| { |
| "best_metric": 0.4471864700317383, |
| "best_model_checkpoint": "output_dir/finetuned_best_stsb/checkpoint-1900", |
| "epoch": 6.111111111111111, |
| "global_step": 2200, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.14, |
| "eval_combined_score": 0.8343080271135384, |
| "eval_loss": 0.7862432599067688, |
| "eval_pearson": 0.829615903016159, |
| "eval_runtime": 7.9747, |
| "eval_samples_per_second": 188.095, |
| "eval_spearmanr": 0.8390001512109179, |
| "eval_steps_per_second": 11.787, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.28, |
| "eval_combined_score": 0.8549898629109538, |
| "eval_loss": 0.6401779055595398, |
| "eval_pearson": 0.8531715498817952, |
| "eval_runtime": 5.6147, |
| "eval_samples_per_second": 267.156, |
| "eval_spearmanr": 0.8568081759401125, |
| "eval_steps_per_second": 16.742, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.42, |
| "eval_combined_score": 0.8443191275483415, |
| "eval_loss": 0.6554257273674011, |
| "eval_pearson": 0.8474110080447523, |
| "eval_runtime": 5.6204, |
| "eval_samples_per_second": 266.887, |
| "eval_spearmanr": 0.8412272470519305, |
| "eval_steps_per_second": 16.725, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.56, |
| "eval_combined_score": 0.8673615699410524, |
| "eval_loss": 0.5701442956924438, |
| "eval_pearson": 0.8692556554827053, |
| "eval_runtime": 5.9834, |
| "eval_samples_per_second": 250.694, |
| "eval_spearmanr": 0.8654674843993996, |
| "eval_steps_per_second": 15.71, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.69, |
| "eval_combined_score": 0.8721712658068579, |
| "eval_loss": 0.5915741920471191, |
| "eval_pearson": 0.8722910960572966, |
| "eval_runtime": 5.5689, |
| "eval_samples_per_second": 269.355, |
| "eval_spearmanr": 0.8720514355564192, |
| "eval_steps_per_second": 16.88, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.83, |
| "eval_combined_score": 0.8812625681870452, |
| "eval_loss": 0.5046865344047546, |
| "eval_pearson": 0.881670075532939, |
| "eval_runtime": 6.0146, |
| "eval_samples_per_second": 249.392, |
| "eval_spearmanr": 0.8808550608411514, |
| "eval_steps_per_second": 15.629, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.97, |
| "eval_combined_score": 0.878795558425897, |
| "eval_loss": 0.5713155269622803, |
| "eval_pearson": 0.8736557247061164, |
| "eval_runtime": 6.0781, |
| "eval_samples_per_second": 246.786, |
| "eval_spearmanr": 0.8839353921456775, |
| "eval_steps_per_second": 15.465, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.11, |
| "eval_combined_score": 0.8778669265351486, |
| "eval_loss": 0.5400073528289795, |
| "eval_pearson": 0.8804620985683852, |
| "eval_runtime": 5.5918, |
| "eval_samples_per_second": 268.248, |
| "eval_spearmanr": 0.875271754501912, |
| "eval_steps_per_second": 16.81, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.25, |
| "eval_combined_score": 0.8833550106755617, |
| "eval_loss": 0.49523210525512695, |
| "eval_pearson": 0.8858630637184036, |
| "eval_runtime": 5.7638, |
| "eval_samples_per_second": 260.246, |
| "eval_spearmanr": 0.8808469576327196, |
| "eval_steps_per_second": 16.309, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 4.1319444444444445e-05, |
| "loss": 0.7828, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.39, |
| "eval_combined_score": 0.8827148836128291, |
| "eval_loss": 0.5279519557952881, |
| "eval_pearson": 0.8845811601256096, |
| "eval_runtime": 5.9061, |
| "eval_samples_per_second": 253.975, |
| "eval_spearmanr": 0.8808486071000486, |
| "eval_steps_per_second": 15.916, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.53, |
| "eval_combined_score": 0.8801981392571121, |
| "eval_loss": 0.5072206258773804, |
| "eval_pearson": 0.8810815176670656, |
| "eval_runtime": 5.5676, |
| "eval_samples_per_second": 269.418, |
| "eval_spearmanr": 0.8793147608471588, |
| "eval_steps_per_second": 16.884, |
| "step": 550 |
| }, |
| { |
| "epoch": 1.67, |
| "eval_combined_score": 0.8851362735229233, |
| "eval_loss": 0.49581071734428406, |
| "eval_pearson": 0.8875092909393476, |
| "eval_runtime": 5.6319, |
| "eval_samples_per_second": 266.34, |
| "eval_spearmanr": 0.8827632561064991, |
| "eval_steps_per_second": 16.691, |
| "step": 600 |
| }, |
| { |
| "epoch": 1.81, |
| "eval_combined_score": 0.8852476609333064, |
| "eval_loss": 0.4841392934322357, |
| "eval_pearson": 0.8870465421559632, |
| "eval_runtime": 5.8286, |
| "eval_samples_per_second": 257.351, |
| "eval_spearmanr": 0.8834487797106495, |
| "eval_steps_per_second": 16.127, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.94, |
| "eval_combined_score": 0.8837498972174722, |
| "eval_loss": 0.5464913845062256, |
| "eval_pearson": 0.8849881470427847, |
| "eval_runtime": 5.5884, |
| "eval_samples_per_second": 268.412, |
| "eval_spearmanr": 0.8825116473921596, |
| "eval_steps_per_second": 16.82, |
| "step": 700 |
| }, |
| { |
| "epoch": 2.08, |
| "eval_combined_score": 0.8858322086355523, |
| "eval_loss": 0.48856478929519653, |
| "eval_pearson": 0.8881076481621831, |
| "eval_runtime": 5.6259, |
| "eval_samples_per_second": 266.624, |
| "eval_spearmanr": 0.8835567691089214, |
| "eval_steps_per_second": 16.708, |
| "step": 750 |
| }, |
| { |
| "epoch": 2.22, |
| "eval_combined_score": 0.8865111243309827, |
| "eval_loss": 0.4807361662387848, |
| "eval_pearson": 0.8892892267625273, |
| "eval_runtime": 5.825, |
| "eval_samples_per_second": 257.512, |
| "eval_spearmanr": 0.8837330218994379, |
| "eval_steps_per_second": 16.137, |
| "step": 800 |
| }, |
| { |
| "epoch": 2.36, |
| "eval_combined_score": 0.8881887854009817, |
| "eval_loss": 0.48988330364227295, |
| "eval_pearson": 0.8898428387864722, |
| "eval_runtime": 6.1177, |
| "eval_samples_per_second": 245.192, |
| "eval_spearmanr": 0.8865347320154912, |
| "eval_steps_per_second": 15.365, |
| "step": 850 |
| }, |
| { |
| "epoch": 2.5, |
| "eval_combined_score": 0.8868479524643782, |
| "eval_loss": 0.4934450387954712, |
| "eval_pearson": 0.8891167810226788, |
| "eval_runtime": 5.6331, |
| "eval_samples_per_second": 266.284, |
| "eval_spearmanr": 0.8845791239060775, |
| "eval_steps_per_second": 16.687, |
| "step": 900 |
| }, |
| { |
| "epoch": 2.64, |
| "eval_combined_score": 0.8889778694122226, |
| "eval_loss": 0.4866905212402344, |
| "eval_pearson": 0.8890192489258524, |
| "eval_runtime": 5.7954, |
| "eval_samples_per_second": 258.824, |
| "eval_spearmanr": 0.8889364898985929, |
| "eval_steps_per_second": 16.22, |
| "step": 950 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 3.263888888888889e-05, |
| "loss": 0.2778, |
| "step": 1000 |
| }, |
| { |
| "epoch": 2.78, |
| "eval_combined_score": 0.88330796900741, |
| "eval_loss": 0.49821072816848755, |
| "eval_pearson": 0.8849717556333476, |
| "eval_runtime": 6.4097, |
| "eval_samples_per_second": 234.021, |
| "eval_spearmanr": 0.8816441823814725, |
| "eval_steps_per_second": 14.665, |
| "step": 1000 |
| }, |
| { |
| "epoch": 2.92, |
| "eval_combined_score": 0.8847183157645039, |
| "eval_loss": 0.5095486044883728, |
| "eval_pearson": 0.8871448241460179, |
| "eval_runtime": 5.596, |
| "eval_samples_per_second": 268.047, |
| "eval_spearmanr": 0.8822918073829898, |
| "eval_steps_per_second": 16.798, |
| "step": 1050 |
| }, |
| { |
| "epoch": 3.06, |
| "eval_combined_score": 0.8874272464175201, |
| "eval_loss": 0.4730277955532074, |
| "eval_pearson": 0.8895831976333978, |
| "eval_runtime": 5.6819, |
| "eval_samples_per_second": 263.994, |
| "eval_spearmanr": 0.8852712952016424, |
