| { | |
| "best_metric": 3.8966124057769775, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/pp-mod-subj/transformer/3/checkpoints/checkpoint-839509", | |
| "epoch": 1.0250002784396635, | |
| "eval_steps": 10, | |
| "global_step": 915828, | |
| "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": 11.0542, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 6.8397, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 6.1966, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 5.9673, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 5.8145, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 5.689, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 5.5976, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994129836742636e-05, | |
| "loss": 5.5186, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993291241991584e-05, | |
| "loss": 5.4556, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992452647240532e-05, | |
| "loss": 5.3853, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99161405248948e-05, | |
| "loss": 5.3386, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990775457738428e-05, | |
| "loss": 5.2945, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989938500867749e-05, | |
| "loss": 5.2612, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989099906116697e-05, | |
| "loss": 5.1969, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988261311365645e-05, | |
| "loss": 5.156, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987422716614593e-05, | |
| "loss": 5.1285, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986584121863541e-05, | |
| "loss": 5.0867, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985745527112489e-05, | |
| "loss": 5.0591, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984906932361437e-05, | |
| "loss": 5.0382, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984068337610385e-05, | |
| "loss": 5.0053, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 4.9929, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9823927859886547e-05, | |
| "loss": 4.965, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9815541912376026e-05, | |
| "loss": 4.9348, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9807155964865506e-05, | |
| "loss": 4.9286, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9798770017354986e-05, | |
| "loss": 4.901, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9790384069844466e-05, | |
| "loss": 4.8779, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9782014501137675e-05, | |
| "loss": 4.8644, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773628553627155e-05, | |
| "loss": 4.8556, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9765242606116635e-05, | |
| "loss": 4.8212, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9756856658606115e-05, | |
| "loss": 4.811, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748487089899324e-05, | |
| "loss": 4.7886, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9740101142388804e-05, | |
| "loss": 4.7793, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9731715194878284e-05, | |
| "loss": 4.7809, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9723329247367764e-05, | |
| "loss": 4.7565, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714943299857244e-05, | |
| "loss": 4.7464, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.970657373115046e-05, | |
| "loss": 4.7362, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.969818778363994e-05, | |
| "loss": 4.7278, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968980183612942e-05, | |
| "loss": 4.7131, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96814158886189e-05, | |
| "loss": 4.6994, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.967304631991211e-05, | |
| "loss": 4.6917, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966466037240159e-05, | |
| "loss": 4.6737, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965627442489107e-05, | |
| "loss": 4.6699, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964788847738054e-05, | |
| "loss": 4.6434, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963950252987002e-05, | |
| "loss": 4.6534, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963113296116324e-05, | |
| "loss": 4.6471, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962274701365272e-05, | |
| "loss": 4.636, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96143610661422e-05, | |
| "loss": 4.6168, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.960597511863168e-05, | |
| "loss": 4.616, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9597605549924894e-05, | |
| "loss": 4.6087, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9589219602414374e-05, | |
| "loss": 4.5917, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958083365490385e-05, | |
| "loss": 4.5816, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.957246408619706e-05, | |
| "loss": 4.5756, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956407813868654e-05, | |
| "loss": 4.5856, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9555692191176016e-05, | |
| "loss": 4.5529, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9547306243665496e-05, | |
| "loss": 4.577, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9538920296154976e-05, | |
| "loss": 4.5391, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953055072744819e-05, | |
| "loss": 4.5471, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.952216477993767e-05, | |
| "loss": 4.5396, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951377883242715e-05, | |
| "loss": 4.5361, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.950539288491663e-05, | |
| "loss": 4.5242, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949700693740611e-05, | |
| "loss": 4.5207, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948863736869932e-05, | |
| "loss": 4.5045, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94802514211888e-05, | |
| "loss": 4.5143, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947186547367828e-05, | |
| "loss": 4.4978, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946347952616776e-05, | |
| "loss": 4.4999, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945510995746097e-05, | |
| "loss": 4.4935, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944672400995045e-05, | |
| "loss": 4.4827, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.943833806243993e-05, | |
| "loss": 4.4648, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942995211492941e-05, | |
| "loss": 4.4785, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9421582546222625e-05, | |
| "loss": 4.4695, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9413196598712105e-05, | |
| "loss": 4.4551, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9404810651201585e-05, | |
| "loss": 4.462, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9396424703691065e-05, | |
| "loss": 4.4524, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388055134984274e-05, | |
| "loss": 4.4533, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379669187473754e-05, | |
| "loss": 4.4536, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371283239963234e-05, | |
| "loss": 4.4467, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.936291367125644e-05, | |
| "loss": 4.4372, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.935452772374592e-05, | |
| "loss": 4.4218, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.93461417762354e-05, | |
| "loss": 4.4294, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.933775582872488e-05, | |
| "loss": 4.426, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932936988121436e-05, | |
| "loss": 4.4269, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932098393370384e-05, | |
| "loss": 4.4225, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.931259798619332e-05, | |
| "loss": 4.4079, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.930421203868281e-05, | |
| "loss": 4.3935, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.929582609117229e-05, | |
| "loss": 4.3973, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.928744014366177e-05, | |
| "loss": 4.3988, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927905419615125e-05, | |
| "loss": 4.4002, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927066824864073e-05, | |
| "loss": 4.395, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.926229867993394e-05, | |
| "loss": 4.3877, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.925392911122715e-05, | |
| "loss": 4.3779, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.924554316371663e-05, | |
| "loss": 4.3743, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923715721620611e-05, | |
| "loss": 4.3734, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922877126869559e-05, | |
| "loss": 4.3801, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.92204016999888e-05, | |
| "loss": 4.3727, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921201575247828e-05, | |
| "loss": 4.3496, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920362980496776e-05, | |
| "loss": 4.3525, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919524385745724e-05, | |
| "loss": 4.3598, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918685790994672e-05, | |
| "loss": 4.3608, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.91784719624362e-05, | |
| "loss": 4.3534, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9170086014925676e-05, | |
| "loss": 4.3487, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9161700067415156e-05, | |
| "loss": 4.3467, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.915333049870837e-05, | |
| "loss": 4.3399, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.914494455119785e-05, | |
| "loss": 4.3468, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.913657498249106e-05, | |
| "loss": 4.3429, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.912820541378427e-05, | |
| "loss": 4.3222, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.911981946627375e-05, | |
| "loss": 4.3342, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.911143351876323e-05, | |
| "loss": 4.3115, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.910304757125272e-05, | |
| "loss": 4.3221, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90946616237422e-05, | |
| "loss": 4.3212, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.908627567623168e-05, | |
| "loss": 4.3143, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907788972872115e-05, | |
| "loss": 4.3165, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906950378121063e-05, | |
| "loss": 4.3029, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906111783370011e-05, | |
| "loss": 4.3006, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905273188618959e-05, | |
| "loss": 4.3037, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904434593867907e-05, | |
| "loss": 4.3188, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903595999116855e-05, | |
| "loss": 4.2902, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.902759042246176e-05, | |
| "loss": 4.2861, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901920447495124e-05, | |
| "loss": 4.2872, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901081852744072e-05, | |
| "loss": 4.3018, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90024325799302e-05, | |
| "loss": 4.2895, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899407939002715e-05, | |
| "loss": 4.291, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8985693442516624e-05, | |
| "loss": 4.2871, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8977307495006104e-05, | |
| "loss": 4.2893, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968921547495584e-05, | |
| "loss": 4.2779, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8960535599985064e-05, | |
| "loss": 4.2769, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8952149652474544e-05, | |
| "loss": 4.2734, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8943763704964024e-05, | |
| "loss": 4.2834, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8935377757453504e-05, | |
| "loss": 4.2619, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.892700818874671e-05, | |
| "loss": 4.273, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.891863862003992e-05, | |
| "loss": 4.2685, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.891025267252941e-05, | |
| "loss": 4.2571, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890186672501889e-05, | |
| "loss": 4.2674, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889348077750837e-05, | |
| "loss": 4.2623, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888509482999785e-05, | |
| "loss": 4.2658, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.887670888248733e-05, | |
| "loss": 4.2633, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.886832293497681e-05, | |
| "loss": 4.2467, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.885993698746629e-05, | |
| "loss": 4.2431, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88515674187595e-05, | |
| "loss": 4.2474, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884318147124898e-05, | |
| "loss": 4.2553, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883479552373846e-05, | |
| "loss": 4.257, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882640957622794e-05, | |
| "loss": 4.2488, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8818040007521146e-05, | |
| "loss": 4.238, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8809654060010626e-05, | |
| "loss": 4.2385, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8801268112500106e-05, | |
| "loss": 4.2412, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8792882164989586e-05, | |
| "loss": 4.2389, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.878452897508653e-05, | |
| "loss": 4.233, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877614302757601e-05, | |
| "loss": 4.2513, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.876775708006549e-05, | |
| "loss": 4.2348, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875937113255497e-05, | |
| "loss": 4.2272, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875098518504445e-05, | |
| "loss": 4.2242, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.240949630737305, | |
| "eval_runtime": 572.3683, | |
| "eval_samples_per_second": 666.688, | |
| "eval_steps_per_second": 20.834, | |
| "step": 76319 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.874261561633766e-05, | |
| "loss": 4.2301, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.873422966882714e-05, | |
| "loss": 4.2271, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.872584372131662e-05, | |
| "loss": 4.2151, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.87174577738061e-05, | |
| "loss": 4.2161, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.870907182629558e-05, | |
| "loss": 4.2071, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.870068587878506e-05, | |
| "loss": 4.2065, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.869229993127454e-05, | |
| "loss": 4.2004, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8683913983764027e-05, | |
| "loss": 4.2012, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8675528036253507e-05, | |
| "loss": 4.2067, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8667158467546716e-05, | |
| "loss": 4.2042, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8658772520036196e-05, | |
| "loss": 4.2019, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8650386572525676e-05, | |
| "loss": 4.205, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8642017003818885e-05, | |
| "loss": 4.1961, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8633631056308365e-05, | |
| "loss": 4.1907, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8625245108797845e-05, | |
| "loss": 4.1879, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8616859161287324e-05, | |
| "loss": 4.1625, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8608473213776804e-05, | |
| "loss": 4.1839, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.860008726626628e-05, | |
| "loss": 4.1816, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8591701318755764e-05, | |
| "loss": 4.1823, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8583315371245244e-05, | |
| "loss": 4.1948, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8574929423734724e-05, | |
| "loss": 4.1758, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8566543476224204e-05, | |
| "loss": 4.1789, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8558157528713684e-05, | |
| "loss": 4.1776, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8549771581203164e-05, | |
| "loss": 4.1762, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.854140201249637e-05, | |
| "loss": 4.1681, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.853301606498585e-05, | |
| "loss": 4.1661, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.852463011747533e-05, | |
| "loss": 4.1715, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.851624416996481e-05, | |
| "loss": 4.1575, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.850785822245429e-05, | |
| "loss": 4.1645, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.849947227494377e-05, | |
| "loss": 4.1495, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.849108632743325e-05, | |
| "loss": 4.1529, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.848270037992273e-05, | |
| "loss": 4.1656, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.847433081121595e-05, | |
| "loss": 4.1565, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.846594486370543e-05, | |
| "loss": 4.1579, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.845755891619491e-05, | |
| "loss": 4.1545, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.844918934748812e-05, | |
| "loss": 4.1561, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.84408033999776e-05, | |
| "loss": 4.1526, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.843241745246708e-05, | |
| "loss": 4.1506, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.842403150495656e-05, | |
| "loss": 4.1492, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.841564555744604e-05, | |
| "loss": 4.1375, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.840725960993552e-05, | |
| "loss": 4.1462, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.839889004122873e-05, | |
| "loss": 4.1303, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.839050409371821e-05, | |
| "loss": 4.1416, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8382118146207687e-05, | |
| "loss": 4.1459, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8373732198697167e-05, | |
| "loss": 4.1426, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.836536262999038e-05, | |
| "loss": 4.1319, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.835697668247986e-05, | |
| "loss": 4.1345, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.834859073496934e-05, | |
| "loss": 4.1398, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.834020478745882e-05, | |
| "loss": 4.1198, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.833183521875203e-05, | |
| "loss": 4.1231, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.832344927124151e-05, | |
| "loss": 4.1218, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.831506332373099e-05, | |
| "loss": 4.1356, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.830667737622047e-05, | |
| "loss": 4.1143, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.829830780751368e-05, | |
| "loss": 4.1358, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.828992186000316e-05, | |
| "loss": 4.1056, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.828153591249264e-05, | |
| "loss": 4.1216, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.827314996498212e-05, | |
| "loss": 4.1181, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.82647640174716e-05, | |
| "loss": 4.1222, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8256394448764816e-05, | |
| "loss": 4.1125, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8248008501254296e-05, | |
| "loss": 4.117, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.823962255374377e-05, | |
| "loss": 4.1055, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.823123660623325e-05, | |
| "loss": 4.1141, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.822285065872273e-05, | |
| "loss": 4.1066, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8214481090015945e-05, | |
| "loss": 4.1103, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.820609514250542e-05, | |
| "loss": 4.1068, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.81977091949949e-05, | |
| "loss": 4.1045, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.818932324748438e-05, | |
| "loss": 4.0875, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8180953678777594e-05, | |
| "loss": 4.1111, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8172567731267074e-05, | |
| "loss": 4.0985, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8164181783756554e-05, | |
| "loss": 4.0917, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8155795836246034e-05, | |
| "loss": 4.1018, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.814742626753924e-05, | |
| "loss": 4.0965, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.813904032002872e-05, | |
| "loss": 4.0968, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.81306543725182e-05, | |
| "loss": 4.1069, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.812226842500768e-05, | |
| "loss": 4.1013, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.811389885630089e-05, | |
| "loss": 4.0959, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.810551290879037e-05, | |
| "loss": 4.0855, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.809712696127985e-05, | |
| "loss": 4.0909, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.808874101376933e-05, | |
| "loss": 4.0939, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.808035506625881e-05, | |
| "loss": 4.101, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.807198549755203e-05, | |
| "loss": 4.0919, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.806359955004151e-05, | |
| "loss": 4.0865, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.805521360253099e-05, | |
| "loss": 4.0706, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.804682765502047e-05, | |
| "loss": 4.0782, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.803844170750995e-05, | |
| "loss": 4.0814, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8030072138803156e-05, | |
| "loss": 4.0831, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8021686191292636e-05, | |
| "loss": 4.0873, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8013300243782116e-05, | |
| "loss": 4.0814, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8004914296271596e-05, | |
| "loss": 4.0712, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7996544727564805e-05, | |
| "loss": 4.0779, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7988158780054285e-05, | |
| "loss": 4.0687, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7979772832543765e-05, | |
| "loss": 4.0787, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7971386885033245e-05, | |
| "loss": 4.0772, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.796301731632646e-05, | |
| "loss": 4.0596, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.795463136881594e-05, | |
| "loss": 4.0579, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.794624542130542e-05, | |
| "loss": 4.0685, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.79378594737949e-05, | |
| "loss": 4.0781, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792947352628438e-05, | |
| "loss": 4.0644, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792108757877386e-05, | |
| "loss": 4.0732, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.791270163126334e-05, | |
| "loss": 4.0646, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.790431568375282e-05, | |
| "loss": 4.0629, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.78959297362423e-05, | |
| "loss": 4.0697, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.788756016753551e-05, | |
| "loss": 4.0671, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.787919059882872e-05, | |
| "loss": 4.0527, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.78708046513182e-05, | |
| "loss": 4.0659, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.786241870380768e-05, | |
| "loss": 4.0474, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7854032756297166e-05, | |
| "loss": 4.0514, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7845646808786646e-05, | |
| "loss": 4.0601, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7837260861276126e-05, | |
| "loss": 4.0493, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7828891292569335e-05, | |
| "loss": 4.0544, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7820505345058815e-05, | |
| "loss": 4.0484, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7812119397548295e-05, | |
| "loss": 4.0442, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7803733450037775e-05, | |
| "loss": 4.0491, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.779534750252725e-05, | |
| "loss": 4.0647, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.778696155501673e-05, | |
| "loss": 4.042, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7778591986309944e-05, | |
| "loss": 4.0415, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7770206038799423e-05, | |
| "loss": 4.0412, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7761820091288903e-05, | |
| "loss": 4.0504, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7753434143778383e-05, | |
| "loss": 4.0481, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.77450645750716e-05, | |
| "loss": 4.0467, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.773667862756108e-05, | |
| "loss": 4.0461, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.772829268005055e-05, | |
| "loss": 4.0489, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.771990673254003e-05, | |
| "loss": 4.0419, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.771152078502951e-05, | |
| "loss": 4.0373, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.770313483751899e-05, | |
| "loss": 4.0389, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.769474889000847e-05, | |
| "loss": 4.0498, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.768636294249795e-05, | |
| "loss": 4.0335, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.767799337379116e-05, | |
| "loss": 4.0356, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.766960742628064e-05, | |
| "loss": 4.0482, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.766122147877012e-05, | |
| "loss": 4.025, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.76528355312596e-05, | |
| "loss": 4.0405, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.764446596255282e-05, | |
| "loss": 4.0336, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.76360800150423e-05, | |
| "loss": 4.0439, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.762769406753178e-05, | |
| "loss": 4.0368, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.761930812002126e-05, | |
| "loss": 4.0242, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7610938551314466e-05, | |
| "loss": 4.0231, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7602552603803946e-05, | |
| "loss": 4.0285, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7594166656293426e-05, | |
| "loss": 4.0367, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7585780708782906e-05, | |
| "loss": 4.0411, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7577411140076115e-05, | |
| "loss": 4.0361, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7569025192565595e-05, | |
| "loss": 4.0224, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7560639245055075e-05, | |
| "loss": 4.0249, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7552253297544555e-05, | |
| "loss": 4.0309, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.754388372883777e-05, | |
| "loss": 4.0282, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.753551416013098e-05, | |
| "loss": 4.0235, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.752712821262046e-05, | |
| "loss": 4.0434, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.751874226510994e-05, | |
| "loss": 4.0209, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.751035631759942e-05, | |
| "loss": 4.025, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.75019703700889e-05, | |
| "loss": 4.02, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.072732448577881, | |
| "eval_runtime": 571.302, | |
| "eval_samples_per_second": 667.932, | |
| "eval_steps_per_second": 20.873, | |
| "step": 152638 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.749358442257838e-05, | |
| "loss": 4.0287, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.748519847506786e-05, | |
| "loss": 4.024, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.747681252755734e-05, | |
| "loss": 4.0159, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.746842658004682e-05, | |
| "loss": 4.0175, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.74600406325363e-05, | |
| "loss": 4.0091, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745165468502578e-05, | |
| "loss": 4.0096, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.744326873751526e-05, | |
| "loss": 4.0081, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.743488279000474e-05, | |
| "loss": 4.0074, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.742649684249422e-05, | |
| "loss": 4.0128, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7418127273787435e-05, | |
| "loss": 4.0138, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.740974132627691e-05, | |
| "loss": 4.0121, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.740135537876639e-05, | |
| "loss": 4.0112, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7392985810059604e-05, | |
| "loss": 4.0055, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7384599862549084e-05, | |
| "loss": 4.0014, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.737621391503856e-05, | |
| "loss": 4.003, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.736782796752804e-05, | |
| "loss": 3.977, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.735944202001752e-05, | |
| "loss": 4.0009, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7351056072507e-05, | |
| "loss": 3.9981, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.734267012499648e-05, | |
| "loss": 3.9981, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.733428417748596e-05, | |
| "loss": 4.0155, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7325898229975444e-05, | |
| "loss": 3.9963, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7317512282464924e-05, | |
| "loss": 4.0027, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7309126334954404e-05, | |
| "loss": 3.9996, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730075676624761e-05, | |
| "loss": 3.9979, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.729237081873709e-05, | |
| "loss": 3.992, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.728398487122657e-05, | |
| "loss": 3.9888, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.727559892371605e-05, | |
| "loss": 3.9993, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.726722935500926e-05, | |
| "loss": 3.9851, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.725884340749874e-05, | |
| "loss": 3.9909, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.725045745998822e-05, | |
| "loss": 3.9757, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.72420715124777e-05, | |
| "loss": 3.9869, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.723370194377091e-05, | |
| "loss": 3.9956, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.72253159962604e-05, | |
| "loss": 3.9842, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.721693004874988e-05, | |
| "loss": 3.9918, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720854410123936e-05, | |
| "loss": 3.9877, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7200174532532566e-05, | |
| "loss": 3.9899, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7191788585022046e-05, | |
| "loss": 3.991, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7183402637511526e-05, | |
| "loss": 3.9867, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7175016690001006e-05, | |
| "loss": 3.9834, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7166647121294215e-05, | |
| "loss": 3.9755, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7158261173783695e-05, | |
| "loss": 3.9835, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7149875226273175e-05, | |
| "loss": 3.9697, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7141489278762655e-05, | |
| "loss": 3.9861, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7133119710055864e-05, | |
| "loss": 3.9817, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.712473376254535e-05, | |
| "loss": 3.9848, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.711634781503483e-05, | |
| "loss": 3.9749, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.710796186752431e-05, | |
| "loss": 3.9727, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.709959229881752e-05, | |
| "loss": 3.9847, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7091206351307e-05, | |
| "loss": 3.9639, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.708282040379648e-05, | |
| "loss": 3.9692, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.707443445628596e-05, | |
| "loss": 3.9649, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.706606488757917e-05, | |
| "loss": 3.9827, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.705767894006865e-05, | |
| "loss": 3.9635, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704929299255813e-05, | |
| "loss": 3.9839, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704090704504761e-05, | |
| "loss": 3.9541, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.703253747634082e-05, | |
| "loss": 3.9671, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7024151528830305e-05, | |
| "loss": 3.9708, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7015765581319785e-05, | |
| "loss": 3.9702, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7007379633809265e-05, | |
| "loss": 3.9656, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6999010065102474e-05, | |
| "loss": 3.9762, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6990624117591954e-05, | |
| "loss": 3.9497, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6982238170081434e-05, | |
| "loss": 3.9638, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6973852222570914e-05, | |
| "loss": 3.9608, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.696548265386412e-05, | |
| "loss": 3.9688, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.69570967063536e-05, | |
| "loss": 3.9572, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.694871075884308e-05, | |
| "loss": 3.9638, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.694032481133256e-05, | |
| "loss": 3.9438, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.693195524262577e-05, | |
| "loss": 3.9653, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.692356929511526e-05, | |
| "loss": 3.9569, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.691518334760474e-05, | |
| "loss": 3.9528, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.690679740009422e-05, | |
| "loss": 3.9581, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.689842783138743e-05, | |
| "loss": 3.9558, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.689004188387691e-05, | |
| "loss": 3.955, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688165593636639e-05, | |
| "loss": 3.9705, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.687326998885587e-05, | |
| "loss": 3.9629, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6864900420149076e-05, | |
| "loss": 3.9587, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6856530851442285e-05, | |
| "loss": 3.9478, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6848144903931765e-05, | |
| "loss": 3.9513, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6839758956421245e-05, | |
| "loss": 3.9582, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6831373008910725e-05, | |
| "loss": 3.9634, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6822987061400205e-05, | |
| "loss": 3.9531, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.681460111388969e-05, | |
| "loss": 3.9546, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6806215166379165e-05, | |
| "loss": 3.9378, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.679784559767238e-05, | |
| "loss": 3.946, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678945965016186e-05, | |
| "loss": 3.9408, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678107370265134e-05, | |
| "loss": 3.9495, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6772687755140814e-05, | |
| "loss": 3.957, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.676431818643403e-05, | |
| "loss": 3.9475, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.675593223892351e-05, | |
| "loss": 3.9385, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.674756267021672e-05, | |
| "loss": 3.9474, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.67391767227062e-05, | |
| "loss": 3.938, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.673079077519568e-05, | |
| "loss": 3.9487, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.672240482768516e-05, | |
| "loss": 3.9459, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.671401888017464e-05, | |
| "loss": 3.9327, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.670563293266412e-05, | |
| "loss": 3.9326, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.66972469851536e-05, | |
| "loss": 3.9374, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6688877416446815e-05, | |
| "loss": 3.9479, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.668049146893629e-05, | |
| "loss": 3.9388, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.667210552142577e-05, | |
| "loss": 3.9479, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.666371957391525e-05, | |
| "loss": 3.9364, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.665533362640473e-05, | |
| "loss": 3.9364, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.664694767889421e-05, | |
| "loss": 3.9425, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663856173138369e-05, | |
| "loss": 3.942, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.66301921626769e-05, | |
| "loss": 3.9307, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6621806215166383e-05, | |
| "loss": 3.9471, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6613420267655863e-05, | |
| "loss": 3.9218, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.660503432014534e-05, | |
| "loss": 3.927, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.659664837263482e-05, | |
| "loss": 3.9401, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.65882624251243e-05, | |
| "loss": 3.9248, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657989285641751e-05, | |
| "loss": 3.937, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657150690890699e-05, | |
| "loss": 3.926, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.656312096139647e-05, | |
| "loss": 3.9204, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.655473501388595e-05, | |
| "loss": 3.9274, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.654634906637543e-05, | |
| "loss": 3.9472, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.653796311886491e-05, | |
| "loss": 3.9223, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.652959355015812e-05, | |
| "loss": 3.9257, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.65212076026476e-05, | |
| "loss": 3.9186, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.651282165513708e-05, | |
| "loss": 3.9285, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.650443570762657e-05, | |
| "loss": 3.9351, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.649606613891978e-05, | |
| "loss": 3.9283, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.648768019140926e-05, | |
| "loss": 3.926, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647929424389874e-05, | |
| "loss": 3.9283, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647090829638822e-05, | |
| "loss": 3.9269, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.64625223488777e-05, | |
| "loss": 3.9251, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.645413640136718e-05, | |
| "loss": 3.9227, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.644575045385665e-05, | |
| "loss": 3.9327, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.643736450634613e-05, | |
| "loss": 3.9193, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6428994937639346e-05, | |
| "loss": 3.9205, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6420625368932555e-05, | |
| "loss": 3.9341, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6412239421422035e-05, | |
| "loss": 3.9115, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.640385347391152e-05, | |
| "loss": 3.9245, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6395467526401e-05, | |
| "loss": 3.9226, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6387081578890475e-05, | |
| "loss": 3.9313, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6378695631379955e-05, | |
| "loss": 3.9241, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6370309683869435e-05, | |
| "loss": 3.9196, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6361923736358915e-05, | |
| "loss": 3.9101, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6353554167652124e-05, | |
| "loss": 3.9158, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6345168220141604e-05, | |
| "loss": 3.9268, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6336782272631084e-05, | |
| "loss": 3.9297, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6328396325120564e-05, | |
| "loss": 3.9266, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.632002675641377e-05, | |
| "loss": 3.9139, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.631164080890326e-05, | |
| "loss": 3.9119, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.630325486139274e-05, | |
| "loss": 3.9269, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.629486891388222e-05, | |
| "loss": 3.9154, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6286515723979164e-05, | |
| "loss": 3.9159, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6278129776468644e-05, | |
| "loss": 3.9312, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.626974382895812e-05, | |
| "loss": 3.9134, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.62613578814476e-05, | |
| "loss": 3.919, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.625297193393708e-05, | |
| "loss": 3.9131, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 3.9988486766815186, | |
| "eval_runtime": 576.8122, | |
| "eval_samples_per_second": 661.552, | |
| "eval_steps_per_second": 20.674, | |
| "step": 228957 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.624460236523029e-05, | |
| "loss": 3.9218, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6236216417719766e-05, | |
| "loss": 3.9198, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6227830470209246e-05, | |
| "loss": 3.9111, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6219444522698726e-05, | |
| "loss": 3.9107, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.621105857518821e-05, | |
| "loss": 3.9063, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.620267262767769e-05, | |
| "loss": 3.9038, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.619428668016717e-05, | |
| "loss": 3.9057, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.618590073265665e-05, | |
| "loss": 3.9013, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.617751478514613e-05, | |
| "loss": 3.908, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.616914521643934e-05, | |
| "loss": 3.9137, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.616075926892882e-05, | |
| "loss": 3.9059, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.61523733214183e-05, | |
| "loss": 3.9073, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.614400375271151e-05, | |
| "loss": 3.9053, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.613561780520099e-05, | |
| "loss": 3.9022, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.612723185769047e-05, | |
| "loss": 3.8991, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.611884591017995e-05, | |
| "loss": 3.8784, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.611045996266943e-05, | |
| "loss": 3.897, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.610207401515891e-05, | |
| "loss": 3.898, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.609368806764839e-05, | |
| "loss": 3.8978, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.608530212013788e-05, | |
| "loss": 3.9141, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.607691617262736e-05, | |
| "loss": 3.8968, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6068546603920566e-05, | |
| "loss": 3.9067, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6060160656410046e-05, | |
| "loss": 3.8961, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6051774708899526e-05, | |
| "loss": 3.8968, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6043388761389006e-05, | |
| "loss": 3.8933, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.603500281387848e-05, | |
| "loss": 3.8955, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.602661686636796e-05, | |
| "loss": 3.8955, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6018247297661175e-05, | |
| "loss": 3.89, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6009861350150655e-05, | |
| "loss": 3.8903, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.600147540264013e-05, | |
| "loss": 3.8783, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5993089455129615e-05, | |
| "loss": 3.8897, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5984703507619095e-05, | |
| "loss": 3.8982, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.597633393891231e-05, | |
| "loss": 3.8913, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5967947991401784e-05, | |
| "loss": 3.8943, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5959562043891264e-05, | |
| "loss": 3.8918, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5951176096380744e-05, | |
| "loss": 3.8944, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5942790148870224e-05, | |
| "loss": 3.8924, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5934404201359704e-05, | |
| "loss": 3.8957, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.592603463265291e-05, | |
| "loss": 3.8884, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.591764868514239e-05, | |
| "loss": 3.881, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.590926273763187e-05, | |
| "loss": 3.8888, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.590087679012135e-05, | |
| "loss": 3.8733, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.589250722141457e-05, | |
| "loss": 3.8953, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.588412127390405e-05, | |
| "loss": 3.8909, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.587573532639353e-05, | |
| "loss": 3.8888, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.586734937888301e-05, | |
| "loss": 3.8824, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.585897981017622e-05, | |
| "loss": 3.881, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.58505938626657e-05, | |
| "loss": 3.8928, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.584220791515518e-05, | |
| "loss": 3.8702, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.583382196764466e-05, | |
| "loss": 3.8782, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.582543602013414e-05, | |
| "loss": 3.8775, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.581705007262362e-05, | |
| "loss": 3.8898, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.580868050391683e-05, | |
| "loss": 3.8713, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.580029455640631e-05, | |
| "loss": 3.8934, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.579190860889579e-05, | |
| "loss": 3.868, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5783522661385267e-05, | |
| "loss": 3.8713, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.577515309267848e-05, | |
| "loss": 3.8844, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.576676714516796e-05, | |
| "loss": 3.8742, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.575838119765744e-05, | |
| "loss": 3.8839, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.574999525014692e-05, | |
| "loss": 3.889, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.57416093026364e-05, | |
| "loss": 3.8647, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.573323973392961e-05, | |
| "loss": 3.8711, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.572485378641909e-05, | |
| "loss": 3.8753, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.571646783890857e-05, | |
| "loss": 3.878, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.570808189139805e-05, | |
| "loss": 3.8685, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.569971232269126e-05, | |
| "loss": 3.874, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.569132637518074e-05, | |
| "loss": 3.8554, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.568294042767022e-05, | |
| "loss": 3.8796, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.567455448015971e-05, | |
| "loss": 3.8678, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.566616853264919e-05, | |
| "loss": 3.866, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.565778258513867e-05, | |
| "loss": 3.8718, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5649413016431876e-05, | |
| "loss": 3.871, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5641027068921356e-05, | |
| "loss": 3.8687, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5632641121410836e-05, | |
| "loss": 3.8801, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5624255173900316e-05, | |
| "loss": 3.8799, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.561586922638979e-05, | |
| "loss": 3.8741, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5607499657683005e-05, | |
| "loss": 3.8651, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5599113710172485e-05, | |
| "loss": 3.8675, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5590727762661965e-05, | |
| "loss": 3.8692, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5582341815151445e-05, | |
| "loss": 3.8813, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5573972246444654e-05, | |
| "loss": 3.872, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.556558629893414e-05, | |
| "loss": 3.8646, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5557200351423614e-05, | |
| "loss": 3.8575, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5548814403913094e-05, | |
| "loss": 3.8618, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.554044483520631e-05, | |
| "loss": 3.8607, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.553205888769579e-05, | |
| "loss": 3.8651, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.552367294018526e-05, | |
| "loss": 3.8672, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.551528699267474e-05, | |
| "loss": 3.8664, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.550690104516422e-05, | |
| "loss": 3.8588, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.549853147645744e-05, | |
| "loss": 3.869, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.549014552894691e-05, | |
| "loss": 3.8571, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.548175958143639e-05, | |
| "loss": 3.8625, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.547337363392588e-05, | |
| "loss": 3.8637, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.546498768641536e-05, | |
| "loss": 3.8518, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.545661811770857e-05, | |
| "loss": 3.8505, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.544823217019805e-05, | |
| "loss": 3.863, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.543984622268753e-05, | |
| "loss": 3.8602, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5431476653980736e-05, | |
| "loss": 3.8649, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5423090706470216e-05, | |
| "loss": 3.8631, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5414704758959696e-05, | |
| "loss": 3.8557, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5406318811449176e-05, | |
| "loss": 3.8555, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5397932863938656e-05, | |
| "loss": 3.8638, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5389563295231865e-05, | |
| "loss": 3.8633, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.538119372652508e-05, | |
| "loss": 3.8516, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.537280777901456e-05, | |
| "loss": 3.8675, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.536442183150404e-05, | |
| "loss": 3.8418, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.535603588399352e-05, | |
| "loss": 3.8486, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5347649936483e-05, | |
| "loss": 3.8635, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.533926398897248e-05, | |
| "loss": 3.8452, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.533087804146196e-05, | |
| "loss": 3.8584, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.532249209395144e-05, | |
| "loss": 3.845, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.531412252524465e-05, | |
| "loss": 3.8416, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.530573657773413e-05, | |
| "loss": 3.8495, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.529735063022361e-05, | |
| "loss": 3.8688, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.528896468271309e-05, | |
| "loss": 3.8491, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.52805951140063e-05, | |
| "loss": 3.8468, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5272209166495786e-05, | |
| "loss": 3.8412, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5263823218985266e-05, | |
| "loss": 3.8503, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5255437271474746e-05, | |
| "loss": 3.8565, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5247067702767955e-05, | |
| "loss": 3.8533, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5238681755257435e-05, | |
| "loss": 3.8468, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5230295807746915e-05, | |
| "loss": 3.8555, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5221909860236395e-05, | |
| "loss": 3.8506, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5213523912725875e-05, | |
| "loss": 3.8484, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5205137965215355e-05, | |
| "loss": 3.8457, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5196752017704834e-05, | |
| "loss": 3.8552, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5188366070194314e-05, | |
| "loss": 3.8437, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5179996501487524e-05, | |
| "loss": 3.8447, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5171610553977003e-05, | |
| "loss": 3.8569, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5163224606466483e-05, | |
| "loss": 3.8366, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.515483865895597e-05, | |
| "loss": 3.8481, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.514645271144545e-05, | |
| "loss": 3.8518, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.513808314273866e-05, | |
| "loss": 3.8558, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.512969719522814e-05, | |
| "loss": 3.8466, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.512131124771762e-05, | |
| "loss": 3.8476, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.51129253002071e-05, | |
| "loss": 3.8367, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.510455573150031e-05, | |
| "loss": 3.8432, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.509616978398979e-05, | |
| "loss": 3.8502, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.508778383647927e-05, | |
| "loss": 3.854, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.507939788896874e-05, | |
| "loss": 3.854, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.507102832026196e-05, | |
| "loss": 3.8427, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.506264237275144e-05, | |
| "loss": 3.8367, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5054256425240924e-05, | |
| "loss": 3.8503, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.50458704777304e-05, | |
| "loss": 3.8414, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.503751728782734e-05, | |
| "loss": 3.8454, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.502913134031682e-05, | |
| "loss": 3.8572, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.50207453928063e-05, | |
| "loss": 3.8444, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.501235944529578e-05, | |
| "loss": 3.8446, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.500397349778526e-05, | |
| "loss": 3.8422, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 3.9590904712677, | |
| "eval_runtime": 583.4335, | |
| "eval_samples_per_second": 654.044, | |
| "eval_steps_per_second": 20.439, | |
| "step": 305276 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.499558755027474e-05, | |
| "loss": 3.8527, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4987201602764215e-05, | |
| "loss": 3.8477, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4978815655253695e-05, | |
| "loss": 3.8395, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4970429707743175e-05, | |
| "loss": 3.8378, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.496204376023266e-05, | |
| "loss": 3.8389, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.495365781272214e-05, | |
| "loss": 3.8325, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.494527186521162e-05, | |
| "loss": 3.8338, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.49368859177011e-05, | |
| "loss": 3.8275, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.492851634899431e-05, | |
| "loss": 3.8377, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.492014678028752e-05, | |
| "loss": 3.8469, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4911760832777e-05, | |
| "loss": 3.8324, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.490337488526648e-05, | |
| "loss": 3.8326, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.489498893775596e-05, | |
| "loss": 3.8377, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.488660299024544e-05, | |
| "loss": 3.8331, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.487821704273492e-05, | |
| "loss": 3.8295, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.48698310952244e-05, | |
| "loss": 3.8087, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486144514771388e-05, | |
| "loss": 3.8263, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4853075579007095e-05, | |
| "loss": 3.8278, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4844689631496575e-05, | |
| "loss": 3.8276, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4836303683986055e-05, | |
| "loss": 3.848, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4827917736475535e-05, | |
| "loss": 3.8248, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4819548167768744e-05, | |
| "loss": 3.8407, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4811162220258224e-05, | |
| "loss": 3.8256, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4802776272747704e-05, | |
| "loss": 3.8332, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4794390325237184e-05, | |
| "loss": 3.8206, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.478602075653039e-05, | |
| "loss": 3.8269, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.477763480901987e-05, | |
| "loss": 3.8287, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.476924886150935e-05, | |
| "loss": 3.8242, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.476086291399883e-05, | |
| "loss": 3.823, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.475249334529205e-05, | |
| "loss": 3.8135, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.474410739778153e-05, | |
| "loss": 3.8171, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.473572145027101e-05, | |
| "loss": 3.8311, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.472733550276049e-05, | |
| "loss": 3.8261, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.47189659340537e-05, | |
| "loss": 3.8277, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.471057998654318e-05, | |
| "loss": 3.8235, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.470219403903266e-05, | |
| "loss": 3.8287, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.469380809152214e-05, | |
| "loss": 3.8256, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.468542214401162e-05, | |
| "loss": 3.8269, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.467705257530483e-05, | |
| "loss": 3.8236, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466866662779431e-05, | |
| "loss": 3.8129, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466028068028379e-05, | |
| "loss": 3.829, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.465189473277327e-05, | |
| "loss": 3.8057, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.464352516406648e-05, | |
| "loss": 3.8277, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.463513921655596e-05, | |
| "loss": 3.8227, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.462675326904544e-05, | |
| "loss": 3.8304, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.461836732153492e-05, | |
| "loss": 3.8187, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.460999775282813e-05, | |
| "loss": 3.8131, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.460161180531761e-05, | |
| "loss": 3.8281, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.459322585780709e-05, | |
| "loss": 3.8079, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.458483991029657e-05, | |
| "loss": 3.8117, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.457647034158978e-05, | |
| "loss": 3.8152, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.456808439407926e-05, | |
| "loss": 3.8201, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.455969844656874e-05, | |
| "loss": 3.8118, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.455131249905822e-05, | |
| "loss": 3.8224, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4542942930351436e-05, | |
| "loss": 3.8059, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4534556982840916e-05, | |
| "loss": 3.8089, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4526171035330396e-05, | |
| "loss": 3.8194, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4517785087819876e-05, | |
| "loss": 3.8064, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4509415519113085e-05, | |
| "loss": 3.8248, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4501029571602565e-05, | |
| "loss": 3.8286, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4492643624092045e-05, | |
| "loss": 3.7972, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4484257676581525e-05, | |
| "loss": 3.8065, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4475888107874734e-05, | |
| "loss": 3.813, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4467502160364214e-05, | |
| "loss": 3.8199, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4459116212853694e-05, | |
| "loss": 3.8061, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4450730265343174e-05, | |
| "loss": 3.8082, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.444236069663639e-05, | |
| "loss": 3.7969, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.443397474912587e-05, | |
| "loss": 3.8127, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.442558880161535e-05, | |
| "loss": 3.8074, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.441721923290856e-05, | |
| "loss": 3.8026, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440883328539804e-05, | |
| "loss": 3.8106, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440044733788752e-05, | |
| "loss": 3.81, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4392061390377e-05, | |
| "loss": 3.8067, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.438369182167021e-05, | |
| "loss": 3.819, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.437530587415969e-05, | |
| "loss": 3.8159, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.436691992664917e-05, | |
| "loss": 3.813, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.435853397913865e-05, | |
| "loss": 3.8056, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4350164410431864e-05, | |
| "loss": 3.8087, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4341778462921344e-05, | |
| "loss": 3.8032, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4333392515410824e-05, | |
| "loss": 3.8224, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.43250065679003e-05, | |
| "loss": 3.8107, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.431663699919351e-05, | |
| "loss": 3.8061, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.430825105168299e-05, | |
| "loss": 3.7914, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.429986510417247e-05, | |
| "loss": 3.808, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4291479156661946e-05, | |
| "loss": 3.7969, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.428310958795516e-05, | |
| "loss": 3.8092, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.427472364044464e-05, | |
| "loss": 3.8041, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.426633769293412e-05, | |
| "loss": 3.8052, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.42579517454236e-05, | |
| "loss": 3.7994, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424958217671682e-05, | |
| "loss": 3.8136, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.42411962292063e-05, | |
| "loss": 3.7956, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.423281028169577e-05, | |
| "loss": 3.8026, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.422442433418525e-05, | |
| "loss": 3.8087, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4216054765478466e-05, | |
| "loss": 3.7883, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4207668817967946e-05, | |
| "loss": 3.7916, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419928287045742e-05, | |
| "loss": 3.806, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.41908969229469e-05, | |
| "loss": 3.7969, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4182527354240115e-05, | |
| "loss": 3.8125, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4174141406729595e-05, | |
| "loss": 3.8033, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.416575545921907e-05, | |
| "loss": 3.7925, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4157369511708555e-05, | |
| "loss": 3.7974, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.414899994300177e-05, | |
| "loss": 3.8044, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4140613995491244e-05, | |
| "loss": 3.8051, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.413224442678446e-05, | |
| "loss": 3.7958, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.412385847927394e-05, | |
| "loss": 3.8081, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.411547253176342e-05, | |
| "loss": 3.7851, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.410708658425289e-05, | |
| "loss": 3.7908, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.409870063674237e-05, | |
| "loss": 3.8036, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.409031468923185e-05, | |
| "loss": 3.7841, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.408192874172133e-05, | |
| "loss": 3.8005, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.407354279421081e-05, | |
| "loss": 3.7966, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.406517322550402e-05, | |
| "loss": 3.7773, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.405678727799351e-05, | |
| "loss": 3.7922, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.404840133048299e-05, | |
| "loss": 3.8097, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.404001538297247e-05, | |
| "loss": 3.7911, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403164581426568e-05, | |
| "loss": 3.7923, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.402325986675516e-05, | |
| "loss": 3.7879, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.401487391924464e-05, | |
| "loss": 3.7912, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.400648797173412e-05, | |
| "loss": 3.8007, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.399813478183106e-05, | |
| "loss": 3.7966, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.398974883432054e-05, | |
| "loss": 3.7898, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3981362886810016e-05, | |
| "loss": 3.7939, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3972976939299496e-05, | |
| "loss": 3.7948, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3964590991788976e-05, | |
| "loss": 3.7925, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3956205044278456e-05, | |
| "loss": 3.7892, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.394781909676794e-05, | |
| "loss": 3.7989, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.393943314925742e-05, | |
| "loss": 3.7836, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.393106358055063e-05, | |
| "loss": 3.7946, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.392267763304011e-05, | |
| "loss": 3.7976, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.391429168552959e-05, | |
| "loss": 3.784, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.390590573801907e-05, | |
| "loss": 3.7889, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.389753616931228e-05, | |
| "loss": 3.7972, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.388915022180176e-05, | |
| "loss": 3.7959, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.388076427429124e-05, | |
| "loss": 3.7952, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.387237832678072e-05, | |
| "loss": 3.7904, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.386400875807393e-05, | |
| "loss": 3.781, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.385562281056341e-05, | |
| "loss": 3.7891, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3847236863052896e-05, | |
| "loss": 3.7929, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3838850915542376e-05, | |
| "loss": 3.7996, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3830481346835585e-05, | |
| "loss": 3.8009, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3822095399325065e-05, | |
| "loss": 3.7896, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3813709451814545e-05, | |
| "loss": 3.7824, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3805323504304025e-05, | |
| "loss": 3.7962, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3796953935597234e-05, | |
| "loss": 3.7854, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.378858436689044e-05, | |
| "loss": 3.79, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.378019841937992e-05, | |
| "loss": 3.8046, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.37718124718694e-05, | |
| "loss": 3.793, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.376342652435888e-05, | |
| "loss": 3.7833, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.375504057684836e-05, | |
| "loss": 3.7903, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 3.93572735786438, | |
| "eval_runtime": 576.4532, | |
| "eval_samples_per_second": 661.964, | |
| "eval_steps_per_second": 20.687, | |
| "step": 381595 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.374665462933785e-05, | |
| "loss": 3.7941, | |
| "step": 381952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.373826868182733e-05, | |
| "loss": 3.7962, | |
| "step": 382464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372988273431681e-05, | |
| "loss": 3.7825, | |
| "step": 382976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372151316561002e-05, | |
| "loss": 3.7804, | |
| "step": 383488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.37131272180995e-05, | |
| "loss": 3.7887, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.370475764939271e-05, | |
| "loss": 3.7806, | |
| "step": 384512 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.369637170188219e-05, | |
| "loss": 3.7819, | |
| "step": 385024 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.368798575437167e-05, | |
| "loss": 3.7712, | |
| "step": 385536 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367959980686115e-05, | |
| "loss": 3.7852, | |
| "step": 386048 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367123023815436e-05, | |
| "loss": 3.7972, | |
| "step": 386560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.366284429064384e-05, | |
| "loss": 3.7765, | |
| "step": 387072 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.365445834313332e-05, | |
| "loss": 3.7784, | |
| "step": 387584 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.36460723956228e-05, | |
| "loss": 3.7864, | |
| "step": 388096 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.363768644811228e-05, | |
| "loss": 3.7781, | |
| "step": 388608 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.362930050060176e-05, | |
| "loss": 3.78, | |
| "step": 389120 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.362091455309124e-05, | |
| "loss": 3.7568, | |
| "step": 389632 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.361252860558072e-05, | |
| "loss": 3.7708, | |
| "step": 390144 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.360415903687393e-05, | |
| "loss": 3.7761, | |
| "step": 390656 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.359577308936341e-05, | |
| "loss": 3.7742, | |
| "step": 391168 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.358738714185289e-05, | |
| "loss": 3.7972, | |
| "step": 391680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357900119434237e-05, | |
| "loss": 3.7729, | |
| "step": 392192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357063162563558e-05, | |
| "loss": 3.788, | |
| "step": 392704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.356224567812506e-05, | |
| "loss": 3.7743, | |
| "step": 393216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.355385973061454e-05, | |
| "loss": 3.7805, | |
| "step": 393728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.354547378310402e-05, | |
| "loss": 3.7681, | |
| "step": 394240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.353710421439724e-05, | |
| "loss": 3.773, | |
| "step": 394752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.352871826688672e-05, | |
| "loss": 3.7779, | |
| "step": 395264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.35203323193762e-05, | |
| "loss": 3.7755, | |
| "step": 395776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.351194637186568e-05, | |
| "loss": 3.7704, | |
| "step": 396288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3503576803158886e-05, | |
| "loss": 3.7667, | |
| "step": 396800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3495190855648366e-05, | |
| "loss": 3.7606, | |
| "step": 397312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3486804908137846e-05, | |
| "loss": 3.782, | |
| "step": 397824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.347841896062732e-05, | |
| "loss": 3.7776, | |
| "step": 398336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.34700330131168e-05, | |
| "loss": 3.7744, | |
| "step": 398848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.346164706560628e-05, | |
| "loss": 3.7746, | |
| "step": 399360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.345326111809576e-05, | |
| "loss": 3.7776, | |
| "step": 399872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.344487517058524e-05, | |
| "loss": 3.7783, | |
| "step": 400384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3436505601878455e-05, | |
| "loss": 3.7729, | |
| "step": 400896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3428119654367935e-05, | |
| "loss": 3.7751, | |
| "step": 401408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3419733706857415e-05, | |
| "loss": 3.7633, | |
| "step": 401920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3411364138150624e-05, | |
| "loss": 3.7757, | |
| "step": 402432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3402978190640104e-05, | |
| "loss": 3.759, | |
| "step": 402944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3394592243129584e-05, | |
| "loss": 3.778, | |
| "step": 403456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3386206295619064e-05, | |
| "loss": 3.7692, | |
| "step": 403968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.337783672691227e-05, | |
| "loss": 3.7781, | |
| "step": 404480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.336945077940175e-05, | |
| "loss": 3.7688, | |
| "step": 404992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.336106483189123e-05, | |
| "loss": 3.7649, | |
| "step": 405504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.335267888438071e-05, | |
| "loss": 3.7807, | |
| "step": 406016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.334430931567393e-05, | |
| "loss": 3.7555, | |
| "step": 406528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.333592336816341e-05, | |
| "loss": 3.7643, | |
| "step": 407040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.332753742065289e-05, | |
| "loss": 3.7644, | |
| "step": 407552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331915147314237e-05, | |
| "loss": 3.7674, | |
| "step": 408064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331078190443558e-05, | |
| "loss": 3.7643, | |
| "step": 408576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.330239595692506e-05, | |
| "loss": 3.7694, | |
| "step": 409088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.329401000941454e-05, | |
| "loss": 3.76, | |
| "step": 409600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.328562406190402e-05, | |
| "loss": 3.7574, | |
| "step": 410112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3277254493197226e-05, | |
| "loss": 3.7711, | |
| "step": 410624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3268868545686706e-05, | |
| "loss": 3.7613, | |
| "step": 411136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3260482598176186e-05, | |
| "loss": 3.7722, | |
| "step": 411648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3252096650665666e-05, | |
| "loss": 3.7768, | |
| "step": 412160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.324372708195888e-05, | |
| "loss": 3.7529, | |
| "step": 412672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.323534113444836e-05, | |
| "loss": 3.7573, | |
| "step": 413184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.322695518693784e-05, | |
| "loss": 3.7599, | |
| "step": 413696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.321856923942732e-05, | |
| "loss": 3.7756, | |
| "step": 414208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.321019967072053e-05, | |
| "loss": 3.7575, | |
| "step": 414720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320181372321001e-05, | |
| "loss": 3.7623, | |
| "step": 415232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.319344415450322e-05, | |
| "loss": 3.749, | |
| "step": 415744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.31850582069927e-05, | |
| "loss": 3.766, | |
| "step": 416256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.317667225948218e-05, | |
| "loss": 3.7553, | |
| "step": 416768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.316828631197166e-05, | |
| "loss": 3.7556, | |
| "step": 417280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315990036446114e-05, | |
| "loss": 3.7614, | |
| "step": 417792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315151441695062e-05, | |
| "loss": 3.7623, | |
| "step": 418304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.31431284694401e-05, | |
| "loss": 3.7582, | |
| "step": 418816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3134758900733316e-05, | |
| "loss": 3.7699, | |
| "step": 419328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3126372953222796e-05, | |
| "loss": 3.7704, | |
| "step": 419840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3117987005712276e-05, | |
| "loss": 3.7647, | |
| "step": 420352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3109601058201756e-05, | |
| "loss": 3.7546, | |
| "step": 420864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3101215110691236e-05, | |
| "loss": 3.765, | |
| "step": 421376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3092829163180715e-05, | |
| "loss": 3.7549, | |
| "step": 421888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3084443215670195e-05, | |
| "loss": 3.7741, | |
| "step": 422400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3076057268159675e-05, | |
| "loss": 3.7594, | |
| "step": 422912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3067687699452885e-05, | |
| "loss": 3.7631, | |
| "step": 423424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3059301751942364e-05, | |
| "loss": 3.7454, | |
| "step": 423936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3050915804431844e-05, | |
| "loss": 3.7603, | |
| "step": 424448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3042529856921324e-05, | |
| "loss": 3.7464, | |
| "step": 424960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.303417666701827e-05, | |
| "loss": 3.7637, | |
| "step": 425472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.302579071950775e-05, | |
| "loss": 3.758, | |
| "step": 425984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.301740477199723e-05, | |
| "loss": 3.7583, | |
| "step": 426496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.300901882448671e-05, | |
| "loss": 3.7571, | |
| "step": 427008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.300063287697619e-05, | |
| "loss": 3.7691, | |
| "step": 427520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.299224692946567e-05, | |
| "loss": 3.742, | |
| "step": 428032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.298386098195515e-05, | |
| "loss": 3.7551, | |
| "step": 428544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.297547503444463e-05, | |
| "loss": 3.7619, | |
| "step": 429056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.296710546573784e-05, | |
| "loss": 3.7442, | |
| "step": 429568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.295871951822732e-05, | |
| "loss": 3.7466, | |
| "step": 430080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.29503335707168e-05, | |
| "loss": 3.7606, | |
| "step": 430592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294194762320628e-05, | |
| "loss": 3.7466, | |
| "step": 431104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.293357805449949e-05, | |
| "loss": 3.7708, | |
| "step": 431616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2925192106988974e-05, | |
| "loss": 3.7544, | |
| "step": 432128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2916806159478454e-05, | |
| "loss": 3.7468, | |
| "step": 432640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.290842021196793e-05, | |
| "loss": 3.7511, | |
| "step": 433152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.290006702206487e-05, | |
| "loss": 3.7599, | |
| "step": 433664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289168107455435e-05, | |
| "loss": 3.7606, | |
| "step": 434176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.288329512704383e-05, | |
| "loss": 3.7477, | |
| "step": 434688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.287490917953331e-05, | |
| "loss": 3.7627, | |
| "step": 435200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.286652323202279e-05, | |
| "loss": 3.7392, | |
| "step": 435712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.285813728451227e-05, | |
| "loss": 3.7457, | |
| "step": 436224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284976771580548e-05, | |
| "loss": 3.7561, | |
| "step": 436736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284138176829496e-05, | |
| "loss": 3.7415, | |
| "step": 437248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.283299582078444e-05, | |
| "loss": 3.7516, | |
| "step": 437760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.282460987327393e-05, | |
| "loss": 3.7565, | |
| "step": 438272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.28162239257634e-05, | |
| "loss": 3.7294, | |
| "step": 438784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.280783797825288e-05, | |
| "loss": 3.7491, | |
| "step": 439296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2799468409546097e-05, | |
| "loss": 3.7604, | |
| "step": 439808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2791082462035576e-05, | |
| "loss": 3.7505, | |
| "step": 440320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.278269651452505e-05, | |
| "loss": 3.7427, | |
| "step": 440832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.277431056701453e-05, | |
| "loss": 3.7407, | |
| "step": 441344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.276592461950401e-05, | |
| "loss": 3.7477, | |
| "step": 441856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.275753867199349e-05, | |
| "loss": 3.7564, | |
| "step": 442368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.27491691032867e-05, | |
| "loss": 3.7534, | |
| "step": 442880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.274078315577618e-05, | |
| "loss": 3.7432, | |
| "step": 443392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2732397208265665e-05, | |
| "loss": 3.7481, | |
| "step": 443904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2724011260755145e-05, | |
| "loss": 3.7499, | |
| "step": 444416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2715625313244625e-05, | |
| "loss": 3.7499, | |
| "step": 444928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2707239365734105e-05, | |
| "loss": 3.7426, | |
| "step": 445440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2698853418223585e-05, | |
| "loss": 3.7505, | |
| "step": 445952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2690467470713065e-05, | |
| "loss": 3.7431, | |
| "step": 446464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.268211428081e-05, | |
| "loss": 3.7512, | |
| "step": 446976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.267372833329948e-05, | |
| "loss": 3.7521, | |
| "step": 447488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.266534238578896e-05, | |
| "loss": 3.7367, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.265695643827844e-05, | |
| "loss": 3.7455, | |
| "step": 448512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.264857049076792e-05, | |
| "loss": 3.7541, | |
| "step": 449024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.26401845432574e-05, | |
| "loss": 3.7517, | |
| "step": 449536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.263181497455062e-05, | |
| "loss": 3.7479, | |
| "step": 450048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.26234290270401e-05, | |
| "loss": 3.7481, | |
| "step": 450560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.261504307952958e-05, | |
| "loss": 3.7359, | |
| "step": 451072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.260665713201906e-05, | |
| "loss": 3.7418, | |
| "step": 451584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.259827118450854e-05, | |
| "loss": 3.7485, | |
| "step": 452096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.258990161580175e-05, | |
| "loss": 3.7603, | |
| "step": 452608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.258151566829123e-05, | |
| "loss": 3.7545, | |
| "step": 453120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.257312972078071e-05, | |
| "loss": 3.7429, | |
| "step": 453632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.256474377327019e-05, | |
| "loss": 3.7451, | |
| "step": 454144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.255635782575967e-05, | |
| "loss": 3.7488, | |
| "step": 454656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.254797187824915e-05, | |
| "loss": 3.7451, | |
| "step": 455168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253960230954236e-05, | |
| "loss": 3.7449, | |
| "step": 455680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253121636203184e-05, | |
| "loss": 3.762, | |
| "step": 456192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.252283041452132e-05, | |
| "loss": 3.7476, | |
| "step": 456704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.25144444670108e-05, | |
| "loss": 3.7391, | |
| "step": 457216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.250605851950028e-05, | |
| "loss": 3.7462, | |
| "step": 457728 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 3.919872999191284, | |
| "eval_runtime": 554.6551, | |
| "eval_samples_per_second": 687.979, | |
| "eval_steps_per_second": 21.5, | |
| "step": 457914 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.249768895079349e-05, | |
| "loss": 3.7548, | |
| "step": 458240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.248930300328297e-05, | |
| "loss": 3.7516, | |
| "step": 458752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.248091705577245e-05, | |
| "loss": 3.7438, | |
| "step": 459264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.247253110826193e-05, | |
| "loss": 3.7407, | |
| "step": 459776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.246416153955514e-05, | |
| "loss": 3.7424, | |
| "step": 460288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.245577559204462e-05, | |
| "loss": 3.7366, | |
| "step": 460800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.24473896445341e-05, | |
| "loss": 3.7381, | |
| "step": 461312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243900369702358e-05, | |
| "loss": 3.7295, | |
| "step": 461824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2430617749513054e-05, | |
| "loss": 3.74, | |
| "step": 462336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2422231802002534e-05, | |
| "loss": 3.7558, | |
| "step": 462848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.241386223329575e-05, | |
| "loss": 3.7358, | |
| "step": 463360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.240547628578524e-05, | |
| "loss": 3.7352, | |
| "step": 463872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.239709033827471e-05, | |
| "loss": 3.7438, | |
| "step": 464384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.238870439076419e-05, | |
| "loss": 3.7353, | |
| "step": 464896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.238031844325367e-05, | |
| "loss": 3.7408, | |
| "step": 465408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2371948874546886e-05, | |
| "loss": 3.7145, | |
| "step": 465920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.236356292703636e-05, | |
| "loss": 3.7262, | |
| "step": 466432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.235517697952584e-05, | |
| "loss": 3.7343, | |
| "step": 466944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.234679103201532e-05, | |
| "loss": 3.7323, | |
| "step": 467456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.23384050845048e-05, | |
| "loss": 3.7578, | |
| "step": 467968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.233001913699428e-05, | |
| "loss": 3.7314, | |
| "step": 468480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.232163318948376e-05, | |
| "loss": 3.7465, | |
| "step": 468992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.231324724197324e-05, | |
| "loss": 3.7288, | |
| "step": 469504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2304877673266455e-05, | |
| "loss": 3.737, | |
| "step": 470016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2296491725755935e-05, | |
| "loss": 3.7269, | |
| "step": 470528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2288105778245415e-05, | |
| "loss": 3.7335, | |
| "step": 471040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2279719830734895e-05, | |
| "loss": 3.7363, | |
| "step": 471552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2271350262028104e-05, | |
| "loss": 3.7298, | |
| "step": 472064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2262964314517584e-05, | |
| "loss": 3.7333, | |
| "step": 472576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2254578367007064e-05, | |
| "loss": 3.7214, | |
| "step": 473088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2246192419496544e-05, | |
| "loss": 3.7186, | |
| "step": 473600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.223782285078975e-05, | |
| "loss": 3.7397, | |
| "step": 474112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.222943690327923e-05, | |
| "loss": 3.7363, | |
| "step": 474624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.222105095576871e-05, | |
| "loss": 3.7369, | |
| "step": 475136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.221266500825819e-05, | |
| "loss": 3.7298, | |
| "step": 475648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.220427906074767e-05, | |
| "loss": 3.7384, | |
| "step": 476160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.219589311323716e-05, | |
| "loss": 3.7323, | |
| "step": 476672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.218750716572664e-05, | |
| "loss": 3.7371, | |
| "step": 477184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.217913759701985e-05, | |
| "loss": 3.7309, | |
| "step": 477696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.217075164950933e-05, | |
| "loss": 3.7245, | |
| "step": 478208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.216236570199881e-05, | |
| "loss": 3.7361, | |
| "step": 478720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.215397975448829e-05, | |
| "loss": 3.7155, | |
| "step": 479232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.214559380697777e-05, | |
| "loss": 3.7369, | |
| "step": 479744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.213722423827098e-05, | |
| "loss": 3.7301, | |
| "step": 480256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.212883829076046e-05, | |
| "loss": 3.7372, | |
| "step": 480768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.212045234324994e-05, | |
| "loss": 3.7301, | |
| "step": 481280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.211206639573942e-05, | |
| "loss": 3.7224, | |
| "step": 481792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.21036804482289e-05, | |
| "loss": 3.7384, | |
| "step": 482304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.209531087952211e-05, | |
| "loss": 3.7192, | |
| "step": 482816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.208692493201159e-05, | |
| "loss": 3.7238, | |
| "step": 483328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.207853898450107e-05, | |
| "loss": 3.7235, | |
| "step": 483840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2070153036990546e-05, | |
| "loss": 3.726, | |
| "step": 484352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.206178346828376e-05, | |
| "loss": 3.7274, | |
| "step": 484864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.205339752077324e-05, | |
| "loss": 3.7248, | |
| "step": 485376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.204501157326272e-05, | |
| "loss": 3.7206, | |
| "step": 485888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2036625625752195e-05, | |
| "loss": 3.7232, | |
| "step": 486400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.202825605704541e-05, | |
| "loss": 3.7264, | |
| "step": 486912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201987010953489e-05, | |
| "loss": 3.7227, | |
| "step": 487424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2011484162024364e-05, | |
| "loss": 3.7319, | |
| "step": 487936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.200309821451385e-05, | |
| "loss": 3.738, | |
| "step": 488448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.199471226700333e-05, | |
| "loss": 3.7156, | |
| "step": 488960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1986342698296547e-05, | |
| "loss": 3.7148, | |
| "step": 489472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.197795675078602e-05, | |
| "loss": 3.7181, | |
| "step": 489984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.19695708032755e-05, | |
| "loss": 3.7393, | |
| "step": 490496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.196118485576498e-05, | |
| "loss": 3.7211, | |
| "step": 491008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.195279890825446e-05, | |
| "loss": 3.7209, | |
| "step": 491520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.194442933954767e-05, | |
| "loss": 3.7115, | |
| "step": 492032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.193604339203715e-05, | |
| "loss": 3.7217, | |
| "step": 492544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.192765744452663e-05, | |
| "loss": 3.7191, | |
| "step": 493056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.191927149701611e-05, | |
| "loss": 3.7166, | |
| "step": 493568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.191088554950559e-05, | |
| "loss": 3.7221, | |
| "step": 494080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.190249960199507e-05, | |
| "loss": 3.7207, | |
| "step": 494592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1894130033288284e-05, | |
| "loss": 3.7204, | |
| "step": 495104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1885744085777764e-05, | |
| "loss": 3.7264, | |
| "step": 495616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1877358138267244e-05, | |
| "loss": 3.7317, | |
| "step": 496128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1868972190756724e-05, | |
| "loss": 3.726, | |
| "step": 496640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.186060262204993e-05, | |
| "loss": 3.7181, | |
| "step": 497152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.185221667453941e-05, | |
| "loss": 3.7227, | |
| "step": 497664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.184383072702889e-05, | |
| "loss": 3.7184, | |
| "step": 498176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.183544477951837e-05, | |
| "loss": 3.7353, | |
| "step": 498688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.182705883200785e-05, | |
| "loss": 3.7223, | |
| "step": 499200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.181868926330106e-05, | |
| "loss": 3.7256, | |
| "step": 499712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.181030331579054e-05, | |
| "loss": 3.7055, | |
| "step": 500224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.180191736828002e-05, | |
| "loss": 3.7261, | |
| "step": 500736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.17935314207695e-05, | |
| "loss": 3.7053, | |
| "step": 501248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.178516185206272e-05, | |
| "loss": 3.7244, | |
| "step": 501760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.17767759045522e-05, | |
| "loss": 3.7204, | |
| "step": 502272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.176838995704168e-05, | |
| "loss": 3.721, | |
| "step": 502784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.176000400953116e-05, | |
| "loss": 3.7173, | |
| "step": 503296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.175161806202064e-05, | |
| "loss": 3.7287, | |
| "step": 503808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.174324849331385e-05, | |
| "loss": 3.7095, | |
| "step": 504320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.173486254580333e-05, | |
| "loss": 3.7174, | |
| "step": 504832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.172647659829281e-05, | |
| "loss": 3.7237, | |
| "step": 505344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.171809065078229e-05, | |
| "loss": 3.7085, | |
| "step": 505856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1709721082075496e-05, | |
| "loss": 3.7096, | |
| "step": 506368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1701335134564976e-05, | |
| "loss": 3.7185, | |
| "step": 506880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1692949187054456e-05, | |
| "loss": 3.7097, | |
| "step": 507392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1684563239543936e-05, | |
| "loss": 3.734, | |
| "step": 507904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.167617729203342e-05, | |
| "loss": 3.7163, | |
| "step": 508416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.166780772332663e-05, | |
| "loss": 3.7119, | |
| "step": 508928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165942177581611e-05, | |
| "loss": 3.7134, | |
| "step": 509440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165103582830559e-05, | |
| "loss": 3.7201, | |
| "step": 509952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.164264988079507e-05, | |
| "loss": 3.7233, | |
| "step": 510464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.163428031208828e-05, | |
| "loss": 3.7108, | |
| "step": 510976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.162589436457776e-05, | |
| "loss": 3.7229, | |
| "step": 511488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.161750841706724e-05, | |
| "loss": 3.7048, | |
| "step": 512000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160912246955672e-05, | |
| "loss": 3.7058, | |
| "step": 512512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160075290084993e-05, | |
| "loss": 3.7158, | |
| "step": 513024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.159236695333941e-05, | |
| "loss": 3.7079, | |
| "step": 513536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.158398100582889e-05, | |
| "loss": 3.7139, | |
| "step": 514048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1575595058318376e-05, | |
| "loss": 3.7176, | |
| "step": 514560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1567225489611585e-05, | |
| "loss": 3.6935, | |
| "step": 515072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1558839542101065e-05, | |
| "loss": 3.7115, | |
| "step": 515584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1550453594590545e-05, | |
| "loss": 3.7277, | |
| "step": 516096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1542067647080025e-05, | |
| "loss": 3.7158, | |
| "step": 516608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.15336816995695e-05, | |
| "loss": 3.7015, | |
| "step": 517120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1525312130862714e-05, | |
| "loss": 3.7037, | |
| "step": 517632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1516926183352194e-05, | |
| "loss": 3.7096, | |
| "step": 518144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1508540235841674e-05, | |
| "loss": 3.7167, | |
| "step": 518656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.150015428833115e-05, | |
| "loss": 3.7165, | |
| "step": 519168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.149178471962436e-05, | |
| "loss": 3.7061, | |
| "step": 519680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.148339877211384e-05, | |
| "loss": 3.7105, | |
| "step": 520192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.147501282460333e-05, | |
| "loss": 3.7143, | |
| "step": 520704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.14666268770928e-05, | |
| "loss": 3.7126, | |
| "step": 521216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.145825730838602e-05, | |
| "loss": 3.7041, | |
| "step": 521728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.14498713608755e-05, | |
| "loss": 3.715, | |
| "step": 522240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.144148541336497e-05, | |
| "loss": 3.707, | |
| "step": 522752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.143309946585445e-05, | |
| "loss": 3.7151, | |
| "step": 523264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.142471351834393e-05, | |
| "loss": 3.7136, | |
| "step": 523776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.141632757083341e-05, | |
| "loss": 3.6993, | |
| "step": 524288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.140795800212662e-05, | |
| "loss": 3.7095, | |
| "step": 524800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.13995720546161e-05, | |
| "loss": 3.7145, | |
| "step": 525312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.139118610710558e-05, | |
| "loss": 3.7163, | |
| "step": 525824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.138280015959507e-05, | |
| "loss": 3.7118, | |
| "step": 526336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.137443059088828e-05, | |
| "loss": 3.714, | |
| "step": 526848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.136604464337776e-05, | |
| "loss": 3.6988, | |
| "step": 527360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1357658695867237e-05, | |
| "loss": 3.7022, | |
| "step": 527872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1349272748356717e-05, | |
| "loss": 3.712, | |
| "step": 528384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1340903179649926e-05, | |
| "loss": 3.7248, | |
| "step": 528896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1332517232139406e-05, | |
| "loss": 3.7176, | |
| "step": 529408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1324131284628886e-05, | |
| "loss": 3.708, | |
| "step": 529920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1315745337118366e-05, | |
| "loss": 3.7039, | |
| "step": 530432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1307375768411575e-05, | |
| "loss": 3.7147, | |
| "step": 530944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1298989820901055e-05, | |
| "loss": 3.7074, | |
| "step": 531456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1290603873390535e-05, | |
| "loss": 3.7087, | |
| "step": 531968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.128221792588002e-05, | |
| "loss": 3.7263, | |
| "step": 532480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.127384835717323e-05, | |
| "loss": 3.7138, | |
| "step": 532992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.126546240966271e-05, | |
| "loss": 3.7031, | |
| "step": 533504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.125707646215219e-05, | |
| "loss": 3.7113, | |
| "step": 534016 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 3.910520076751709, | |
| "eval_runtime": 572.6756, | |
| "eval_samples_per_second": 666.33, | |
| "eval_steps_per_second": 20.823, | |
| "step": 534233 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.124869051464167e-05, | |
| "loss": 3.7176, | |
| "step": 534528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.124030456713115e-05, | |
| "loss": 3.7165, | |
| "step": 535040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.123191861962063e-05, | |
| "loss": 3.7093, | |
| "step": 535552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.122353267211011e-05, | |
| "loss": 3.7046, | |
| "step": 536064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.121516310340332e-05, | |
| "loss": 3.7036, | |
| "step": 536576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.12067771558928e-05, | |
| "loss": 3.7056, | |
| "step": 537088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.119839120838228e-05, | |
| "loss": 3.7012, | |
| "step": 537600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.119000526087176e-05, | |
| "loss": 3.6956, | |
| "step": 538112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.118163569216497e-05, | |
| "loss": 3.7024, | |
| "step": 538624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1173249744654455e-05, | |
| "loss": 3.7172, | |
| "step": 539136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1164880175947664e-05, | |
| "loss": 3.7013, | |
| "step": 539648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1156494228437144e-05, | |
| "loss": 3.6992, | |
| "step": 540160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1148108280926624e-05, | |
| "loss": 3.7123, | |
| "step": 540672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1139722333416104e-05, | |
| "loss": 3.6965, | |
| "step": 541184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1131336385905584e-05, | |
| "loss": 3.7088, | |
| "step": 541696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1122950438395064e-05, | |
| "loss": 3.6817, | |
| "step": 542208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1114564490884544e-05, | |
| "loss": 3.6859, | |
| "step": 542720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1106178543374024e-05, | |
| "loss": 3.7033, | |
| "step": 543232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.109780897466723e-05, | |
| "loss": 3.6948, | |
| "step": 543744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.108942302715671e-05, | |
| "loss": 3.7212, | |
| "step": 544256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.108103707964619e-05, | |
| "loss": 3.6988, | |
| "step": 544768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.107265113213567e-05, | |
| "loss": 3.7093, | |
| "step": 545280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.106428156342889e-05, | |
| "loss": 3.6953, | |
| "step": 545792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.105589561591837e-05, | |
| "loss": 3.7063, | |
| "step": 546304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.104750966840785e-05, | |
| "loss": 3.6892, | |
| "step": 546816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.103912372089733e-05, | |
| "loss": 3.698, | |
| "step": 547328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.103075415219054e-05, | |
| "loss": 3.7017, | |
| "step": 547840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.102236820468002e-05, | |
| "loss": 3.6983, | |
| "step": 548352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.10139822571695e-05, | |
| "loss": 3.6939, | |
| "step": 548864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.100559630965898e-05, | |
| "loss": 3.6918, | |
| "step": 549376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0997226740952186e-05, | |
| "loss": 3.6826, | |
| "step": 549888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0988840793441666e-05, | |
| "loss": 3.7051, | |
| "step": 550400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0980454845931146e-05, | |
| "loss": 3.7007, | |
| "step": 550912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0972068898420626e-05, | |
| "loss": 3.6987, | |
| "step": 551424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0963682950910106e-05, | |
| "loss": 3.6983, | |
| "step": 551936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.095531338220332e-05, | |
| "loss": 3.7044, | |
| "step": 552448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.09469274346928e-05, | |
| "loss": 3.696, | |
| "step": 552960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.093854148718228e-05, | |
| "loss": 3.6978, | |
| "step": 553472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0930155539671755e-05, | |
| "loss": 3.6966, | |
| "step": 553984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.092178597096497e-05, | |
| "loss": 3.6968, | |
| "step": 554496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.091340002345445e-05, | |
| "loss": 3.6975, | |
| "step": 555008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.090501407594393e-05, | |
| "loss": 3.6843, | |
| "step": 555520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0896628128433404e-05, | |
| "loss": 3.7069, | |
| "step": 556032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.088825855972662e-05, | |
| "loss": 3.6928, | |
| "step": 556544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.08798726122161e-05, | |
| "loss": 3.7049, | |
| "step": 557056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.087148666470558e-05, | |
| "loss": 3.6982, | |
| "step": 557568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.086310071719506e-05, | |
| "loss": 3.6878, | |
| "step": 558080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.085471476968454e-05, | |
| "loss": 3.7056, | |
| "step": 558592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0846345200977756e-05, | |
| "loss": 3.6839, | |
| "step": 559104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.083795925346723e-05, | |
| "loss": 3.692, | |
| "step": 559616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.082957330595671e-05, | |
| "loss": 3.6843, | |
| "step": 560128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.082118735844619e-05, | |
| "loss": 3.6944, | |
| "step": 560640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0812817789739405e-05, | |
| "loss": 3.6925, | |
| "step": 561152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.080443184222888e-05, | |
| "loss": 3.6904, | |
| "step": 561664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.079604589471836e-05, | |
| "loss": 3.6884, | |
| "step": 562176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.078765994720784e-05, | |
| "loss": 3.6884, | |
| "step": 562688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0779290378501054e-05, | |
| "loss": 3.692, | |
| "step": 563200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0770904430990534e-05, | |
| "loss": 3.6921, | |
| "step": 563712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0762518483480014e-05, | |
| "loss": 3.701, | |
| "step": 564224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0754132535969494e-05, | |
| "loss": 3.6991, | |
| "step": 564736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.07457629672627e-05, | |
| "loss": 3.6852, | |
| "step": 565248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.073737701975218e-05, | |
| "loss": 3.6791, | |
| "step": 565760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.072899107224166e-05, | |
| "loss": 3.6886, | |
| "step": 566272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.072060512473114e-05, | |
| "loss": 3.7021, | |
| "step": 566784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.071223555602435e-05, | |
| "loss": 3.6893, | |
| "step": 567296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.070384960851383e-05, | |
| "loss": 3.683, | |
| "step": 567808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.069546366100331e-05, | |
| "loss": 3.6833, | |
| "step": 568320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.068707771349279e-05, | |
| "loss": 3.6875, | |
| "step": 568832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.067870814478601e-05, | |
| "loss": 3.6818, | |
| "step": 569344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.067032219727549e-05, | |
| "loss": 3.6858, | |
| "step": 569856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.066193624976497e-05, | |
| "loss": 3.6891, | |
| "step": 570368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.065355030225445e-05, | |
| "loss": 3.6843, | |
| "step": 570880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.064516435474393e-05, | |
| "loss": 3.6909, | |
| "step": 571392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0636794786037136e-05, | |
| "loss": 3.6954, | |
| "step": 571904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0628408838526616e-05, | |
| "loss": 3.6959, | |
| "step": 572416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0620022891016096e-05, | |
| "loss": 3.6909, | |
| "step": 572928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0611636943505576e-05, | |
| "loss": 3.6872, | |
| "step": 573440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0603267374798785e-05, | |
| "loss": 3.6896, | |
| "step": 573952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0594881427288265e-05, | |
| "loss": 3.6865, | |
| "step": 574464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0586495479777745e-05, | |
| "loss": 3.6994, | |
| "step": 574976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0578109532267225e-05, | |
| "loss": 3.6948, | |
| "step": 575488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.056973996356044e-05, | |
| "loss": 3.693, | |
| "step": 576000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.056135401604992e-05, | |
| "loss": 3.6733, | |
| "step": 576512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.05529680685394e-05, | |
| "loss": 3.6884, | |
| "step": 577024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.054458212102888e-05, | |
| "loss": 3.6741, | |
| "step": 577536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.053621255232209e-05, | |
| "loss": 3.6887, | |
| "step": 578048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.052782660481157e-05, | |
| "loss": 3.6881, | |
| "step": 578560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.051944065730105e-05, | |
| "loss": 3.6859, | |
| "step": 579072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.051105470979053e-05, | |
| "loss": 3.6871, | |
| "step": 579584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.050268514108374e-05, | |
| "loss": 3.6936, | |
| "step": 580096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.049429919357322e-05, | |
| "loss": 3.6783, | |
| "step": 580608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.04859132460627e-05, | |
| "loss": 3.6849, | |
| "step": 581120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.047752729855218e-05, | |
| "loss": 3.6907, | |
| "step": 581632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0469157729845395e-05, | |
| "loss": 3.6786, | |
| "step": 582144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0460771782334875e-05, | |
| "loss": 3.6782, | |
| "step": 582656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0452385834824355e-05, | |
| "loss": 3.6799, | |
| "step": 583168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0443999887313835e-05, | |
| "loss": 3.6823, | |
| "step": 583680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0435630318607044e-05, | |
| "loss": 3.7017, | |
| "step": 584192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0427244371096524e-05, | |
| "loss": 3.6814, | |
| "step": 584704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0418858423586004e-05, | |
| "loss": 3.6834, | |
| "step": 585216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0410472476075483e-05, | |
| "loss": 3.68, | |
| "step": 585728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.040210290736869e-05, | |
| "loss": 3.6929, | |
| "step": 586240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.039371695985817e-05, | |
| "loss": 3.6862, | |
| "step": 586752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.038533101234765e-05, | |
| "loss": 3.6781, | |
| "step": 587264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.037694506483713e-05, | |
| "loss": 3.6923, | |
| "step": 587776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.036857549613035e-05, | |
| "loss": 3.6704, | |
| "step": 588288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.036018954861983e-05, | |
| "loss": 3.6738, | |
| "step": 588800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.035180360110931e-05, | |
| "loss": 3.6858, | |
| "step": 589312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.034341765359879e-05, | |
| "loss": 3.6771, | |
| "step": 589824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0335048084892e-05, | |
| "loss": 3.6828, | |
| "step": 590336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.032666213738148e-05, | |
| "loss": 3.6832, | |
| "step": 590848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.031827618987096e-05, | |
| "loss": 3.6578, | |
| "step": 591360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.030989024236044e-05, | |
| "loss": 3.6819, | |
| "step": 591872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0301520673653646e-05, | |
| "loss": 3.6942, | |
| "step": 592384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0293134726143126e-05, | |
| "loss": 3.6829, | |
| "step": 592896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0284748778632606e-05, | |
| "loss": 3.674, | |
| "step": 593408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0276362831122086e-05, | |
| "loss": 3.6728, | |
| "step": 593920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0267976883611566e-05, | |
| "loss": 3.6755, | |
| "step": 594432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.025960731490478e-05, | |
| "loss": 3.6862, | |
| "step": 594944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.025123774619799e-05, | |
| "loss": 3.6854, | |
| "step": 595456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.024285179868747e-05, | |
| "loss": 3.6759, | |
| "step": 595968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.023446585117695e-05, | |
| "loss": 3.6755, | |
| "step": 596480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.022607990366643e-05, | |
| "loss": 3.6859, | |
| "step": 596992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.021769395615591e-05, | |
| "loss": 3.6804, | |
| "step": 597504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.020930800864539e-05, | |
| "loss": 3.6711, | |
| "step": 598016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.020092206113487e-05, | |
| "loss": 3.6848, | |
| "step": 598528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.019253611362435e-05, | |
| "loss": 3.6769, | |
| "step": 599040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.018415016611383e-05, | |
| "loss": 3.6774, | |
| "step": 599552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.017578059740704e-05, | |
| "loss": 3.6893, | |
| "step": 600064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.016739464989652e-05, | |
| "loss": 3.6654, | |
| "step": 600576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0159008702386e-05, | |
| "loss": 3.6791, | |
| "step": 601088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0150622754875486e-05, | |
| "loss": 3.6868, | |
| "step": 601600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0142253186168696e-05, | |
| "loss": 3.6832, | |
| "step": 602112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0133867238658175e-05, | |
| "loss": 3.6786, | |
| "step": 602624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0125481291147655e-05, | |
| "loss": 3.6827, | |
| "step": 603136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.011709534363713e-05, | |
| "loss": 3.6695, | |
| "step": 603648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0108725774930344e-05, | |
| "loss": 3.6712, | |
| "step": 604160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0100339827419824e-05, | |
| "loss": 3.6826, | |
| "step": 604672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0091953879909304e-05, | |
| "loss": 3.687, | |
| "step": 605184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.008356793239878e-05, | |
| "loss": 3.6904, | |
| "step": 605696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0075198363691993e-05, | |
| "loss": 3.6778, | |
| "step": 606208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.006681241618147e-05, | |
| "loss": 3.6737, | |
| "step": 606720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.005842646867095e-05, | |
| "loss": 3.684, | |
| "step": 607232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.005004052116043e-05, | |
| "loss": 3.6725, | |
| "step": 607744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.004167095245365e-05, | |
| "loss": 3.681, | |
| "step": 608256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.003328500494313e-05, | |
| "loss": 3.6911, | |
| "step": 608768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.00248990574326e-05, | |
| "loss": 3.6867, | |
| "step": 609280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.001651310992208e-05, | |
| "loss": 3.6714, | |
| "step": 609792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.00081435412153e-05, | |
| "loss": 3.6838, | |
| "step": 610304 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 3.9045441150665283, | |
| "eval_runtime": 564.5634, | |
| "eval_samples_per_second": 675.905, | |
| "eval_steps_per_second": 21.123, | |
| "step": 610552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.999975759370478e-05, | |
| "loss": 3.6758, | |
| "step": 610816 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.999137164619425e-05, | |
| "loss": 3.6878, | |
| "step": 611328 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.998298569868373e-05, | |
| "loss": 3.6765, | |
| "step": 611840 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.997461612997695e-05, | |
| "loss": 3.6774, | |
| "step": 612352 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.996623018246643e-05, | |
| "loss": 3.6705, | |
| "step": 612864 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.995784423495591e-05, | |
| "loss": 3.676, | |
| "step": 613376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.994945828744539e-05, | |
| "loss": 3.6737, | |
| "step": 613888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.99410887187386e-05, | |
| "loss": 3.6608, | |
| "step": 614400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9932702771228076e-05, | |
| "loss": 3.671, | |
| "step": 614912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9924316823717556e-05, | |
| "loss": 3.6883, | |
| "step": 615424 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.991594725501077e-05, | |
| "loss": 3.6733, | |
| "step": 615936 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.990756130750025e-05, | |
| "loss": 3.6678, | |
| "step": 616448 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9899175359989725e-05, | |
| "loss": 3.6823, | |
| "step": 616960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9890789412479205e-05, | |
| "loss": 3.6677, | |
| "step": 617472 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9882403464968685e-05, | |
| "loss": 3.6777, | |
| "step": 617984 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9874017517458165e-05, | |
| "loss": 3.6532, | |
| "step": 618496 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9865631569947645e-05, | |
| "loss": 3.654, | |
| "step": 619008 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.985724562243713e-05, | |
| "loss": 3.6748, | |
| "step": 619520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.984887605373034e-05, | |
| "loss": 3.6627, | |
| "step": 620032 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.984049010621982e-05, | |
| "loss": 3.6901, | |
| "step": 620544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.98321041587093e-05, | |
| "loss": 3.6707, | |
| "step": 621056 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.982371821119878e-05, | |
| "loss": 3.6767, | |
| "step": 621568 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.981534864249199e-05, | |
| "loss": 3.6654, | |
| "step": 622080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.980696269498147e-05, | |
| "loss": 3.678, | |
| "step": 622592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.979857674747095e-05, | |
| "loss": 3.661, | |
| "step": 623104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.979019079996043e-05, | |
| "loss": 3.6666, | |
| "step": 623616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.978182123125364e-05, | |
| "loss": 3.6687, | |
| "step": 624128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.977343528374312e-05, | |
| "loss": 3.6715, | |
| "step": 624640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.97650493362326e-05, | |
| "loss": 3.6664, | |
| "step": 625152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9756663388722085e-05, | |
| "loss": 3.659, | |
| "step": 625664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9748293820015294e-05, | |
| "loss": 3.6544, | |
| "step": 626176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9739907872504774e-05, | |
| "loss": 3.6728, | |
| "step": 626688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9731521924994254e-05, | |
| "loss": 3.6741, | |
| "step": 627200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9723135977483734e-05, | |
| "loss": 3.6679, | |
| "step": 627712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.971476640877694e-05, | |
| "loss": 3.6682, | |
| "step": 628224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.970638046126642e-05, | |
| "loss": 3.6729, | |
| "step": 628736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.96979945137559e-05, | |
| "loss": 3.6686, | |
| "step": 629248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.968960856624538e-05, | |
| "loss": 3.6694, | |
| "step": 629760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.968123899753859e-05, | |
| "loss": 3.6662, | |
| "step": 630272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.967285305002807e-05, | |
| "loss": 3.6635, | |
| "step": 630784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.966446710251755e-05, | |
| "loss": 3.6697, | |
| "step": 631296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.965608115500703e-05, | |
| "loss": 3.6531, | |
| "step": 631808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.964771158630025e-05, | |
| "loss": 3.6724, | |
| "step": 632320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.963932563878973e-05, | |
| "loss": 3.67, | |
| "step": 632832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.963093969127921e-05, | |
| "loss": 3.672, | |
| "step": 633344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.962255374376869e-05, | |
| "loss": 3.6699, | |
| "step": 633856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.96141841750619e-05, | |
| "loss": 3.6555, | |
| "step": 634368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.960579822755138e-05, | |
| "loss": 3.678, | |
| "step": 634880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.959741228004086e-05, | |
| "loss": 3.6561, | |
| "step": 635392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.958902633253034e-05, | |
| "loss": 3.6641, | |
| "step": 635904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9580656763823546e-05, | |
| "loss": 3.6563, | |
| "step": 636416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9572270816313026e-05, | |
| "loss": 3.6609, | |
| "step": 636928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9563884868802506e-05, | |
| "loss": 3.6714, | |
| "step": 637440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9555498921291986e-05, | |
| "loss": 3.6562, | |
| "step": 637952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.95471293525852e-05, | |
| "loss": 3.6619, | |
| "step": 638464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.953874340507468e-05, | |
| "loss": 3.6572, | |
| "step": 638976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.953035745756416e-05, | |
| "loss": 3.6673, | |
| "step": 639488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.952197151005364e-05, | |
| "loss": 3.6602, | |
| "step": 640000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.951360194134685e-05, | |
| "loss": 3.6722, | |
| "step": 640512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.950521599383633e-05, | |
| "loss": 3.666, | |
| "step": 641024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.949683004632581e-05, | |
| "loss": 3.6577, | |
| "step": 641536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.948844409881529e-05, | |
| "loss": 3.6539, | |
| "step": 642048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.94800745301085e-05, | |
| "loss": 3.6604, | |
| "step": 642560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.947168858259798e-05, | |
| "loss": 3.6664, | |
| "step": 643072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.946330263508746e-05, | |
| "loss": 3.6621, | |
| "step": 643584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.945491668757694e-05, | |
| "loss": 3.6592, | |
| "step": 644096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9446547118870155e-05, | |
| "loss": 3.655, | |
| "step": 644608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9438161171359635e-05, | |
| "loss": 3.6544, | |
| "step": 645120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9429775223849115e-05, | |
| "loss": 3.6546, | |
| "step": 645632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9421389276338595e-05, | |
| "loss": 3.6567, | |
| "step": 646144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9413019707631804e-05, | |
| "loss": 3.6596, | |
| "step": 646656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9404633760121284e-05, | |
| "loss": 3.6581, | |
| "step": 647168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9396247812610764e-05, | |
| "loss": 3.6614, | |
| "step": 647680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9387861865100244e-05, | |
| "loss": 3.671, | |
| "step": 648192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.937949229639345e-05, | |
| "loss": 3.6659, | |
| "step": 648704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.937110634888293e-05, | |
| "loss": 3.6601, | |
| "step": 649216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.936272040137241e-05, | |
| "loss": 3.661, | |
| "step": 649728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.935433445386189e-05, | |
| "loss": 3.6626, | |
| "step": 650240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.934596488515511e-05, | |
| "loss": 3.6548, | |
| "step": 650752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.933757893764459e-05, | |
| "loss": 3.671, | |
| "step": 651264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.932919299013407e-05, | |
| "loss": 3.667, | |
| "step": 651776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.932080704262355e-05, | |
| "loss": 3.6652, | |
| "step": 652288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.931242109511303e-05, | |
| "loss": 3.6481, | |
| "step": 652800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.930405152640624e-05, | |
| "loss": 3.6579, | |
| "step": 653312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.929566557889572e-05, | |
| "loss": 3.6452, | |
| "step": 653824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.92872796313852e-05, | |
| "loss": 3.6666, | |
| "step": 654336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.927889368387468e-05, | |
| "loss": 3.66, | |
| "step": 654848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.927052411516789e-05, | |
| "loss": 3.6542, | |
| "step": 655360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.926213816765737e-05, | |
| "loss": 3.6595, | |
| "step": 655872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.925375222014685e-05, | |
| "loss": 3.6644, | |
| "step": 656384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9245366272636333e-05, | |
| "loss": 3.6508, | |
| "step": 656896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.923699670392954e-05, | |
| "loss": 3.6538, | |
| "step": 657408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.922861075641902e-05, | |
| "loss": 3.6619, | |
| "step": 657920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.92202248089085e-05, | |
| "loss": 3.6522, | |
| "step": 658432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9211838861397976e-05, | |
| "loss": 3.6516, | |
| "step": 658944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.920346929269119e-05, | |
| "loss": 3.6498, | |
| "step": 659456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.919508334518067e-05, | |
| "loss": 3.6548, | |
| "step": 659968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.918669739767015e-05, | |
| "loss": 3.6737, | |
| "step": 660480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9178311450159625e-05, | |
| "loss": 3.6544, | |
| "step": 660992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.916994188145284e-05, | |
| "loss": 3.6574, | |
| "step": 661504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.916155593394232e-05, | |
| "loss": 3.6548, | |
| "step": 662016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.91531699864318e-05, | |
| "loss": 3.6601, | |
| "step": 662528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.914478403892128e-05, | |
| "loss": 3.6596, | |
| "step": 663040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9136414470214496e-05, | |
| "loss": 3.6529, | |
| "step": 663552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9128028522703976e-05, | |
| "loss": 3.6628, | |
| "step": 664064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.911964257519345e-05, | |
| "loss": 3.6409, | |
| "step": 664576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.911125662768293e-05, | |
| "loss": 3.6473, | |
| "step": 665088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9102887058976145e-05, | |
| "loss": 3.6577, | |
| "step": 665600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9094501111465625e-05, | |
| "loss": 3.6543, | |
| "step": 666112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.90861151639551e-05, | |
| "loss": 3.6499, | |
| "step": 666624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.907772921644458e-05, | |
| "loss": 3.6559, | |
| "step": 667136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9069359647737794e-05, | |
| "loss": 3.6331, | |
| "step": 667648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9060973700227274e-05, | |
| "loss": 3.6567, | |
| "step": 668160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9052587752716754e-05, | |
| "loss": 3.6601, | |
| "step": 668672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9044201805206234e-05, | |
| "loss": 3.6564, | |
| "step": 669184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.903583223649945e-05, | |
| "loss": 3.6444, | |
| "step": 669696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.902744628898892e-05, | |
| "loss": 3.6473, | |
| "step": 670208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.90190603414784e-05, | |
| "loss": 3.6459, | |
| "step": 670720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.901067439396788e-05, | |
| "loss": 3.6598, | |
| "step": 671232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.90023048252611e-05, | |
| "loss": 3.6548, | |
| "step": 671744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.899391887775057e-05, | |
| "loss": 3.6525, | |
| "step": 672256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.898553293024005e-05, | |
| "loss": 3.6481, | |
| "step": 672768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.897714698272953e-05, | |
| "loss": 3.6593, | |
| "step": 673280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.896876103521901e-05, | |
| "loss": 3.6523, | |
| "step": 673792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.896037508770849e-05, | |
| "loss": 3.6455, | |
| "step": 674304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.895198914019798e-05, | |
| "loss": 3.6523, | |
| "step": 674816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.894360319268746e-05, | |
| "loss": 3.6507, | |
| "step": 675328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.893523362398067e-05, | |
| "loss": 3.652, | |
| "step": 675840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.892684767647015e-05, | |
| "loss": 3.6591, | |
| "step": 676352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.891846172895963e-05, | |
| "loss": 3.6438, | |
| "step": 676864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.891007578144911e-05, | |
| "loss": 3.6498, | |
| "step": 677376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.890170621274232e-05, | |
| "loss": 3.6604, | |
| "step": 677888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8893320265231797e-05, | |
| "loss": 3.6519, | |
| "step": 678400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8884934317721277e-05, | |
| "loss": 3.6516, | |
| "step": 678912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8876548370210757e-05, | |
| "loss": 3.6551, | |
| "step": 679424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8868178801503966e-05, | |
| "loss": 3.6454, | |
| "step": 679936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8859792853993446e-05, | |
| "loss": 3.6454, | |
| "step": 680448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.885140690648293e-05, | |
| "loss": 3.653, | |
| "step": 680960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.884302095897241e-05, | |
| "loss": 3.664, | |
| "step": 681472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.883465139026562e-05, | |
| "loss": 3.6592, | |
| "step": 681984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.88262654427551e-05, | |
| "loss": 3.65, | |
| "step": 682496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.881787949524458e-05, | |
| "loss": 3.6504, | |
| "step": 683008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.880949354773406e-05, | |
| "loss": 3.6522, | |
| "step": 683520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.880112397902727e-05, | |
| "loss": 3.6492, | |
| "step": 684032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.879273803151675e-05, | |
| "loss": 3.6536, | |
| "step": 684544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.878435208400623e-05, | |
| "loss": 3.6613, | |
| "step": 685056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.877596613649571e-05, | |
| "loss": 3.66, | |
| "step": 685568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.876759656778892e-05, | |
| "loss": 3.6434, | |
| "step": 686080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.87592106202784e-05, | |
| "loss": 3.6564, | |
| "step": 686592 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 3.8993594646453857, | |
| "eval_runtime": 551.7316, | |
| "eval_samples_per_second": 691.624, | |
| "eval_steps_per_second": 21.614, | |
| "step": 686871 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.875082467276788e-05, | |
| "loss": 3.6594, | |
| "step": 687104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8742438725257366e-05, | |
| "loss": 3.6602, | |
| "step": 687616 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8734052777746846e-05, | |
| "loss": 3.6521, | |
| "step": 688128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8725666830236326e-05, | |
| "loss": 3.6474, | |
| "step": 688640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8717297261529535e-05, | |
| "loss": 3.6448, | |
| "step": 689152 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8708911314019015e-05, | |
| "loss": 3.6474, | |
| "step": 689664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8700525366508495e-05, | |
| "loss": 3.648, | |
| "step": 690176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8692139418997975e-05, | |
| "loss": 3.6348, | |
| "step": 690688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8683769850291184e-05, | |
| "loss": 3.6423, | |
| "step": 691200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8675383902780664e-05, | |
| "loss": 3.6615, | |
| "step": 691712 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.866701433407387e-05, | |
| "loss": 3.646, | |
| "step": 692224 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.865862838656335e-05, | |
| "loss": 3.6397, | |
| "step": 692736 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.865024243905283e-05, | |
| "loss": 3.6583, | |
| "step": 693248 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.864185649154232e-05, | |
| "loss": 3.6405, | |
| "step": 693760 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.86334705440318e-05, | |
| "loss": 3.6505, | |
| "step": 694272 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.862508459652128e-05, | |
| "loss": 3.6267, | |
| "step": 694784 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.861669864901076e-05, | |
| "loss": 3.6285, | |
| "step": 695296 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.860831270150023e-05, | |
| "loss": 3.6485, | |
| "step": 695808 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.859994313279345e-05, | |
| "loss": 3.6378, | |
| "step": 696320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.859155718528293e-05, | |
| "loss": 3.6607, | |
| "step": 696832 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.858317123777241e-05, | |
| "loss": 3.6432, | |
| "step": 697344 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.857478529026188e-05, | |
| "loss": 3.6572, | |
| "step": 697856 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.85664157215551e-05, | |
| "loss": 3.6387, | |
| "step": 698368 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.855802977404458e-05, | |
| "loss": 3.6461, | |
| "step": 698880 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.854964382653406e-05, | |
| "loss": 3.6365, | |
| "step": 699392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.854125787902354e-05, | |
| "loss": 3.643, | |
| "step": 699904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.853288831031675e-05, | |
| "loss": 3.6436, | |
| "step": 700416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.852450236280623e-05, | |
| "loss": 3.6458, | |
| "step": 700928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8516116415295706e-05, | |
| "loss": 3.638, | |
| "step": 701440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8507730467785186e-05, | |
| "loss": 3.6341, | |
| "step": 701952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.84993608990784e-05, | |
| "loss": 3.6279, | |
| "step": 702464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.849097495156788e-05, | |
| "loss": 3.6487, | |
| "step": 702976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8482589004057355e-05, | |
| "loss": 3.645, | |
| "step": 703488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8474203056546835e-05, | |
| "loss": 3.6421, | |
| "step": 704000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.846583348784005e-05, | |
| "loss": 3.6423, | |
| "step": 704512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8457447540329524e-05, | |
| "loss": 3.6474, | |
| "step": 705024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.844906159281901e-05, | |
| "loss": 3.6398, | |
| "step": 705536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.844067564530849e-05, | |
| "loss": 3.6426, | |
| "step": 706048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.843230607660171e-05, | |
| "loss": 3.642, | |
| "step": 706560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.842392012909118e-05, | |
| "loss": 3.6376, | |
| "step": 707072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.841553418158066e-05, | |
| "loss": 3.6416, | |
| "step": 707584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8407164612873876e-05, | |
| "loss": 3.6347, | |
| "step": 708096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8398778665363356e-05, | |
| "loss": 3.643, | |
| "step": 708608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.839039271785283e-05, | |
| "loss": 3.6463, | |
| "step": 709120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.838200677034231e-05, | |
| "loss": 3.6461, | |
| "step": 709632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8373637201635525e-05, | |
| "loss": 3.6447, | |
| "step": 710144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8365251254125e-05, | |
| "loss": 3.631, | |
| "step": 710656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.835686530661448e-05, | |
| "loss": 3.6504, | |
| "step": 711168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8348479359103965e-05, | |
| "loss": 3.6337, | |
| "step": 711680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.834010979039718e-05, | |
| "loss": 3.6363, | |
| "step": 712192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8331723842886654e-05, | |
| "loss": 3.6316, | |
| "step": 712704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8323337895376134e-05, | |
| "loss": 3.6322, | |
| "step": 713216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8314951947865614e-05, | |
| "loss": 3.65, | |
| "step": 713728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.830658237915883e-05, | |
| "loss": 3.6281, | |
| "step": 714240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.82981964316483e-05, | |
| "loss": 3.6418, | |
| "step": 714752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.828981048413778e-05, | |
| "loss": 3.6255, | |
| "step": 715264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.828142453662726e-05, | |
| "loss": 3.6414, | |
| "step": 715776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.827305496792047e-05, | |
| "loss": 3.6335, | |
| "step": 716288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.826466902040995e-05, | |
| "loss": 3.6485, | |
| "step": 716800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.825628307289943e-05, | |
| "loss": 3.6432, | |
| "step": 717312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.824789712538892e-05, | |
| "loss": 3.6323, | |
| "step": 717824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.823952755668213e-05, | |
| "loss": 3.6295, | |
| "step": 718336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.823114160917161e-05, | |
| "loss": 3.6317, | |
| "step": 718848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.822275566166109e-05, | |
| "loss": 3.6368, | |
| "step": 719360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.821436971415057e-05, | |
| "loss": 3.6397, | |
| "step": 719872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8206000145443776e-05, | |
| "loss": 3.6333, | |
| "step": 720384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8197614197933256e-05, | |
| "loss": 3.629, | |
| "step": 720896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8189228250422736e-05, | |
| "loss": 3.632, | |
| "step": 721408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8180842302912216e-05, | |
| "loss": 3.6292, | |
| "step": 721920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8172472734205425e-05, | |
| "loss": 3.6341, | |
| "step": 722432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8164086786694905e-05, | |
| "loss": 3.6332, | |
| "step": 722944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8155700839184385e-05, | |
| "loss": 3.6303, | |
| "step": 723456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.814731489167387e-05, | |
| "loss": 3.6405, | |
| "step": 723968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.813894532296708e-05, | |
| "loss": 3.6424, | |
| "step": 724480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.813055937545656e-05, | |
| "loss": 3.6403, | |
| "step": 724992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.812217342794604e-05, | |
| "loss": 3.6388, | |
| "step": 725504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.811378748043552e-05, | |
| "loss": 3.633, | |
| "step": 726016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.810541791172873e-05, | |
| "loss": 3.6408, | |
| "step": 726528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.809703196421821e-05, | |
| "loss": 3.6303, | |
| "step": 727040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.808864601670769e-05, | |
| "loss": 3.642, | |
| "step": 727552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.808026006919717e-05, | |
| "loss": 3.6441, | |
| "step": 728064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.807189050049038e-05, | |
| "loss": 3.6415, | |
| "step": 728576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.806350455297986e-05, | |
| "loss": 3.6252, | |
| "step": 729088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.805511860546934e-05, | |
| "loss": 3.6291, | |
| "step": 729600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8046732657958826e-05, | |
| "loss": 3.6178, | |
| "step": 730112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8038363089252035e-05, | |
| "loss": 3.6423, | |
| "step": 730624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8029977141741515e-05, | |
| "loss": 3.6353, | |
| "step": 731136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8021591194230995e-05, | |
| "loss": 3.634, | |
| "step": 731648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8013205246720475e-05, | |
| "loss": 3.636, | |
| "step": 732160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8004835678013684e-05, | |
| "loss": 3.6397, | |
| "step": 732672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7996449730503164e-05, | |
| "loss": 3.6257, | |
| "step": 733184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7988063782992644e-05, | |
| "loss": 3.628, | |
| "step": 733696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7979677835482124e-05, | |
| "loss": 3.6417, | |
| "step": 734208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.797130826677533e-05, | |
| "loss": 3.6232, | |
| "step": 734720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.796292231926481e-05, | |
| "loss": 3.6268, | |
| "step": 735232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.795453637175429e-05, | |
| "loss": 3.6252, | |
| "step": 735744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.794615042424377e-05, | |
| "loss": 3.6314, | |
| "step": 736256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.793778085553699e-05, | |
| "loss": 3.6451, | |
| "step": 736768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.792939490802647e-05, | |
| "loss": 3.6272, | |
| "step": 737280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.792100896051595e-05, | |
| "loss": 3.6343, | |
| "step": 737792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.791262301300543e-05, | |
| "loss": 3.6284, | |
| "step": 738304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.790425344429864e-05, | |
| "loss": 3.6351, | |
| "step": 738816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.789586749678812e-05, | |
| "loss": 3.6329, | |
| "step": 739328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.78874815492776e-05, | |
| "loss": 3.6299, | |
| "step": 739840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.787909560176708e-05, | |
| "loss": 3.6336, | |
| "step": 740352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7870726033060286e-05, | |
| "loss": 3.6207, | |
| "step": 740864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7862340085549766e-05, | |
| "loss": 3.6228, | |
| "step": 741376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7853954138039246e-05, | |
| "loss": 3.629, | |
| "step": 741888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7845568190528726e-05, | |
| "loss": 3.6287, | |
| "step": 742400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.783719862182194e-05, | |
| "loss": 3.6238, | |
| "step": 742912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.782881267431142e-05, | |
| "loss": 3.6347, | |
| "step": 743424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.78204267268009e-05, | |
| "loss": 3.6082, | |
| "step": 743936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.781204077929038e-05, | |
| "loss": 3.6306, | |
| "step": 744448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.780367121058359e-05, | |
| "loss": 3.6356, | |
| "step": 744960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.779528526307307e-05, | |
| "loss": 3.6333, | |
| "step": 745472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.778689931556255e-05, | |
| "loss": 3.6233, | |
| "step": 745984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.777851336805203e-05, | |
| "loss": 3.6222, | |
| "step": 746496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.777014379934524e-05, | |
| "loss": 3.6218, | |
| "step": 747008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.776175785183472e-05, | |
| "loss": 3.6318, | |
| "step": 747520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7753388283127936e-05, | |
| "loss": 3.6315, | |
| "step": 748032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7745002335617416e-05, | |
| "loss": 3.6326, | |
| "step": 748544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7736616388106896e-05, | |
| "loss": 3.6166, | |
| "step": 749056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7728230440596376e-05, | |
| "loss": 3.6374, | |
| "step": 749568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7719844493085856e-05, | |
| "loss": 3.6257, | |
| "step": 750080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7711458545575336e-05, | |
| "loss": 3.6233, | |
| "step": 750592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7703072598064816e-05, | |
| "loss": 3.6244, | |
| "step": 751104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7694686650554296e-05, | |
| "loss": 3.6313, | |
| "step": 751616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7686317081847505e-05, | |
| "loss": 3.6264, | |
| "step": 752128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7677931134336985e-05, | |
| "loss": 3.6325, | |
| "step": 752640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7669545186826465e-05, | |
| "loss": 3.6206, | |
| "step": 753152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7661159239315945e-05, | |
| "loss": 3.6255, | |
| "step": 753664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7652789670609154e-05, | |
| "loss": 3.6371, | |
| "step": 754176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7644403723098634e-05, | |
| "loss": 3.6255, | |
| "step": 754688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.763601777558812e-05, | |
| "loss": 3.6265, | |
| "step": 755200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.76276318280776e-05, | |
| "loss": 3.6325, | |
| "step": 755712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.761926225937081e-05, | |
| "loss": 3.6188, | |
| "step": 756224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.761087631186029e-05, | |
| "loss": 3.6227, | |
| "step": 756736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.760249036434977e-05, | |
| "loss": 3.6265, | |
| "step": 757248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.759410441683925e-05, | |
| "loss": 3.6322, | |
| "step": 757760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.758573484813246e-05, | |
| "loss": 3.6421, | |
| "step": 758272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.757734890062194e-05, | |
| "loss": 3.6242, | |
| "step": 758784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.756896295311142e-05, | |
| "loss": 3.6279, | |
| "step": 759296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.75605770056009e-05, | |
| "loss": 3.6267, | |
| "step": 759808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.755220743689411e-05, | |
| "loss": 3.6256, | |
| "step": 760320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.754382148938359e-05, | |
| "loss": 3.6309, | |
| "step": 760832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7535435541873074e-05, | |
| "loss": 3.6353, | |
| "step": 761344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7527049594362554e-05, | |
| "loss": 3.6362, | |
| "step": 761856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.751868002565576e-05, | |
| "loss": 3.6218, | |
| "step": 762368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.751029407814524e-05, | |
| "loss": 3.63, | |
| "step": 762880 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 3.897984266281128, | |
| "eval_runtime": 578.1247, | |
| "eval_samples_per_second": 660.05, | |
| "eval_steps_per_second": 20.627, | |
| "step": 763190 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.750190813063472e-05, | |
| "loss": 3.6407, | |
| "step": 763392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.74935221831242e-05, | |
| "loss": 3.6345, | |
| "step": 763904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7485136235613676e-05, | |
| "loss": 3.6286, | |
| "step": 764416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7476750288103156e-05, | |
| "loss": 3.6216, | |
| "step": 764928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.746838071939637e-05, | |
| "loss": 3.6244, | |
| "step": 765440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7459994771885845e-05, | |
| "loss": 3.6226, | |
| "step": 765952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7451608824375325e-05, | |
| "loss": 3.6261, | |
| "step": 766464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.744322287686481e-05, | |
| "loss": 3.6061, | |
| "step": 766976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.743485330815803e-05, | |
| "loss": 3.6205, | |
| "step": 767488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.74264673606475e-05, | |
| "loss": 3.6378, | |
| "step": 768000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.741809779194072e-05, | |
| "loss": 3.6244, | |
| "step": 768512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.74097118444302e-05, | |
| "loss": 3.6192, | |
| "step": 769024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.740132589691968e-05, | |
| "loss": 3.6308, | |
| "step": 769536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.739293994940915e-05, | |
| "loss": 3.6164, | |
| "step": 770048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.738455400189863e-05, | |
| "loss": 3.6269, | |
| "step": 770560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.737616805438811e-05, | |
| "loss": 3.6017, | |
| "step": 771072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.736778210687759e-05, | |
| "loss": 3.6036, | |
| "step": 771584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.735939615936707e-05, | |
| "loss": 3.6258, | |
| "step": 772096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.735102659066028e-05, | |
| "loss": 3.6151, | |
| "step": 772608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7342640643149765e-05, | |
| "loss": 3.6387, | |
| "step": 773120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7334254695639245e-05, | |
| "loss": 3.6153, | |
| "step": 773632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7325868748128725e-05, | |
| "loss": 3.6332, | |
| "step": 774144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7317499179421934e-05, | |
| "loss": 3.6202, | |
| "step": 774656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7309113231911414e-05, | |
| "loss": 3.6208, | |
| "step": 775168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7300727284400894e-05, | |
| "loss": 3.6118, | |
| "step": 775680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7292341336890374e-05, | |
| "loss": 3.6201, | |
| "step": 776192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7283971768183583e-05, | |
| "loss": 3.621, | |
| "step": 776704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.727558582067306e-05, | |
| "loss": 3.624, | |
| "step": 777216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.726719987316254e-05, | |
| "loss": 3.6113, | |
| "step": 777728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.725881392565202e-05, | |
| "loss": 3.6155, | |
| "step": 778240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.725044435694523e-05, | |
| "loss": 3.5986, | |
| "step": 778752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.724205840943472e-05, | |
| "loss": 3.6268, | |
| "step": 779264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.72336724619242e-05, | |
| "loss": 3.6207, | |
| "step": 779776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.722530289321741e-05, | |
| "loss": 3.6186, | |
| "step": 780288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.721691694570689e-05, | |
| "loss": 3.6207, | |
| "step": 780800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.720853099819637e-05, | |
| "loss": 3.6232, | |
| "step": 781312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.720014505068585e-05, | |
| "loss": 3.6203, | |
| "step": 781824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.719177548197906e-05, | |
| "loss": 3.6204, | |
| "step": 782336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.718338953446854e-05, | |
| "loss": 3.6188, | |
| "step": 782848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.717500358695802e-05, | |
| "loss": 3.6159, | |
| "step": 783360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.71666176394475e-05, | |
| "loss": 3.6189, | |
| "step": 783872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7158248070740706e-05, | |
| "loss": 3.6135, | |
| "step": 784384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7149862123230186e-05, | |
| "loss": 3.6189, | |
| "step": 784896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.714147617571967e-05, | |
| "loss": 3.6217, | |
| "step": 785408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.713309022820915e-05, | |
| "loss": 3.6204, | |
| "step": 785920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.712472065950236e-05, | |
| "loss": 3.6217, | |
| "step": 786432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.711633471199184e-05, | |
| "loss": 3.6078, | |
| "step": 786944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.710794876448132e-05, | |
| "loss": 3.6267, | |
| "step": 787456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.70995628169708e-05, | |
| "loss": 3.6131, | |
| "step": 787968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.709119324826401e-05, | |
| "loss": 3.611, | |
| "step": 788480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.708280730075349e-05, | |
| "loss": 3.612, | |
| "step": 788992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.707442135324297e-05, | |
| "loss": 3.6098, | |
| "step": 789504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.706603540573245e-05, | |
| "loss": 3.6231, | |
| "step": 790016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.705766583702566e-05, | |
| "loss": 3.6085, | |
| "step": 790528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.704927988951514e-05, | |
| "loss": 3.6158, | |
| "step": 791040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.704089394200462e-05, | |
| "loss": 3.6032, | |
| "step": 791552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7032507994494106e-05, | |
| "loss": 3.6199, | |
| "step": 792064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7024138425787316e-05, | |
| "loss": 3.6098, | |
| "step": 792576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7015752478276795e-05, | |
| "loss": 3.6233, | |
| "step": 793088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7007366530766275e-05, | |
| "loss": 3.6206, | |
| "step": 793600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6998980583255755e-05, | |
| "loss": 3.6127, | |
| "step": 794112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6990611014548964e-05, | |
| "loss": 3.6088, | |
| "step": 794624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6982225067038444e-05, | |
| "loss": 3.6047, | |
| "step": 795136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6973839119527924e-05, | |
| "loss": 3.6125, | |
| "step": 795648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6965453172017404e-05, | |
| "loss": 3.6195, | |
| "step": 796160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6957083603310613e-05, | |
| "loss": 3.6137, | |
| "step": 796672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6948697655800093e-05, | |
| "loss": 3.6054, | |
| "step": 797184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.694031170828957e-05, | |
| "loss": 3.608, | |
| "step": 797696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.693192576077906e-05, | |
| "loss": 3.6078, | |
| "step": 798208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.692355619207227e-05, | |
| "loss": 3.6114, | |
| "step": 798720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.691517024456175e-05, | |
| "loss": 3.6104, | |
| "step": 799232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.690678429705123e-05, | |
| "loss": 3.6096, | |
| "step": 799744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.689839834954071e-05, | |
| "loss": 3.6124, | |
| "step": 800256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.689002878083392e-05, | |
| "loss": 3.6198, | |
| "step": 800768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.68816428333234e-05, | |
| "loss": 3.6178, | |
| "step": 801280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.687325688581288e-05, | |
| "loss": 3.6132, | |
| "step": 801792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.686487093830236e-05, | |
| "loss": 3.6116, | |
| "step": 802304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.685650136959557e-05, | |
| "loss": 3.6179, | |
| "step": 802816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.684811542208505e-05, | |
| "loss": 3.6116, | |
| "step": 803328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.683972947457453e-05, | |
| "loss": 3.618, | |
| "step": 803840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6831343527064014e-05, | |
| "loss": 3.6227, | |
| "step": 804352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.682297395835722e-05, | |
| "loss": 3.6197, | |
| "step": 804864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.68145880108467e-05, | |
| "loss": 3.6047, | |
| "step": 805376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.680620206333618e-05, | |
| "loss": 3.6029, | |
| "step": 805888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.679781611582566e-05, | |
| "loss": 3.597, | |
| "step": 806400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.678944654711887e-05, | |
| "loss": 3.619, | |
| "step": 806912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.678106059960835e-05, | |
| "loss": 3.6105, | |
| "step": 807424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.677267465209783e-05, | |
| "loss": 3.6153, | |
| "step": 807936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.676428870458731e-05, | |
| "loss": 3.613, | |
| "step": 808448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.675591913588052e-05, | |
| "loss": 3.6164, | |
| "step": 808960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.674753318837e-05, | |
| "loss": 3.6045, | |
| "step": 809472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.673914724085948e-05, | |
| "loss": 3.6047, | |
| "step": 809984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.673076129334897e-05, | |
| "loss": 3.6173, | |
| "step": 810496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6722391724642177e-05, | |
| "loss": 3.6014, | |
| "step": 811008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6714005777131656e-05, | |
| "loss": 3.603, | |
| "step": 811520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6705619829621136e-05, | |
| "loss": 3.6025, | |
| "step": 812032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6697233882110616e-05, | |
| "loss": 3.6133, | |
| "step": 812544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6688864313403825e-05, | |
| "loss": 3.617, | |
| "step": 813056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6680478365893305e-05, | |
| "loss": 3.6105, | |
| "step": 813568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6672092418382785e-05, | |
| "loss": 3.613, | |
| "step": 814080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6663706470872265e-05, | |
| "loss": 3.6055, | |
| "step": 814592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6655336902165474e-05, | |
| "loss": 3.6136, | |
| "step": 815104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6646950954654954e-05, | |
| "loss": 3.6091, | |
| "step": 815616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6638565007144434e-05, | |
| "loss": 3.6118, | |
| "step": 816128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.663017905963392e-05, | |
| "loss": 3.6097, | |
| "step": 816640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.662180949092713e-05, | |
| "loss": 3.5969, | |
| "step": 817152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.661342354341661e-05, | |
| "loss": 3.601, | |
| "step": 817664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.660503759590609e-05, | |
| "loss": 3.6095, | |
| "step": 818176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.659665164839557e-05, | |
| "loss": 3.6077, | |
| "step": 818688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.658828207968878e-05, | |
| "loss": 3.6012, | |
| "step": 819200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.657989613217826e-05, | |
| "loss": 3.613, | |
| "step": 819712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.657151018466774e-05, | |
| "loss": 3.5905, | |
| "step": 820224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.656312423715722e-05, | |
| "loss": 3.6041, | |
| "step": 820736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.655475466845043e-05, | |
| "loss": 3.6113, | |
| "step": 821248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.654636872093991e-05, | |
| "loss": 3.6142, | |
| "step": 821760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.653798277342939e-05, | |
| "loss": 3.601, | |
| "step": 822272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6529596825918875e-05, | |
| "loss": 3.6038, | |
| "step": 822784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6521227257212084e-05, | |
| "loss": 3.5969, | |
| "step": 823296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6512841309701564e-05, | |
| "loss": 3.6092, | |
| "step": 823808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.650447174099477e-05, | |
| "loss": 3.6117, | |
| "step": 824320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.649608579348425e-05, | |
| "loss": 3.6089, | |
| "step": 824832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.648769984597373e-05, | |
| "loss": 3.5967, | |
| "step": 825344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.647931389846321e-05, | |
| "loss": 3.6125, | |
| "step": 825856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.647092795095269e-05, | |
| "loss": 3.6055, | |
| "step": 826368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.646254200344217e-05, | |
| "loss": 3.6015, | |
| "step": 826880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6454156055931646e-05, | |
| "loss": 3.5988, | |
| "step": 827392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6445770108421126e-05, | |
| "loss": 3.6132, | |
| "step": 827904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.643740053971434e-05, | |
| "loss": 3.6037, | |
| "step": 828416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.642901459220382e-05, | |
| "loss": 3.6097, | |
| "step": 828928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.64206286446933e-05, | |
| "loss": 3.6006, | |
| "step": 829440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.641224269718278e-05, | |
| "loss": 3.6017, | |
| "step": 829952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6403873128476e-05, | |
| "loss": 3.6117, | |
| "step": 830464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.639548718096547e-05, | |
| "loss": 3.6078, | |
| "step": 830976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.638710123345495e-05, | |
| "loss": 3.6048, | |
| "step": 831488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.637871528594443e-05, | |
| "loss": 3.6137, | |
| "step": 832000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6370345717237646e-05, | |
| "loss": 3.5969, | |
| "step": 832512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.636195976972712e-05, | |
| "loss": 3.5983, | |
| "step": 833024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.63535738222166e-05, | |
| "loss": 3.6053, | |
| "step": 833536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.634518787470608e-05, | |
| "loss": 3.6133, | |
| "step": 834048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6336818305999295e-05, | |
| "loss": 3.6205, | |
| "step": 834560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6328432358488775e-05, | |
| "loss": 3.6039, | |
| "step": 835072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6320046410978255e-05, | |
| "loss": 3.6066, | |
| "step": 835584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6311660463467735e-05, | |
| "loss": 3.6032, | |
| "step": 836096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6303290894760944e-05, | |
| "loss": 3.6046, | |
| "step": 836608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6294904947250424e-05, | |
| "loss": 3.6082, | |
| "step": 837120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6286518999739904e-05, | |
| "loss": 3.6131, | |
| "step": 837632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6278133052229384e-05, | |
| "loss": 3.6179, | |
| "step": 838144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.626976348352259e-05, | |
| "loss": 3.5994, | |
| "step": 838656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.626137753601207e-05, | |
| "loss": 3.6086, | |
| "step": 839168 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 3.8966124057769775, | |
| "eval_runtime": 330.182, | |
| "eval_samples_per_second": 1155.699, | |
| "eval_steps_per_second": 36.116, | |
| "step": 839509 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.625299158850155e-05, | |
| "loss": 3.607, | |
| "step": 839680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.624460564099103e-05, | |
| "loss": 3.615, | |
| "step": 840192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.623623607228425e-05, | |
| "loss": 3.6052, | |
| "step": 840704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.622785012477373e-05, | |
| "loss": 3.6021, | |
| "step": 841216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.621946417726321e-05, | |
| "loss": 3.6048, | |
| "step": 841728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.621107822975269e-05, | |
| "loss": 3.6009, | |
| "step": 842240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.62027086610459e-05, | |
| "loss": 3.6039, | |
| "step": 842752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.619432271353538e-05, | |
| "loss": 3.5869, | |
| "step": 843264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.618593676602486e-05, | |
| "loss": 3.6016, | |
| "step": 843776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.617755081851434e-05, | |
| "loss": 3.6141, | |
| "step": 844288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.616918124980755e-05, | |
| "loss": 3.601, | |
| "step": 844800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.616079530229703e-05, | |
| "loss": 3.5975, | |
| "step": 845312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.615240935478651e-05, | |
| "loss": 3.6119, | |
| "step": 845824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.614402340727599e-05, | |
| "loss": 3.5965, | |
| "step": 846336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.613563745976547e-05, | |
| "loss": 3.6023, | |
| "step": 846848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6127251512254953e-05, | |
| "loss": 3.5846, | |
| "step": 847360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6118865564744433e-05, | |
| "loss": 3.5784, | |
| "step": 847872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6110479617233913e-05, | |
| "loss": 3.6076, | |
| "step": 848384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.610211004852712e-05, | |
| "loss": 3.5931, | |
| "step": 848896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.60937241010166e-05, | |
| "loss": 3.6146, | |
| "step": 849408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.608533815350608e-05, | |
| "loss": 3.5978, | |
| "step": 849920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.607695220599556e-05, | |
| "loss": 3.6069, | |
| "step": 850432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.606858263728877e-05, | |
| "loss": 3.5967, | |
| "step": 850944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.606019668977825e-05, | |
| "loss": 3.6058, | |
| "step": 851456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.605181074226773e-05, | |
| "loss": 3.5871, | |
| "step": 851968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.604342479475721e-05, | |
| "loss": 3.6002, | |
| "step": 852480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.603505522605042e-05, | |
| "loss": 3.5998, | |
| "step": 852992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.602666927853991e-05, | |
| "loss": 3.6011, | |
| "step": 853504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.601828333102939e-05, | |
| "loss": 3.5889, | |
| "step": 854016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.600989738351887e-05, | |
| "loss": 3.5949, | |
| "step": 854528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.6001527814812076e-05, | |
| "loss": 3.5787, | |
| "step": 855040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5993141867301556e-05, | |
| "loss": 3.6014, | |
| "step": 855552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5984755919791036e-05, | |
| "loss": 3.602, | |
| "step": 856064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5976369972280516e-05, | |
| "loss": 3.5969, | |
| "step": 856576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5968000403573725e-05, | |
| "loss": 3.598, | |
| "step": 857088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5959614456063205e-05, | |
| "loss": 3.6057, | |
| "step": 857600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5951228508552685e-05, | |
| "loss": 3.5952, | |
| "step": 858112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5942858939845894e-05, | |
| "loss": 3.6004, | |
| "step": 858624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5934472992335374e-05, | |
| "loss": 3.5982, | |
| "step": 859136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.592608704482486e-05, | |
| "loss": 3.5981, | |
| "step": 859648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.591770109731434e-05, | |
| "loss": 3.5969, | |
| "step": 860160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.590933152860755e-05, | |
| "loss": 3.5883, | |
| "step": 860672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.590094558109703e-05, | |
| "loss": 3.5983, | |
| "step": 861184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.589255963358651e-05, | |
| "loss": 3.6014, | |
| "step": 861696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.588417368607599e-05, | |
| "loss": 3.599, | |
| "step": 862208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.58758041173692e-05, | |
| "loss": 3.5987, | |
| "step": 862720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.586741816985868e-05, | |
| "loss": 3.5895, | |
| "step": 863232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.585903222234816e-05, | |
| "loss": 3.6077, | |
| "step": 863744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.585064627483764e-05, | |
| "loss": 3.5938, | |
| "step": 864256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.584227670613085e-05, | |
| "loss": 3.5885, | |
| "step": 864768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.583389075862033e-05, | |
| "loss": 3.586, | |
| "step": 865280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5825504811109814e-05, | |
| "loss": 3.5933, | |
| "step": 865792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5817118863599294e-05, | |
| "loss": 3.601, | |
| "step": 866304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5808749294892504e-05, | |
| "loss": 3.5878, | |
| "step": 866816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5800363347381984e-05, | |
| "loss": 3.5962, | |
| "step": 867328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5791977399871463e-05, | |
| "loss": 3.5826, | |
| "step": 867840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5783591452360943e-05, | |
| "loss": 3.5998, | |
| "step": 868352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.577522188365415e-05, | |
| "loss": 3.594, | |
| "step": 868864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.576683593614363e-05, | |
| "loss": 3.6007, | |
| "step": 869376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.575844998863311e-05, | |
| "loss": 3.5967, | |
| "step": 869888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.575006404112259e-05, | |
| "loss": 3.5917, | |
| "step": 870400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.57416944724158e-05, | |
| "loss": 3.5884, | |
| "step": 870912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.573330852490528e-05, | |
| "loss": 3.5858, | |
| "step": 871424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.572492257739477e-05, | |
| "loss": 3.5904, | |
| "step": 871936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.571653662988425e-05, | |
| "loss": 3.6013, | |
| "step": 872448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.570816706117746e-05, | |
| "loss": 3.5939, | |
| "step": 872960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5699797492470666e-05, | |
| "loss": 3.5798, | |
| "step": 873472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5691411544960146e-05, | |
| "loss": 3.5891, | |
| "step": 873984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5683025597449626e-05, | |
| "loss": 3.588, | |
| "step": 874496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5674639649939106e-05, | |
| "loss": 3.5911, | |
| "step": 875008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5666253702428586e-05, | |
| "loss": 3.5856, | |
| "step": 875520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5657867754918066e-05, | |
| "loss": 3.5933, | |
| "step": 876032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5649481807407546e-05, | |
| "loss": 3.5908, | |
| "step": 876544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.564109585989702e-05, | |
| "loss": 3.5968, | |
| "step": 877056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5632726291190235e-05, | |
| "loss": 3.5991, | |
| "step": 877568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.562434034367972e-05, | |
| "loss": 3.5965, | |
| "step": 878080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.56159543961692e-05, | |
| "loss": 3.5859, | |
| "step": 878592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5607568448658675e-05, | |
| "loss": 3.5951, | |
| "step": 879104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.559919887995189e-05, | |
| "loss": 3.5887, | |
| "step": 879616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.559081293244137e-05, | |
| "loss": 3.6035, | |
| "step": 880128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5582426984930844e-05, | |
| "loss": 3.5997, | |
| "step": 880640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5574041037420324e-05, | |
| "loss": 3.6029, | |
| "step": 881152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.556567146871354e-05, | |
| "loss": 3.5844, | |
| "step": 881664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.555728552120302e-05, | |
| "loss": 3.5843, | |
| "step": 882176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.554889957369249e-05, | |
| "loss": 3.5769, | |
| "step": 882688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.554051362618197e-05, | |
| "loss": 3.5971, | |
| "step": 883200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.553214405747519e-05, | |
| "loss": 3.59, | |
| "step": 883712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.552375810996467e-05, | |
| "loss": 3.5932, | |
| "step": 884224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.551537216245415e-05, | |
| "loss": 3.5942, | |
| "step": 884736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5507002593747365e-05, | |
| "loss": 3.5974, | |
| "step": 885248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5498616646236845e-05, | |
| "loss": 3.5811, | |
| "step": 885760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.549023069872632e-05, | |
| "loss": 3.5881, | |
| "step": 886272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.54818447512158e-05, | |
| "loss": 3.5943, | |
| "step": 886784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5473475182509014e-05, | |
| "loss": 3.5825, | |
| "step": 887296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5465089234998493e-05, | |
| "loss": 3.5785, | |
| "step": 887808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.545670328748797e-05, | |
| "loss": 3.5852, | |
| "step": 888320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5448317339977447e-05, | |
| "loss": 3.5949, | |
| "step": 888832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.543994777127066e-05, | |
| "loss": 3.5943, | |
| "step": 889344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.543156182376014e-05, | |
| "loss": 3.5919, | |
| "step": 889856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.542317587624962e-05, | |
| "loss": 3.5946, | |
| "step": 890368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.54147899287391e-05, | |
| "loss": 3.5839, | |
| "step": 890880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.540642036003232e-05, | |
| "loss": 3.5919, | |
| "step": 891392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.539803441252179e-05, | |
| "loss": 3.5925, | |
| "step": 891904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.538964846501127e-05, | |
| "loss": 3.5871, | |
| "step": 892416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.538126251750075e-05, | |
| "loss": 3.5864, | |
| "step": 892928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.537289294879397e-05, | |
| "loss": 3.5827, | |
| "step": 893440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.536450700128344e-05, | |
| "loss": 3.5816, | |
| "step": 893952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.535612105377292e-05, | |
| "loss": 3.584, | |
| "step": 894464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.53477351062624e-05, | |
| "loss": 3.5861, | |
| "step": 894976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5339365537555616e-05, | |
| "loss": 3.5821, | |
| "step": 895488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5330979590045096e-05, | |
| "loss": 3.5911, | |
| "step": 896000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5322593642534576e-05, | |
| "loss": 3.5732, | |
| "step": 896512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5314207695024056e-05, | |
| "loss": 3.5819, | |
| "step": 897024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5305838126317265e-05, | |
| "loss": 3.5912, | |
| "step": 897536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5297452178806745e-05, | |
| "loss": 3.5953, | |
| "step": 898048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5289066231296225e-05, | |
| "loss": 3.58, | |
| "step": 898560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5280680283785705e-05, | |
| "loss": 3.5844, | |
| "step": 899072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5272310715078914e-05, | |
| "loss": 3.5795, | |
| "step": 899584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5263924767568394e-05, | |
| "loss": 3.5869, | |
| "step": 900096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5255538820057874e-05, | |
| "loss": 3.5915, | |
| "step": 900608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.524716925135109e-05, | |
| "loss": 3.5885, | |
| "step": 901120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.523878330384057e-05, | |
| "loss": 3.5768, | |
| "step": 901632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.523039735633005e-05, | |
| "loss": 3.5931, | |
| "step": 902144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.522201140881953e-05, | |
| "loss": 3.5845, | |
| "step": 902656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.521362546130901e-05, | |
| "loss": 3.587, | |
| "step": 903168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.520523951379849e-05, | |
| "loss": 3.577, | |
| "step": 903680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.519685356628797e-05, | |
| "loss": 3.5919, | |
| "step": 904192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.518848399758118e-05, | |
| "loss": 3.5831, | |
| "step": 904704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.518009805007066e-05, | |
| "loss": 3.5893, | |
| "step": 905216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.517171210256014e-05, | |
| "loss": 3.581, | |
| "step": 905728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.516332615504962e-05, | |
| "loss": 3.5828, | |
| "step": 906240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.515495658634283e-05, | |
| "loss": 3.5891, | |
| "step": 906752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.514657063883231e-05, | |
| "loss": 3.5916, | |
| "step": 907264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.513818469132179e-05, | |
| "loss": 3.5822, | |
| "step": 907776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.512979874381127e-05, | |
| "loss": 3.592, | |
| "step": 908288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5121429175104483e-05, | |
| "loss": 3.5794, | |
| "step": 908800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.511304322759396e-05, | |
| "loss": 3.5804, | |
| "step": 909312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.510465728008344e-05, | |
| "loss": 3.5859, | |
| "step": 909824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.509627133257292e-05, | |
| "loss": 3.5901, | |
| "step": 910336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.508790176386613e-05, | |
| "loss": 3.6014, | |
| "step": 910848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.507951581635561e-05, | |
| "loss": 3.5868, | |
| "step": 911360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.507112986884509e-05, | |
| "loss": 3.5871, | |
| "step": 911872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.506274392133457e-05, | |
| "loss": 3.5844, | |
| "step": 912384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.505437435262778e-05, | |
| "loss": 3.5817, | |
| "step": 912896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.504598840511726e-05, | |
| "loss": 3.5906, | |
| "step": 913408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.503760245760674e-05, | |
| "loss": 3.5962, | |
| "step": 913920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.502921651009622e-05, | |
| "loss": 3.5976, | |
| "step": 914432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.502084694138944e-05, | |
| "loss": 3.5796, | |
| "step": 914944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.501246099387892e-05, | |
| "loss": 3.5902, | |
| "step": 915456 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 3.897824764251709, | |
| "eval_runtime": 316.0109, | |
| "eval_samples_per_second": 1207.525, | |
| "eval_steps_per_second": 37.736, | |
| "step": 915828 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 5.936951604989583e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |