{ "best_metric": 3.952998399734497, "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-c-command/transformer/4/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": 10.971, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.999161405248948e-05, "loss": 6.8154, "step": 512 }, { "epoch": 0.0, "learning_rate": 4.998322810497896e-05, "loss": 6.1847, "step": 1024 }, { "epoch": 0.0, "learning_rate": 4.997484215746844e-05, "loss": 5.9857, "step": 1536 }, { "epoch": 0.0, "learning_rate": 4.996645620995792e-05, "loss": 5.8193, "step": 2048 }, { "epoch": 0.0, "learning_rate": 4.99580702624474e-05, "loss": 5.7079, "step": 2560 }, { "epoch": 0.0, "learning_rate": 4.994968431493688e-05, "loss": 5.6091, "step": 3072 }, { "epoch": 0.0, "learning_rate": 4.994129836742636e-05, "loss": 5.5446, "step": 3584 }, { "epoch": 0.0, "learning_rate": 4.993291241991584e-05, "loss": 5.4504, "step": 4096 }, { "epoch": 0.0, "learning_rate": 4.992452647240532e-05, "loss": 5.4209, "step": 4608 }, { "epoch": 0.0, "learning_rate": 4.99161405248948e-05, "loss": 5.36, "step": 5120 }, { "epoch": 0.0, "learning_rate": 4.990775457738428e-05, "loss": 5.3199, "step": 5632 }, { "epoch": 0.0, "learning_rate": 4.989938500867749e-05, "loss": 5.2787, "step": 6144 }, { "epoch": 0.0, "learning_rate": 4.989099906116697e-05, "loss": 5.2129, "step": 6656 }, { "epoch": 0.0, "learning_rate": 4.988261311365645e-05, "loss": 5.1939, "step": 7168 }, { "epoch": 0.0, "learning_rate": 4.987422716614593e-05, "loss": 5.1454, "step": 7680 }, { "epoch": 0.0, "learning_rate": 4.986584121863541e-05, "loss": 5.1243, "step": 8192 }, { "epoch": 0.0, "learning_rate": 4.985745527112489e-05, "loss": 5.096, "step": 8704 }, { "epoch": 0.0, "learning_rate": 4.984906932361437e-05, "loss": 5.0567, "step": 9216 }, { "epoch": 0.0, "learning_rate": 4.984068337610385e-05, "loss": 5.0328, "step": 9728 }, { "epoch": 0.0, "learning_rate": 4.983229742859333e-05, "loss": 5.0169, "step": 10240 }, { "epoch": 0.0, "learning_rate": 4.9823927859886547e-05, "loss": 4.9887, "step": 10752 }, { "epoch": 0.0, "learning_rate": 4.9815541912376026e-05, "loss": 4.9587, "step": 11264 }, { "epoch": 0.0, "learning_rate": 4.9807155964865506e-05, "loss": 4.9377, "step": 11776 }, { "epoch": 0.0, "learning_rate": 4.9798770017354986e-05, "loss": 4.9283, "step": 12288 }, { "epoch": 0.0, "learning_rate": 4.9790384069844466e-05, "loss": 4.9015, "step": 12800 }, { "epoch": 0.0, "learning_rate": 4.9782014501137675e-05, "loss": 4.8741, "step": 13312 }, { "epoch": 0.0, "learning_rate": 4.9773628553627155e-05, "loss": 4.8642, "step": 13824 }, { "epoch": 0.0, "learning_rate": 4.9765242606116635e-05, "loss": 4.8442, "step": 14336 }, { "epoch": 0.0, "learning_rate": 4.9756856658606115e-05, "loss": 4.8307, "step": 14848 }, { "epoch": 0.01, "learning_rate": 4.9748487089899324e-05, "loss": 4.8157, "step": 15360 }, { "epoch": 0.01, "learning_rate": 4.9740101142388804e-05, "loss": 4.8021, "step": 15872 }, { "epoch": 0.01, "learning_rate": 4.9731715194878284e-05, "loss": 4.7877, "step": 16384 }, { "epoch": 0.01, "learning_rate": 4.97233456261715e-05, "loss": 4.7701, "step": 16896 }, { "epoch": 0.01, "learning_rate": 4.971495967866098e-05, "loss": 4.766, "step": 17408 }, { "epoch": 0.01, "learning_rate": 4.970657373115046e-05, "loss": 4.7659, "step": 17920 }, { "epoch": 0.01, "learning_rate": 4.969818778363994e-05, "loss": 4.7421, "step": 18432 }, { "epoch": 0.01, "learning_rate": 4.968980183612942e-05, "loss": 4.7039, "step": 18944 }, { "epoch": 0.01, "learning_rate": 4.96814158886189e-05, "loss": 4.709, "step": 19456 }, { "epoch": 0.01, "learning_rate": 4.967302994110837e-05, "loss": 4.6884, "step": 19968 }, { "epoch": 0.01, "learning_rate": 4.966466037240159e-05, "loss": 4.6933, "step": 20480 }, { "epoch": 0.01, "learning_rate": 4.965627442489107e-05, "loss": 4.679, "step": 20992 }, { "epoch": 0.01, "learning_rate": 4.964788847738054e-05, "loss": 4.6728, "step": 21504 }, { "epoch": 0.01, "learning_rate": 4.963950252987002e-05, "loss": 4.6509, "step": 22016 }, { "epoch": 0.01, "learning_rate": 4.96311165823595e-05, "loss": 4.66, "step": 22528 }, { "epoch": 0.01, "learning_rate": 4.962274701365272e-05, "loss": 4.6436, "step": 23040 }, { "epoch": 0.01, "learning_rate": 4.96143610661422e-05, "loss": 4.6429, "step": 23552 }, { "epoch": 0.01, "learning_rate": 4.960597511863168e-05, "loss": 4.6296, "step": 24064 }, { "epoch": 0.01, "learning_rate": 4.959758917112116e-05, "loss": 4.6088, "step": 24576 }, { "epoch": 0.01, "learning_rate": 4.95892359812181e-05, "loss": 4.6106, "step": 25088 }, { "epoch": 0.01, "learning_rate": 4.958085003370758e-05, "loss": 4.6125, "step": 25600 }, { "epoch": 0.01, "learning_rate": 4.957246408619706e-05, "loss": 4.59, "step": 26112 }, { "epoch": 0.01, "learning_rate": 4.956407813868654e-05, "loss": 4.5812, "step": 26624 }, { "epoch": 0.01, "learning_rate": 4.9555692191176016e-05, "loss": 4.5749, "step": 27136 }, { "epoch": 0.01, "learning_rate": 4.9547306243665496e-05, "loss": 4.571, "step": 27648 }, { "epoch": 0.01, "learning_rate": 4.9538920296154976e-05, "loss": 4.5556, "step": 28160 }, { "epoch": 0.01, "learning_rate": 4.9530534348644456e-05, "loss": 4.5736, "step": 28672 }, { "epoch": 0.01, "learning_rate": 4.952216477993767e-05, "loss": 4.5415, "step": 29184 }, { "epoch": 0.01, "learning_rate": 4.951377883242715e-05, "loss": 4.5508, "step": 29696 }, { "epoch": 0.01, "learning_rate": 4.950539288491663e-05, "loss": 4.5421, "step": 30208 }, { "epoch": 0.01, "learning_rate": 4.949700693740611e-05, "loss": 4.5168, "step": 30720 }, { "epoch": 0.01, "learning_rate": 4.948863736869932e-05, "loss": 4.5256, "step": 31232 }, { "epoch": 0.01, "learning_rate": 4.94802514211888e-05, "loss": 4.515, "step": 31744 }, { "epoch": 0.01, "learning_rate": 4.947186547367828e-05, "loss": 4.5005, "step": 32256 }, { "epoch": 0.01, "learning_rate": 4.946347952616776e-05, "loss": 4.4936, "step": 32768 }, { "epoch": 0.01, "learning_rate": 4.945510995746097e-05, "loss": 4.5088, "step": 33280 }, { "epoch": 0.01, "learning_rate": 4.944672400995045e-05, "loss": 4.4867, "step": 33792 }, { "epoch": 0.01, "learning_rate": 4.943833806243993e-05, "loss": 4.4814, "step": 34304 }, { "epoch": 0.01, "learning_rate": 4.942995211492941e-05, "loss": 4.4686, "step": 34816 }, { "epoch": 0.01, "learning_rate": 4.9421582546222625e-05, "loss": 4.4814, "step": 35328 }, { "epoch": 0.01, "learning_rate": 4.9413196598712105e-05, "loss": 4.4871, "step": 35840 }, { "epoch": 0.01, "learning_rate": 4.9404810651201585e-05, "loss": 4.4712, "step": 36352 }, { "epoch": 0.01, "learning_rate": 4.9396424703691065e-05, "loss": 4.4646, "step": 36864 }, { "epoch": 0.01, "learning_rate": 4.9388038756180545e-05, "loss": 4.4687, "step": 37376 }, { "epoch": 0.01, "learning_rate": 4.9379652808670025e-05, "loss": 4.4669, "step": 37888 }, { "epoch": 0.01, "learning_rate": 4.9371266861159505e-05, "loss": 4.458, "step": 38400 }, { "epoch": 0.01, "learning_rate": 4.9362897292452714e-05, "loss": 4.4395, "step": 38912 }, { "epoch": 0.01, "learning_rate": 4.9354511344942194e-05, "loss": 4.4467, "step": 39424 }, { "epoch": 0.01, "learning_rate": 4.9346125397431674e-05, "loss": 4.435, "step": 39936 }, { "epoch": 0.01, "learning_rate": 4.9337739449921154e-05, "loss": 4.4315, "step": 40448 }, { "epoch": 0.01, "learning_rate": 4.9329353502410634e-05, "loss": 4.4216, "step": 40960 }, { "epoch": 0.01, "learning_rate": 4.9320967554900114e-05, "loss": 4.4283, "step": 41472 }, { "epoch": 0.01, "learning_rate": 4.9312581607389594e-05, "loss": 4.4326, "step": 41984 }, { "epoch": 0.01, "learning_rate": 4.9304195659879074e-05, "loss": 4.4228, "step": 42496 }, { "epoch": 0.01, "learning_rate": 4.929582609117229e-05, "loss": 4.3977, "step": 43008 }, { "epoch": 0.01, "learning_rate": 4.928744014366177e-05, "loss": 4.4055, "step": 43520 }, { "epoch": 0.01, "learning_rate": 4.927905419615125e-05, "loss": 4.4086, "step": 44032 }, { "epoch": 0.01, "learning_rate": 4.927066824864073e-05, "loss": 4.4042, "step": 44544 }, { "epoch": 0.01, "learning_rate": 4.926229867993394e-05, "loss": 4.3994, "step": 45056 }, { "epoch": 0.01, "learning_rate": 4.925392911122715e-05, "loss": 4.3937, "step": 45568 }, { "epoch": 0.02, "learning_rate": 4.924554316371663e-05, "loss": 4.3892, "step": 46080 }, { "epoch": 0.02, "learning_rate": 4.923715721620611e-05, "loss": 4.3766, "step": 46592 }, { "epoch": 0.02, "learning_rate": 4.922877126869559e-05, "loss": 4.3817, "step": 47104 }, { "epoch": 0.02, "learning_rate": 4.922038532118507e-05, "loss": 4.3821, "step": 47616 }, { "epoch": 0.02, "learning_rate": 4.921199937367455e-05, "loss": 4.3737, "step": 48128 }, { "epoch": 0.02, "learning_rate": 4.920361342616403e-05, "loss": 4.3833, "step": 48640 }, { "epoch": 0.02, "learning_rate": 4.919522747865351e-05, "loss": 4.3616, "step": 49152 }, { "epoch": 0.02, "learning_rate": 4.918685790994672e-05, "loss": 4.3614, "step": 49664 }, { "epoch": 0.02, "learning_rate": 4.91784719624362e-05, "loss": 4.3594, "step": 50176 }, { "epoch": 0.02, "learning_rate": 4.9170086014925676e-05, "loss": 4.3576, "step": 50688 }, { "epoch": 0.02, "learning_rate": 4.9161700067415156e-05, "loss": 4.3627, "step": 51200 }, { "epoch": 0.02, "learning_rate": 4.915333049870837e-05, "loss": 4.347, "step": 51712 }, { "epoch": 0.02, "learning_rate": 4.914494455119785e-05, "loss": 4.3416, "step": 52224 }, { "epoch": 0.02, "learning_rate": 4.9136558603687325e-05, "loss": 4.3399, "step": 52736 }, { "epoch": 0.02, "learning_rate": 4.912818903498054e-05, "loss": 4.3441, "step": 53248 }, { "epoch": 0.02, "learning_rate": 4.911980308747002e-05, "loss": 4.3255, "step": 53760 }, { "epoch": 0.02, "learning_rate": 4.9111449897566966e-05, "loss": 4.3178, "step": 54272 }, { "epoch": 0.02, "learning_rate": 4.9103063950056446e-05, "loss": 4.3277, "step": 54784 }, { "epoch": 0.02, "learning_rate": 4.9094678002545926e-05, "loss": 4.3325, "step": 55296 }, { "epoch": 0.02, "learning_rate": 4.9086292055035406e-05, "loss": 4.3254, "step": 55808 }, { "epoch": 0.02, "learning_rate": 4.9077906107524886e-05, "loss": 4.3222, "step": 56320 }, { "epoch": 0.02, "learning_rate": 4.9069520160014366e-05, "loss": 4.3114, "step": 56832 }, { "epoch": 0.02, "learning_rate": 4.9061134212503846e-05, "loss": 4.3221, "step": 57344 }, { "epoch": 0.02, "learning_rate": 4.9052748264993326e-05, "loss": 4.3049, "step": 57856 }, { "epoch": 0.02, "learning_rate": 4.90443623174828e-05, "loss": 4.3119, "step": 58368 }, { "epoch": 0.02, "learning_rate": 4.9035992748776015e-05, "loss": 4.3134, "step": 58880 }, { "epoch": 0.02, "learning_rate": 4.9027606801265495e-05, "loss": 4.3124, "step": 59392 }, { "epoch": 0.02, "learning_rate": 4.9019220853754975e-05, "loss": 4.3014, "step": 59904 }, { "epoch": 0.02, "learning_rate": 4.9010834906244455e-05, "loss": 4.3005, "step": 60416 }, { "epoch": 0.02, "learning_rate": 4.9002448958733935e-05, "loss": 4.3, "step": 60928 }, { "epoch": 0.02, "learning_rate": 4.8994063011223415e-05, "loss": 4.3007, "step": 61440 }, { "epoch": 0.02, "learning_rate": 4.8985693442516624e-05, "loss": 4.2895, "step": 61952 }, { "epoch": 0.02, "learning_rate": 4.8977307495006104e-05, "loss": 4.2936, "step": 62464 }, { "epoch": 0.02, "learning_rate": 4.8968921547495584e-05, "loss": 4.2884, "step": 62976 }, { "epoch": 0.02, "learning_rate": 4.8960535599985064e-05, "loss": 4.2913, "step": 63488 }, { "epoch": 0.02, "learning_rate": 4.8952149652474544e-05, "loss": 4.2749, "step": 64000 }, { "epoch": 0.02, "learning_rate": 4.8943763704964024e-05, "loss": 4.2714, "step": 64512 }, { "epoch": 0.02, "learning_rate": 4.8935377757453504e-05, "loss": 4.2777, "step": 65024 }, { "epoch": 0.02, "learning_rate": 4.892700818874671e-05, "loss": 4.279, "step": 65536 }, { "epoch": 0.02, "learning_rate": 4.891862224123619e-05, "loss": 4.2693, "step": 66048 }, { "epoch": 0.02, "learning_rate": 4.891023629372567e-05, "loss": 4.2824, "step": 66560 }, { "epoch": 0.02, "learning_rate": 4.890185034621515e-05, "loss": 4.2736, "step": 67072 }, { "epoch": 0.02, "learning_rate": 4.889346439870464e-05, "loss": 4.2597, "step": 67584 }, { "epoch": 0.02, "learning_rate": 4.888507845119412e-05, "loss": 4.2663, "step": 68096 }, { "epoch": 0.02, "learning_rate": 4.887670888248733e-05, "loss": 4.2757, "step": 68608 }, { "epoch": 0.02, "learning_rate": 4.886832293497681e-05, "loss": 4.2652, "step": 69120 }, { "epoch": 0.02, "learning_rate": 4.885993698746629e-05, "loss": 4.2612, "step": 69632 }, { "epoch": 0.02, "learning_rate": 4.885155103995577e-05, "loss": 4.2524, "step": 70144 }, { "epoch": 0.02, "learning_rate": 4.884316509244525e-05, "loss": 4.2599, "step": 70656 }, { "epoch": 0.02, "learning_rate": 4.883477914493473e-05, "loss": 4.2648, "step": 71168 }, { "epoch": 0.02, "learning_rate": 4.882639319742421e-05, "loss": 4.2503, "step": 71680 }, { "epoch": 0.02, "learning_rate": 4.881802362871742e-05, "loss": 4.2432, "step": 72192 }, { "epoch": 0.02, "learning_rate": 4.88096376812069e-05, "loss": 4.2483, "step": 72704 }, { "epoch": 0.02, "learning_rate": 4.880125173369638e-05, "loss": 4.24, "step": 73216 }, { "epoch": 0.02, "learning_rate": 4.879286578618586e-05, "loss": 4.238, "step": 73728 }, { "epoch": 0.02, "learning_rate": 4.878449621747907e-05, "loss": 4.2466, "step": 74240 }, { "epoch": 0.02, "learning_rate": 4.877611026996855e-05, "loss": 4.236, "step": 74752 }, { "epoch": 0.02, "learning_rate": 4.876772432245803e-05, "loss": 4.2366, "step": 75264 }, { "epoch": 0.02, "learning_rate": 4.875933837494751e-05, "loss": 4.2339, "step": 75776 }, { "epoch": 0.02, "learning_rate": 4.8750952427436986e-05, "loss": 4.2297, "step": 76288 }, { "epoch": 0.03, "eval_loss": 4.198172569274902, "eval_runtime": 303.0817, "eval_samples_per_second": 1259.037, "eval_steps_per_second": 39.346, "step": 76320 }, { "epoch": 1.0, "learning_rate": 4.87425828587302e-05, "loss": 4.2177, "step": 76800 }, { "epoch": 1.0, "learning_rate": 4.873419691121968e-05, "loss": 4.2251, "step": 77312 }, { "epoch": 1.0, "learning_rate": 4.872581096370916e-05, "loss": 4.2321, "step": 77824 }, { "epoch": 1.0, "learning_rate": 4.8717425016198635e-05, "loss": 4.2205, "step": 78336 }, { "epoch": 1.0, "learning_rate": 4.870905544749185e-05, "loss": 4.2239, "step": 78848 }, { "epoch": 1.0, "learning_rate": 4.870066949998133e-05, "loss": 4.2168, "step": 79360 }, { "epoch": 1.0, "learning_rate": 4.869228355247081e-05, "loss": 4.2159, "step": 79872 }, { "epoch": 1.0, "learning_rate": 4.868389760496029e-05, "loss": 4.2005, "step": 80384 }, { "epoch": 1.0, "learning_rate": 4.8675528036253507e-05, "loss": 4.2256, "step": 80896 }, { "epoch": 1.0, "learning_rate": 4.8667142088742986e-05, "loss": 4.2091, "step": 81408 }, { "epoch": 1.0, "learning_rate": 4.865875614123246e-05, "loss": 4.2195, "step": 81920 }, { "epoch": 1.0, "learning_rate": 4.865037019372194e-05, "loss": 4.2165, "step": 82432 }, { "epoch": 1.0, "learning_rate": 4.8642000625015155e-05, "loss": 4.1987, "step": 82944 }, { "epoch": 1.0, "learning_rate": 4.8633614677504635e-05, "loss": 4.1981, "step": 83456 }, { "epoch": 1.0, "learning_rate": 4.862522872999411e-05, "loss": 4.1957, "step": 83968 }, { "epoch": 1.0, "learning_rate": 4.861684278248359e-05, "loss": 4.1873, "step": 84480 }, { "epoch": 1.0, "learning_rate": 4.860845683497307e-05, "loss": 4.1943, "step": 84992 }, { "epoch": 1.0, "learning_rate": 4.860007088746255e-05, "loss": 4.1865, "step": 85504 }, { "epoch": 1.0, "learning_rate": 4.859168493995203e-05, "loss": 4.1822, "step": 86016 }, { "epoch": 1.0, "learning_rate": 4.858329899244151e-05, "loss": 4.2057, "step": 86528 }, { "epoch": 1.0, "learning_rate": 4.8574929423734724e-05, "loss": 4.1886, "step": 87040 }, { "epoch": 1.0, "learning_rate": 4.8566543476224204e-05, "loss": 4.1817, "step": 87552 }, { "epoch": 1.0, "learning_rate": 4.8558157528713684e-05, "loss": 4.1789, "step": 88064 }, { "epoch": 1.0, "learning_rate": 4.8549771581203164e-05, "loss": 4.1914, "step": 88576 }, { "epoch": 1.0, "learning_rate": 4.854140201249637e-05, "loss": 4.1695, "step": 89088 }, { "epoch": 1.0, "learning_rate": 4.853301606498585e-05, "loss": 4.1674, "step": 89600 }, { "epoch": 1.0, "learning_rate": 4.852463011747533e-05, "loss": 4.1716, "step": 90112 }, { "epoch": 1.0, "learning_rate": 4.851624416996481e-05, "loss": 4.1716, "step": 90624 }, { "epoch": 1.0, "learning_rate": 4.850785822245429e-05, "loss": 4.1654, "step": 91136 }, { "epoch": 1.01, "learning_rate": 4.84994886537475e-05, "loss": 4.1652, "step": 91648 }, { "epoch": 1.01, "learning_rate": 4.849110270623698e-05, "loss": 4.1663, "step": 92160 }, { "epoch": 1.01, "learning_rate": 4.848271675872646e-05, "loss": 4.1691, "step": 92672 }, { "epoch": 1.01, "learning_rate": 4.847433081121595e-05, "loss": 4.1558, "step": 93184 }, { "epoch": 1.01, "learning_rate": 4.846596124250916e-05, "loss": 4.1647, "step": 93696 }, { "epoch": 1.01, "learning_rate": 4.845757529499864e-05, "loss": 4.1718, "step": 94208 }, { "epoch": 1.01, "learning_rate": 4.844918934748812e-05, "loss": 4.164, "step": 94720 }, { "epoch": 1.01, "learning_rate": 4.84408033999776e-05, "loss": 4.1396, "step": 95232 }, { "epoch": 1.01, "learning_rate": 4.843243383127081e-05, "loss": 4.1464, "step": 95744 }, { "epoch": 1.01, "learning_rate": 4.842404788376029e-05, "loss": 4.1435, "step": 96256 }, { "epoch": 1.01, "learning_rate": 4.841566193624977e-05, "loss": 4.1504, "step": 96768 }, { "epoch": 1.01, "learning_rate": 4.840727598873925e-05, "loss": 4.1424, "step": 97280 }, { "epoch": 1.01, "learning_rate": 4.839889004122873e-05, "loss": 4.1507, "step": 97792 }, { "epoch": 1.01, "learning_rate": 4.8390520472521936e-05, "loss": 4.1367, "step": 98304 }, { "epoch": 1.01, "learning_rate": 4.8382134525011416e-05, "loss": 4.1469, "step": 98816 }, { "epoch": 1.01, "learning_rate": 4.83737485775009e-05, "loss": 4.1451, "step": 99328 }, { "epoch": 1.01, "learning_rate": 4.836536262999038e-05, "loss": 4.1454, "step": 99840 }, { "epoch": 1.01, "learning_rate": 4.835699306128359e-05, "loss": 4.1422, "step": 100352 }, { "epoch": 1.01, "learning_rate": 4.834860711377307e-05, "loss": 4.1227, "step": 100864 }, { "epoch": 1.01, "learning_rate": 4.834022116626255e-05, "loss": 4.139, "step": 101376 }, { "epoch": 1.01, "learning_rate": 4.833183521875203e-05, "loss": 4.1418, "step": 101888 }, { "epoch": 1.01, "learning_rate": 4.832346565004524e-05, "loss": 4.134, "step": 102400 }, { "epoch": 1.01, "learning_rate": 4.831507970253472e-05, "loss": 4.1267, "step": 102912 }, { "epoch": 1.01, "learning_rate": 4.83066937550242e-05, "loss": 4.1205, "step": 103424 }, { "epoch": 1.01, "learning_rate": 4.829830780751368e-05, "loss": 4.1279, "step": 103936 }, { "epoch": 1.01, "learning_rate": 4.828993823880689e-05, "loss": 4.1106, "step": 104448 }, { "epoch": 1.01, "learning_rate": 4.828155229129637e-05, "loss": 4.1408, "step": 104960 }, { "epoch": 1.01, "learning_rate": 4.8273166343785856e-05, "loss": 4.1188, "step": 105472 }, { "epoch": 1.01, "learning_rate": 4.8264780396275336e-05, "loss": 4.1304, "step": 105984 }, { "epoch": 1.01, "learning_rate": 4.8256410827568545e-05, "loss": 4.1292, "step": 106496 }, { "epoch": 1.01, "learning_rate": 4.8248024880058025e-05, "loss": 4.1034, "step": 107008 }, { "epoch": 1.01, "learning_rate": 4.8239638932547505e-05, "loss": 4.1196, "step": 107520 }, { "epoch": 1.01, "learning_rate": 4.8231252985036985e-05, "loss": 4.1123, "step": 108032 }, { "epoch": 1.01, "learning_rate": 4.8222883416330194e-05, "loss": 4.1013, "step": 108544 }, { "epoch": 1.01, "learning_rate": 4.8214497468819674e-05, "loss": 4.0992, "step": 109056 }, { "epoch": 1.01, "learning_rate": 4.8206111521309154e-05, "loss": 4.1171, "step": 109568 }, { "epoch": 1.01, "learning_rate": 4.8197725573798634e-05, "loss": 4.1066, "step": 110080 }, { "epoch": 1.01, "learning_rate": 4.818935600509184e-05, "loss": 4.1019, "step": 110592 }, { "epoch": 1.01, "learning_rate": 4.818097005758132e-05, "loss": 4.0912, "step": 111104 }, { "epoch": 1.01, "learning_rate": 4.817258411007081e-05, "loss": 4.1102, "step": 111616 }, { "epoch": 1.01, "learning_rate": 4.816419816256029e-05, "loss": 4.1136, "step": 112128 }, { "epoch": 1.01, "learning_rate": 4.81558285938535e-05, "loss": 4.1068, "step": 112640 }, { "epoch": 1.01, "learning_rate": 4.814744264634298e-05, "loss": 4.1034, "step": 113152 }, { "epoch": 1.01, "learning_rate": 4.813905669883246e-05, "loss": 4.1147, "step": 113664 }, { "epoch": 1.01, "learning_rate": 4.813067075132194e-05, "loss": 4.1088, "step": 114176 }, { "epoch": 1.01, "learning_rate": 4.812230118261515e-05, "loss": 4.1085, "step": 114688 }, { "epoch": 1.01, "learning_rate": 4.811391523510463e-05, "loss": 4.0917, "step": 115200 }, { "epoch": 1.01, "learning_rate": 4.810552928759411e-05, "loss": 4.1033, "step": 115712 }, { "epoch": 1.01, "learning_rate": 4.809714334008359e-05, "loss": 4.0952, "step": 116224 }, { "epoch": 1.01, "learning_rate": 4.80887737713768e-05, "loss": 4.0935, "step": 116736 }, { "epoch": 1.01, "learning_rate": 4.808038782386628e-05, "loss": 4.09, "step": 117248 }, { "epoch": 1.01, "learning_rate": 4.807200187635576e-05, "loss": 4.0953, "step": 117760 }, { "epoch": 1.01, "learning_rate": 4.8063615928845243e-05, "loss": 4.1048, "step": 118272 }, { "epoch": 1.01, "learning_rate": 4.805524636013845e-05, "loss": 4.0998, "step": 118784 }, { "epoch": 1.01, "learning_rate": 4.804686041262793e-05, "loss": 4.0741, "step": 119296 }, { "epoch": 1.01, "learning_rate": 4.803847446511741e-05, "loss": 4.0838, "step": 119808 }, { "epoch": 1.01, "learning_rate": 4.8030088517606886e-05, "loss": 4.0892, "step": 120320 }, { "epoch": 1.01, "learning_rate": 4.80217189489001e-05, "loss": 4.0875, "step": 120832 }, { "epoch": 1.01, "learning_rate": 4.801333300138958e-05, "loss": 4.0875, "step": 121344 }, { "epoch": 1.01, "learning_rate": 4.800496343268279e-05, "loss": 4.0828, "step": 121856 }, { "epoch": 1.02, "learning_rate": 4.799657748517227e-05, "loss": 4.0834, "step": 122368 }, { "epoch": 1.02, "learning_rate": 4.798819153766175e-05, "loss": 4.0721, "step": 122880 }, { "epoch": 1.02, "learning_rate": 4.797980559015123e-05, "loss": 4.0812, "step": 123392 }, { "epoch": 1.02, "learning_rate": 4.797141964264071e-05, "loss": 4.0808, "step": 123904 }, { "epoch": 1.02, "learning_rate": 4.796303369513019e-05, "loss": 4.0778, "step": 124416 }, { "epoch": 1.02, "learning_rate": 4.795464774761967e-05, "loss": 4.086, "step": 124928 }, { "epoch": 1.02, "learning_rate": 4.794626180010915e-05, "loss": 4.0761, "step": 125440 }, { "epoch": 1.02, "learning_rate": 4.793789223140236e-05, "loss": 4.0694, "step": 125952 }, { "epoch": 1.02, "learning_rate": 4.792950628389184e-05, "loss": 4.0744, "step": 126464 }, { "epoch": 1.02, "learning_rate": 4.792112033638132e-05, "loss": 4.0743, "step": 126976 }, { "epoch": 1.02, "learning_rate": 4.7912750767674535e-05, "loss": 4.0787, "step": 127488 }, { "epoch": 1.02, "learning_rate": 4.790436482016401e-05, "loss": 4.0665, "step": 128000 }, { "epoch": 1.02, "learning_rate": 4.789597887265349e-05, "loss": 4.0637, "step": 128512 }, { "epoch": 1.02, "learning_rate": 4.788759292514297e-05, "loss": 4.0647, "step": 129024 }, { "epoch": 1.02, "learning_rate": 4.787920697763245e-05, "loss": 4.0715, "step": 129536 }, { "epoch": 1.02, "learning_rate": 4.7870821030121935e-05, "loss": 4.0566, "step": 130048 }, { "epoch": 1.02, "learning_rate": 4.7862451461415144e-05, "loss": 4.0463, "step": 130560 }, { "epoch": 1.02, "learning_rate": 4.7854065513904624e-05, "loss": 4.0621, "step": 131072 }, { "epoch": 1.02, "learning_rate": 4.7845679566394104e-05, "loss": 4.0635, "step": 131584 }, { "epoch": 1.02, "learning_rate": 4.7837293618883584e-05, "loss": 4.057, "step": 132096 }, { "epoch": 1.02, "learning_rate": 4.7828907671373064e-05, "loss": 4.0587, "step": 132608 }, { "epoch": 1.02, "learning_rate": 4.7820521723862544e-05, "loss": 4.0541, "step": 133120 }, { "epoch": 1.02, "learning_rate": 4.7812135776352024e-05, "loss": 4.0613, "step": 133632 }, { "epoch": 1.02, "learning_rate": 4.7803749828841504e-05, "loss": 4.0482, "step": 134144 }, { "epoch": 1.02, "learning_rate": 4.779538026013471e-05, "loss": 4.0581, "step": 134656 }, { "epoch": 1.02, "learning_rate": 4.778699431262419e-05, "loss": 4.0636, "step": 135168 }, { "epoch": 1.02, "learning_rate": 4.777860836511367e-05, "loss": 4.0628, "step": 135680 }, { "epoch": 1.02, "learning_rate": 4.777023879640689e-05, "loss": 4.0497, "step": 136192 }, { "epoch": 1.02, "learning_rate": 4.776185284889637e-05, "loss": 4.0498, "step": 136704 }, { "epoch": 1.02, "learning_rate": 4.775346690138585e-05, "loss": 4.0543, "step": 137216 }, { "epoch": 1.02, "learning_rate": 4.774508095387533e-05, "loss": 4.0564, "step": 137728 }, { "epoch": 1.02, "learning_rate": 4.773669500636481e-05, "loss": 4.041, "step": 138240 }, { "epoch": 1.02, "learning_rate": 4.772830905885429e-05, "loss": 4.0543, "step": 138752 }, { "epoch": 1.02, "learning_rate": 4.771992311134377e-05, "loss": 4.0433, "step": 139264 }, { "epoch": 1.02, "learning_rate": 4.771153716383325e-05, "loss": 4.0564, "step": 139776 }, { "epoch": 1.02, "learning_rate": 4.770316759512646e-05, "loss": 4.0379, "step": 140288 }, { "epoch": 1.02, "learning_rate": 4.769478164761594e-05, "loss": 4.0366, "step": 140800 }, { "epoch": 1.02, "learning_rate": 4.768639570010542e-05, "loss": 4.0411, "step": 141312 }, { "epoch": 1.02, "learning_rate": 4.76780097525949e-05, "loss": 4.0449, "step": 141824 }, { "epoch": 1.02, "learning_rate": 4.7669640183888106e-05, "loss": 4.0361, "step": 142336 }, { "epoch": 1.02, "learning_rate": 4.7661254236377586e-05, "loss": 4.0557, "step": 142848 }, { "epoch": 1.02, "learning_rate": 4.765286828886707e-05, "loss": 4.0441, "step": 143360 }, { "epoch": 1.02, "learning_rate": 4.764448234135655e-05, "loss": 4.0343, "step": 143872 }, { "epoch": 1.02, "learning_rate": 4.763611277264976e-05, "loss": 4.0356, "step": 144384 }, { "epoch": 1.02, "learning_rate": 4.762772682513924e-05, "loss": 4.0524, "step": 144896 }, { "epoch": 1.02, "learning_rate": 4.761934087762872e-05, "loss": 4.0442, "step": 145408 }, { "epoch": 1.02, "learning_rate": 4.7610954930118195e-05, "loss": 4.0373, "step": 145920 }, { "epoch": 1.02, "learning_rate": 4.760258536141141e-05, "loss": 4.0327, "step": 146432 }, { "epoch": 1.02, "learning_rate": 4.759419941390089e-05, "loss": 4.0427, "step": 146944 }, { "epoch": 1.02, "learning_rate": 4.758581346639037e-05, "loss": 4.0435, "step": 147456 }, { "epoch": 1.02, "learning_rate": 4.7577427518879844e-05, "loss": 4.0324, "step": 147968 }, { "epoch": 1.02, "learning_rate": 4.756905795017306e-05, "loss": 4.0302, "step": 148480 }, { "epoch": 1.02, "learning_rate": 4.756067200266254e-05, "loss": 4.0327, "step": 148992 }, { "epoch": 1.02, "learning_rate": 4.755228605515202e-05, "loss": 4.0274, "step": 149504 }, { "epoch": 1.02, "learning_rate": 4.7543932865248965e-05, "loss": 4.025, "step": 150016 }, { "epoch": 1.02, "learning_rate": 4.7535546917738445e-05, "loss": 4.04, "step": 150528 }, { "epoch": 1.02, "learning_rate": 4.7527160970227925e-05, "loss": 4.0271, "step": 151040 }, { "epoch": 1.02, "learning_rate": 4.7518775022717405e-05, "loss": 4.0274, "step": 151552 }, { "epoch": 1.02, "learning_rate": 4.7510389075206885e-05, "loss": 4.026, "step": 152064 }, { "epoch": 1.02, "learning_rate": 4.7502019506500094e-05, "loss": 4.0224, "step": 152576 }, { "epoch": 1.03, "eval_loss": 4.027204513549805, "eval_runtime": 303.964, "eval_samples_per_second": 1255.382, "eval_steps_per_second": 39.232, "step": 152640 }, { "epoch": 0.0, "learning_rate": 4.7493633558989574e-05, "loss": 4.0115, "step": 153088 }, { "epoch": 0.0, "learning_rate": 4.7485247611479054e-05, "loss": 4.0217, "step": 153600 }, { "epoch": 0.0, "learning_rate": 4.7476861663968534e-05, "loss": 4.032, "step": 154112 }, { "epoch": 0.0, "learning_rate": 4.7468475716458014e-05, "loss": 4.0213, "step": 154624 }, { "epoch": 0.0, "learning_rate": 4.7460089768947494e-05, "loss": 4.0255, "step": 155136 }, { "epoch": 0.0, "learning_rate": 4.7451703821436974e-05, "loss": 4.0163, "step": 155648 }, { "epoch": 0.0, "learning_rate": 4.7443317873926453e-05, "loss": 4.0214, "step": 156160 }, { "epoch": 0.0, "learning_rate": 4.7434931926415933e-05, "loss": 4.0029, "step": 156672 }, { "epoch": 0.0, "learning_rate": 4.7426545978905413e-05, "loss": 4.0278, "step": 157184 }, { "epoch": 0.0, "learning_rate": 4.741816003139489e-05, "loss": 4.0225, "step": 157696 }, { "epoch": 0.0, "learning_rate": 4.740977408388437e-05, "loss": 4.0199, "step": 158208 }, { "epoch": 0.0, "learning_rate": 4.740140451517758e-05, "loss": 4.0273, "step": 158720 }, { "epoch": 0.0, "learning_rate": 4.739303494647079e-05, "loss": 4.0059, "step": 159232 }, { "epoch": 0.0, "learning_rate": 4.738464899896027e-05, "loss": 4.0075, "step": 159744 }, { "epoch": 0.0, "learning_rate": 4.737626305144975e-05, "loss": 4.0141, "step": 160256 }, { "epoch": 0.0, "learning_rate": 4.736787710393923e-05, "loss": 3.998, "step": 160768 }, { "epoch": 0.0, "learning_rate": 4.735949115642871e-05, "loss": 4.0051, "step": 161280 }, { "epoch": 0.0, "learning_rate": 4.735112158772193e-05, "loss": 4.0006, "step": 161792 }, { "epoch": 0.0, "learning_rate": 4.734273564021141e-05, "loss": 3.9996, "step": 162304 }, { "epoch": 0.0, "learning_rate": 4.733434969270089e-05, "loss": 4.026, "step": 162816 }, { "epoch": 0.0, "learning_rate": 4.732596374519037e-05, "loss": 4.0046, "step": 163328 }, { "epoch": 0.0, "learning_rate": 4.731757779767985e-05, "loss": 4.0082, "step": 163840 }, { "epoch": 0.0, "learning_rate": 4.730919185016933e-05, "loss": 3.9973, "step": 164352 }, { "epoch": 0.0, "learning_rate": 4.730080590265881e-05, "loss": 4.0145, "step": 164864 }, { "epoch": 0.0, "learning_rate": 4.7292436333952016e-05, "loss": 3.9873, "step": 165376 }, { "epoch": 0.0, "learning_rate": 4.7284050386441496e-05, "loss": 3.9943, "step": 165888 }, { "epoch": 0.0, "learning_rate": 4.7275664438930976e-05, "loss": 3.9933, "step": 166400 }, { "epoch": 0.0, "learning_rate": 4.7267294870224185e-05, "loss": 4.0017, "step": 166912 }, { "epoch": 0.0, "learning_rate": 4.7258908922713665e-05, "loss": 3.9908, "step": 167424 }, { "epoch": 0.01, "learning_rate": 4.725052297520315e-05, "loss": 3.9908, "step": 167936 }, { "epoch": 0.01, "learning_rate": 4.724213702769263e-05, "loss": 3.9973, "step": 168448 }, { "epoch": 0.01, "learning_rate": 4.723375108018211e-05, "loss": 4.0005, "step": 168960 }, { "epoch": 0.01, "learning_rate": 4.722536513267159e-05, "loss": 3.9818, "step": 169472 }, { "epoch": 0.01, "learning_rate": 4.721697918516107e-05, "loss": 3.9961, "step": 169984 }, { "epoch": 0.01, "learning_rate": 4.720859323765055e-05, "loss": 3.9991, "step": 170496 }, { "epoch": 0.01, "learning_rate": 4.720022366894376e-05, "loss": 3.9947, "step": 171008 }, { "epoch": 0.01, "learning_rate": 4.719183772143324e-05, "loss": 3.9745, "step": 171520 }, { "epoch": 0.01, "learning_rate": 4.718345177392272e-05, "loss": 3.9842, "step": 172032 }, { "epoch": 0.01, "learning_rate": 4.71750658264122e-05, "loss": 3.971, "step": 172544 }, { "epoch": 0.01, "learning_rate": 4.716667987890168e-05, "loss": 3.9886, "step": 173056 }, { "epoch": 0.01, "learning_rate": 4.7158293931391154e-05, "loss": 3.9811, "step": 173568 }, { "epoch": 0.01, "learning_rate": 4.714992436268437e-05, "loss": 3.9887, "step": 174080 }, { "epoch": 0.01, "learning_rate": 4.714153841517385e-05, "loss": 3.9778, "step": 174592 }, { "epoch": 0.01, "learning_rate": 4.713315246766333e-05, "loss": 3.9827, "step": 175104 }, { "epoch": 0.01, "learning_rate": 4.712476652015281e-05, "loss": 3.987, "step": 175616 }, { "epoch": 0.01, "learning_rate": 4.7116396951446025e-05, "loss": 3.9873, "step": 176128 }, { "epoch": 0.01, "learning_rate": 4.7108011003935505e-05, "loss": 3.9844, "step": 176640 }, { "epoch": 0.01, "learning_rate": 4.709962505642498e-05, "loss": 3.9627, "step": 177152 }, { "epoch": 0.01, "learning_rate": 4.709123910891446e-05, "loss": 3.9782, "step": 177664 }, { "epoch": 0.01, "learning_rate": 4.708285316140394e-05, "loss": 3.9893, "step": 178176 }, { "epoch": 0.01, "learning_rate": 4.7074483592697154e-05, "loss": 3.9735, "step": 178688 }, { "epoch": 0.01, "learning_rate": 4.706609764518663e-05, "loss": 3.9714, "step": 179200 }, { "epoch": 0.01, "learning_rate": 4.705771169767611e-05, "loss": 3.9695, "step": 179712 }, { "epoch": 0.01, "learning_rate": 4.704932575016559e-05, "loss": 3.9735, "step": 180224 }, { "epoch": 0.01, "learning_rate": 4.7040939802655074e-05, "loss": 3.9591, "step": 180736 }, { "epoch": 0.01, "learning_rate": 4.703257023394828e-05, "loss": 3.9822, "step": 181248 }, { "epoch": 0.01, "learning_rate": 4.702418428643776e-05, "loss": 3.9716, "step": 181760 }, { "epoch": 0.01, "learning_rate": 4.701579833892724e-05, "loss": 3.977, "step": 182272 }, { "epoch": 0.01, "learning_rate": 4.700741239141672e-05, "loss": 3.9828, "step": 182784 }, { "epoch": 0.01, "learning_rate": 4.69990264439062e-05, "loss": 3.9553, "step": 183296 }, { "epoch": 0.01, "learning_rate": 4.699064049639568e-05, "loss": 3.9699, "step": 183808 }, { "epoch": 0.01, "learning_rate": 4.698227092768889e-05, "loss": 3.964, "step": 184320 }, { "epoch": 0.01, "learning_rate": 4.697388498017837e-05, "loss": 3.9575, "step": 184832 }, { "epoch": 0.01, "learning_rate": 4.696549903266785e-05, "loss": 3.9523, "step": 185344 }, { "epoch": 0.01, "learning_rate": 4.695711308515733e-05, "loss": 3.968, "step": 185856 }, { "epoch": 0.01, "learning_rate": 4.694874351645054e-05, "loss": 3.9632, "step": 186368 }, { "epoch": 0.01, "learning_rate": 4.694035756894003e-05, "loss": 3.9556, "step": 186880 }, { "epoch": 0.01, "learning_rate": 4.693197162142951e-05, "loss": 3.947, "step": 187392 }, { "epoch": 0.01, "learning_rate": 4.692358567391899e-05, "loss": 3.9676, "step": 187904 }, { "epoch": 0.01, "learning_rate": 4.691519972640847e-05, "loss": 3.9724, "step": 188416 }, { "epoch": 0.01, "learning_rate": 4.690681377889795e-05, "loss": 3.9636, "step": 188928 }, { "epoch": 0.01, "learning_rate": 4.6898444210191157e-05, "loss": 3.9626, "step": 189440 }, { "epoch": 0.01, "learning_rate": 4.6890058262680636e-05, "loss": 3.9729, "step": 189952 }, { "epoch": 0.01, "learning_rate": 4.6881672315170116e-05, "loss": 3.9663, "step": 190464 }, { "epoch": 0.01, "learning_rate": 4.6873286367659596e-05, "loss": 3.9706, "step": 190976 }, { "epoch": 0.01, "learning_rate": 4.6864916798952806e-05, "loss": 3.9496, "step": 191488 }, { "epoch": 0.01, "learning_rate": 4.6856530851442285e-05, "loss": 3.9671, "step": 192000 }, { "epoch": 0.01, "learning_rate": 4.6848144903931765e-05, "loss": 3.9611, "step": 192512 }, { "epoch": 0.01, "learning_rate": 4.6839758956421245e-05, "loss": 3.9549, "step": 193024 }, { "epoch": 0.01, "learning_rate": 4.683138938771446e-05, "loss": 3.9569, "step": 193536 }, { "epoch": 0.01, "learning_rate": 4.682300344020394e-05, "loss": 3.9589, "step": 194048 }, { "epoch": 0.01, "learning_rate": 4.681461749269342e-05, "loss": 3.9697, "step": 194560 }, { "epoch": 0.01, "learning_rate": 4.68062315451829e-05, "loss": 3.9609, "step": 195072 }, { "epoch": 0.01, "learning_rate": 4.679784559767238e-05, "loss": 3.9398, "step": 195584 }, { "epoch": 0.01, "learning_rate": 4.678947602896559e-05, "loss": 3.9477, "step": 196096 }, { "epoch": 0.01, "learning_rate": 4.678109008145507e-05, "loss": 3.954, "step": 196608 }, { "epoch": 0.01, "learning_rate": 4.677270413394455e-05, "loss": 3.9576, "step": 197120 }, { "epoch": 0.01, "learning_rate": 4.676431818643403e-05, "loss": 3.9541, "step": 197632 }, { "epoch": 0.01, "learning_rate": 4.675593223892351e-05, "loss": 3.9485, "step": 198144 }, { "epoch": 0.02, "learning_rate": 4.674756267021672e-05, "loss": 3.9517, "step": 198656 }, { "epoch": 0.02, "learning_rate": 4.67391767227062e-05, "loss": 3.941, "step": 199168 }, { "epoch": 0.02, "learning_rate": 4.673079077519568e-05, "loss": 3.9508, "step": 199680 }, { "epoch": 0.02, "learning_rate": 4.672240482768516e-05, "loss": 3.947, "step": 200192 }, { "epoch": 0.02, "learning_rate": 4.671401888017464e-05, "loss": 3.9491, "step": 200704 }, { "epoch": 0.02, "learning_rate": 4.6705649311467855e-05, "loss": 3.9578, "step": 201216 }, { "epoch": 0.02, "learning_rate": 4.6697263363957335e-05, "loss": 3.9509, "step": 201728 }, { "epoch": 0.02, "learning_rate": 4.6688877416446815e-05, "loss": 3.9403, "step": 202240 }, { "epoch": 0.02, "learning_rate": 4.668049146893629e-05, "loss": 3.9473, "step": 202752 }, { "epoch": 0.02, "learning_rate": 4.6672121900229504e-05, "loss": 3.9461, "step": 203264 }, { "epoch": 0.02, "learning_rate": 4.6663735952718984e-05, "loss": 3.9502, "step": 203776 }, { "epoch": 0.02, "learning_rate": 4.665535000520846e-05, "loss": 3.9418, "step": 204288 }, { "epoch": 0.02, "learning_rate": 4.664696405769794e-05, "loss": 3.9387, "step": 204800 }, { "epoch": 0.02, "learning_rate": 4.663857811018742e-05, "loss": 3.9352, "step": 205312 }, { "epoch": 0.02, "learning_rate": 4.66301921626769e-05, "loss": 3.9465, "step": 205824 }, { "epoch": 0.02, "learning_rate": 4.662182259397011e-05, "loss": 3.9356, "step": 206336 }, { "epoch": 0.02, "learning_rate": 4.661343664645959e-05, "loss": 3.9235, "step": 206848 }, { "epoch": 0.02, "learning_rate": 4.660505069894907e-05, "loss": 3.9368, "step": 207360 }, { "epoch": 0.02, "learning_rate": 4.659666475143855e-05, "loss": 3.9425, "step": 207872 }, { "epoch": 0.02, "learning_rate": 4.658829518273176e-05, "loss": 3.9314, "step": 208384 }, { "epoch": 0.02, "learning_rate": 4.657990923522124e-05, "loss": 3.9422, "step": 208896 }, { "epoch": 0.02, "learning_rate": 4.657152328771072e-05, "loss": 3.9277, "step": 209408 }, { "epoch": 0.02, "learning_rate": 4.65631373402002e-05, "loss": 3.9433, "step": 209920 }, { "epoch": 0.02, "learning_rate": 4.655476777149341e-05, "loss": 3.9286, "step": 210432 }, { "epoch": 0.02, "learning_rate": 4.654638182398289e-05, "loss": 3.9319, "step": 210944 }, { "epoch": 0.02, "learning_rate": 4.653799587647237e-05, "loss": 3.9468, "step": 211456 }, { "epoch": 0.02, "learning_rate": 4.652960992896185e-05, "loss": 3.9419, "step": 211968 }, { "epoch": 0.02, "learning_rate": 4.6521240360255066e-05, "loss": 3.9322, "step": 212480 }, { "epoch": 0.02, "learning_rate": 4.6512854412744546e-05, "loss": 3.9341, "step": 212992 }, { "epoch": 0.02, "learning_rate": 4.6504468465234026e-05, "loss": 3.9302, "step": 213504 }, { "epoch": 0.02, "learning_rate": 4.6496082517723506e-05, "loss": 3.9365, "step": 214016 }, { "epoch": 0.02, "learning_rate": 4.6487712949016715e-05, "loss": 3.9251, "step": 214528 }, { "epoch": 0.02, "learning_rate": 4.6479327001506195e-05, "loss": 3.9393, "step": 215040 }, { "epoch": 0.02, "learning_rate": 4.6470941053995675e-05, "loss": 3.9264, "step": 215552 }, { "epoch": 0.02, "learning_rate": 4.6462555106485155e-05, "loss": 3.9387, "step": 216064 }, { "epoch": 0.02, "learning_rate": 4.6454185537778364e-05, "loss": 3.9178, "step": 216576 }, { "epoch": 0.02, "learning_rate": 4.6445799590267844e-05, "loss": 3.9266, "step": 217088 }, { "epoch": 0.02, "learning_rate": 4.6437413642757324e-05, "loss": 3.9226, "step": 217600 }, { "epoch": 0.02, "learning_rate": 4.6429027695246804e-05, "loss": 3.9267, "step": 218112 }, { "epoch": 0.02, "learning_rate": 4.642064174773629e-05, "loss": 3.9222, "step": 218624 }, { "epoch": 0.02, "learning_rate": 4.64122721790295e-05, "loss": 3.9442, "step": 219136 }, { "epoch": 0.02, "learning_rate": 4.640388623151898e-05, "loss": 3.9297, "step": 219648 }, { "epoch": 0.02, "learning_rate": 4.639550028400846e-05, "loss": 3.9231, "step": 220160 }, { "epoch": 0.02, "learning_rate": 4.638711433649794e-05, "loss": 3.9163, "step": 220672 }, { "epoch": 0.02, "learning_rate": 4.637874476779115e-05, "loss": 3.9367, "step": 221184 }, { "epoch": 0.02, "learning_rate": 4.637035882028063e-05, "loss": 3.9337, "step": 221696 }, { "epoch": 0.02, "learning_rate": 4.636197287277011e-05, "loss": 3.9284, "step": 222208 }, { "epoch": 0.02, "learning_rate": 4.635358692525959e-05, "loss": 3.9213, "step": 222720 }, { "epoch": 0.02, "learning_rate": 4.63452173565528e-05, "loss": 3.9296, "step": 223232 }, { "epoch": 0.02, "learning_rate": 4.633683140904228e-05, "loss": 3.9319, "step": 223744 }, { "epoch": 0.02, "learning_rate": 4.632844546153176e-05, "loss": 3.9225, "step": 224256 }, { "epoch": 0.02, "learning_rate": 4.6320059514021244e-05, "loss": 3.9198, "step": 224768 }, { "epoch": 0.02, "learning_rate": 4.6311673566510724e-05, "loss": 3.9232, "step": 225280 }, { "epoch": 0.02, "learning_rate": 4.6303303997803934e-05, "loss": 3.9211, "step": 225792 }, { "epoch": 0.02, "learning_rate": 4.6294918050293413e-05, "loss": 3.9132, "step": 226304 }, { "epoch": 0.02, "learning_rate": 4.6286532102782893e-05, "loss": 3.9266, "step": 226816 }, { "epoch": 0.02, "learning_rate": 4.6278146155272373e-05, "loss": 3.9238, "step": 227328 }, { "epoch": 0.02, "learning_rate": 4.626976020776185e-05, "loss": 3.9165, "step": 227840 }, { "epoch": 0.02, "learning_rate": 4.626139063905506e-05, "loss": 3.9151, "step": 228352 }, { "epoch": 0.02, "learning_rate": 4.625300469154454e-05, "loss": 3.9115, "step": 228864 }, { "epoch": 0.03, "eval_loss": 3.952998399734497, "eval_runtime": 307.7424, "eval_samples_per_second": 1239.969, "eval_steps_per_second": 38.75, "step": 228960 } ], "logging_steps": 512, "max_steps": 3052726, "num_train_epochs": 9223372036854775807, "save_steps": 10, "total_flos": 1.6001973826759066e+17, "trial_name": null, "trial_params": null }