| "eval_steps_per_second": 16.544, |
| "step": 1100 |
| }, |
| { |
| "epoch": 3.19, |
| "eval_combined_score": 0.886644981813258, |
| "eval_loss": 0.4820166826248169, |
| "eval_pearson": 0.8884315334781125, |
| "eval_runtime": 6.0521, |
| "eval_samples_per_second": 247.847, |
| "eval_spearmanr": 0.8848584301484034, |
| "eval_steps_per_second": 15.532, |
| "step": 1150 |
| }, |
| { |
| "epoch": 3.33, |
| "eval_combined_score": 0.8894357915147134, |
| "eval_loss": 0.47540178894996643, |
| "eval_pearson": 0.8913766457258069, |
| "eval_runtime": 5.6258, |
| "eval_samples_per_second": 266.631, |
| "eval_spearmanr": 0.88749493730362, |
| "eval_steps_per_second": 16.709, |
| "step": 1200 |
| }, |
| { |
| "epoch": 3.47, |
| "eval_combined_score": 0.8882337266236713, |
| "eval_loss": 0.47845378518104553, |
| "eval_pearson": 0.8898157282686552, |
| "eval_runtime": 5.9023, |
| "eval_samples_per_second": 254.137, |
| "eval_spearmanr": 0.8866517249786874, |
| "eval_steps_per_second": 15.926, |
| "step": 1250 |
| }, |
| { |
| "epoch": 3.61, |
| "eval_combined_score": 0.88903651674165, |
| "eval_loss": 0.4937942624092102, |
| "eval_pearson": 0.8910047139114365, |
| "eval_runtime": 6.1737, |
| "eval_samples_per_second": 242.964, |
| "eval_spearmanr": 0.8870683195718634, |
| "eval_steps_per_second": 15.226, |
| "step": 1300 |
| }, |
| { |
| "epoch": 3.75, |
| "eval_combined_score": 0.8901558681204902, |
| "eval_loss": 0.46190693974494934, |
| "eval_pearson": 0.8927358825114844, |
| "eval_runtime": 5.7562, |
| "eval_samples_per_second": 260.588, |
| "eval_spearmanr": 0.887575853729496, |
| "eval_steps_per_second": 16.33, |
| "step": 1350 |
| }, |
| { |
| "epoch": 3.89, |
| "eval_combined_score": 0.8889376492175258, |
| "eval_loss": 0.49563419818878174, |
| "eval_pearson": 0.8915857107166335, |
| "eval_runtime": 5.5793, |
| "eval_samples_per_second": 268.853, |
| "eval_spearmanr": 0.8862895877184183, |
| "eval_steps_per_second": 16.848, |
| "step": 1400 |
| }, |
| { |
| "epoch": 4.03, |
| "eval_combined_score": 0.8911112435797486, |
| "eval_loss": 0.4565975069999695, |
| "eval_pearson": 0.8938483143744439, |
| "eval_runtime": 6.0592, |
| "eval_samples_per_second": 247.559, |
| "eval_spearmanr": 0.8883741727850532, |
| "eval_steps_per_second": 15.514, |
| "step": 1450 |
| }, |
| { |
| "epoch": 4.17, |
| "learning_rate": 2.3958333333333334e-05, |
| "loss": 0.141, |
| "step": 1500 |
| }, |
| { |
| "epoch": 4.17, |
| "eval_combined_score": 0.890231579373417, |
| "eval_loss": 0.4669683873653412, |
| "eval_pearson": 0.8928537444959865, |
| "eval_runtime": 6.214, |
| "eval_samples_per_second": 241.389, |
| "eval_spearmanr": 0.8876094142508474, |
| "eval_steps_per_second": 15.127, |
| "step": 1500 |
| }, |
| { |
| "epoch": 4.31, |
| "eval_combined_score": 0.8917620927758609, |
| "eval_loss": 0.4545550048351288, |
| "eval_pearson": 0.8941284924896612, |
| "eval_runtime": 5.6167, |
| "eval_samples_per_second": 267.062, |
| "eval_spearmanr": 0.8893956930620608, |
| "eval_steps_per_second": 16.736, |
| "step": 1550 |
| }, |
| { |
| "epoch": 4.44, |
| "eval_combined_score": 0.8919076890360074, |
| "eval_loss": 0.46239447593688965, |
| "eval_pearson": 0.8939787505423513, |
| "eval_runtime": 6.1392, |
| "eval_samples_per_second": 244.33, |
| "eval_spearmanr": 0.8898366275296635, |
| "eval_steps_per_second": 15.311, |
| "step": 1600 |
| }, |
| { |
| "epoch": 4.58, |
| "eval_combined_score": 0.8905705624659631, |
| "eval_loss": 0.4560569226741791, |
| "eval_pearson": 0.8930309291971393, |
| "eval_runtime": 5.7715, |
| "eval_samples_per_second": 259.896, |
| "eval_spearmanr": 0.8881101957347869, |
| "eval_steps_per_second": 16.287, |
| "step": 1650 |
| }, |
| { |
| "epoch": 4.72, |
| "eval_combined_score": 0.89164078789547, |
| "eval_loss": 0.47078728675842285, |
| "eval_pearson": 0.8931105725859777, |
| "eval_runtime": 5.5689, |
| "eval_samples_per_second": 269.352, |
| "eval_spearmanr": 0.8901710032049623, |
| "eval_steps_per_second": 16.879, |
| "step": 1700 |
| }, |
| { |
| "epoch": 4.86, |
| "eval_combined_score": 0.8922518284784362, |
| "eval_loss": 0.5126345157623291, |
| "eval_pearson": 0.8948084725851928, |
| "eval_runtime": 5.5913, |
| "eval_samples_per_second": 268.275, |
| "eval_spearmanr": 0.8896951843716796, |
| "eval_steps_per_second": 16.812, |
| "step": 1750 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_combined_score": 0.8936765208721354, |
| "eval_loss": 0.4546430706977844, |
| "eval_pearson": 0.8960807975667862, |
| "eval_runtime": 5.6821, |
| "eval_samples_per_second": 263.988, |
| "eval_spearmanr": 0.8912722441774847, |
| "eval_steps_per_second": 16.543, |
| "step": 1800 |
| }, |
| { |
| "epoch": 5.14, |
| "eval_combined_score": 0.8930431248747293, |
| "eval_loss": 0.44920727610588074, |
| "eval_pearson": 0.8947890019960217, |
| "eval_runtime": 5.9173, |
| "eval_samples_per_second": 253.493, |
| "eval_spearmanr": 0.8912972477534367, |
| "eval_steps_per_second": 15.886, |
| "step": 1850 |
| }, |
| { |
| "epoch": 5.28, |
| "eval_combined_score": 0.894053370656966, |
| "eval_loss": 0.4471864700317383, |
| "eval_pearson": 0.8965503528124947, |
| "eval_runtime": 5.6396, |
| "eval_samples_per_second": 265.978, |
| "eval_spearmanr": 0.8915563885014371, |
| "eval_steps_per_second": 16.668, |
| "step": 1900 |
| }, |
| { |
| "epoch": 5.42, |
| "eval_combined_score": 0.8948116388641144, |
| "eval_loss": 0.470526784658432, |
| "eval_pearson": 0.8971476181305765, |
| "eval_runtime": 5.7724, |
| "eval_samples_per_second": 259.855, |
| "eval_spearmanr": 0.8924756595976523, |
| "eval_steps_per_second": 16.284, |
| "step": 1950 |
| }, |
| { |
| "epoch": 5.56, |
| "learning_rate": 1.527777777777778e-05, |
| "loss": 0.0844, |
| "step": 2000 |
| }, |
| { |
| "epoch": 5.56, |
| "eval_combined_score": 0.8922658461057144, |
| "eval_loss": 0.46638524532318115, |
| "eval_pearson": 0.8939377420957451, |
| "eval_runtime": 5.7907, |
| "eval_samples_per_second": 259.036, |
| "eval_spearmanr": 0.8905939501156837, |
| "eval_steps_per_second": 16.233, |
| "step": 2000 |
| }, |
| { |
| "epoch": 5.69, |
| "eval_combined_score": 0.8915829503407375, |
| "eval_loss": 0.49491751194000244, |
| "eval_pearson": 0.8941690530900368, |
| "eval_runtime": 5.6201, |
| "eval_samples_per_second": 266.9, |
| "eval_spearmanr": 0.8889968475914382, |
| "eval_steps_per_second": 16.726, |
| "step": 2050 |
| }, |
| { |
| "epoch": 5.83, |
| "eval_combined_score": 0.8940547207791253, |
| "eval_loss": 0.451761394739151, |
| "eval_pearson": 0.8965362985968673, |
| "eval_runtime": 6.0011, |
| "eval_samples_per_second": 249.956, |
| "eval_spearmanr": 0.8915731429613832, |
| "eval_steps_per_second": 15.664, |
| "step": 2100 |
| }, |
| { |
| "epoch": 5.97, |
| "eval_combined_score": 0.8923553049941013, |
| "eval_loss": 0.45326724648475647, |
| "eval_pearson": 0.8945634893619782, |
| "eval_runtime": 5.6075, |
| "eval_samples_per_second": 267.501, |
| "eval_spearmanr": 0.8901471206262244, |
| "eval_steps_per_second": 16.763, |
| "step": 2150 |
| }, |
| { |
| "epoch": 6.11, |
| "eval_combined_score": 0.8949367985553012, |
| "eval_loss": 0.44921526312828064, |
| "eval_pearson": 0.8970393518565526, |
| "eval_runtime": 5.5418, |
| "eval_samples_per_second": 270.668, |
| "eval_spearmanr": 0.8928342452540498, |
| "eval_steps_per_second": 16.962, |
| "step": 2200 |
| } |
| ], |
| "max_steps": 2880, |
| "num_train_epochs": 8, |
| "total_flos": 2311015204895232.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
