| { | |
| "best_metric": 0.1043073907494545, | |
| "best_model_checkpoint": "flan-t5-base-nlrebel\\checkpoint-110235", | |
| "epoch": 5.0, | |
| "global_step": 110235, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9773211774844656e-05, | |
| "loss": 1.1883, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.95464235496893e-05, | |
| "loss": 0.4505, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9319635324533955e-05, | |
| "loss": 0.3265, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.90928470993786e-05, | |
| "loss": 0.2999, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8866058874223255e-05, | |
| "loss": 0.282, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.86392706490679e-05, | |
| "loss": 0.2608, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.8412482423912555e-05, | |
| "loss": 0.2377, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.81856941987572e-05, | |
| "loss": 0.2276, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.7958905973601855e-05, | |
| "loss": 0.2064, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.77321177484465e-05, | |
| "loss": 0.2141, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.7505329523291154e-05, | |
| "loss": 0.193, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.72785412981358e-05, | |
| "loss": 0.1955, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.7051753072980454e-05, | |
| "loss": 0.1827, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.68249648478251e-05, | |
| "loss": 0.183, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.6598176622669754e-05, | |
| "loss": 0.1834, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.63713883975144e-05, | |
| "loss": 0.1812, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.6144600172359054e-05, | |
| "loss": 0.1808, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.591781194720371e-05, | |
| "loss": 0.1779, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.569102372204835e-05, | |
| "loss": 0.1567, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.5464235496893007e-05, | |
| "loss": 0.1667, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.523744727173765e-05, | |
| "loss": 0.1579, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.5010659046582306e-05, | |
| "loss": 0.1621, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.478387082142695e-05, | |
| "loss": 0.1631, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.4557082596271606e-05, | |
| "loss": 0.1511, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.433029437111625e-05, | |
| "loss": 0.1549, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.4103506145960906e-05, | |
| "loss": 0.1528, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.387671792080555e-05, | |
| "loss": 0.1518, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.3649929695650205e-05, | |
| "loss": 0.15, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.342314147049485e-05, | |
| "loss": 0.1457, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.3196353245339505e-05, | |
| "loss": 0.1467, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.296956502018415e-05, | |
| "loss": 0.1528, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.2742776795028805e-05, | |
| "loss": 0.1459, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.251598856987346e-05, | |
| "loss": 0.1369, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.2289200344718105e-05, | |
| "loss": 0.144, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.206241211956276e-05, | |
| "loss": 0.1387, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.1835623894407404e-05, | |
| "loss": 0.1375, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.160883566925206e-05, | |
| "loss": 0.1445, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.1382047444096704e-05, | |
| "loss": 0.1436, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.115525921894136e-05, | |
| "loss": 0.1373, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.0928470993786004e-05, | |
| "loss": 0.1466, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.070168276863066e-05, | |
| "loss": 0.1342, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.0474894543475304e-05, | |
| "loss": 0.1322, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.024810631831996e-05, | |
| "loss": 0.1367, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.00213180931646e-05, | |
| "loss": 0.1309, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_gen_len": 18.65612244897959, | |
| "eval_loss": 0.12737533450126648, | |
| "eval_rouge1": 50.7462, | |
| "eval_rouge2": 38.4276, | |
| "eval_rougeL": 48.9889, | |
| "eval_rougeLsum": 49.009, | |
| "eval_runtime": 527.904, | |
| "eval_samples_per_second": 9.282, | |
| "eval_steps_per_second": 2.32, | |
| "step": 22047 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.979452986800926e-05, | |
| "loss": 0.123, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.95677416428539e-05, | |
| "loss": 0.1248, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.9340953417698556e-05, | |
| "loss": 0.1154, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.91141651925432e-05, | |
| "loss": 0.1254, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.8887376967387856e-05, | |
| "loss": 0.1269, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.866058874223251e-05, | |
| "loss": 0.1198, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.8433800517077156e-05, | |
| "loss": 0.1186, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.820701229192181e-05, | |
| "loss": 0.1208, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.7980224066766456e-05, | |
| "loss": 0.1223, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.775343584161111e-05, | |
| "loss": 0.125, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.7526647616455755e-05, | |
| "loss": 0.1147, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.729985939130041e-05, | |
| "loss": 0.1266, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.7073071166145055e-05, | |
| "loss": 0.1197, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.684628294098971e-05, | |
| "loss": 0.1208, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.6619494715834355e-05, | |
| "loss": 0.1166, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.639270649067901e-05, | |
| "loss": 0.1195, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.6165918265523655e-05, | |
| "loss": 0.1153, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 3.593913004036831e-05, | |
| "loss": 0.1156, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.5712341815212954e-05, | |
| "loss": 0.1215, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.548555359005761e-05, | |
| "loss": 0.1126, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.525876536490226e-05, | |
| "loss": 0.1176, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.503197713974691e-05, | |
| "loss": 0.1161, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 3.480518891459156e-05, | |
| "loss": 0.1131, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 3.457840068943621e-05, | |
| "loss": 0.1182, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.435161246428086e-05, | |
| "loss": 0.1111, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.412482423912551e-05, | |
| "loss": 0.1166, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.389803601397016e-05, | |
| "loss": 0.1171, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.3671247788814806e-05, | |
| "loss": 0.1218, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.344445956365946e-05, | |
| "loss": 0.1088, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.3217671338504106e-05, | |
| "loss": 0.1139, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 3.299088311334876e-05, | |
| "loss": 0.1111, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.2764094888193406e-05, | |
| "loss": 0.1184, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.253730666303806e-05, | |
| "loss": 0.11, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.2310518437882706e-05, | |
| "loss": 0.1131, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.208373021272736e-05, | |
| "loss": 0.111, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.1856941987572005e-05, | |
| "loss": 0.107, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.163015376241666e-05, | |
| "loss": 0.1122, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.140336553726131e-05, | |
| "loss": 0.1087, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.117657731210596e-05, | |
| "loss": 0.1149, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.094978908695061e-05, | |
| "loss": 0.1049, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.072300086179526e-05, | |
| "loss": 0.1127, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.0496212636639908e-05, | |
| "loss": 0.103, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.0269424411484558e-05, | |
| "loss": 0.1097, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.0042636186329208e-05, | |
| "loss": 0.113, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_gen_len": 18.55, | |
| "eval_loss": 0.11201398819684982, | |
| "eval_rouge1": 51.3852, | |
| "eval_rouge2": 39.4979, | |
| "eval_rougeL": 49.6625, | |
| "eval_rougeLsum": 49.6776, | |
| "eval_runtime": 524.7178, | |
| "eval_samples_per_second": 9.338, | |
| "eval_steps_per_second": 2.335, | |
| "step": 44094 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.981584796117386e-05, | |
| "loss": 0.0999, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.958905973601851e-05, | |
| "loss": 0.0997, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.936227151086316e-05, | |
| "loss": 0.102, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.913548328570781e-05, | |
| "loss": 0.1088, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.890869506055246e-05, | |
| "loss": 0.0993, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.868190683539711e-05, | |
| "loss": 0.1009, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 2.845511861024176e-05, | |
| "loss": 0.0922, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 2.822833038508641e-05, | |
| "loss": 0.0964, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 2.800154215993106e-05, | |
| "loss": 0.0985, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 2.777475393477571e-05, | |
| "loss": 0.099, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.754796570962036e-05, | |
| "loss": 0.0978, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.732117748446501e-05, | |
| "loss": 0.0977, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 2.709438925930966e-05, | |
| "loss": 0.1012, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 2.686760103415431e-05, | |
| "loss": 0.0967, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.664081280899896e-05, | |
| "loss": 0.101, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.641402458384361e-05, | |
| "loss": 0.0961, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.6187236358688262e-05, | |
| "loss": 0.0939, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.5960448133532912e-05, | |
| "loss": 0.093, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.5733659908377562e-05, | |
| "loss": 0.0939, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 2.5506871683222212e-05, | |
| "loss": 0.0991, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.5280083458066862e-05, | |
| "loss": 0.0911, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 2.505329523291151e-05, | |
| "loss": 0.1011, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 2.4826507007756158e-05, | |
| "loss": 0.0989, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 2.4599718782600808e-05, | |
| "loss": 0.0978, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.4372930557445458e-05, | |
