| { | |
| "best_metric": 1.4270155429840088, | |
| "best_model_checkpoint": "autotrain-7t7rk-gfqs1/checkpoint-7366", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 7366, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 17.693096160888672, | |
| "learning_rate": 5.088195386702849e-06, | |
| "loss": 3.139, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 22.15139389038086, | |
| "learning_rate": 1.0515603799185889e-05, | |
| "loss": 2.5999, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 14.272784233093262, | |
| "learning_rate": 1.5943012211668928e-05, | |
| "loss": 2.3391, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 17.81722640991211, | |
| "learning_rate": 2.1438263229308007e-05, | |
| "loss": 2.1259, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 9.983675956726074, | |
| "learning_rate": 2.6933514246947083e-05, | |
| "loss": 1.8274, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 10.129060745239258, | |
| "learning_rate": 3.2428765264586165e-05, | |
| "loss": 1.963, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 7.1694655418396, | |
| "learning_rate": 3.792401628222524e-05, | |
| "loss": 1.8109, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 11.632865905761719, | |
| "learning_rate": 4.3419267299864316e-05, | |
| "loss": 1.8193, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 42.26719665527344, | |
| "learning_rate": 4.891451831750339e-05, | |
| "loss": 1.7714, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 15.210801124572754, | |
| "learning_rate": 4.950972997435511e-05, | |
| "loss": 1.7169, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 25.737319946289062, | |
| "learning_rate": 4.889877809624378e-05, | |
| "loss": 1.7059, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 7.352960109710693, | |
| "learning_rate": 4.828782621813245e-05, | |
| "loss": 1.7726, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 20.277687072753906, | |
| "learning_rate": 4.767687434002112e-05, | |
| "loss": 1.6683, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 11.7589750289917, | |
| "learning_rate": 4.706592246190979e-05, | |
| "loss": 1.73, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 12.03089714050293, | |
| "learning_rate": 4.6454970583798465e-05, | |
| "loss": 1.6849, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 7.4564056396484375, | |
| "learning_rate": 4.5844018705687136e-05, | |
| "loss": 1.6458, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 16.240528106689453, | |
| "learning_rate": 4.523306682757581e-05, | |
| "loss": 1.6597, | |
| "step": 1377 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 8.320143699645996, | |
| "learning_rate": 4.4629657565243625e-05, | |
| "loss": 1.6376, | |
| "step": 1458 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 7.235590934753418, | |
| "learning_rate": 4.40187056871323e-05, | |
| "loss": 1.6217, | |
| "step": 1539 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 5.213665008544922, | |
| "learning_rate": 4.340775380902097e-05, | |
| "loss": 1.6757, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 7.416939735412598, | |
| "learning_rate": 4.279680193090964e-05, | |
| "loss": 1.6158, | |
| "step": 1701 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 6.5677103996276855, | |
| "learning_rate": 4.218585005279832e-05, | |
| "loss": 1.589, | |
| "step": 1782 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 11.56302261352539, | |
| "learning_rate": 4.157489817468698e-05, | |
| "loss": 1.6241, | |
| "step": 1863 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 10.844891548156738, | |
| "learning_rate": 4.096394629657566e-05, | |
| "loss": 1.6057, | |
| "step": 1944 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 7.110713958740234, | |
| "learning_rate": 4.0352994418464324e-05, | |
| "loss": 1.6382, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 10.134081840515137, | |
| "learning_rate": 3.9742042540352995e-05, | |
| "loss": 1.6126, | |
| "step": 2106 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 40.32931137084961, | |
| "learning_rate": 3.9131090662241666e-05, | |
| "loss": 1.5994, | |
| "step": 2187 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 15.259111404418945, | |
| "learning_rate": 3.852013878413034e-05, | |
| "loss": 1.6828, | |
| "step": 2268 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 8.93104362487793, | |
| "learning_rate": 3.7909186906019015e-05, | |
| "loss": 1.6281, | |
| "step": 2349 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 7.933257102966309, | |
| "learning_rate": 3.729823502790768e-05, | |
| "loss": 1.586, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 7.876591205596924, | |
| "learning_rate": 3.668728314979635e-05, | |
| "loss": 1.7194, | |
| "step": 2511 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 10.981273651123047, | |
| "learning_rate": 3.607633127168502e-05, | |
| "loss": 1.7405, | |
| "step": 2592 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 5.999240398406982, | |
| "learning_rate": 3.546537939357369e-05, | |
| "loss": 1.6577, | |
| "step": 2673 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 6.238893985748291, | |
| "learning_rate": 3.4854427515462364e-05, | |
| "loss": 1.6111, | |
| "step": 2754 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 8.094114303588867, | |
| "learning_rate": 3.4243475637351035e-05, | |
| "loss": 1.5915, | |
| "step": 2835 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 6.022202014923096, | |
| "learning_rate": 3.363252375923971e-05, | |
| "loss": 1.6025, | |
| "step": 2916 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 12.310922622680664, | |
| "learning_rate": 3.302157188112838e-05, | |
| "loss": 1.5482, | |
| "step": 2997 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 5.6983256340026855, | |
| "learning_rate": 3.241062000301705e-05, | |
| "loss": 1.6938, | |
| "step": 3078 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 11.397314071655273, | |
| "learning_rate": 3.179966812490572e-05, | |
| "loss": 1.5931, | |
| "step": 3159 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 10.81894588470459, | |
| "learning_rate": 3.118871624679439e-05, | |
| "loss": 1.6076, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 13.12124252319336, | |
| "learning_rate": 3.0585306984462217e-05, | |
| "loss": 1.6337, | |
| "step": 3321 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 10.692020416259766, | |
| "learning_rate": 2.997435510635088e-05, | |
| "loss": 1.5236, | |
| "step": 3402 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 8.20348834991455, | |
| "learning_rate": 2.9363403228239556e-05, | |
| "loss": 1.665, | |
| "step": 3483 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 9.92470932006836, | |
| "learning_rate": 2.8752451350128223e-05, | |
| "loss": 1.5857, | |
| "step": 3564 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 20.7886962890625, | |
| "learning_rate": 2.8141499472016898e-05, | |
| "loss": 1.5629, | |
| "step": 3645 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 16.494325637817383, | |
| "learning_rate": 2.753054759390557e-05, | |
| "loss": 1.6066, | |
| "step": 3726 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 6.577916145324707, | |
| "learning_rate": 2.6919595715794237e-05, | |
| "loss": 1.571, | |
| "step": 3807 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 7.186463356018066, | |
| "learning_rate": 2.630864383768291e-05, | |
| "loss": 1.5828, | |
| "step": 3888 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 12.30034065246582, | |
| "learning_rate": 2.569769195957158e-05, | |
| "loss": 1.6485, | |
| "step": 3969 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 8.444236755371094, | |
| "learning_rate": 2.5086740081460254e-05, | |
| "loss": 1.5417, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 11.730199813842773, | |
| "learning_rate": 2.447578820334892e-05, | |
| "loss": 1.5634, | |
| "step": 4131 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 9.82486629486084, | |
| "learning_rate": 2.3864836325237593e-05, | |
| "loss": 1.6321, | |
| "step": 4212 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 9.35803508758545, | |
| "learning_rate": 2.3253884447126264e-05, | |
| "loss": 1.5536, | |
| "step": 4293 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 8.086362838745117, | |
| "learning_rate": 2.2642932569014935e-05, | |
| "loss": 1.6023, | |
| "step": 4374 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 12.126749038696289, | |
| "learning_rate": 2.2031980690903606e-05, | |
| "loss": 1.5755, | |
| "step": 4455 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 4.729997158050537, | |
| "learning_rate": 2.1421028812792278e-05, | |
| "loss": 1.6292, | |
| "step": 4536 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 5.383386135101318, | |
| "learning_rate": 2.0810076934680945e-05, | |
| "loss": 1.6119, | |
| "step": 4617 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 7.945587635040283, | |
| "learning_rate": 2.019912505656962e-05, | |
| "loss": 1.4745, | |
| "step": 4698 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 10.771860122680664, | |
| "learning_rate": 1.958817317845829e-05, | |
| "loss": 1.5889, | |
| "step": 4779 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 7.598873615264893, | |
| "learning_rate": 1.8977221300346962e-05, | |
| "loss": 1.5232, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 9.717412948608398, | |
| "learning_rate": 1.8366269422235634e-05, | |
| "loss": 1.6095, | |
| "step": 4941 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 6.822420597076416, | |
| "learning_rate": 1.77553175441243e-05, | |
| "loss": 1.5692, | |
| "step": 5022 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 9.613419532775879, | |
| "learning_rate": 1.7144365666012973e-05, | |
| "loss": 1.5622, | |
| "step": 5103 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 8.61678409576416, | |
| "learning_rate": 1.6533413787901644e-05, | |
| "loss": 1.5005, | |
| "step": 5184 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 9.557565689086914, | |
| "learning_rate": 1.5922461909790315e-05, | |
| "loss": 1.5246, | |
| "step": 5265 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 6.297652721405029, | |
| "learning_rate": 1.531151003167899e-05, | |
| "loss": 1.5551, | |
| "step": 5346 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 7.161550521850586, | |
| "learning_rate": 1.4700558153567659e-05, | |
| "loss": 1.5132, | |
| "step": 5427 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 5.5995635986328125, | |
| "learning_rate": 1.408960627545633e-05, | |
| "loss": 1.5309, | |
| "step": 5508 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 7.648403644561768, | |
| "learning_rate": 1.3478654397345e-05, | |
| "loss": 1.5823, | |
| "step": 5589 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 12.299725532531738, | |
| "learning_rate": 1.2867702519233671e-05, | |
| "loss": 1.5483, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 18.328304290771484, | |
| "learning_rate": 1.2256750641122342e-05, | |
| "loss": 1.5496, | |
| "step": 5751 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 7.374316692352295, | |
| "learning_rate": 1.1645798763011012e-05, | |
| "loss": 1.497, | |
| "step": 5832 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 6.491637706756592, | |
| "learning_rate": 1.1034846884899685e-05, | |
| "loss": 1.6182, | |
| "step": 5913 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 8.2967529296875, | |
| "learning_rate": 1.0423895006788354e-05, | |
| "loss": 1.5618, | |
| "step": 5994 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 8.186725616455078, | |
| "learning_rate": 9.812943128677025e-06, | |
| "loss": 1.5187, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 14.838460922241211, | |
| "learning_rate": 9.201991250565696e-06, | |
| "loss": 1.5244, | |
| "step": 6156 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 17.547962188720703, | |
| "learning_rate": 8.591039372454368e-06, | |
| "loss": 1.5382, | |
| "step": 6237 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 10.257816314697266, | |
| "learning_rate": 7.980087494343039e-06, | |
| "loss": 1.5117, | |
| "step": 6318 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 8.583636283874512, | |
| "learning_rate": 7.369135616231709e-06, | |
| "loss": 1.5942, | |
| "step": 6399 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 10.24096393585205, | |
| "learning_rate": 6.75818373812038e-06, | |
| "loss": 1.4925, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 71.38775634765625, | |
| "learning_rate": 6.1472318600090515e-06, | |
| "loss": 1.5429, | |
| "step": 6561 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 6.424124240875244, | |
| "learning_rate": 5.536279981897723e-06, | |
| "loss": 1.5851, | |
| "step": 6642 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 9.493167877197266, | |
| "learning_rate": 4.925328103786393e-06, | |
| "loss": 1.5089, | |
| "step": 6723 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 7.073258399963379, | |
| "learning_rate": 4.314376225675064e-06, | |
| "loss": 1.4964, | |
| "step": 6804 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 20.42180061340332, | |
| "learning_rate": 3.703424347563735e-06, | |
| "loss": 1.5215, | |
| "step": 6885 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 12.12634563446045, | |
| "learning_rate": 3.0924724694524062e-06, | |
| "loss": 1.551, | |
| "step": 6966 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 11.100313186645508, | |
| "learning_rate": 2.481520591341077e-06, | |
| "loss": 1.4721, | |
| "step": 7047 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 8.091376304626465, | |
| "learning_rate": 1.8705687132297482e-06, | |
| "loss": 1.4702, | |
| "step": 7128 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 7.810707092285156, | |
| "learning_rate": 1.2596168351184192e-06, | |
| "loss": 1.5667, | |
| "step": 7209 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 8.03673267364502, | |
| "learning_rate": 6.486649570070901e-07, | |
| "loss": 1.477, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_gen_len": 35.9694, | |
| "eval_loss": 1.4270155429840088, | |
| "eval_rouge1": 46.4301, | |
| "eval_rouge2": 23.4668, | |
| "eval_rougeL": 37.0224, | |
| "eval_rougeLsum": 42.8893, | |
| "eval_runtime": 468.1997, | |
| "eval_samples_per_second": 1.747, | |
| "eval_steps_per_second": 0.438, | |
| "step": 7366 | |
| } | |
| ], | |
| "logging_steps": 81, | |
| "max_steps": 7366, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 7678847620005888.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |