norm": 0.43148595094680786, | |
| "learning_rate": 3.378437060203357e-05, | |
| "loss": 0.774, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 0.42241978645324707, | |
| "learning_rate": 3.272542485937369e-05, | |
| "loss": 0.747, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "eval_loss": 0.7737195491790771, | |
| "eval_runtime": 158.4242, | |
| "eval_samples_per_second": 5.391, | |
| "eval_steps_per_second": 1.351, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.875, | |
| "grad_norm": 0.4389643371105194, | |
| "learning_rate": 3.165092113916688e-05, | |
| "loss": 0.7535, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.4277195930480957, | |
| "learning_rate": 3.056302334890786e-05, | |
| "loss": 0.7539, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 4.125, | |
| "grad_norm": 0.39998236298561096, | |
| "learning_rate": 2.9463922369965917e-05, | |
| "loss": 0.7722, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "grad_norm": 0.42156070470809937, | |
| "learning_rate": 2.8355831645441388e-05, | |
| "loss": 0.7584, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 4.375, | |
| "grad_norm": 0.47488293051719666, | |
| "learning_rate": 2.724098272258584e-05, | |
| "loss": 0.7432, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "grad_norm": 0.4431322515010834, | |
| "learning_rate": 2.6121620758762877e-05, | |
| "loss": 0.7454, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 4.625, | |
| "grad_norm": 0.5012672543525696, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.7285, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "grad_norm": 0.4605712592601776, | |
| "learning_rate": 2.3878379241237136e-05, | |
| "loss": 0.7352, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 4.875, | |
| "grad_norm": 0.4491554796695709, | |
| "learning_rate": 2.2759017277414166e-05, | |
| "loss": 0.7569, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.4903900921344757, | |
| "learning_rate": 2.164416835455862e-05, | |
| "loss": 0.7507, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.763729989528656, | |
| "eval_runtime": 158.4134, | |
| "eval_samples_per_second": 5.391, | |
| "eval_steps_per_second": 1.351, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 5.125, | |
| "grad_norm": 0.4337986707687378, | |
| "learning_rate": 2.0536077630034086e-05, | |
| "loss": 0.7354, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "grad_norm": 0.4212823212146759, | |
| "learning_rate": 1.9436976651092144e-05, | |
| "loss": 0.7396, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 5.375, | |
| "grad_norm": 0.482803612947464, | |
| "learning_rate": 1.8349078860833123e-05, | |
| "loss": 0.7592, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "grad_norm": 0.4545280933380127, | |
| "learning_rate": 1.7274575140626318e-05, | |
| "loss": 0.744, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 5.625, | |
| "grad_norm": 0.46052929759025574, | |
| "learning_rate": 1.621562939796643e-05, | |
| "loss": 0.7484, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "grad_norm": 0.4419727027416229, | |
| "learning_rate": 1.5174374208651912e-05, | |
| "loss": 0.7268, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 5.875, | |
| "grad_norm": 0.48050615191459656, | |
| "learning_rate": 1.4152906522061048e-05, | |
| "loss": 0.7317, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 0.46696364879608154, | |
| "learning_rate": 1.3153283438175034e-05, | |
| "loss": 0.7432, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 6.125, | |
| "grad_norm": 0.5167751908302307, | |
| "learning_rate": 1.217751806485235e-05, | |
| "loss": 0.7472, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "grad_norm": 0.5040733814239502, | |
| "learning_rate": 1.122757546369744e-05, | |
| "loss": 0.7327, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "eval_loss": 0.7590272426605225, | |
| "eval_runtime": 158.301, | |
| "eval_samples_per_second": 5.395, | |
| "eval_steps_per_second": 1.352, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 6.375, | |
| "grad_norm": 0.45372575521469116, | |
| "learning_rate": 1.0305368692688174e-05, | |
| "loss": 0.7301, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "grad_norm": 0.46018218994140625, | |
| "learning_rate": 9.412754953531663e-06, | |
| "loss": 0.7409, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 6.625, | |
| "grad_norm": 0.4666420817375183, | |
| "learning_rate": 8.551531851507186e-06, | |
| "loss": 0.7433, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "grad_norm": 0.49834486842155457, | |
| "learning_rate": 7.723433775328384e-06, | |
| "loss": 0.7326, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 6.875, | |
| "grad_norm": 0.4643003046512604, | |
| "learning_rate": 6.930128404315214e-06, | |
| "loss": 0.731, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.47693514823913574, | |
| "learning_rate": 6.173213349909729e-06, | |
| "loss": 0.7306, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 7.125, | |
| "grad_norm": 0.49089315533638, | |
| "learning_rate": 5.454212938299255e-06, | |
| "loss": 0.7367, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "grad_norm": 0.4702819585800171, | |
| "learning_rate": 4.7745751406263165e-06, | |
| "loss": 0.7283, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 7.375, | |
| "grad_norm": 0.489766001701355, | |
| "learning_rate": 4.135668656967434e-06, | |
| "loss": 0.7387, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "grad_norm": 0.45880013704299927, | |
| "learning_rate": 3.5387801599533475e-06, | |
| "loss": 0.7331, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "eval_loss": 0.756921648979187, | |
| "eval_runtime": 158.3438, | |
| "eval_samples_per_second": 5.393, | |
| "eval_steps_per_second": 1.351, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 7.625, | |
| "grad_norm": 0.47875136137008667, | |
| "learning_rate": 2.98511170358155e-06, | |
| "loss": 0.7366, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "grad_norm": 0.44256335496902466, | |
| "learning_rate": 2.475778302439524e-06, | |
| "loss": 0.7419, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 7.875, | |
| "grad_norm": 0.4561857581138611, | |
| "learning_rate": 2.0118056862137357e-06, | |
| "loss": 0.7216, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 0.44974443316459656, | |
| "learning_rate": 1.59412823400657e-06, | |
| "loss": 0.7312, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 8.125, | |
| "grad_norm": 0.44745445251464844, | |
| "learning_rate": 1.2235870926211619e-06, | |
| "loss": 0.7365, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "grad_norm": 0.4579036235809326, | |
| "learning_rate": 9.009284826036691e-07, | |
| "loss": 0.7199, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 8.375, | |
| "grad_norm": 0.4414173662662506, | |
| "learning_rate": 6.268021954544096e-07, | |
| "loss": 0.734, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "grad_norm": 0.4644961953163147, | |
| "learning_rate": 4.0176028503425835e-07, | |
| "loss": 0.725, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 8.625, | |
| "grad_norm": 0.4445784389972687, | |
| "learning_rate": 2.262559558016325e-07, | |
| "loss": 0.7288, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "grad_norm": 0.4681429862976074, | |
| "learning_rate": 1.006426501190233e-07, | |
| "loss": 0.7396, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "eval_loss": 0.7564328908920288, | |
| "eval_runtime": 158.3416, | |
| "eval_samples_per_second": 5.393, | |
| "eval_steps_per_second": 1.352, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 8.875, | |
| "grad_norm": 0.4473038911819458, | |
| "learning_rate": 2.5173336467135267e-08, | |
| "loss": 0.7469, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "grad_norm": 0.47429242730140686, | |
| "learning_rate": 0.0, | |
| "loss": 0.7312, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "step": 720, | |
| "total_flos": 3.187342666522165e+18, | |
| "train_loss": 0.7853862928019629, | |
| "train_runtime": 44891.6571, | |
| "train_samples_per_second": 1.539, | |
| "train_steps_per_second": 0.016 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 720, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9, | |
| "save_steps": 100, | |
| "total_flos": 3.187342666522165e+18, | |
| "train_batch_size": 12, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |