{ "best_metric": 4.342891693115234, "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-sim-ques/lstm/3/checkpoints/checkpoint-228960", "epoch": 0.025000606015738065, "eval_steps": 10, "global_step": 228960, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.999998362119627e-05, "loss": 10.8208, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.999161405248948e-05, "loss": 7.5517, "step": 512 }, { "epoch": 0.0, "learning_rate": 4.998322810497896e-05, "loss": 7.0563, "step": 1024 }, { "epoch": 0.0, "learning_rate": 4.997484215746844e-05, "loss": 6.99, "step": 1536 }, { "epoch": 0.0, "learning_rate": 4.996645620995792e-05, "loss": 6.9429, "step": 2048 }, { "epoch": 0.0, "learning_rate": 4.99580702624474e-05, "loss": 6.8728, "step": 2560 }, { "epoch": 0.0, "learning_rate": 4.994968431493688e-05, "loss": 6.7168, "step": 3072 }, { "epoch": 0.0, "learning_rate": 4.994129836742636e-05, "loss": 6.6108, "step": 3584 }, { "epoch": 0.0, "learning_rate": 4.993291241991584e-05, "loss": 6.5186, "step": 4096 }, { "epoch": 0.0, "learning_rate": 4.992452647240532e-05, "loss": 6.4481, "step": 4608 }, { "epoch": 0.0, "learning_rate": 4.99161405248948e-05, "loss": 6.391, "step": 5120 }, { "epoch": 0.0, "learning_rate": 4.990775457738428e-05, "loss": 6.348, "step": 5632 }, { "epoch": 0.0, "learning_rate": 4.989936862987376e-05, "loss": 6.282, "step": 6144 }, { "epoch": 0.0, "learning_rate": 4.989098268236324e-05, "loss": 6.2102, "step": 6656 }, { "epoch": 0.0, "learning_rate": 4.988259673485272e-05, "loss": 6.1618, "step": 7168 }, { "epoch": 0.0, "learning_rate": 4.98742107873422e-05, "loss": 6.104, "step": 7680 }, { "epoch": 0.0, "learning_rate": 4.986582483983168e-05, "loss": 6.0698, "step": 8192 }, { "epoch": 0.0, "learning_rate": 4.985743889232116e-05, "loss": 6.0137, "step": 8704 }, { "epoch": 0.0, "learning_rate": 4.98490857024181e-05, "loss": 5.9697, "step": 9216 }, { "epoch": 0.0, "learning_rate": 4.984069975490758e-05, "loss": 5.9382, "step": 9728 }, { "epoch": 0.0, "learning_rate": 4.983231380739706e-05, "loss": 5.8979, "step": 10240 }, { "epoch": 0.0, "learning_rate": 4.9823927859886547e-05, "loss": 5.867, "step": 10752 }, { "epoch": 0.0, "learning_rate": 4.9815541912376026e-05, "loss": 5.8406, "step": 11264 }, { "epoch": 0.0, "learning_rate": 4.9807155964865506e-05, "loss": 5.808, "step": 11776 }, { "epoch": 0.0, "learning_rate": 4.9798770017354986e-05, "loss": 5.7775, "step": 12288 }, { "epoch": 0.0, "learning_rate": 4.9790384069844466e-05, "loss": 5.7532, "step": 12800 }, { "epoch": 0.0, "learning_rate": 4.9782014501137675e-05, "loss": 5.7227, "step": 13312 }, { "epoch": 0.0, "learning_rate": 4.9773628553627155e-05, "loss": 5.6951, "step": 13824 }, { "epoch": 0.0, "learning_rate": 4.9765242606116635e-05, "loss": 5.6682, "step": 14336 }, { "epoch": 0.0, "learning_rate": 4.9756856658606115e-05, "loss": 5.6546, "step": 14848 }, { "epoch": 0.01, "learning_rate": 4.9748470711095595e-05, "loss": 5.6318, "step": 15360 }, { "epoch": 0.01, "learning_rate": 4.974008476358507e-05, "loss": 5.6101, "step": 15872 }, { "epoch": 0.01, "learning_rate": 4.9731715194878284e-05, "loss": 5.5869, "step": 16384 }, { "epoch": 0.01, "learning_rate": 4.9723329247367764e-05, "loss": 5.5685, "step": 16896 }, { "epoch": 0.01, "learning_rate": 4.9714943299857244e-05, "loss": 5.5597, "step": 17408 }, { "epoch": 0.01, "learning_rate": 4.9706557352346724e-05, "loss": 5.5485, "step": 17920 }, { "epoch": 0.01, "learning_rate": 4.969820416244367e-05, "loss": 5.5247, "step": 18432 }, { "epoch": 0.01, "learning_rate": 4.968981821493315e-05, "loss": 5.4856, "step": 18944 }, { "epoch": 0.01, "learning_rate": 4.968143226742263e-05, "loss": 5.4922, "step": 19456 }, { "epoch": 0.01, "learning_rate": 4.967304631991211e-05, "loss": 5.4567, "step": 19968 }, { "epoch": 0.01, "learning_rate": 4.966466037240159e-05, "loss": 5.4491, "step": 20480 }, { "epoch": 0.01, "learning_rate": 4.965627442489107e-05, "loss": 5.4399, "step": 20992 }, { "epoch": 0.01, "learning_rate": 4.964788847738054e-05, "loss": 5.4303, "step": 21504 }, { "epoch": 0.01, "learning_rate": 4.963950252987002e-05, "loss": 5.4086, "step": 22016 }, { "epoch": 0.01, "learning_rate": 4.96311165823595e-05, "loss": 5.4033, "step": 22528 }, { "epoch": 0.01, "learning_rate": 4.962273063484898e-05, "loss": 5.3974, "step": 23040 }, { "epoch": 0.01, "learning_rate": 4.961434468733847e-05, "loss": 5.3855, "step": 23552 }, { "epoch": 0.01, "learning_rate": 4.960595873982795e-05, "loss": 5.3738, "step": 24064 }, { "epoch": 0.01, "learning_rate": 4.959758917112116e-05, "loss": 5.3541, "step": 24576 }, { "epoch": 0.01, "learning_rate": 4.958920322361064e-05, "loss": 5.341, "step": 25088 }, { "epoch": 0.01, "learning_rate": 4.958081727610012e-05, "loss": 5.3443, "step": 25600 }, { "epoch": 0.01, "learning_rate": 4.95724313285896e-05, "loss": 5.3236, "step": 26112 }, { "epoch": 0.01, "learning_rate": 4.956407813868654e-05, "loss": 5.3126, "step": 26624 }, { "epoch": 0.01, "learning_rate": 4.9555692191176016e-05, "loss": 5.2938, "step": 27136 }, { "epoch": 0.01, "learning_rate": 4.9547306243665496e-05, "loss": 5.2978, "step": 27648 }, { "epoch": 0.01, "learning_rate": 4.9538920296154976e-05, "loss": 5.2791, "step": 28160 }, { "epoch": 0.01, "learning_rate": 4.9530534348644456e-05, "loss": 5.2928, "step": 28672 }, { "epoch": 0.01, "learning_rate": 4.9522148401133936e-05, "loss": 5.2641, "step": 29184 }, { "epoch": 0.01, "learning_rate": 4.951376245362342e-05, "loss": 5.2604, "step": 29696 }, { "epoch": 0.01, "learning_rate": 4.95053765061129e-05, "loss": 5.254, "step": 30208 }, { "epoch": 0.01, "learning_rate": 4.949700693740611e-05, "loss": 5.2305, "step": 30720 }, { "epoch": 0.01, "learning_rate": 4.948862098989559e-05, "loss": 5.2343, "step": 31232 }, { "epoch": 0.01, "learning_rate": 4.948023504238507e-05, "loss": 5.2187, "step": 31744 }, { "epoch": 0.01, "learning_rate": 4.947184909487455e-05, "loss": 5.2077, "step": 32256 }, { "epoch": 0.01, "learning_rate": 4.946346314736403e-05, "loss": 5.2039, "step": 32768 }, { "epoch": 0.01, "learning_rate": 4.945507719985351e-05, "loss": 5.2199, "step": 33280 }, { "epoch": 0.01, "learning_rate": 4.944669125234299e-05, "loss": 5.1943, "step": 33792 }, { "epoch": 0.01, "learning_rate": 4.943830530483247e-05, "loss": 5.1815, "step": 34304 }, { "epoch": 0.01, "learning_rate": 4.942993573612568e-05, "loss": 5.1627, "step": 34816 }, { "epoch": 0.01, "learning_rate": 4.942154978861516e-05, "loss": 5.1609, "step": 35328 }, { "epoch": 0.01, "learning_rate": 4.941316384110464e-05, "loss": 5.1744, "step": 35840 }, { "epoch": 0.01, "learning_rate": 4.940477789359412e-05, "loss": 5.171, "step": 36352 }, { "epoch": 0.01, "learning_rate": 4.9396408324887336e-05, "loss": 5.1446, "step": 36864 }, { "epoch": 0.01, "learning_rate": 4.9388022377376816e-05, "loss": 5.1604, "step": 37376 }, { "epoch": 0.01, "learning_rate": 4.9379652808670025e-05, "loss": 5.1524, "step": 37888 }, { "epoch": 0.01, "learning_rate": 4.9371266861159505e-05, "loss": 5.1377, "step": 38400 }, { "epoch": 0.01, "learning_rate": 4.9362880913648985e-05, "loss": 5.1178, "step": 38912 }, { "epoch": 0.01, "learning_rate": 4.9354494966138465e-05, "loss": 5.121, "step": 39424 }, { "epoch": 0.01, "learning_rate": 4.9346109018627945e-05, "loss": 5.1054, "step": 39936 }, { "epoch": 0.01, "learning_rate": 4.9337723071117425e-05, "loss": 5.1034, "step": 40448 }, { "epoch": 0.01, "learning_rate": 4.9329337123606905e-05, "loss": 5.1024, "step": 40960 }, { "epoch": 0.01, "learning_rate": 4.932095117609638e-05, "loss": 5.0926, "step": 41472 }, { "epoch": 0.01, "learning_rate": 4.9312581607389594e-05, "loss": 5.0958, "step": 41984 }, { "epoch": 0.01, "learning_rate": 4.9304195659879074e-05, "loss": 5.0889, "step": 42496 }, { "epoch": 0.01, "learning_rate": 4.9295809712368554e-05, "loss": 5.0684, "step": 43008 }, { "epoch": 0.01, "learning_rate": 4.9287423764858034e-05, "loss": 5.0709, "step": 43520 }, { "epoch": 0.01, "learning_rate": 4.927907057495498e-05, "loss": 5.071, "step": 44032 }, { "epoch": 0.01, "learning_rate": 4.927068462744446e-05, "loss": 5.0623, "step": 44544 }, { "epoch": 0.01, "learning_rate": 4.926229867993394e-05, "loss": 5.0511, "step": 45056 }, { "epoch": 0.01, "learning_rate": 4.925391273242342e-05, "loss": 5.0483, "step": 45568 }, { "epoch": 0.02, "learning_rate": 4.92455267849129e-05, "loss": 5.0465, "step": 46080 }, { "epoch": 0.02, "learning_rate": 4.923715721620611e-05, "loss": 5.0331, "step": 46592 }, { "epoch": 0.02, "learning_rate": 4.922877126869559e-05, "loss": 5.0366, "step": 47104 }, { "epoch": 0.02, "learning_rate": 4.922038532118507e-05, "loss": 5.0225, "step": 47616 }, { "epoch": 0.02, "learning_rate": 4.921199937367455e-05, "loss": 5.0247, "step": 48128 }, { "epoch": 0.02, "learning_rate": 4.920361342616403e-05, "loss": 5.0191, "step": 48640 }, { "epoch": 0.02, "learning_rate": 4.919522747865351e-05, "loss": 5.0115, "step": 49152 }, { "epoch": 0.02, "learning_rate": 4.918684153114299e-05, "loss": 4.9925, "step": 49664 }, { "epoch": 0.02, "learning_rate": 4.917845558363247e-05, "loss": 4.9902, "step": 50176 }, { "epoch": 0.02, "learning_rate": 4.9170086014925676e-05, "loss": 4.9897, "step": 50688 }, { "epoch": 0.02, "learning_rate": 4.9161700067415156e-05, "loss": 5.0017, "step": 51200 }, { "epoch": 0.02, "learning_rate": 4.9153314119904636e-05, "loss": 4.9876, "step": 51712 }, { "epoch": 0.02, "learning_rate": 4.914494455119785e-05, "loss": 4.9748, "step": 52224 }, { "epoch": 0.02, "learning_rate": 4.9136558603687325e-05, "loss": 4.9678, "step": 52736 }, { "epoch": 0.02, "learning_rate": 4.9128172656176805e-05, "loss": 4.9723, "step": 53248 }, { "epoch": 0.02, "learning_rate": 4.9119786708666285e-05, "loss": 4.9466, "step": 53760 }, { "epoch": 0.02, "learning_rate": 4.91114171399595e-05, "loss": 4.9493, "step": 54272 }, { "epoch": 0.02, "learning_rate": 4.910303119244898e-05, "loss": 4.9458, "step": 54784 }, { "epoch": 0.02, "learning_rate": 4.909464524493846e-05, "loss": 4.9488, "step": 55296 }, { "epoch": 0.02, "learning_rate": 4.908625929742794e-05, "loss": 4.9383, "step": 55808 }, { "epoch": 0.02, "learning_rate": 4.907788972872115e-05, "loss": 4.9456, "step": 56320 }, { "epoch": 0.02, "learning_rate": 4.906950378121063e-05, "loss": 4.9232, "step": 56832 }, { "epoch": 0.02, "learning_rate": 4.906111783370011e-05, "loss": 4.9442, "step": 57344 }, { "epoch": 0.02, "learning_rate": 4.9052748264993326e-05, "loss": 4.924, "step": 57856 }, { "epoch": 0.02, "learning_rate": 4.90443623174828e-05, "loss": 4.9267, "step": 58368 }, { "epoch": 0.02, "learning_rate": 4.903597636997228e-05, "loss": 4.926, "step": 58880 }, { "epoch": 0.02, "learning_rate": 4.902759042246176e-05, "loss": 4.9258, "step": 59392 }, { "epoch": 0.02, "learning_rate": 4.901920447495124e-05, "loss": 4.9119, "step": 59904 }, { "epoch": 0.02, "learning_rate": 4.901081852744072e-05, "loss": 4.9, "step": 60416 }, { "epoch": 0.02, "learning_rate": 4.9002448958733935e-05, "loss": 4.9076, "step": 60928 }, { "epoch": 0.02, "learning_rate": 4.8994063011223415e-05, "loss": 4.8999, "step": 61440 }, { "epoch": 0.02, "learning_rate": 4.8985677063712895e-05, "loss": 4.9007, "step": 61952 }, { "epoch": 0.02, "learning_rate": 4.8977291116202375e-05, "loss": 4.8958, "step": 62464 }, { "epoch": 0.02, "learning_rate": 4.8968921547495584e-05, "loss": 4.8834, "step": 62976 }, { "epoch": 0.02, "learning_rate": 4.8960535599985064e-05, "loss": 4.8963, "step": 63488 }, { "epoch": 0.02, "learning_rate": 4.8952149652474544e-05, "loss": 4.8753, "step": 64000 }, { "epoch": 0.02, "learning_rate": 4.8943763704964024e-05, "loss": 4.8706, "step": 64512 }, { "epoch": 0.02, "learning_rate": 4.8935377757453504e-05, "loss": 4.8648, "step": 65024 }, { "epoch": 0.02, "learning_rate": 4.892700818874671e-05, "loss": 4.8713, "step": 65536 }, { "epoch": 0.02, "learning_rate": 4.891862224123619e-05, "loss": 4.8561, "step": 66048 }, { "epoch": 0.02, "learning_rate": 4.891023629372567e-05, "loss": 4.8744, "step": 66560 }, { "epoch": 0.02, "learning_rate": 4.890185034621515e-05, "loss": 4.8645, "step": 67072 }, { "epoch": 0.02, "learning_rate": 4.889346439870464e-05, "loss": 4.8527, "step": 67584 }, { "epoch": 0.02, "learning_rate": 4.888507845119412e-05, "loss": 4.8446, "step": 68096 }, { "epoch": 0.02, "learning_rate": 4.88766925036836e-05, "loss": 4.8467, "step": 68608 }, { "epoch": 0.02, "learning_rate": 4.886830655617308e-05, "loss": 4.8441, "step": 69120 }, { "epoch": 0.02, "learning_rate": 4.885993698746629e-05, "loss": 4.8461, "step": 69632 }, { "epoch": 0.02, "learning_rate": 4.885155103995577e-05, "loss": 4.8443, "step": 70144 }, { "epoch": 0.02, "learning_rate": 4.884316509244525e-05, "loss": 4.8407, "step": 70656 }, { "epoch": 0.02, "learning_rate": 4.883477914493473e-05, "loss": 4.8341, "step": 71168 }, { "epoch": 0.02, "learning_rate": 4.882639319742421e-05, "loss": 4.8285, "step": 71680 }, { "epoch": 0.02, "learning_rate": 4.881802362871742e-05, "loss": 4.8168, "step": 72192 }, { "epoch": 0.02, "learning_rate": 4.88096376812069e-05, "loss": 4.8147, "step": 72704 }, { "epoch": 0.02, "learning_rate": 4.880125173369638e-05, "loss": 4.8217, "step": 73216 }, { "epoch": 0.02, "learning_rate": 4.879286578618586e-05, "loss": 4.8012, "step": 73728 }, { "epoch": 0.02, "learning_rate": 4.878449621747907e-05, "loss": 4.8147, "step": 74240 }, { "epoch": 0.02, "learning_rate": 4.877611026996855e-05, "loss": 4.8137, "step": 74752 }, { "epoch": 0.02, "learning_rate": 4.876772432245803e-05, "loss": 4.8021, "step": 75264 }, { "epoch": 0.02, "learning_rate": 4.875933837494751e-05, "loss": 4.8081, "step": 75776 }, { "epoch": 0.02, "learning_rate": 4.8750952427436986e-05, "loss": 4.8041, "step": 76288 }, { "epoch": 0.03, "eval_loss": 4.758626461029053, "eval_runtime": 286.4264, "eval_samples_per_second": 1332.248, "eval_steps_per_second": 41.634, "step": 76320 }, { "epoch": 1.0, "learning_rate": 4.87425828587302e-05, "loss": 4.7813, "step": 76800 }, { "epoch": 1.0, "learning_rate": 4.873419691121968e-05, "loss": 4.7877, "step": 77312 }, { "epoch": 1.0, "learning_rate": 4.872581096370916e-05, "loss": 4.7998, "step": 77824 }, { "epoch": 1.0, "learning_rate": 4.871744139500237e-05, "loss": 4.7809, "step": 78336 }, { "epoch": 1.0, "learning_rate": 4.870905544749185e-05, "loss": 4.7866, "step": 78848 }, { "epoch": 1.0, "learning_rate": 4.870066949998133e-05, "loss": 4.7755, "step": 79360 }, { "epoch": 1.0, "learning_rate": 4.869228355247081e-05, "loss": 4.7759, "step": 79872 }, { "epoch": 1.0, "learning_rate": 4.868389760496029e-05, "loss": 4.7649, "step": 80384 }, { "epoch": 1.0, "learning_rate": 4.867551165744977e-05, "loss": 4.7728, "step": 80896 }, { "epoch": 1.0, "learning_rate": 4.866712570993925e-05, "loss": 4.7604, "step": 81408 }, { "epoch": 1.0, "learning_rate": 4.865873976242873e-05, "loss": 4.7825, "step": 81920 }, { "epoch": 1.0, "learning_rate": 4.865037019372194e-05, "loss": 4.7727, "step": 82432 }, { "epoch": 1.0, "learning_rate": 4.8642000625015155e-05, "loss": 4.7532, "step": 82944 }, { "epoch": 1.0, "learning_rate": 4.8633614677504635e-05, "loss": 4.7504, "step": 83456 }, { "epoch": 1.0, "learning_rate": 4.862522872999411e-05, "loss": 4.7474, "step": 83968 }, { "epoch": 1.0, "learning_rate": 4.861684278248359e-05, "loss": 4.748, "step": 84480 }, { "epoch": 1.0, "learning_rate": 4.860845683497307e-05, "loss": 4.7429, "step": 84992 }, { "epoch": 1.0, "learning_rate": 4.860007088746255e-05, "loss": 4.7417, "step": 85504 }, { "epoch": 1.0, "learning_rate": 4.859168493995203e-05, "loss": 4.7375, "step": 86016 }, { "epoch": 1.0, "learning_rate": 4.8583315371245244e-05, "loss": 4.7482, "step": 86528 }, { "epoch": 1.0, "learning_rate": 4.8574929423734724e-05, "loss": 4.7383, "step": 87040 }, { "epoch": 1.0, "learning_rate": 4.8566543476224204e-05, "loss": 4.7433, "step": 87552 }, { "epoch": 1.0, "learning_rate": 4.8558157528713684e-05, "loss": 4.7383, "step": 88064 }, { "epoch": 1.0, "learning_rate": 4.8549771581203164e-05, "loss": 4.7438, "step": 88576 }, { "epoch": 1.0, "learning_rate": 4.854140201249637e-05, "loss": 4.7214, "step": 89088 }, { "epoch": 1.0, "learning_rate": 4.853301606498585e-05, "loss": 4.7208, "step": 89600 }, { "epoch": 1.0, "learning_rate": 4.852463011747533e-05, "loss": 4.7157, "step": 90112 }, { "epoch": 1.0, "learning_rate": 4.851624416996481e-05, "loss": 4.7154, "step": 90624 }, { "epoch": 1.0, "learning_rate": 4.850787460125802e-05, "loss": 4.7162, "step": 91136 }, { "epoch": 1.01, "learning_rate": 4.84994886537475e-05, "loss": 4.7131, "step": 91648 }, { "epoch": 1.01, "learning_rate": 4.849110270623698e-05, "loss": 4.714, "step": 92160 }, { "epoch": 1.01, "learning_rate": 4.84827331375302e-05, "loss": 4.7097, "step": 92672 }, { "epoch": 1.01, "learning_rate": 4.847434719001968e-05, "loss": 4.7, "step": 93184 }, { "epoch": 1.01, "learning_rate": 4.846596124250916e-05, "loss": 4.7137, "step": 93696 }, { "epoch": 1.01, "learning_rate": 4.845757529499864e-05, "loss": 4.7135, "step": 94208 }, { "epoch": 1.01, "learning_rate": 4.844920572629185e-05, "loss": 4.699, "step": 94720 }, { "epoch": 1.01, "learning_rate": 4.844081977878133e-05, "loss": 4.6831, "step": 95232 }, { "epoch": 1.01, "learning_rate": 4.843243383127081e-05, "loss": 4.6972, "step": 95744 }, { "epoch": 1.01, "learning_rate": 4.842404788376029e-05, "loss": 4.6758, "step": 96256 }, { "epoch": 1.01, "learning_rate": 4.8415678315053496e-05, "loss": 4.6856, "step": 96768 }, { "epoch": 1.01, "learning_rate": 4.8407292367542976e-05, "loss": 4.6852, "step": 97280 }, { "epoch": 1.01, "learning_rate": 4.8398906420032456e-05, "loss": 4.6888, "step": 97792 }, { "epoch": 1.01, "learning_rate": 4.8390520472521936e-05, "loss": 4.6778, "step": 98304 }, { "epoch": 1.01, "learning_rate": 4.838215090381515e-05, "loss": 4.6829, "step": 98816 }, { "epoch": 1.01, "learning_rate": 4.837376495630463e-05, "loss": 4.6848, "step": 99328 }, { "epoch": 1.01, "learning_rate": 4.836537900879411e-05, "loss": 4.6784, "step": 99840 }, { "epoch": 1.01, "learning_rate": 4.835699306128359e-05, "loss": 4.6795, "step": 100352 }, { "epoch": 1.01, "learning_rate": 4.834860711377307e-05, "loss": 4.6631, "step": 100864 }, { "epoch": 1.01, "learning_rate": 4.834022116626255e-05, "loss": 4.6638, "step": 101376 }, { "epoch": 1.01, "learning_rate": 4.833183521875203e-05, "loss": 4.6732, "step": 101888 }, { "epoch": 1.01, "learning_rate": 4.832344927124151e-05, "loss": 4.6624, "step": 102400 }, { "epoch": 1.01, "learning_rate": 4.831506332373099e-05, "loss": 4.6629, "step": 102912 }, { "epoch": 1.01, "learning_rate": 4.83066937550242e-05, "loss": 4.6447, "step": 103424 }, { "epoch": 1.01, "learning_rate": 4.829832418631741e-05, "loss": 4.6567, "step": 103936 }, { "epoch": 1.01, "learning_rate": 4.828993823880689e-05, "loss": 4.6457, "step": 104448 }, { "epoch": 1.01, "learning_rate": 4.828155229129637e-05, "loss": 4.6632, "step": 104960 }, { "epoch": 1.01, "learning_rate": 4.8273166343785856e-05, "loss": 4.6467, "step": 105472 }, { "epoch": 1.01, "learning_rate": 4.8264780396275336e-05, "loss": 4.6526, "step": 105984 }, { "epoch": 1.01, "learning_rate": 4.8256394448764816e-05, "loss": 4.6516, "step": 106496 }, { "epoch": 1.01, "learning_rate": 4.8248008501254296e-05, "loss": 4.6294, "step": 107008 }, { "epoch": 1.01, "learning_rate": 4.823962255374377e-05, "loss": 4.6426, "step": 107520 }, { "epoch": 1.01, "learning_rate": 4.823123660623325e-05, "loss": 4.6403, "step": 108032 }, { "epoch": 1.01, "learning_rate": 4.8222883416330194e-05, "loss": 4.6263, "step": 108544 }, { "epoch": 1.01, "learning_rate": 4.8214497468819674e-05, "loss": 4.6309, "step": 109056 }, { "epoch": 1.01, "learning_rate": 4.8206111521309154e-05, "loss": 4.6491, "step": 109568 }, { "epoch": 1.01, "learning_rate": 4.8197725573798634e-05, "loss": 4.634, "step": 110080 }, { "epoch": 1.01, "learning_rate": 4.8189339626288114e-05, "loss": 4.6195, "step": 110592 }, { "epoch": 1.01, "learning_rate": 4.8180953678777594e-05, "loss": 4.6154, "step": 111104 }, { "epoch": 1.01, "learning_rate": 4.8172567731267074e-05, "loss": 4.6183, "step": 111616 }, { "epoch": 1.01, "learning_rate": 4.8164181783756554e-05, "loss": 4.6367, "step": 112128 }, { "epoch": 1.01, "learning_rate": 4.8155795836246034e-05, "loss": 4.6356, "step": 112640 }, { "epoch": 1.01, "learning_rate": 4.814742626753924e-05, "loss": 4.6174, "step": 113152 }, { "epoch": 1.01, "learning_rate": 4.813904032002872e-05, "loss": 4.6338, "step": 113664 }, { "epoch": 1.01, "learning_rate": 4.81306543725182e-05, "loss": 4.6322, "step": 114176 }, { "epoch": 1.01, "learning_rate": 4.812226842500768e-05, "loss": 4.6221, "step": 114688 }, { "epoch": 1.01, "learning_rate": 4.811389885630089e-05, "loss": 4.6076, "step": 115200 }, { "epoch": 1.01, "learning_rate": 4.810551290879037e-05, "loss": 4.6211, "step": 115712 }, { "epoch": 1.01, "learning_rate": 4.809712696127985e-05, "loss": 4.6081, "step": 116224 }, { "epoch": 1.01, "learning_rate": 4.808874101376933e-05, "loss": 4.6062, "step": 116736 }, { "epoch": 1.01, "learning_rate": 4.808037144506255e-05, "loss": 4.6106, "step": 117248 }, { "epoch": 1.01, "learning_rate": 4.807198549755203e-05, "loss": 4.6023, "step": 117760 }, { "epoch": 1.01, "learning_rate": 4.806359955004151e-05, "loss": 4.6182, "step": 118272 }, { "epoch": 1.01, "learning_rate": 4.805521360253099e-05, "loss": 4.6048, "step": 118784 }, { "epoch": 1.01, "learning_rate": 4.8046844033824197e-05, "loss": 4.5969, "step": 119296 }, { "epoch": 1.01, "learning_rate": 4.8038458086313677e-05, "loss": 4.6024, "step": 119808 }, { "epoch": 1.01, "learning_rate": 4.8030072138803156e-05, "loss": 4.6016, "step": 120320 }, { "epoch": 1.01, "learning_rate": 4.8021686191292636e-05, "loss": 4.5994, "step": 120832 }, { "epoch": 1.01, "learning_rate": 4.8013300243782116e-05, "loss": 4.5951, "step": 121344 }, { "epoch": 1.01, "learning_rate": 4.8004914296271596e-05, "loss": 4.5898, "step": 121856 }, { "epoch": 1.02, "learning_rate": 4.7996544727564805e-05, "loss": 4.6004, "step": 122368 }, { "epoch": 1.02, "learning_rate": 4.7988158780054285e-05, "loss": 4.5869, "step": 122880 }, { "epoch": 1.02, "learning_rate": 4.7979772832543765e-05, "loss": 4.5927, "step": 123392 }, { "epoch": 1.02, "learning_rate": 4.7971386885033245e-05, "loss": 4.5799, "step": 123904 }, { "epoch": 1.02, "learning_rate": 4.7963000937522725e-05, "loss": 4.5923, "step": 124416 }, { "epoch": 1.02, "learning_rate": 4.795461499001221e-05, "loss": 4.5891, "step": 124928 }, { "epoch": 1.02, "learning_rate": 4.794624542130542e-05, "loss": 4.5831, "step": 125440 }, { "epoch": 1.02, "learning_rate": 4.79378594737949e-05, "loss": 4.5682, "step": 125952 }, { "epoch": 1.02, "learning_rate": 4.792947352628438e-05, "loss": 4.5742, "step": 126464 }, { "epoch": 1.02, "learning_rate": 4.792108757877386e-05, "loss": 4.5779, "step": 126976 }, { "epoch": 1.02, "learning_rate": 4.791270163126334e-05, "loss": 4.5878, "step": 127488 }, { "epoch": 1.02, "learning_rate": 4.790431568375282e-05, "loss": 4.5793, "step": 128000 }, { "epoch": 1.02, "learning_rate": 4.78959297362423e-05, "loss": 4.5694, "step": 128512 }, { "epoch": 1.02, "learning_rate": 4.7887543788731774e-05, "loss": 4.5646, "step": 129024 }, { "epoch": 1.02, "learning_rate": 4.7879157841221254e-05, "loss": 4.5723, "step": 129536 }, { "epoch": 1.02, "learning_rate": 4.787078827251447e-05, "loss": 4.5463, "step": 130048 }, { "epoch": 1.02, "learning_rate": 4.786240232500395e-05, "loss": 4.5589, "step": 130560 }, { "epoch": 1.02, "learning_rate": 4.7854032756297166e-05, "loss": 4.5595, "step": 131072 }, { "epoch": 1.02, "learning_rate": 4.7845646808786646e-05, "loss": 4.5606, "step": 131584 }, { "epoch": 1.02, "learning_rate": 4.7837260861276126e-05, "loss": 4.5527, "step": 132096 }, { "epoch": 1.02, "learning_rate": 4.7828874913765606e-05, "loss": 4.562, "step": 132608 }, { "epoch": 1.02, "learning_rate": 4.782048896625508e-05, "loss": 4.5503, "step": 133120 }, { "epoch": 1.02, "learning_rate": 4.781210301874456e-05, "loss": 4.5702, "step": 133632 }, { "epoch": 1.02, "learning_rate": 4.7803733450037775e-05, "loss": 4.5539, "step": 134144 }, { "epoch": 1.02, "learning_rate": 4.779534750252725e-05, "loss": 4.5517, "step": 134656 }, { "epoch": 1.02, "learning_rate": 4.778696155501673e-05, "loss": 4.5688, "step": 135168 }, { "epoch": 1.02, "learning_rate": 4.777857560750621e-05, "loss": 4.5643, "step": 135680 }, { "epoch": 1.02, "learning_rate": 4.777018965999569e-05, "loss": 4.55, "step": 136192 }, { "epoch": 1.02, "learning_rate": 4.7761820091288903e-05, "loss": 4.5431, "step": 136704 }, { "epoch": 1.02, "learning_rate": 4.7753434143778383e-05, "loss": 4.5575, "step": 137216 }, { "epoch": 1.02, "learning_rate": 4.774504819626786e-05, "loss": 4.5457, "step": 137728 }, { "epoch": 1.02, "learning_rate": 4.773666224875734e-05, "loss": 4.5458, "step": 138240 }, { "epoch": 1.02, "learning_rate": 4.772827630124682e-05, "loss": 4.5536, "step": 138752 }, { "epoch": 1.02, "learning_rate": 4.77198903537363e-05, "loss": 4.5387, "step": 139264 }, { "epoch": 1.02, "learning_rate": 4.771150440622578e-05, "loss": 4.5533, "step": 139776 }, { "epoch": 1.02, "learning_rate": 4.770313483751899e-05, "loss": 4.5362, "step": 140288 }, { "epoch": 1.02, "learning_rate": 4.769474889000847e-05, "loss": 4.5384, "step": 140800 }, { "epoch": 1.02, "learning_rate": 4.768636294249795e-05, "loss": 4.5282, "step": 141312 }, { "epoch": 1.02, "learning_rate": 4.767797699498743e-05, "loss": 4.5392, "step": 141824 }, { "epoch": 1.02, "learning_rate": 4.766959104747691e-05, "loss": 4.5298, "step": 142336 }, { "epoch": 1.02, "learning_rate": 4.766120509996639e-05, "loss": 4.5454, "step": 142848 }, { "epoch": 1.02, "learning_rate": 4.76528355312596e-05, "loss": 4.5423, "step": 143360 }, { "epoch": 1.02, "learning_rate": 4.764444958374909e-05, "loss": 4.5325, "step": 143872 }, { "epoch": 1.02, "learning_rate": 4.763606363623857e-05, "loss": 4.5164, "step": 144384 }, { "epoch": 1.02, "learning_rate": 4.762767768872805e-05, "loss": 4.5338, "step": 144896 }, { "epoch": 1.02, "learning_rate": 4.761929174121753e-05, "loss": 4.5266, "step": 145408 }, { "epoch": 1.02, "learning_rate": 4.761090579370701e-05, "loss": 4.5338, "step": 145920 }, { "epoch": 1.02, "learning_rate": 4.760251984619649e-05, "loss": 4.5319, "step": 146432 }, { "epoch": 1.02, "learning_rate": 4.759413389868596e-05, "loss": 4.5284, "step": 146944 }, { "epoch": 1.02, "learning_rate": 4.758574795117544e-05, "loss": 4.5262, "step": 147456 }, { "epoch": 1.02, "learning_rate": 4.757736200366492e-05, "loss": 4.5293, "step": 147968 }, { "epoch": 1.02, "learning_rate": 4.75689760561544e-05, "loss": 4.5123, "step": 148480 }, { "epoch": 1.02, "learning_rate": 4.756060648744761e-05, "loss": 4.5139, "step": 148992 }, { "epoch": 1.02, "learning_rate": 4.755222053993709e-05, "loss": 4.5219, "step": 149504 }, { "epoch": 1.02, "learning_rate": 4.754383459242657e-05, "loss": 4.509, "step": 150016 }, { "epoch": 1.02, "learning_rate": 4.753544864491605e-05, "loss": 4.5175, "step": 150528 }, { "epoch": 1.02, "learning_rate": 4.7527062697405536e-05, "loss": 4.5259, "step": 151040 }, { "epoch": 1.02, "learning_rate": 4.751870950750248e-05, "loss": 4.5115, "step": 151552 }, { "epoch": 1.02, "learning_rate": 4.751032355999196e-05, "loss": 4.5134, "step": 152064 }, { "epoch": 1.02, "learning_rate": 4.7501937612481435e-05, "loss": 4.5148, "step": 152576 }, { "epoch": 1.03, "eval_loss": 4.477108001708984, "eval_runtime": 284.887, "eval_samples_per_second": 1339.447, "eval_steps_per_second": 41.859, "step": 152640 }, { "epoch": 0.0, "learning_rate": 4.7493551664970914e-05, "loss": 4.4986, "step": 153088 }, { "epoch": 0.0, "learning_rate": 4.7485165717460394e-05, "loss": 4.5038, "step": 153600 }, { "epoch": 0.0, "learning_rate": 4.7476779769949874e-05, "loss": 4.5148, "step": 154112 }, { "epoch": 0.0, "learning_rate": 4.7468393822439354e-05, "loss": 4.5037, "step": 154624 }, { "epoch": 0.0, "learning_rate": 4.7460007874928834e-05, "loss": 4.5096, "step": 155136 }, { "epoch": 0.0, "learning_rate": 4.7451621927418314e-05, "loss": 4.4938, "step": 155648 }, { "epoch": 0.0, "learning_rate": 4.7443235979907794e-05, "loss": 4.5066, "step": 156160 }, { "epoch": 0.0, "learning_rate": 4.7434850032397274e-05, "loss": 4.4883, "step": 156672 }, { "epoch": 0.0, "learning_rate": 4.7426464084886754e-05, "loss": 4.5024, "step": 157184 }, { "epoch": 0.0, "learning_rate": 4.7418078137376234e-05, "loss": 4.4958, "step": 157696 }, { "epoch": 0.0, "learning_rate": 4.740969218986572e-05, "loss": 4.5084, "step": 158208 }, { "epoch": 0.0, "learning_rate": 4.74013062423552e-05, "loss": 4.5101, "step": 158720 }, { "epoch": 0.0, "learning_rate": 4.739292029484468e-05, "loss": 4.4856, "step": 159232 }, { "epoch": 0.0, "learning_rate": 4.7384534347334154e-05, "loss": 4.4903, "step": 159744 }, { "epoch": 0.0, "learning_rate": 4.7376148399823634e-05, "loss": 4.4916, "step": 160256 }, { "epoch": 0.0, "learning_rate": 4.7367762452313114e-05, "loss": 4.485, "step": 160768 }, { "epoch": 0.0, "learning_rate": 4.7359376504802594e-05, "loss": 4.4848, "step": 161280 }, { "epoch": 0.0, "learning_rate": 4.73510069360958e-05, "loss": 4.4828, "step": 161792 }, { "epoch": 0.0, "learning_rate": 4.734262098858528e-05, "loss": 4.4803, "step": 162304 }, { "epoch": 0.0, "learning_rate": 4.73342514198785e-05, "loss": 4.5015, "step": 162816 }, { "epoch": 0.0, "learning_rate": 4.732586547236797e-05, "loss": 4.4824, "step": 163328 }, { "epoch": 0.0, "learning_rate": 4.731747952485746e-05, "loss": 4.4945, "step": 163840 }, { "epoch": 0.0, "learning_rate": 4.730909357734694e-05, "loss": 4.4892, "step": 164352 }, { "epoch": 0.0, "learning_rate": 4.730070762983642e-05, "loss": 4.4977, "step": 164864 }, { "epoch": 0.0, "learning_rate": 4.72923216823259e-05, "loss": 4.4709, "step": 165376 }, { "epoch": 0.0, "learning_rate": 4.728393573481538e-05, "loss": 4.4788, "step": 165888 }, { "epoch": 0.0, "learning_rate": 4.727554978730486e-05, "loss": 4.4703, "step": 166400 }, { "epoch": 0.0, "learning_rate": 4.726716383979434e-05, "loss": 4.4774, "step": 166912 }, { "epoch": 0.0, "learning_rate": 4.725879427108755e-05, "loss": 4.4697, "step": 167424 }, { "epoch": 0.01, "learning_rate": 4.725040832357703e-05, "loss": 4.4731, "step": 167936 }, { "epoch": 0.01, "learning_rate": 4.724202237606651e-05, "loss": 4.4815, "step": 168448 }, { "epoch": 0.01, "learning_rate": 4.723363642855599e-05, "loss": 4.4742, "step": 168960 }, { "epoch": 0.01, "learning_rate": 4.722525048104547e-05, "loss": 4.4632, "step": 169472 }, { "epoch": 0.01, "learning_rate": 4.7216880912338676e-05, "loss": 4.4805, "step": 169984 }, { "epoch": 0.01, "learning_rate": 4.720851134363189e-05, "loss": 4.4744, "step": 170496 }, { "epoch": 0.01, "learning_rate": 4.720012539612137e-05, "loss": 4.4696, "step": 171008 }, { "epoch": 0.01, "learning_rate": 4.719173944861085e-05, "loss": 4.457, "step": 171520 }, { "epoch": 0.01, "learning_rate": 4.718335350110033e-05, "loss": 4.4666, "step": 172032 }, { "epoch": 0.01, "learning_rate": 4.717496755358981e-05, "loss": 4.4495, "step": 172544 }, { "epoch": 0.01, "learning_rate": 4.716658160607929e-05, "loss": 4.4621, "step": 173056 }, { "epoch": 0.01, "learning_rate": 4.715819565856877e-05, "loss": 4.4566, "step": 173568 }, { "epoch": 0.01, "learning_rate": 4.714980971105825e-05, "loss": 4.4638, "step": 174080 }, { "epoch": 0.01, "learning_rate": 4.714144014235146e-05, "loss": 4.4578, "step": 174592 }, { "epoch": 0.01, "learning_rate": 4.713305419484094e-05, "loss": 4.4602, "step": 175104 }, { "epoch": 0.01, "learning_rate": 4.712468462613415e-05, "loss": 4.4652, "step": 175616 }, { "epoch": 0.01, "learning_rate": 4.711629867862363e-05, "loss": 4.4606, "step": 176128 }, { "epoch": 0.01, "learning_rate": 4.710791273111311e-05, "loss": 4.4631, "step": 176640 }, { "epoch": 0.01, "learning_rate": 4.709952678360259e-05, "loss": 4.4449, "step": 177152 }, { "epoch": 0.01, "learning_rate": 4.709114083609208e-05, "loss": 4.4445, "step": 177664 }, { "epoch": 0.01, "learning_rate": 4.708275488858156e-05, "loss": 4.4606, "step": 178176 }, { "epoch": 0.01, "learning_rate": 4.7074385319874766e-05, "loss": 4.4486, "step": 178688 }, { "epoch": 0.01, "learning_rate": 4.7065999372364246e-05, "loss": 4.4471, "step": 179200 }, { "epoch": 0.01, "learning_rate": 4.7057613424853726e-05, "loss": 4.4352, "step": 179712 }, { "epoch": 0.01, "learning_rate": 4.7049227477343206e-05, "loss": 4.4478, "step": 180224 }, { "epoch": 0.01, "learning_rate": 4.7040841529832686e-05, "loss": 4.4333, "step": 180736 }, { "epoch": 0.01, "learning_rate": 4.703245558232216e-05, "loss": 4.4524, "step": 181248 }, { "epoch": 0.01, "learning_rate": 4.7024102392419104e-05, "loss": 4.4387, "step": 181760 }, { "epoch": 0.01, "learning_rate": 4.7015716444908584e-05, "loss": 4.4471, "step": 182272 }, { "epoch": 0.01, "learning_rate": 4.7007330497398064e-05, "loss": 4.4439, "step": 182784 }, { "epoch": 0.01, "learning_rate": 4.6998944549887544e-05, "loss": 4.4278, "step": 183296 }, { "epoch": 0.01, "learning_rate": 4.699055860237703e-05, "loss": 4.4364, "step": 183808 }, { "epoch": 0.01, "learning_rate": 4.698217265486651e-05, "loss": 4.4367, "step": 184320 }, { "epoch": 0.01, "learning_rate": 4.6973786707355983e-05, "loss": 4.4253, "step": 184832 }, { "epoch": 0.01, "learning_rate": 4.6965400759845463e-05, "loss": 4.4348, "step": 185344 }, { "epoch": 0.01, "learning_rate": 4.6957014812334943e-05, "loss": 4.4414, "step": 185856 }, { "epoch": 0.01, "learning_rate": 4.694862886482442e-05, "loss": 4.4366, "step": 186368 }, { "epoch": 0.01, "learning_rate": 4.69402429173139e-05, "loss": 4.4252, "step": 186880 }, { "epoch": 0.01, "learning_rate": 4.693187334860711e-05, "loss": 4.4169, "step": 187392 }, { "epoch": 0.01, "learning_rate": 4.692348740109659e-05, "loss": 4.423, "step": 187904 }, { "epoch": 0.01, "learning_rate": 4.691510145358607e-05, "loss": 4.4393, "step": 188416 }, { "epoch": 0.01, "learning_rate": 4.690671550607555e-05, "loss": 4.4429, "step": 188928 }, { "epoch": 0.01, "learning_rate": 4.689832955856503e-05, "loss": 4.4267, "step": 189440 }, { "epoch": 0.01, "learning_rate": 4.688994361105451e-05, "loss": 4.4402, "step": 189952 }, { "epoch": 0.01, "learning_rate": 4.6881557663544e-05, "loss": 4.4399, "step": 190464 }, { "epoch": 0.01, "learning_rate": 4.687317171603348e-05, "loss": 4.4355, "step": 190976 }, { "epoch": 0.01, "learning_rate": 4.686480214732669e-05, "loss": 4.4138, "step": 191488 }, { "epoch": 0.01, "learning_rate": 4.685641619981617e-05, "loss": 4.4325, "step": 192000 }, { "epoch": 0.01, "learning_rate": 4.684803025230565e-05, "loss": 4.4263, "step": 192512 }, { "epoch": 0.01, "learning_rate": 4.683966068359886e-05, "loss": 4.4158, "step": 193024 }, { "epoch": 0.01, "learning_rate": 4.683127473608834e-05, "loss": 4.4275, "step": 193536 }, { "epoch": 0.01, "learning_rate": 4.682288878857782e-05, "loss": 4.4165, "step": 194048 }, { "epoch": 0.01, "learning_rate": 4.68145028410673e-05, "loss": 4.4298, "step": 194560 }, { "epoch": 0.01, "learning_rate": 4.6806133272360506e-05, "loss": 4.4219, "step": 195072 }, { "epoch": 0.01, "learning_rate": 4.6797747324849986e-05, "loss": 4.4121, "step": 195584 }, { "epoch": 0.01, "learning_rate": 4.6789361377339466e-05, "loss": 4.422, "step": 196096 }, { "epoch": 0.01, "learning_rate": 4.678097542982895e-05, "loss": 4.4172, "step": 196608 }, { "epoch": 0.01, "learning_rate": 4.677258948231843e-05, "loss": 4.4188, "step": 197120 }, { "epoch": 0.01, "learning_rate": 4.676420353480791e-05, "loss": 4.4148, "step": 197632 }, { "epoch": 0.01, "learning_rate": 4.675583396610112e-05, "loss": 4.4114, "step": 198144 }, { "epoch": 0.02, "learning_rate": 4.67474480185906e-05, "loss": 4.4187, "step": 198656 }, { "epoch": 0.02, "learning_rate": 4.673906207108008e-05, "loss": 4.4091, "step": 199168 }, { "epoch": 0.02, "learning_rate": 4.673067612356956e-05, "loss": 4.4146, "step": 199680 }, { "epoch": 0.02, "learning_rate": 4.672229017605904e-05, "loss": 4.4044, "step": 200192 }, { "epoch": 0.02, "learning_rate": 4.671390422854852e-05, "loss": 4.4198, "step": 200704 }, { "epoch": 0.02, "learning_rate": 4.6705518281037995e-05, "loss": 4.4147, "step": 201216 }, { "epoch": 0.02, "learning_rate": 4.669714871233121e-05, "loss": 4.4076, "step": 201728 }, { "epoch": 0.02, "learning_rate": 4.668876276482069e-05, "loss": 4.4006, "step": 202240 }, { "epoch": 0.02, "learning_rate": 4.668037681731017e-05, "loss": 4.4075, "step": 202752 }, { "epoch": 0.02, "learning_rate": 4.667199086979965e-05, "loss": 4.401, "step": 203264 }, { "epoch": 0.02, "learning_rate": 4.666360492228913e-05, "loss": 4.4193, "step": 203776 }, { "epoch": 0.02, "learning_rate": 4.665521897477861e-05, "loss": 4.4048, "step": 204288 }, { "epoch": 0.02, "learning_rate": 4.664684940607182e-05, "loss": 4.4002, "step": 204800 }, { "epoch": 0.02, "learning_rate": 4.66384634585613e-05, "loss": 4.3951, "step": 205312 }, { "epoch": 0.02, "learning_rate": 4.663007751105078e-05, "loss": 4.4023, "step": 205824 }, { "epoch": 0.02, "learning_rate": 4.662169156354026e-05, "loss": 4.382, "step": 206336 }, { "epoch": 0.02, "learning_rate": 4.661330561602974e-05, "loss": 4.3924, "step": 206848 }, { "epoch": 0.02, "learning_rate": 4.660493604732295e-05, "loss": 4.3932, "step": 207360 }, { "epoch": 0.02, "learning_rate": 4.659655009981243e-05, "loss": 4.3966, "step": 207872 }, { "epoch": 0.02, "learning_rate": 4.6588180531105644e-05, "loss": 4.3857, "step": 208384 }, { "epoch": 0.02, "learning_rate": 4.6579794583595124e-05, "loss": 4.4011, "step": 208896 }, { "epoch": 0.02, "learning_rate": 4.6571408636084604e-05, "loss": 4.3826, "step": 209408 }, { "epoch": 0.02, "learning_rate": 4.6563022688574084e-05, "loss": 4.4074, "step": 209920 }, { "epoch": 0.02, "learning_rate": 4.6554636741063564e-05, "loss": 4.393, "step": 210432 }, { "epoch": 0.02, "learning_rate": 4.6546250793553044e-05, "loss": 4.3883, "step": 210944 }, { "epoch": 0.02, "learning_rate": 4.6537864846042524e-05, "loss": 4.4064, "step": 211456 }, { "epoch": 0.02, "learning_rate": 4.6529478898532004e-05, "loss": 4.4066, "step": 211968 }, { "epoch": 0.02, "learning_rate": 4.652110932982521e-05, "loss": 4.3911, "step": 212480 }, { "epoch": 0.02, "learning_rate": 4.651272338231469e-05, "loss": 4.3836, "step": 212992 }, { "epoch": 0.02, "learning_rate": 4.650433743480417e-05, "loss": 4.3917, "step": 213504 }, { "epoch": 0.02, "learning_rate": 4.649595148729365e-05, "loss": 4.3889, "step": 214016 }, { "epoch": 0.02, "learning_rate": 4.648756553978313e-05, "loss": 4.387, "step": 214528 }, { "epoch": 0.02, "learning_rate": 4.647917959227261e-05, "loss": 4.4008, "step": 215040 }, { "epoch": 0.02, "learning_rate": 4.647079364476209e-05, "loss": 4.3802, "step": 215552 }, { "epoch": 0.02, "learning_rate": 4.646240769725157e-05, "loss": 4.3957, "step": 216064 }, { "epoch": 0.02, "learning_rate": 4.645403812854479e-05, "loss": 4.3823, "step": 216576 }, { "epoch": 0.02, "learning_rate": 4.644565218103427e-05, "loss": 4.3862, "step": 217088 }, { "epoch": 0.02, "learning_rate": 4.643728261232748e-05, "loss": 4.3716, "step": 217600 }, { "epoch": 0.02, "learning_rate": 4.642889666481696e-05, "loss": 4.3895, "step": 218112 }, { "epoch": 0.02, "learning_rate": 4.642051071730644e-05, "loss": 4.3733, "step": 218624 }, { "epoch": 0.02, "learning_rate": 4.641212476979592e-05, "loss": 4.3938, "step": 219136 }, { "epoch": 0.02, "learning_rate": 4.64037388222854e-05, "loss": 4.3868, "step": 219648 }, { "epoch": 0.02, "learning_rate": 4.6395369253578606e-05, "loss": 4.3815, "step": 220160 }, { "epoch": 0.02, "learning_rate": 4.6386983306068086e-05, "loss": 4.3656, "step": 220672 }, { "epoch": 0.02, "learning_rate": 4.6378597358557566e-05, "loss": 4.3845, "step": 221184 }, { "epoch": 0.02, "learning_rate": 4.6370211411047046e-05, "loss": 4.3807, "step": 221696 }, { "epoch": 0.02, "learning_rate": 4.6361825463536526e-05, "loss": 4.3825, "step": 222208 }, { "epoch": 0.02, "learning_rate": 4.6353439516026006e-05, "loss": 4.3778, "step": 222720 }, { "epoch": 0.02, "learning_rate": 4.6345053568515486e-05, "loss": 4.3852, "step": 223232 }, { "epoch": 0.02, "learning_rate": 4.63366839998087e-05, "loss": 4.3794, "step": 223744 }, { "epoch": 0.02, "learning_rate": 4.632831443110191e-05, "loss": 4.385, "step": 224256 }, { "epoch": 0.02, "learning_rate": 4.631992848359139e-05, "loss": 4.3644, "step": 224768 }, { "epoch": 0.02, "learning_rate": 4.631154253608087e-05, "loss": 4.3698, "step": 225280 }, { "epoch": 0.02, "learning_rate": 4.630315658857035e-05, "loss": 4.3756, "step": 225792 }, { "epoch": 0.02, "learning_rate": 4.629477064105983e-05, "loss": 4.3634, "step": 226304 }, { "epoch": 0.02, "learning_rate": 4.6286384693549304e-05, "loss": 4.3735, "step": 226816 }, { "epoch": 0.02, "learning_rate": 4.6277998746038784e-05, "loss": 4.3836, "step": 227328 }, { "epoch": 0.02, "learning_rate": 4.6269612798528264e-05, "loss": 4.3673, "step": 227840 }, { "epoch": 0.02, "learning_rate": 4.6261226851017744e-05, "loss": 4.3673, "step": 228352 }, { "epoch": 0.02, "learning_rate": 4.625284090350723e-05, "loss": 4.3758, "step": 228864 }, { "epoch": 0.03, "eval_loss": 4.342891693115234, "eval_runtime": 299.3676, "eval_samples_per_second": 1274.657, "eval_steps_per_second": 39.834, "step": 228960 } ], "logging_steps": 512, "max_steps": 3052726, "num_train_epochs": 9223372036854775807, "save_steps": 10, "total_flos": 9.525304855346558e+16, "trial_name": null, "trial_params": null }