| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 1730, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.136181831359863, | |
| "learning_rate": 2.8901734104046245e-05, | |
| "loss": 3.3442, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 2.936872959136963, | |
| "learning_rate": 5.780346820809249e-05, | |
| "loss": 1.9328, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 2.958111047744751, | |
| "learning_rate": 8.670520231213874e-05, | |
| "loss": 1.0726, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 2.6158933639526367, | |
| "learning_rate": 0.00011560693641618498, | |
| "loss": 0.843, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 2.173802375793457, | |
| "learning_rate": 0.00014450867052023122, | |
| "loss": 0.7975, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 2.179237127304077, | |
| "learning_rate": 0.00017341040462427748, | |
| "loss": 0.7632, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 2.171823024749756, | |
| "learning_rate": 0.00019974309569685292, | |
| "loss": 0.7023, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 1.7833949327468872, | |
| "learning_rate": 0.00019653179190751445, | |
| "loss": 0.7452, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 1.8641424179077148, | |
| "learning_rate": 0.00019332048811817598, | |
| "loss": 0.6986, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 2.020118474960327, | |
| "learning_rate": 0.0001901091843288375, | |
| "loss": 0.6461, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 1.5515590906143188, | |
| "learning_rate": 0.00018689788053949903, | |
| "loss": 0.6287, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 1.8482364416122437, | |
| "learning_rate": 0.00018368657675016056, | |
| "loss": 0.5941, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 1.901448130607605, | |
| "learning_rate": 0.00018047527296082209, | |
| "loss": 0.5869, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 1.3397067785263062, | |
| "learning_rate": 0.0001772639691714836, | |
| "loss": 0.576, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 1.5321860313415527, | |
| "learning_rate": 0.00017405266538214514, | |
| "loss": 0.5475, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 1.7586708068847656, | |
| "learning_rate": 0.0001708413615928067, | |
| "loss": 0.5372, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 1.6688508987426758, | |
| "learning_rate": 0.00016763005780346822, | |
| "loss": 0.5429, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "grad_norm": 1.8364487886428833, | |
| "learning_rate": 0.00016441875401412975, | |
| "loss": 0.539, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 1.682422399520874, | |
| "learning_rate": 0.00016120745022479128, | |
| "loss": 0.4863, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 1.8813568353652954, | |
| "learning_rate": 0.0001579961464354528, | |
| "loss": 0.5587, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 1.593272089958191, | |
| "learning_rate": 0.00015478484264611433, | |
| "loss": 0.5121, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "grad_norm": 1.3057204484939575, | |
| "learning_rate": 0.00015157353885677586, | |
| "loss": 0.4958, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "grad_norm": 1.764782190322876, | |
| "learning_rate": 0.00014836223506743738, | |
| "loss": 0.4538, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "grad_norm": 1.9039567708969116, | |
| "learning_rate": 0.0001451509312780989, | |
| "loss": 0.4897, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "grad_norm": 1.793358325958252, | |
| "learning_rate": 0.00014193962748876044, | |
| "loss": 0.4601, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 1.4797160625457764, | |
| "learning_rate": 0.00013872832369942197, | |
| "loss": 0.5012, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "grad_norm": 1.8706116676330566, | |
| "learning_rate": 0.0001355170199100835, | |
| "loss": 0.463, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 1.2917760610580444, | |
| "learning_rate": 0.00013230571612074502, | |
| "loss": 0.4608, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 1.67048978805542, | |
| "learning_rate": 0.00012909441233140655, | |
| "loss": 0.4104, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "grad_norm": 1.5852957963943481, | |
| "learning_rate": 0.00012588310854206808, | |
| "loss": 0.3789, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 1.431355595588684, | |
| "learning_rate": 0.0001226718047527296, | |
| "loss": 0.3824, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "grad_norm": 1.3853150606155396, | |
| "learning_rate": 0.00011946050096339114, | |
| "loss": 0.4049, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "grad_norm": 1.5978338718414307, | |
| "learning_rate": 0.00011624919717405267, | |
| "loss": 0.3867, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "grad_norm": 1.5443283319473267, | |
| "learning_rate": 0.0001130378933847142, | |
| "loss": 0.3646, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 1.928194284439087, | |
| "learning_rate": 0.00010982658959537572, | |
| "loss": 0.3898, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 1.7459748983383179, | |
| "learning_rate": 0.00010661528580603725, | |