| "loss": 0.0994, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 2.4146142332290108e-05, | |
| "loss": 0.098, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 2.3919354107134758e-05, | |
| "loss": 0.0907, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.3692565881979407e-05, | |
| "loss": 0.0993, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 2.3465777656824057e-05, | |
| "loss": 0.0956, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.323898943166871e-05, | |
| "loss": 0.0997, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.301220120651336e-05, | |
| "loss": 0.0947, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 2.278541298135801e-05, | |
| "loss": 0.0969, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 2.255862475620266e-05, | |
| "loss": 0.0994, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 2.233183653104731e-05, | |
| "loss": 0.0927, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 2.210504830589196e-05, | |
| "loss": 0.0975, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 2.187826008073661e-05, | |
| "loss": 0.0953, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.165147185558126e-05, | |
| "loss": 0.0936, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.142468363042591e-05, | |
| "loss": 0.0944, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.119789540527056e-05, | |
| "loss": 0.1008, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 2.097110718011521e-05, | |
| "loss": 0.0945, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 2.074431895495986e-05, | |
| "loss": 0.0916, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 2.051753072980451e-05, | |
| "loss": 0.0998, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.029074250464916e-05, | |
| "loss": 0.0932, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.006395427949381e-05, | |
| "loss": 0.0925, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_gen_len": 18.567551020408164, | |
| "eval_loss": 0.10631391406059265, | |
| "eval_rouge1": 51.5769, | |
| "eval_rouge2": 39.9334, | |
| "eval_rougeL": 49.9305, | |
| "eval_rougeLsum": 49.9362, | |
| "eval_runtime": 524.369, | |
| "eval_samples_per_second": 9.345, | |
| "eval_steps_per_second": 2.336, | |
| "step": 66141 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.983716605433846e-05, | |
| "loss": 0.0874, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.9610377829183112e-05, | |
| "loss": 0.084, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.9383589604027762e-05, | |
| "loss": 0.0907, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 1.915680137887241e-05, | |
| "loss": 0.0955, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.893001315371706e-05, | |
| "loss": 0.0803, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 1.870322492856171e-05, | |
| "loss": 0.0962, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 1.847643670340636e-05, | |
| "loss": 0.0806, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.824964847825101e-05, | |
| "loss": 0.0887, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.802286025309566e-05, | |
| "loss": 0.0869, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.779607202794031e-05, | |
| "loss": 0.0908, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.756928380278496e-05, | |
| "loss": 0.0828, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.734249557762961e-05, | |
| "loss": 0.086, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.711570735247426e-05, | |
| "loss": 0.0827, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.688891912731891e-05, | |
| "loss": 0.085, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.666213090216356e-05, | |
| "loss": 0.0894, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.643534267700821e-05, | |
| "loss": 0.0834, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.620855445185286e-05, | |
| "loss": 0.0842, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.5981766226697513e-05, | |
| "loss": 0.0878, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 1.5754978001542163e-05, | |
| "loss": 0.0889, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 1.5528189776386813e-05, | |
| "loss": 0.0875, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 1.5301401551231463e-05, | |
| "loss": 0.0855, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 1.5074613326076113e-05, | |
| "loss": 0.0903, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 1.4847825100920762e-05, | |
| "loss": 0.0878, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 1.4621036875765412e-05, | |
| "loss": 0.0838, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 1.4394248650610062e-05, | |
| "loss": 0.0857, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 1.4167460425454712e-05, | |
| "loss": 0.0838, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 1.3940672200299362e-05, | |
| "loss": 0.0894, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 1.3713883975144012e-05, | |
| "loss": 0.0793, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 1.3487095749988662e-05, | |
| "loss": 0.0859, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 1.3260307524833313e-05, | |
| "loss": 0.0851, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 1.3033519299677963e-05, | |
| "loss": 0.0847, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 1.2806731074522613e-05, | |
| "loss": 0.0812, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 1.2579942849367263e-05, | |
| "loss": 0.0878, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 1.2353154624211911e-05, | |
| "loss": 0.0878, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.2126366399056561e-05, | |
| "loss": 0.0872, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.1899578173901212e-05, | |
| "loss": 0.082, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 1.1672789948745862e-05, | |
| "loss": 0.0844, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 1.1446001723590512e-05, | |
| "loss": 0.0872, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 1.1219213498435162e-05, | |
| "loss": 0.0878, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 1.0992425273279812e-05, | |
| "loss": 0.0854, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 1.0765637048124462e-05, | |
| "loss": 0.0825, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 1.0538848822969112e-05, | |
| "loss": 0.0817, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.0312060597813762e-05, | |
| "loss": 0.0831, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 1.0085272372658413e-05, | |
| "loss": 0.0878, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_gen_len": 18.59469387755102, | |
| "eval_loss": 0.10489419847726822, | |
| "eval_rouge1": 51.7274, | |
| "eval_rouge2": 40.2559, | |
| "eval_rougeL": 50.122, | |
| "eval_rougeLsum": 50.147, | |
| "eval_runtime": 522.9353, | |
| "eval_samples_per_second": 9.37, | |
| "eval_steps_per_second": 2.343, | |
| "step": 88188 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.858484147503063e-06, | |
| "loss": 0.0782, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 9.631695922347713e-06, | |
| "loss": 0.0791, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 9.404907697192363e-06, | |
| "loss": 0.0764, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 9.178119472037013e-06, | |
| "loss": 0.0786, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 8.951331246881662e-06, | |
| "loss": 0.075, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 8.724543021726312e-06, | |
| "loss": 0.0819, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 8.497754796570962e-06, | |
| "loss": 0.086, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 8.270966571415614e-06, | |
| "loss": 0.0778, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 8.044178346260264e-06, | |
| "loss": 0.0781, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 7.817390121104913e-06, | |
| "loss": 0.0791, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 7.590601895949563e-06, | |
| "loss": 0.0827, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 7.363813670794213e-06, | |
| "loss": 0.0812, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 7.137025445638863e-06, | |
| "loss": 0.0755, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 6.910237220483514e-06, | |
| "loss": 0.0807, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 6.683448995328164e-06, | |
| "loss": 0.0822, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 6.4566607701728135e-06, | |
| "loss": 0.0786, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 6.2298725450174625e-06, | |
| "loss": 0.0833, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 6.003084319862113e-06, | |
| "loss": 0.078, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 5.776296094706763e-06, | |
| "loss": 0.079, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 5.549507869551413e-06, | |
| "loss": 0.0769, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 5.322719644396063e-06, | |
| "loss": 0.077, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 5.0959314192407136e-06, | |
| "loss": 0.0839, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 4.8691431940853634e-06, | |
| "loss": 0.0808, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 4.642354968930013e-06, | |
| "loss": 0.081, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 4.415566743774663e-06, | |
| "loss": 0.0763, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 4.188778518619314e-06, | |
| "loss": 0.0785, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 3.961990293463964e-06, | |
| "loss": 0.0756, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 3.7352020683086136e-06, | |
| "loss": 0.0779, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.508413843153264e-06, | |
| "loss": 0.0801, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 3.2816256179979138e-06, | |
| "loss": 0.0782, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 3.0548373928425636e-06, | |
| "loss": 0.0826, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.828049167687214e-06, | |
| "loss": 0.088, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.601260942531864e-06, | |
| "loss": 0.078, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.374472717376514e-06, | |
| "loss": 0.0761, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 2.147684492221164e-06, | |
| "loss": 0.0772, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 1.9208962670658143e-06, | |
| "loss": 0.0777, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 1.6941080419104641e-06, | |
| "loss": 0.0788, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 1.4673198167551142e-06, | |
| "loss": 0.079, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 1.240531591599764e-06, | |
| "loss": 0.0799, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 1.0137433664444142e-06, | |
| "loss": 0.0765, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 7.869551412890642e-07, | |
| "loss": 0.081, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.601669161337143e-07, | |
| "loss": 0.0819, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.333786909783644e-07, | |
| "loss": 0.0801, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 1.0659046582301448e-07, | |
| "loss": 0.0813, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_gen_len": 18.58265306122449, | |
| "eval_loss": 0.1043073907494545, | |
| "eval_rouge1": 51.9039, | |
| "eval_rouge2": 40.5297, | |
| "eval_rougeL": 50.2805, | |
| "eval_rougeLsum": 50.2968, | |
| "eval_runtime": 522.6723, | |
| "eval_samples_per_second": 9.375, | |
| "eval_steps_per_second": 2.344, | |
| "step": 110235 | |
| } | |
| ], | |
| "max_steps": 110235, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.509503144067072e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |