| { | |
| "best_metric": 0.41303551197052, | |
| "best_model_checkpoint": "deberta-v3-base-dnd/checkpoint-3341", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 3341, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0074827895839568994, | |
| "grad_norm": 3.0921874046325684, | |
| "learning_rate": 3.7313432835820893e-06, | |
| "loss": 0.6928, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.014965579167913799, | |
| "grad_norm": 1.82194185256958, | |
| "learning_rate": 7.4626865671641785e-06, | |
| "loss": 0.6825, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.022448368751870697, | |
| "grad_norm": 3.4050981998443604, | |
| "learning_rate": 1.119402985074627e-05, | |
| "loss": 0.6506, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.029931158335827598, | |
| "grad_norm": 4.51128625869751, | |
| "learning_rate": 1.4925373134328357e-05, | |
| "loss": 0.6525, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.037413947919784495, | |
| "grad_norm": 4.009546756744385, | |
| "learning_rate": 1.865671641791045e-05, | |
| "loss": 0.6028, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.044896737503741395, | |
| "grad_norm": 9.321969985961914, | |
| "learning_rate": 2.238805970149254e-05, | |
| "loss": 0.6125, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.052379527087698295, | |
| "grad_norm": 3.1312828063964844, | |
| "learning_rate": 2.6119402985074626e-05, | |
| "loss": 0.6783, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.059862316671655195, | |
| "grad_norm": 4.284193992614746, | |
| "learning_rate": 2.9850746268656714e-05, | |
| "loss": 0.5675, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06734510625561209, | |
| "grad_norm": 6.1560869216918945, | |
| "learning_rate": 3.358208955223881e-05, | |
| "loss": 0.5775, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.07482789583956899, | |
| "grad_norm": 6.232644557952881, | |
| "learning_rate": 3.73134328358209e-05, | |
| "loss": 0.5287, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.08231068542352589, | |
| "grad_norm": 4.5780439376831055, | |
| "learning_rate": 4.104477611940299e-05, | |
| "loss": 0.5788, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.08979347500748279, | |
| "grad_norm": 4.855668544769287, | |
| "learning_rate": 4.477611940298508e-05, | |
| "loss": 0.5477, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09727626459143969, | |
| "grad_norm": 10.779150009155273, | |
| "learning_rate": 4.850746268656717e-05, | |
| "loss": 0.538, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.10475905417539659, | |
| "grad_norm": 9.850800514221191, | |
| "learning_rate": 4.975049900199601e-05, | |
| "loss": 0.6012, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.11224184375935349, | |
| "grad_norm": 5.283133029937744, | |
| "learning_rate": 4.933466400532269e-05, | |
| "loss": 0.5796, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.11972463334331039, | |
| "grad_norm": 3.158041477203369, | |
| "learning_rate": 4.891882900864937e-05, | |
| "loss": 0.4967, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.12720742292726728, | |
| "grad_norm": 5.8787713050842285, | |
| "learning_rate": 4.8502994011976046e-05, | |
| "loss": 0.4407, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.13469021251122418, | |
| "grad_norm": 3.417243003845215, | |
| "learning_rate": 4.808715901530273e-05, | |
| "loss": 0.5061, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.14217300209518108, | |
| "grad_norm": 3.5104575157165527, | |
| "learning_rate": 4.767132401862941e-05, | |
| "loss": 0.5225, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.14965579167913798, | |
| "grad_norm": 3.441028594970703, | |
| "learning_rate": 4.725548902195609e-05, | |
| "loss": 0.4702, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.15713858126309488, | |
| "grad_norm": 5.702778339385986, | |
| "learning_rate": 4.683965402528277e-05, | |
| "loss": 0.5491, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.16462137084705178, | |
| "grad_norm": 9.593718528747559, | |
| "learning_rate": 4.642381902860945e-05, | |
| "loss": 0.4963, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.17210416043100868, | |
| "grad_norm": 17.2470760345459, | |
| "learning_rate": 4.600798403193613e-05, | |
| "loss": 0.5948, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.17958695001496558, | |
| "grad_norm": 11.078958511352539, | |
| "learning_rate": 4.559214903526281e-05, | |
| "loss": 0.5469, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.18706973959892248, | |
| "grad_norm": 4.973213195800781, | |
| "learning_rate": 4.517631403858949e-05, | |
| "loss": 0.5423, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.19455252918287938, | |
| "grad_norm": 2.1011242866516113, | |
| "learning_rate": 4.476047904191617e-05, | |
| "loss": 0.4844, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.20203531876683628, | |
| "grad_norm": 3.7748775482177734, | |
| "learning_rate": 4.434464404524285e-05, | |
| "loss": 0.475, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.20951810835079318, | |
| "grad_norm": 3.1729724407196045, | |
| "learning_rate": 4.392880904856953e-05, | |
| "loss": 0.4022, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.21700089793475008, | |
| "grad_norm": 6.466404438018799, | |
| "learning_rate": 4.351297405189621e-05, | |
| "loss": 0.5645, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.22448368751870698, | |
| "grad_norm": 3.1774260997772217, | |
| "learning_rate": 4.3097139055222893e-05, | |
| "loss": 0.5004, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.23196647710266388, | |
| "grad_norm": 4.647167205810547, | |
| "learning_rate": 4.268130405854957e-05, | |
| "loss": 0.4604, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.23944926668662078, | |
| "grad_norm": 8.287752151489258, | |
| "learning_rate": 4.226546906187625e-05, | |
| "loss": 0.4739, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.24693205627057768, | |
| "grad_norm": 4.331120014190674, | |
| "learning_rate": 4.184963406520293e-05, | |
| "loss": 0.4093, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.25441484585453455, | |
| "grad_norm": 3.7586264610290527, | |
| "learning_rate": 4.143379906852961e-05, | |
| "loss": 0.3685, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.2618976354384915, | |
| "grad_norm": 2.5479791164398193, | |
| "learning_rate": 4.101796407185629e-05, | |
| "loss": 0.5314, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.26938042502244836, | |
| "grad_norm": 2.888842821121216, | |
| "learning_rate": 4.060212907518297e-05, | |
| "loss": 0.6073, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.2768632146064053, | |
| "grad_norm": 1.649104356765747, | |
| "learning_rate": 4.018629407850965e-05, | |
| "loss": 0.4649, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.28434600419036216, | |
| "grad_norm": 2.926392078399658, | |
| "learning_rate": 3.977045908183633e-05, | |
| "loss": 0.5067, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2918287937743191, | |
| "grad_norm": 1.7140668630599976, | |
| "learning_rate": 3.935462408516301e-05, | |
| "loss": 0.4471, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.29931158335827596, | |
| "grad_norm": 15.356063842773438, | |
| "learning_rate": 3.893878908848969e-05, | |
| "loss": 0.5652, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3067943729422329, | |
| "grad_norm": 6.341818332672119, | |
| "learning_rate": 3.852295409181637e-05, | |
| "loss": 0.5331, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.31427716252618976, | |
| "grad_norm": 3.6557815074920654, | |
| "learning_rate": 3.810711909514305e-05, | |
| "loss": 0.538, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.3217599521101467, | |
| "grad_norm": 3.872098207473755, | |
| "learning_rate": 3.7691284098469734e-05, | |
| "loss": 0.4805, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.32924274169410356, | |
| "grad_norm": 2.6456832885742188, | |
| "learning_rate": 3.727544910179641e-05, | |
| "loss": 0.5127, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.3367255312780605, | |
| "grad_norm": 4.959244728088379, | |
| "learning_rate": 3.685961410512309e-05, | |
| "loss": 0.5048, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.34420832086201736, | |
| "grad_norm": 4.114342212677002, | |
| "learning_rate": 3.644377910844977e-05, | |
| "loss": 0.4545, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.3516911104459743, | |
| "grad_norm": 6.731595039367676, | |
| "learning_rate": 3.6027944111776455e-05, | |
| "loss": 0.4655, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.35917390002993116, | |
| "grad_norm": 2.69596791267395, | |
| "learning_rate": 3.5612109115103126e-05, | |
| "loss": 0.443, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.36665668961388803, | |
| "grad_norm": 7.755747318267822, | |
| "learning_rate": 3.5196274118429805e-05, | |
| "loss": 0.5292, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.37413947919784496, | |
| "grad_norm": 9.343152046203613, | |
| "learning_rate": 3.478043912175648e-05, | |
| "loss": 0.5741, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.38162226878180183, | |
| "grad_norm": 8.174920082092285, | |
| "learning_rate": 3.436460412508317e-05, | |
| "loss": 0.5172, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.38910505836575876, | |
| "grad_norm": 3.547642946243286, | |
| "learning_rate": 3.394876912840985e-05, | |
| "loss": 0.5739, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.39658784794971563, | |
| "grad_norm": 1.9605724811553955, | |
| "learning_rate": 3.3532934131736525e-05, | |
| "loss": 0.4965, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.40407063753367256, | |
| "grad_norm": 2.454697370529175, | |
| "learning_rate": 3.3117099135063204e-05, | |
| "loss": 0.4717, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.41155342711762943, | |
| "grad_norm": 1.8937394618988037, | |
| "learning_rate": 3.270126413838989e-05, | |
| "loss": 0.5098, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.41903621670158636, | |
| "grad_norm": 1.9308140277862549, | |
| "learning_rate": 3.228542914171657e-05, | |
| "loss": 0.4482, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.42651900628554323, | |
| "grad_norm": 6.394463062286377, | |
| "learning_rate": 3.1869594145043246e-05, | |
| "loss": 0.4902, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.43400179586950016, | |
| "grad_norm": 2.807293653488159, | |
| "learning_rate": 3.1453759148369925e-05, | |
| "loss": 0.5226, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.44148458545345703, | |
| "grad_norm": 5.654155254364014, | |
| "learning_rate": 3.103792415169661e-05, | |
| "loss": 0.439, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.44896737503741396, | |
| "grad_norm": 1.9263148307800293, | |
| "learning_rate": 3.062208915502329e-05, | |
| "loss": 0.5519, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.45645016462137084, | |
| "grad_norm": 6.3616437911987305, | |
| "learning_rate": 3.0206254158349967e-05, | |
| "loss": 0.4668, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.46393295420532776, | |
| "grad_norm": 2.5046446323394775, | |
| "learning_rate": 2.979041916167665e-05, | |
| "loss": 0.4642, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.47141574378928464, | |
| "grad_norm": 3.0230417251586914, | |
| "learning_rate": 2.9374584165003327e-05, | |
| "loss": 0.4973, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.47889853337324156, | |
| "grad_norm": 2.627425193786621, | |
| "learning_rate": 2.8958749168330006e-05, | |
| "loss": 0.5057, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.48638132295719844, | |
| "grad_norm": 2.3020899295806885, | |
| "learning_rate": 2.8542914171656687e-05, | |
| "loss": 0.4433, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.49386411254115536, | |
| "grad_norm": 5.167375564575195, | |
| "learning_rate": 2.8127079174983366e-05, | |
| "loss": 0.3782, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.5013469021251122, | |
| "grad_norm": 0.9771930575370789, | |
| "learning_rate": 2.7711244178310048e-05, | |
| "loss": 0.5208, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.5088296917090691, | |
| "grad_norm": 3.793553113937378, | |
| "learning_rate": 2.7295409181636726e-05, | |
| "loss": 0.517, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.5163124812930261, | |
| "grad_norm": 20.930782318115234, | |
| "learning_rate": 2.6879574184963408e-05, | |
| "loss": 0.4343, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.523795270876983, | |
| "grad_norm": 1.3979058265686035, | |
| "learning_rate": 2.6463739188290087e-05, | |
| "loss": 0.3221, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.5312780604609398, | |
| "grad_norm": 2.7222259044647217, | |
| "learning_rate": 2.604790419161677e-05, | |
| "loss": 0.5571, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.5387608500448967, | |
| "grad_norm": 3.1218860149383545, | |
| "learning_rate": 2.5632069194943447e-05, | |
| "loss": 0.4573, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.5462436396288536, | |
| "grad_norm": 15.449758529663086, | |
| "learning_rate": 2.521623419827013e-05, | |
| "loss": 0.4923, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.5537264292128106, | |
| "grad_norm": 1.3911386728286743, | |
| "learning_rate": 2.4800399201596807e-05, | |
| "loss": 0.4064, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.5612092187967674, | |
| "grad_norm": 4.126830577850342, | |
| "learning_rate": 2.438456420492349e-05, | |
| "loss": 0.4475, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.5686920083807243, | |
| "grad_norm": 11.039737701416016, | |
| "learning_rate": 2.3968729208250168e-05, | |
| "loss": 0.4749, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.5761747979646812, | |
| "grad_norm": 2.969331979751587, | |
| "learning_rate": 2.355289421157685e-05, | |
| "loss": 0.4823, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.5836575875486382, | |
| "grad_norm": 1.880447506904602, | |
| "learning_rate": 2.3137059214903528e-05, | |
| "loss": 0.3848, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.591140377132595, | |
| "grad_norm": 3.7677900791168213, | |
| "learning_rate": 2.272122421823021e-05, | |
| "loss": 0.4666, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.5986231667165519, | |
| "grad_norm": 5.5123701095581055, | |
| "learning_rate": 2.2305389221556888e-05, | |
| "loss": 0.4615, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.6061059563005088, | |
| "grad_norm": 4.495054244995117, | |
| "learning_rate": 2.1889554224883567e-05, | |
| "loss": 0.4972, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.6135887458844658, | |
| "grad_norm": 3.059657335281372, | |
| "learning_rate": 2.147371922821025e-05, | |
| "loss": 0.4191, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.6210715354684226, | |
| "grad_norm": 1.6755062341690063, | |
| "learning_rate": 2.1057884231536927e-05, | |
| "loss": 0.375, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.6285543250523795, | |
| "grad_norm": 4.203754901885986, | |
| "learning_rate": 2.0642049234863606e-05, | |
| "loss": 0.4363, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.6360371146363364, | |
| "grad_norm": 3.1495301723480225, | |
| "learning_rate": 2.0226214238190287e-05, | |
| "loss": 0.5009, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.6435199042202934, | |
| "grad_norm": 1.5389057397842407, | |
| "learning_rate": 1.9810379241516966e-05, | |
| "loss": 0.5188, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.6510026938042502, | |
| "grad_norm": 5.250295162200928, | |
| "learning_rate": 1.9394544244843644e-05, | |
| "loss": 0.4371, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.6584854833882071, | |
| "grad_norm": 4.908881187438965, | |
| "learning_rate": 1.8978709248170326e-05, | |
| "loss": 0.3896, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.665968272972164, | |
| "grad_norm": 2.80906081199646, | |
| "learning_rate": 1.8562874251497005e-05, | |
| "loss": 0.4146, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.673451062556121, | |
| "grad_norm": 5.563296794891357, | |
| "learning_rate": 1.8147039254823687e-05, | |
| "loss": 0.4671, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.6809338521400778, | |
| "grad_norm": 2.8475704193115234, | |
| "learning_rate": 1.7731204258150365e-05, | |
| "loss": 0.4635, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.6884166417240347, | |
| "grad_norm": 19.94379425048828, | |
| "learning_rate": 1.7315369261477047e-05, | |
| "loss": 0.4447, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.6958994313079916, | |
| "grad_norm": 4.219967842102051, | |
| "learning_rate": 1.6899534264803725e-05, | |
| "loss": 0.5094, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.7033822208919486, | |
| "grad_norm": 5.59881591796875, | |
| "learning_rate": 1.6483699268130407e-05, | |
| "loss": 0.4474, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.7108650104759054, | |
| "grad_norm": 4.04728364944458, | |
| "learning_rate": 1.6067864271457086e-05, | |
| "loss": 0.4025, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.7183478000598623, | |
| "grad_norm": 3.3200647830963135, | |
| "learning_rate": 1.5652029274783768e-05, | |
| "loss": 0.3932, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.7258305896438192, | |
| "grad_norm": 3.505768060684204, | |
| "learning_rate": 1.5236194278110446e-05, | |
| "loss": 0.4847, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.7333133792277761, | |
| "grad_norm": 1.7178373336791992, | |
| "learning_rate": 1.4820359281437126e-05, | |
| "loss": 0.464, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.740796168811733, | |
| "grad_norm": 7.6995463371276855, | |
| "learning_rate": 1.4404524284763806e-05, | |
| "loss": 0.4037, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.7482789583956899, | |
| "grad_norm": 4.5850982666015625, | |
| "learning_rate": 1.3988689288090487e-05, | |
| "loss": 0.4475, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.7557617479796468, | |
| "grad_norm": 2.429030418395996, | |
| "learning_rate": 1.3572854291417167e-05, | |
| "loss": 0.4272, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.7632445375636037, | |
| "grad_norm": 9.057367324829102, | |
| "learning_rate": 1.3157019294743847e-05, | |
| "loss": 0.5257, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.7707273271475606, | |
| "grad_norm": 2.6991724967956543, | |
| "learning_rate": 1.2741184298070527e-05, | |
| "loss": 0.445, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.7782101167315175, | |
| "grad_norm": 7.034119606018066, | |
| "learning_rate": 1.2325349301397205e-05, | |
| "loss": 0.3666, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.7856929063154744, | |
| "grad_norm": 11.517420768737793, | |
| "learning_rate": 1.1909514304723886e-05, | |
| "loss": 0.4286, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.7931756958994313, | |
| "grad_norm": 2.4604222774505615, | |
| "learning_rate": 1.1493679308050566e-05, | |
| "loss": 0.5487, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.8006584854833882, | |
| "grad_norm": 3.8169350624084473, | |
| "learning_rate": 1.1077844311377246e-05, | |
| "loss": 0.4114, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.8081412750673451, | |
| "grad_norm": 2.442943572998047, | |
| "learning_rate": 1.0662009314703926e-05, | |
| "loss": 0.4117, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.815624064651302, | |
| "grad_norm": 2.7049098014831543, | |
| "learning_rate": 1.0246174318030606e-05, | |