| "loss": 0.3728, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "grad_norm": 1.5855330228805542, | |
| "learning_rate": 0.00010340398201669879, | |
| "loss": 0.3829, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 1.3970378637313843, | |
| "learning_rate": 0.00010019267822736032, | |
| "loss": 0.361, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 1.7334144115447998, | |
| "learning_rate": 9.698137443802185e-05, | |
| "loss": 0.3767, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "grad_norm": 1.5465258359909058, | |
| "learning_rate": 9.377007064868337e-05, | |
| "loss": 0.3614, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 1.5643867254257202, | |
| "learning_rate": 9.05587668593449e-05, | |
| "loss": 0.3956, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "grad_norm": 1.9123274087905884, | |
| "learning_rate": 8.734746307000643e-05, | |
| "loss": 0.3449, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "grad_norm": 2.7245140075683594, | |
| "learning_rate": 8.413615928066796e-05, | |
| "loss": 0.2816, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "grad_norm": 1.991822600364685, | |
| "learning_rate": 8.092485549132948e-05, | |
| "loss": 0.2756, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "grad_norm": 2.225388765335083, | |
| "learning_rate": 7.771355170199101e-05, | |
| "loss": 0.2869, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "grad_norm": 2.127760648727417, | |
| "learning_rate": 7.450224791265255e-05, | |
| "loss": 0.2746, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "grad_norm": 2.051787853240967, | |
| "learning_rate": 7.129094412331408e-05, | |
| "loss": 0.2723, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "grad_norm": 2.1885359287261963, | |
| "learning_rate": 6.80796403339756e-05, | |
| "loss": 0.2849, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "grad_norm": 1.6504746675491333, | |
| "learning_rate": 6.486833654463712e-05, | |
| "loss": 0.2578, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "grad_norm": 1.753282904624939, | |
| "learning_rate": 6.165703275529865e-05, | |
| "loss": 0.3017, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "grad_norm": 2.5774452686309814, | |
| "learning_rate": 5.844572896596018e-05, | |
| "loss": 0.275, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 1.954413652420044, | |
| "learning_rate": 5.523442517662171e-05, | |
| "loss": 0.2934, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "grad_norm": 1.5767195224761963, | |
| "learning_rate": 5.2023121387283234e-05, | |
| "loss": 0.2885, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "grad_norm": 1.8420324325561523, | |
| "learning_rate": 4.881181759794477e-05, | |
| "loss": 0.2735, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "grad_norm": 2.003744602203369, | |
| "learning_rate": 4.56005138086063e-05, | |
| "loss": 0.2692, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "grad_norm": 1.7634247541427612, | |
| "learning_rate": 4.238921001926782e-05, | |
| "loss": 0.253, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "grad_norm": 2.338707447052002, | |
| "learning_rate": 3.917790622992935e-05, | |
| "loss": 0.1929, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "grad_norm": 2.5318188667297363, | |
| "learning_rate": 3.596660244059088e-05, | |
| "loss": 0.2114, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "grad_norm": 1.8917136192321777, | |
| "learning_rate": 3.275529865125241e-05, | |
| "loss": 0.2274, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "grad_norm": 1.8130145072937012, | |
| "learning_rate": 2.9543994861913938e-05, | |
| "loss": 0.2101, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "grad_norm": 1.923666000366211, | |
| "learning_rate": 2.6332691072575465e-05, | |
| "loss": 0.1908, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "grad_norm": 1.6899950504302979, | |
| "learning_rate": 2.3121387283236996e-05, | |
| "loss": 0.198, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "grad_norm": 2.0312962532043457, | |
| "learning_rate": 1.9910083493898523e-05, | |
| "loss": 0.2109, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "grad_norm": 2.349602460861206, | |
| "learning_rate": 1.6698779704560053e-05, | |
| "loss": 0.2136, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "grad_norm": 1.9838098287582397, | |
| "learning_rate": 1.348747591522158e-05, | |
| "loss": 0.2251, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "grad_norm": 3.0191967487335205, | |
| "learning_rate": 1.027617212588311e-05, | |
| "loss": 0.2107, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "grad_norm": 1.7804679870605469, | |
| "learning_rate": 7.064868336544637e-06, | |
| "loss": 0.1865, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "grad_norm": 2.1056630611419678, | |
| "learning_rate": 3.853564547206165e-06, | |
| "loss": 0.2099, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "grad_norm": 1.8963791131973267, | |
| "learning_rate": 6.422607578676943e-07, | |
| "loss": 0.201, | |
| "step": 1725 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 1730, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 7.5248089823232e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |