| { |
| "best_global_step": 955, |
| "best_metric": 0.3973127603530884, |
| "best_model_checkpoint": "finbert_model/checkpoint-955", |
| "epoch": 3.0, |
| "eval_steps": 500, |
| "global_step": 2865, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.02617801047120419, |
| "grad_norm": 5.389065742492676, |
| "learning_rate": 1.9832460732984295e-05, |
| "loss": 1.0684, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.05235602094240838, |
| "grad_norm": 10.32921314239502, |
| "learning_rate": 1.9657940663176267e-05, |
| "loss": 0.7284, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.07853403141361257, |
| "grad_norm": 6.734121322631836, |
| "learning_rate": 1.948342059336824e-05, |
| "loss": 0.7981, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.10471204188481675, |
| "grad_norm": 8.204632759094238, |
| "learning_rate": 1.930890052356021e-05, |
| "loss": 0.5878, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.13089005235602094, |
| "grad_norm": 10.211732864379883, |
| "learning_rate": 1.9134380453752185e-05, |
| "loss": 0.7182, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.15706806282722513, |
| "grad_norm": 9.390113830566406, |
| "learning_rate": 1.8959860383944157e-05, |
| "loss": 0.72, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.18324607329842932, |
| "grad_norm": 11.717000007629395, |
| "learning_rate": 1.8785340314136128e-05, |
| "loss": 0.6367, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.2094240837696335, |
| "grad_norm": 8.935040473937988, |
| "learning_rate": 1.86108202443281e-05, |
| "loss": 0.6106, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.2356020942408377, |
| "grad_norm": 11.59050464630127, |
| "learning_rate": 1.843630017452007e-05, |
| "loss": 0.6202, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.2617801047120419, |
| "grad_norm": 7.747163772583008, |
| "learning_rate": 1.8261780104712043e-05, |
| "loss": 0.669, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.2879581151832461, |
| "grad_norm": 16.7243595123291, |
| "learning_rate": 1.8087260034904015e-05, |
| "loss": 0.4931, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.31413612565445026, |
| "grad_norm": 7.23421049118042, |
| "learning_rate": 1.7912739965095986e-05, |
| "loss": 0.5974, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.3403141361256545, |
| "grad_norm": 6.768913269042969, |
| "learning_rate": 1.773821989528796e-05, |
| "loss": 0.4926, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.36649214659685864, |
| "grad_norm": 3.186237096786499, |
| "learning_rate": 1.7563699825479933e-05, |
| "loss": 0.4709, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.39267015706806285, |
| "grad_norm": 12.499926567077637, |
| "learning_rate": 1.7389179755671904e-05, |
| "loss": 0.5128, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.418848167539267, |
| "grad_norm": 8.35361099243164, |
| "learning_rate": 1.7214659685863876e-05, |
| "loss": 0.5878, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.44502617801047123, |
| "grad_norm": 6.956220626831055, |
| "learning_rate": 1.7040139616055848e-05, |
| "loss": 0.4794, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.4712041884816754, |
| "grad_norm": 15.914446830749512, |
| "learning_rate": 1.686561954624782e-05, |
| "loss": 0.5827, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.4973821989528796, |
| "grad_norm": 15.149320602416992, |
| "learning_rate": 1.669109947643979e-05, |
| "loss": 0.5467, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.5235602094240838, |
| "grad_norm": 10.100659370422363, |
| "learning_rate": 1.6516579406631763e-05, |
| "loss": 0.525, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.5497382198952879, |
| "grad_norm": 8.821247100830078, |
| "learning_rate": 1.6342059336823738e-05, |
| "loss": 0.5345, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.5759162303664922, |
| "grad_norm": 14.301637649536133, |
| "learning_rate": 1.616753926701571e-05, |
| "loss": 0.5115, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.6020942408376964, |
| "grad_norm": 14.779314994812012, |
| "learning_rate": 1.599301919720768e-05, |
| "loss": 0.5034, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.6282722513089005, |
| "grad_norm": 9.956971168518066, |
| "learning_rate": 1.5818499127399652e-05, |
| "loss": 0.3719, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.6544502617801047, |
| "grad_norm": 17.58684730529785, |
| "learning_rate": 1.5643979057591624e-05, |
| "loss": 0.5269, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.680628272251309, |
| "grad_norm": 10.585662841796875, |
| "learning_rate": 1.5469458987783596e-05, |
| "loss": 0.574, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.7068062827225131, |
| "grad_norm": 17.19683074951172, |
| "learning_rate": 1.5294938917975567e-05, |
| "loss": 0.4294, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.7329842931937173, |
| "grad_norm": 14.906782150268555, |
| "learning_rate": 1.5120418848167539e-05, |
| "loss": 0.4861, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.7591623036649214, |
| "grad_norm": 12.943304061889648, |
| "learning_rate": 1.4945898778359514e-05, |
| "loss": 0.4121, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.7853403141361257, |
| "grad_norm": 0.5662004351615906, |
| "learning_rate": 1.4771378708551485e-05, |
| "loss": 0.5198, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.8115183246073299, |
| "grad_norm": 7.927789211273193, |
| "learning_rate": 1.4596858638743457e-05, |
| "loss": 0.4721, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.837696335078534, |
| "grad_norm": 6.3558855056762695, |
| "learning_rate": 1.4422338568935429e-05, |
| "loss": 0.3977, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.8638743455497382, |
| "grad_norm": 17.170700073242188, |
| "learning_rate": 1.4247818499127402e-05, |
| "loss": 0.3493, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.8900523560209425, |
| "grad_norm": 15.060056686401367, |
| "learning_rate": 1.4073298429319373e-05, |
| "loss": 0.5364, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.9162303664921466, |
| "grad_norm": 11.084234237670898, |
| "learning_rate": 1.3898778359511345e-05, |
| "loss": 0.4815, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.9424083769633508, |
| "grad_norm": 11.328404426574707, |
| "learning_rate": 1.3724258289703317e-05, |
| "loss": 0.5914, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.9685863874345549, |
| "grad_norm": 14.335894584655762, |
| "learning_rate": 1.354973821989529e-05, |
| "loss": 0.4111, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.9947643979057592, |
| "grad_norm": 11.200068473815918, |
| "learning_rate": 1.3375218150087262e-05, |
| "loss": 0.4069, |
| "step": 950 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_accuracy": 0.8690413829229964, |
| "eval_f1": 0.8665004404377881, |
| "eval_loss": 0.3973127603530884, |
| "eval_runtime": 4.1083, |
| "eval_samples_per_second": 464.669, |
| "eval_steps_per_second": 58.175, |
| "step": 955 |
| }, |
| { |
| "epoch": 1.0209424083769634, |
| "grad_norm": 10.518115043640137, |
| "learning_rate": 1.3200698080279233e-05, |
| "loss": 0.3101, |
| "step": 975 |
| }, |
| { |
| "epoch": 1.0471204188481675, |
| "grad_norm": 7.9896440505981445, |
| "learning_rate": 1.3026178010471205e-05, |
| "loss": 0.3442, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.0732984293193717, |
| "grad_norm": 2.440387010574341, |
| "learning_rate": 1.2851657940663178e-05, |
| "loss": 0.2582, |
| "step": 1025 |
| }, |
| { |
| "epoch": 1.0994764397905759, |
| "grad_norm": 4.903307914733887, |
| "learning_rate": 1.267713787085515e-05, |
| "loss": 0.2512, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.12565445026178, |
| "grad_norm": 4.360142707824707, |
| "learning_rate": 1.2502617801047121e-05, |
| "loss": 0.3125, |
| "step": 1075 |
| }, |
| { |
| "epoch": 1.1518324607329844, |
| "grad_norm": 0.4692545533180237, |
| "learning_rate": 1.2328097731239093e-05, |
| "loss": 0.3161, |
| "step": 1100 |
| }, |
| { |
| "epoch": 1.1780104712041886, |
| "grad_norm": 1.2277106046676636, |
| "learning_rate": 1.2153577661431066e-05, |
| "loss": 0.2237, |
| "step": 1125 |
| }, |
| { |
| "epoch": 1.2041884816753927, |
| "grad_norm": 13.60217571258545, |
| "learning_rate": 1.1979057591623038e-05, |
| "loss": 0.3578, |
| "step": 1150 |
| }, |
| { |
| "epoch": 1.2303664921465969, |
| "grad_norm": 8.050612449645996, |
| "learning_rate": 1.180453752181501e-05, |
| "loss": 0.3301, |
| "step": 1175 |
| }, |
| { |
| "epoch": 1.256544502617801, |
| "grad_norm": 23.87043571472168, |
| "learning_rate": 1.1630017452006981e-05, |
| "loss": 0.2843, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.2827225130890052, |
| "grad_norm": 0.21612492203712463, |
| "learning_rate": 1.1455497382198954e-05, |
| "loss": 0.2529, |
| "step": 1225 |
| }, |
| { |
| "epoch": 1.3089005235602094, |
| "grad_norm": 15.835173606872559, |
| "learning_rate": 1.1280977312390926e-05, |
| "loss": 0.2231, |
| "step": 1250 |
| }, |
| { |
| "epoch": 1.3350785340314135, |
| "grad_norm": 49.70761489868164, |
| "learning_rate": 1.1106457242582898e-05, |
| "loss": 0.3135, |
| "step": 1275 |
| }, |
| { |
| "epoch": 1.3612565445026177, |
| "grad_norm": 0.1663261204957962, |
| "learning_rate": 1.093193717277487e-05, |
| "loss": 0.3561, |
| "step": 1300 |
| }, |
| { |
| "epoch": 1.387434554973822, |
| "grad_norm": 17.380788803100586, |
| "learning_rate": 1.0757417102966842e-05, |
| "loss": 0.3588, |
| "step": 1325 |
| }, |
| { |
| "epoch": 1.4136125654450262, |
| "grad_norm": 11.443221092224121, |
| "learning_rate": 1.0582897033158814e-05, |
| "loss": 0.2319, |
| "step": 1350 |
| }, |
| { |
| "epoch": 1.4397905759162304, |
| "grad_norm": 12.970437049865723, |
| "learning_rate": 1.0408376963350786e-05, |
| "loss": 0.3858, |
| "step": 1375 |
| }, |
| { |
| "epoch": 1.4659685863874345, |
| "grad_norm": 8.48019790649414, |
| "learning_rate": 1.0233856893542757e-05, |
| "loss": 0.3213, |
| "step": 1400 |
| }, |
| { |
| "epoch": 1.4921465968586387, |
| "grad_norm": 0.29926174879074097, |
| "learning_rate": 1.005933682373473e-05, |
| "loss": 0.1976, |
| "step": 1425 |
| }, |
| { |
| "epoch": 1.518324607329843, |
| "grad_norm": 7.1454339027404785, |
| "learning_rate": 9.884816753926702e-06, |
| "loss": 0.2965, |
| "step": 1450 |
| }, |
| { |
| "epoch": 1.5445026178010473, |
| "grad_norm": 9.631463050842285, |
| "learning_rate": 9.710296684118674e-06, |
| "loss": 0.2411, |
| "step": 1475 |
| }, |
| { |
| "epoch": 1.5706806282722514, |
| "grad_norm": 72.68104553222656, |
| "learning_rate": 9.535776614310647e-06, |
| "loss": 0.209, |
| "step": 1500 |
| }, |
| { |
| "epoch": 1.5968586387434556, |
| "grad_norm": 72.5035400390625, |
| "learning_rate": 9.361256544502619e-06, |
| "loss": 0.3374, |
| "step": 1525 |
| }, |
| { |
| "epoch": 1.6230366492146597, |
| "grad_norm": 20.452966690063477, |
| "learning_rate": 9.18673647469459e-06, |
| "loss": 0.3733, |
| "step": 1550 |
| }, |
| { |
| "epoch": 1.649214659685864, |
| "grad_norm": 9.723285675048828, |
| "learning_rate": 9.012216404886562e-06, |
| "loss": 0.259, |
| "step": 1575 |
| }, |
| { |
| "epoch": 1.675392670157068, |
| "grad_norm": 0.682996928691864, |
| "learning_rate": 8.837696335078535e-06, |
| "loss": 0.4751, |
| "step": 1600 |
| }, |
| { |
| "epoch": 1.7015706806282722, |
| "grad_norm": 0.5499667525291443, |
| "learning_rate": 8.663176265270507e-06, |
| "loss": 0.3404, |
| "step": 1625 |
| }, |
| { |
| "epoch": 1.7277486910994764, |
| "grad_norm": 31.961536407470703, |
| "learning_rate": 8.488656195462478e-06, |
| "loss": 0.2747, |
| "step": 1650 |
| }, |
| { |
| "epoch": 1.7539267015706805, |
| "grad_norm": 21.48028564453125, |
| "learning_rate": 8.31413612565445e-06, |
| "loss": 0.2676, |
| "step": 1675 |
| }, |
| { |
| "epoch": 1.7801047120418847, |
| "grad_norm": 8.918591499328613, |
| "learning_rate": 8.139616055846423e-06, |
| "loss": 0.3828, |
| "step": 1700 |
| }, |
| { |
| "epoch": 1.8062827225130889, |
| "grad_norm": 19.257600784301758, |
| "learning_rate": 7.965095986038395e-06, |
| "loss": 0.3434, |
| "step": 1725 |
| }, |
| { |
| "epoch": 1.8324607329842932, |
| "grad_norm": 11.988317489624023, |
| "learning_rate": 7.790575916230367e-06, |
| "loss": 0.3589, |
| "step": 1750 |
| }, |
| { |
| "epoch": 1.8586387434554974, |
| "grad_norm": 27.76133155822754, |
| "learning_rate": 7.616055846422339e-06, |
| "loss": 0.3216, |
| "step": 1775 |
| }, |
| { |
| "epoch": 1.8848167539267016, |
| "grad_norm": 31.528942108154297, |
| "learning_rate": 7.4415357766143115e-06, |
| "loss": 0.2637, |
| "step": 1800 |
| }, |
| { |
| "epoch": 1.9109947643979057, |
| "grad_norm": 0.13127724826335907, |
| "learning_rate": 7.267015706806283e-06, |
| "loss": 0.1963, |
| "step": 1825 |
| }, |
| { |
| "epoch": 1.93717277486911, |
| "grad_norm": 1.6614102125167847, |
| "learning_rate": 7.0924956369982556e-06, |
| "loss": 0.2581, |
| "step": 1850 |
| }, |
| { |
| "epoch": 1.9633507853403143, |
| "grad_norm": 1.648483395576477, |
| "learning_rate": 6.917975567190227e-06, |
| "loss": 0.3178, |
| "step": 1875 |
| }, |
| { |
| "epoch": 1.9895287958115184, |
| "grad_norm": 0.3059515953063965, |
| "learning_rate": 6.7434554973822e-06, |
| "loss": 0.2418, |
| "step": 1900 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_accuracy": 0.8727082242011525, |
| "eval_f1": 0.8728823235536903, |
| "eval_loss": 0.4636137783527374, |
| "eval_runtime": 4.2893, |
| "eval_samples_per_second": 445.066, |
| "eval_steps_per_second": 55.721, |
| "step": 1910 |
| }, |
| { |
| "epoch": 2.0157068062827226, |
| "grad_norm": 15.890205383300781, |
| "learning_rate": 6.568935427574171e-06, |
| "loss": 0.1975, |
| "step": 1925 |
| }, |
| { |
| "epoch": 2.0418848167539267, |
| "grad_norm": 4.748979091644287, |
| "learning_rate": 6.394415357766144e-06, |
| "loss": 0.1375, |
| "step": 1950 |
| }, |
| { |
| "epoch": 2.068062827225131, |
| "grad_norm": 12.78852653503418, |
| "learning_rate": 6.219895287958115e-06, |
| "loss": 0.1926, |
| "step": 1975 |
| }, |
| { |
| "epoch": 2.094240837696335, |
| "grad_norm": 9.572379112243652, |
| "learning_rate": 6.045375218150088e-06, |
| "loss": 0.1087, |
| "step": 2000 |
| }, |
| { |
| "epoch": 2.1204188481675392, |
| "grad_norm": 0.5682252049446106, |
| "learning_rate": 5.870855148342059e-06, |
| "loss": 0.1792, |
| "step": 2025 |
| }, |
| { |
| "epoch": 2.1465968586387434, |
| "grad_norm": 0.7415192723274231, |
| "learning_rate": 5.696335078534032e-06, |
| "loss": 0.2378, |
| "step": 2050 |
| }, |
| { |
| "epoch": 2.1727748691099475, |
| "grad_norm": 0.07994470000267029, |
| "learning_rate": 5.521815008726003e-06, |
| "loss": 0.2135, |
| "step": 2075 |
| }, |
| { |
| "epoch": 2.1989528795811517, |
| "grad_norm": 0.22101280093193054, |
| "learning_rate": 5.347294938917976e-06, |
| "loss": 0.0965, |
| "step": 2100 |
| }, |
| { |
| "epoch": 2.225130890052356, |
| "grad_norm": 0.17456702888011932, |
| "learning_rate": 5.1727748691099475e-06, |
| "loss": 0.2084, |
| "step": 2125 |
| }, |
| { |
| "epoch": 2.25130890052356, |
| "grad_norm": 18.24934959411621, |
| "learning_rate": 4.998254799301921e-06, |
| "loss": 0.0802, |
| "step": 2150 |
| }, |
| { |
| "epoch": 2.2774869109947646, |
| "grad_norm": 40.49116897583008, |
| "learning_rate": 4.823734729493892e-06, |
| "loss": 0.2103, |
| "step": 2175 |
| }, |
| { |
| "epoch": 2.303664921465969, |
| "grad_norm": 0.37025186419487, |
| "learning_rate": 4.649214659685865e-06, |
| "loss": 0.1339, |
| "step": 2200 |
| }, |
| { |
| "epoch": 2.329842931937173, |
| "grad_norm": 0.07061374187469482, |
| "learning_rate": 4.4746945898778364e-06, |
| "loss": 0.1267, |
| "step": 2225 |
| }, |
| { |
| "epoch": 2.356020942408377, |
| "grad_norm": 22.344051361083984, |
| "learning_rate": 4.300174520069809e-06, |
| "loss": 0.0843, |
| "step": 2250 |
| }, |
| { |
| "epoch": 2.3821989528795813, |
| "grad_norm": 0.05722755193710327, |
| "learning_rate": 4.1256544502617805e-06, |
| "loss": 0.0903, |
| "step": 2275 |
| }, |
| { |
| "epoch": 2.4083769633507854, |
| "grad_norm": 0.10248207300901413, |
| "learning_rate": 3.951134380453753e-06, |
| "loss": 0.211, |
| "step": 2300 |
| }, |
| { |
| "epoch": 2.4345549738219896, |
| "grad_norm": 0.12041915953159332, |
| "learning_rate": 3.776614310645725e-06, |
| "loss": 0.1836, |
| "step": 2325 |
| }, |
| { |
| "epoch": 2.4607329842931938, |
| "grad_norm": 0.15190765261650085, |
| "learning_rate": 3.602094240837697e-06, |
| "loss": 0.2645, |
| "step": 2350 |
| }, |
| { |
| "epoch": 2.486910994764398, |
| "grad_norm": 0.6992888450622559, |
| "learning_rate": 3.427574171029669e-06, |
| "loss": 0.1515, |
| "step": 2375 |
| }, |
| { |
| "epoch": 2.513089005235602, |
| "grad_norm": 0.07796134799718857, |
| "learning_rate": 3.253054101221641e-06, |
| "loss": 0.0545, |
| "step": 2400 |
| }, |
| { |
| "epoch": 2.5392670157068062, |
| "grad_norm": 0.41577911376953125, |
| "learning_rate": 3.078534031413613e-06, |
| "loss": 0.101, |
| "step": 2425 |
| }, |
| { |
| "epoch": 2.5654450261780104, |
| "grad_norm": 0.14312124252319336, |
| "learning_rate": 2.904013961605585e-06, |
| "loss": 0.2121, |
| "step": 2450 |
| }, |
| { |
| "epoch": 2.5916230366492146, |
| "grad_norm": 6.699955463409424, |
| "learning_rate": 2.729493891797557e-06, |
| "loss": 0.1265, |
| "step": 2475 |
| }, |
| { |
| "epoch": 2.6178010471204187, |
| "grad_norm": 0.11033662408590317, |
| "learning_rate": 2.554973821989529e-06, |
| "loss": 0.1985, |
| "step": 2500 |
| }, |
| { |
| "epoch": 2.643979057591623, |
| "grad_norm": 0.09589897096157074, |
| "learning_rate": 2.3804537521815012e-06, |
| "loss": 0.1419, |
| "step": 2525 |
| }, |
| { |
| "epoch": 2.670157068062827, |
| "grad_norm": 1.7485756874084473, |
| "learning_rate": 2.2059336823734733e-06, |
| "loss": 0.2441, |
| "step": 2550 |
| }, |
| { |
| "epoch": 2.696335078534031, |
| "grad_norm": 0.28520870208740234, |
| "learning_rate": 2.0314136125654453e-06, |
| "loss": 0.1209, |
| "step": 2575 |
| }, |
| { |
| "epoch": 2.7225130890052354, |
| "grad_norm": 0.26048529148101807, |
| "learning_rate": 1.8568935427574173e-06, |
| "loss": 0.2296, |
| "step": 2600 |
| }, |
| { |
| "epoch": 2.7486910994764395, |
| "grad_norm": 0.07016175985336304, |
| "learning_rate": 1.6823734729493894e-06, |
| "loss": 0.0609, |
| "step": 2625 |
| }, |
| { |
| "epoch": 2.774869109947644, |
| "grad_norm": 4.9815592765808105, |
| "learning_rate": 1.5078534031413614e-06, |
| "loss": 0.0824, |
| "step": 2650 |
| }, |
| { |
| "epoch": 2.8010471204188483, |
| "grad_norm": 0.11184385418891907, |
| "learning_rate": 1.3333333333333334e-06, |
| "loss": 0.1331, |
| "step": 2675 |
| }, |
| { |
| "epoch": 2.8272251308900525, |
| "grad_norm": 0.6429989337921143, |
| "learning_rate": 1.1588132635253054e-06, |
| "loss": 0.2019, |
| "step": 2700 |
| }, |
| { |
| "epoch": 2.8534031413612566, |
| "grad_norm": 21.87600326538086, |
| "learning_rate": 9.842931937172775e-07, |
| "loss": 0.0632, |
| "step": 2725 |
| }, |
| { |
| "epoch": 2.8795811518324608, |
| "grad_norm": 0.03200078383088112, |
| "learning_rate": 8.097731239092496e-07, |
| "loss": 0.2638, |
| "step": 2750 |
| }, |
| { |
| "epoch": 2.905759162303665, |
| "grad_norm": 47.9969596862793, |
| "learning_rate": 6.352530541012216e-07, |
| "loss": 0.188, |
| "step": 2775 |
| }, |
| { |
| "epoch": 2.931937172774869, |
| "grad_norm": 0.9299336075782776, |
| "learning_rate": 4.6073298429319373e-07, |
| "loss": 0.064, |
| "step": 2800 |
| }, |
| { |
| "epoch": 2.9581151832460733, |
| "grad_norm": 0.059168703854084015, |
| "learning_rate": 2.862129144851658e-07, |
| "loss": 0.1886, |
| "step": 2825 |
| }, |
| { |
| "epoch": 2.9842931937172774, |
| "grad_norm": 0.6567233800888062, |
| "learning_rate": 1.1169284467713788e-07, |
| "loss": 0.1826, |
| "step": 2850 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_accuracy": 0.8779465688842326, |
| "eval_f1": 0.8773471210523544, |
| "eval_loss": 0.5452297925949097, |
| "eval_runtime": 4.1246, |
| "eval_samples_per_second": 462.832, |
| "eval_steps_per_second": 57.945, |
| "step": 2865 |
| } |
| ], |
| "logging_steps": 25, |
| "max_steps": 2865, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1506455873201664.0, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