| "loss": 0.4082, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.8231068542352589, | |
| "grad_norm": 7.977570056915283, | |
| "learning_rate": 9.830339321357286e-06, | |
| "loss": 0.3953, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.8305896438192159, | |
| "grad_norm": 2.9819185733795166, | |
| "learning_rate": 9.414504324683967e-06, | |
| "loss": 0.3916, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.8380724334031727, | |
| "grad_norm": 13.983453750610352, | |
| "learning_rate": 8.998669328010647e-06, | |
| "loss": 0.4052, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.8455552229871296, | |
| "grad_norm": 4.8795247077941895, | |
| "learning_rate": 8.582834331337327e-06, | |
| "loss": 0.4146, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.8530380125710865, | |
| "grad_norm": 4.371706962585449, | |
| "learning_rate": 8.166999334664007e-06, | |
| "loss": 0.367, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.8605208021550435, | |
| "grad_norm": 2.1439707279205322, | |
| "learning_rate": 7.751164337990686e-06, | |
| "loss": 0.3967, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.8680035917390003, | |
| "grad_norm": 9.88875961303711, | |
| "learning_rate": 7.335329341317365e-06, | |
| "loss": 0.4283, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.8754863813229572, | |
| "grad_norm": 3.421769618988037, | |
| "learning_rate": 6.919494344644045e-06, | |
| "loss": 0.3436, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.8829691709069141, | |
| "grad_norm": 5.392879009246826, | |
| "learning_rate": 6.503659347970725e-06, | |
| "loss": 0.4977, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.8904519604908709, | |
| "grad_norm": 1.4103248119354248, | |
| "learning_rate": 6.0878243512974054e-06, | |
| "loss": 0.3044, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.8979347500748279, | |
| "grad_norm": 10.673860549926758, | |
| "learning_rate": 5.671989354624086e-06, | |
| "loss": 0.3868, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.9054175396587848, | |
| "grad_norm": 1.0500974655151367, | |
| "learning_rate": 5.256154357950766e-06, | |
| "loss": 0.4401, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.9129003292427417, | |
| "grad_norm": 4.299657821655273, | |
| "learning_rate": 4.840319361277446e-06, | |
| "loss": 0.392, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.9203831188266985, | |
| "grad_norm": 2.322474956512451, | |
| "learning_rate": 4.424484364604125e-06, | |
| "loss": 0.4357, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.9278659084106555, | |
| "grad_norm": 4.478582859039307, | |
| "learning_rate": 4.008649367930805e-06, | |
| "loss": 0.3292, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.9353486979946124, | |
| "grad_norm": 7.355320930480957, | |
| "learning_rate": 3.592814371257485e-06, | |
| "loss": 0.483, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.9428314875785693, | |
| "grad_norm": 3.229567766189575, | |
| "learning_rate": 3.1769793745841653e-06, | |
| "loss": 0.3812, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.9503142771625261, | |
| "grad_norm": 6.137628555297852, | |
| "learning_rate": 2.761144377910845e-06, | |
| "loss": 0.3236, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.9577970667464831, | |
| "grad_norm": 9.047451972961426, | |
| "learning_rate": 2.345309381237525e-06, | |
| "loss": 0.4797, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.96527985633044, | |
| "grad_norm": 9.233771324157715, | |
| "learning_rate": 1.929474384564205e-06, | |
| "loss": 0.3397, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.9727626459143969, | |
| "grad_norm": 4.4976935386657715, | |
| "learning_rate": 1.513639387890885e-06, | |
| "loss": 0.3884, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.9802454354983537, | |
| "grad_norm": 5.986300945281982, | |
| "learning_rate": 1.097804391217565e-06, | |
| "loss": 0.4094, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.9877282250823107, | |
| "grad_norm": 4.511610507965088, | |
| "learning_rate": 6.819693945442449e-07, | |
| "loss": 0.4267, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.9952110146662676, | |
| "grad_norm": 7.771286964416504, | |
| "learning_rate": 2.661343978709248e-07, | |
| "loss": 0.3932, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8155964675946714, | |
| "eval_auc": 0.8995583741711224, | |
| "eval_f1": 0.8469565217391304, | |
| "eval_loss": 0.41303551197052, | |
| "eval_precision": 0.8563175081637779, | |
| "eval_recall": 0.837797984762841, | |
| "eval_runtime": 1439.6587, | |
| "eval_samples_per_second": 4.641, | |
| "eval_steps_per_second": 0.29, | |
| "step": 3341 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 3341, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.01 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1757876485773312.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |