| { | |
| "best_metric": 4.053233623504639, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl/lstm/2/checkpoints/checkpoint-915828", | |
| "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": 10.8203, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 7.5687, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 7.0534, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 6.9801, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 6.9528, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 6.8928, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994970069374061e-05, | |
| "loss": 6.7305, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994131474623009e-05, | |
| "loss": 6.6212, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993292879871958e-05, | |
| "loss": 6.531, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992454285120906e-05, | |
| "loss": 6.4382, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.991615690369854e-05, | |
| "loss": 6.3957, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990777095618801e-05, | |
| "loss": 6.3302, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989938500867749e-05, | |
| "loss": 6.261, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989099906116697e-05, | |
| "loss": 6.1908, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988261311365645e-05, | |
| "loss": 6.1396, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987422716614593e-05, | |
| "loss": 6.0984, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986584121863541e-05, | |
| "loss": 6.0549, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985745527112489e-05, | |
| "loss": 5.997, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984906932361437e-05, | |
| "loss": 5.9588, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984068337610385e-05, | |
| "loss": 5.9206, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.8828, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.982391148108281e-05, | |
| "loss": 5.8553, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.981552553357229e-05, | |
| "loss": 5.8258, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.980713958606178e-05, | |
| "loss": 5.7928, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.979875363855125e-05, | |
| "loss": 5.7648, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.979036769104073e-05, | |
| "loss": 5.7474, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9781998122333946e-05, | |
| "loss": 5.7141, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773628553627155e-05, | |
| "loss": 5.6924, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9765242606116635e-05, | |
| "loss": 5.6582, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9756856658606115e-05, | |
| "loss": 5.6537, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748470711095595e-05, | |
| "loss": 5.6206, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9740101142388804e-05, | |
| "loss": 5.604, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9731715194878284e-05, | |
| "loss": 5.5936, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9723329247367764e-05, | |
| "loss": 5.5759, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714943299857244e-05, | |
| "loss": 5.5452, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.970657373115046e-05, | |
| "loss": 5.5326, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.969820416244367e-05, | |
| "loss": 5.5245, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968981821493315e-05, | |
| "loss": 5.5017, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968143226742263e-05, | |
| "loss": 5.4895, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.967304631991211e-05, | |
| "loss": 5.4689, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966466037240159e-05, | |
| "loss": 5.4454, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965627442489107e-05, | |
| "loss": 5.4333, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964788847738054e-05, | |
| "loss": 5.4194, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963950252987002e-05, | |
| "loss": 5.4124, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963113296116324e-05, | |
| "loss": 5.4004, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962274701365272e-05, | |
| "loss": 5.3951, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9614377444945934e-05, | |
| "loss": 5.3712, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9605991497435414e-05, | |
| "loss": 5.3707, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9597605549924894e-05, | |
| "loss": 5.3748, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9589219602414374e-05, | |
| "loss": 5.347, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958083365490385e-05, | |
| "loss": 5.3368, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.957244770739333e-05, | |
| "loss": 5.3234, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956406175988281e-05, | |
| "loss": 5.3324, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955567581237229e-05, | |
| "loss": 5.295, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9547306243665496e-05, | |
| "loss": 5.3127, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953893667495871e-05, | |
| "loss": 5.2827, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953055072744819e-05, | |
| "loss": 5.2793, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.952216477993767e-05, | |
| "loss": 5.2744, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951377883242715e-05, | |
| "loss": 5.2637, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.950539288491663e-05, | |
| "loss": 5.2516, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949700693740611e-05, | |
| "loss": 5.2388, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948862098989559e-05, | |
| "loss": 5.2241, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948023504238507e-05, | |
| "loss": 5.2312, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947184909487455e-05, | |
| "loss": 5.22, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946346314736403e-05, | |
| "loss": 5.2093, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945509357865724e-05, | |
| "loss": 5.2011, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944670763114672e-05, | |
| "loss": 5.1917, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94383216836362e-05, | |
| "loss": 5.1736, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942993573612568e-05, | |
| "loss": 5.1913, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942156616741889e-05, | |
| "loss": 5.1683, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.941318021990837e-05, | |
| "loss": 5.1615, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9404794272397856e-05, | |
| "loss": 5.1706, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9396408324887336e-05, | |
| "loss": 5.1465, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388022377376816e-05, | |
| "loss": 5.1592, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379652808670025e-05, | |
| "loss": 5.1496, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371266861159505e-05, | |
| "loss": 5.139, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362897292452714e-05, | |
| "loss": 5.137, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9354511344942194e-05, | |
| "loss": 5.1119, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.93461417762354e-05, | |
| "loss": 5.108, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.933775582872488e-05, | |
| "loss": 5.1074, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932936988121436e-05, | |
| "loss": 5.1066, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932098393370384e-05, | |
| "loss": 5.0962, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.931259798619332e-05, | |
| "loss": 5.0821, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.930421203868281e-05, | |
| "loss": 5.0683, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.929582609117229e-05, | |
| "loss": 5.0755, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.928744014366177e-05, | |
| "loss": 5.0744, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927905419615125e-05, | |
| "loss": 5.0703, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927066824864073e-05, | |
| "loss": 5.0674, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.92622823011302e-05, | |
| "loss": 5.0616, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.925389635361968e-05, | |
| "loss": 5.0502, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.92455267849129e-05, | |
| "loss": 5.0315, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923714083740238e-05, | |
| "loss": 5.0285, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922875488989185e-05, | |
| "loss": 5.0339, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922038532118507e-05, | |
| "loss": 5.0307, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921199937367455e-05, | |
| "loss": 5.0084, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920361342616403e-05, | |
| "loss": 5.0133, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919522747865351e-05, | |
| "loss": 5.0087, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918684153114299e-05, | |
| "loss": 5.0097, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.91784719624362e-05, | |
| "loss": 5.0005, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9170086014925676e-05, | |
| "loss": 5.0025, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9161700067415156e-05, | |
| "loss": 4.9942, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9153314119904636e-05, | |
| "loss": 4.9846, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9144928172394116e-05, | |
| "loss": 4.9828, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9136542224883596e-05, | |
| "loss": 4.9844, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9128156277373076e-05, | |
| "loss": 4.9684, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9119770329862556e-05, | |
| "loss": 4.9716, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9111400761155765e-05, | |
| "loss": 4.9488, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9103014813645245e-05, | |
| "loss": 4.9452, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.909462886613473e-05, | |
| "loss": 4.9431, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.908624291862421e-05, | |
| "loss": 4.9489, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907785697111369e-05, | |
| "loss": 4.941, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906947102360317e-05, | |
| "loss": 4.9176, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906108507609265e-05, | |
| "loss": 4.9233, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905271550738586e-05, | |
| "loss": 4.9261, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904432955987534e-05, | |
| "loss": 4.9335, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903594361236482e-05, | |
| "loss": 4.9202, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90275576648543e-05, | |
| "loss": 4.9187, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901917171734378e-05, | |
| "loss": 4.915, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901078576983326e-05, | |
| "loss": 4.9177, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.900239982232274e-05, | |
| "loss": 4.9009, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899403025361595e-05, | |
| "loss": 4.9012, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.898564430610543e-05, | |
| "loss": 4.8918, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.897725835859491e-05, | |
| "loss": 4.8958, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968872411084396e-05, | |
| "loss": 4.8777, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.896048646357387e-05, | |
| "loss": 4.8901, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.895210051606335e-05, | |
| "loss": 4.881, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.894371456855283e-05, | |
| "loss": 4.8882, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.893532862104231e-05, | |
| "loss": 4.8592, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.892695905233552e-05, | |
| "loss": 4.8706, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918573104825e-05, | |
| "loss": 4.8804, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.891018715731448e-05, | |
| "loss": 4.853, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890180120980396e-05, | |
| "loss": 4.8728, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889341526229344e-05, | |
| "loss": 4.8565, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888504569358665e-05, | |
| "loss": 4.8597, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8876659746076134e-05, | |
| "loss": 4.8493, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8868273798565614e-05, | |
| "loss": 4.8443, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8859887851055094e-05, | |
| "loss": 4.8456, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88515182823483e-05, | |
| "loss": 4.844, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884313233483778e-05, | |
| "loss": 4.845, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883474638732726e-05, | |
| "loss": 4.8369, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882636043981674e-05, | |
| "loss": 4.8388, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881797449230622e-05, | |
| "loss": 4.8373, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880960492359943e-05, | |
| "loss": 4.8183, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880123535489264e-05, | |
| "loss": 4.8198, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879284940738212e-05, | |
| "loss": 4.8097, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.87844634598716e-05, | |
| "loss": 4.8113, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877607751236109e-05, | |
| "loss": 4.8115, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.876769156485057e-05, | |
| "loss": 4.8039, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875930561734005e-05, | |
| "loss": 4.8016, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875093604863326e-05, | |
| "loss": 4.8008, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.762693881988525, | |
| "eval_runtime": 547.5599, | |
| "eval_samples_per_second": 696.894, | |
| "eval_steps_per_second": 21.778, | |
| "step": 76319 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.874255010112274e-05, | |
| "loss": 4.8123, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.873416415361222e-05, | |
| "loss": 4.8018, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.87257782061017e-05, | |
| "loss": 4.7914, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.871739225859118e-05, | |
| "loss": 4.7917, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8709006311080657e-05, | |
| "loss": 4.7865, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8700620363570137e-05, | |
| "loss": 4.7638, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8692234416059616e-05, | |
| "loss": 4.778, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8683848468549096e-05, | |
| "loss": 4.7801, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8675462521038576e-05, | |
| "loss": 4.7616, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8667092952331785e-05, | |
| "loss": 4.7768, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.865870700482127e-05, | |
| "loss": 4.7818, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.865032105731075e-05, | |
| "loss": 4.7651, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8641935109800225e-05, | |
| "loss": 4.7583, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8633549162289705e-05, | |
| "loss": 4.7605, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8625163214779185e-05, | |
| "loss": 4.7607, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8616777267268665e-05, | |
| "loss": 4.7545, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8608391319758145e-05, | |
| "loss": 4.75, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8600005372247625e-05, | |
| "loss": 4.7488, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8591619424737105e-05, | |
| "loss": 4.7441, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8583233477226585e-05, | |
| "loss": 4.7474, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8574847529716065e-05, | |
| "loss": 4.7475, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8566461582205545e-05, | |
| "loss": 4.747, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8558092013498754e-05, | |
| "loss": 4.7408, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.854970606598824e-05, | |
| "loss": 4.7411, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.854132011847772e-05, | |
| "loss": 4.7342, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.85329341709672e-05, | |
| "loss": 4.7323, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.852456460226041e-05, | |
| "loss": 4.7334, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.851617865474989e-05, | |
| "loss": 4.716, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.850779270723937e-05, | |
| "loss": 4.7295, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.849942313853258e-05, | |
| "loss": 4.7101, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.849103719102206e-05, | |
| "loss": 4.7153, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.848265124351154e-05, | |
| "loss": 4.7248, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.847426529600102e-05, | |
| "loss": 4.7218, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.84658793484905e-05, | |
| "loss": 4.708, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.845749340097998e-05, | |
| "loss": 4.7001, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.844910745346946e-05, | |
| "loss": 4.718, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.844072150595894e-05, | |
| "loss": 4.7053, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.843233555844842e-05, | |
| "loss": 4.7041, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.84239496109379e-05, | |
| "loss": 4.7015, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8415580042231114e-05, | |
| "loss": 4.6859, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.840719409472059e-05, | |
| "loss": 4.683, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.839880814721007e-05, | |
| "loss": 4.6846, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.839042219969955e-05, | |
| "loss": 4.6891, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.838203625218903e-05, | |
| "loss": 4.6821, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.837365030467851e-05, | |
| "loss": 4.6854, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.836526435716799e-05, | |
| "loss": 4.6781, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8356894788461196e-05, | |
| "loss": 4.6801, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8348508840950676e-05, | |
| "loss": 4.6916, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8340122893440156e-05, | |
| "loss": 4.6734, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.833173694592964e-05, | |
| "loss": 4.6745, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.832335099841912e-05, | |
| "loss": 4.6637, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.831498142971233e-05, | |
| "loss": 4.6853, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.830659548220181e-05, | |
| "loss": 4.6546, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.829820953469129e-05, | |
| "loss": 4.6767, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.828982358718077e-05, | |
| "loss": 4.6536, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.828143763967025e-05, | |
| "loss": 4.662, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.827305169215973e-05, | |
| "loss": 4.6628, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.826468212345294e-05, | |
| "loss": 4.6577, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.825629617594242e-05, | |
| "loss": 4.6493, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.82479102284319e-05, | |
| "loss": 4.6494, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.823952428092138e-05, | |
| "loss": 4.6365, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.823113833341086e-05, | |
| "loss": 4.6537, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.822275238590034e-05, | |
| "loss": 4.6452, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8214382817193557e-05, | |
| "loss": 4.6394, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8205996869683037e-05, | |
| "loss": 4.6377, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8197610922172516e-05, | |
| "loss": 4.6367, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8189224974661996e-05, | |
| "loss": 4.6197, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8180839027151476e-05, | |
| "loss": 4.6439, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.817245307964095e-05, | |
| "loss": 4.6283, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8164083510934165e-05, | |
| "loss": 4.6296, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8155697563423645e-05, | |
| "loss": 4.6379, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8147311615913125e-05, | |
| "loss": 4.6233, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8138942047206334e-05, | |
| "loss": 4.6383, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8130556099695814e-05, | |
| "loss": 4.6384, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8122170152185294e-05, | |
| "loss": 4.6325, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8113784204674774e-05, | |
| "loss": 4.6309, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8105398257164254e-05, | |
| "loss": 4.6124, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8097012309653734e-05, | |
| "loss": 4.6112, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8088626362143214e-05, | |
| "loss": 4.6203, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8080240414632694e-05, | |
| "loss": 4.6231, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8071854467122174e-05, | |
| "loss": 4.6138, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8063468519611654e-05, | |
| "loss": 4.6085, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.805509895090486e-05, | |
| "loss": 4.5931, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.804671300339434e-05, | |
| "loss": 4.6024, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.803832705588382e-05, | |
| "loss": 4.6073, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.80299411083733e-05, | |
| "loss": 4.6061, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.802155516086278e-05, | |
| "loss": 4.6146, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8013185592156e-05, | |
| "loss": 4.6105, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.800479964464548e-05, | |
| "loss": 4.5996, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.799643007593869e-05, | |
| "loss": 4.5937, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.798804412842817e-05, | |
| "loss": 4.5871, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.797965818091765e-05, | |
| "loss": 4.5926, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.797127223340713e-05, | |
| "loss": 4.5957, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.796288628589661e-05, | |
| "loss": 4.5838, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.795450033838609e-05, | |
| "loss": 4.5833, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.794611439087557e-05, | |
| "loss": 4.5923, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.793772844336505e-05, | |
| "loss": 4.5951, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792934249585453e-05, | |
| "loss": 4.5842, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792097292714774e-05, | |
| "loss": 4.59, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7912586979637217e-05, | |
| "loss": 4.5911, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.790421741093043e-05, | |
| "loss": 4.5834, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.789583146341991e-05, | |
| "loss": 4.579, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.788744551590939e-05, | |
| "loss": 4.5886, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.787905956839887e-05, | |
| "loss": 4.5753, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.787067362088835e-05, | |
| "loss": 4.5868, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.786228767337783e-05, | |
| "loss": 4.5671, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.785390172586731e-05, | |
| "loss": 4.5559, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7845515778356785e-05, | |
| "loss": 4.5642, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7837129830846265e-05, | |
| "loss": 4.5728, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7828743883335745e-05, | |
| "loss": 4.5648, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7820357935825225e-05, | |
| "loss": 4.5506, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7811971988314705e-05, | |
| "loss": 4.5516, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.780360241960792e-05, | |
| "loss": 4.5602, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.779523285090114e-05, | |
| "loss": 4.5708, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.778684690339061e-05, | |
| "loss": 4.5587, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.777846095588009e-05, | |
| "loss": 4.5675, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.777007500836957e-05, | |
| "loss": 4.5612, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.776168906085905e-05, | |
| "loss": 4.5589, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7753335870955995e-05, | |
| "loss": 4.5543, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7744949923445475e-05, | |
| "loss": 4.5512, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7736563975934955e-05, | |
| "loss": 4.5459, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7728178028424435e-05, | |
| "loss": 4.5565, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.771979208091391e-05, | |
| "loss": 4.5422, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.771140613340339e-05, | |
| "loss": 4.5447, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7703020185892875e-05, | |
| "loss": 4.5482, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7694650617186084e-05, | |
| "loss": 4.5548, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7686264669675564e-05, | |
| "loss": 4.5325, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7677878722165044e-05, | |
| "loss": 4.5379, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7669492774654524e-05, | |
| "loss": 4.5592, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7661106827144004e-05, | |
| "loss": 4.5323, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.765273725843721e-05, | |
| "loss": 4.5536, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.764435131092669e-05, | |
| "loss": 4.5361, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.763596536341617e-05, | |
| "loss": 4.5473, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.762757941590565e-05, | |
| "loss": 4.5339, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.761919346839513e-05, | |
| "loss": 4.5286, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.761080752088461e-05, | |
| "loss": 4.5373, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.760242157337409e-05, | |
| "loss": 4.5353, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.759403562586357e-05, | |
| "loss": 4.5423, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.758564967835306e-05, | |
| "loss": 4.5374, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.757728010964627e-05, | |
| "loss": 4.5371, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.756889416213575e-05, | |
| "loss": 4.5334, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.756050821462523e-05, | |
| "loss": 4.5196, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.755212226711471e-05, | |
| "loss": 4.5304, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.754373631960419e-05, | |
| "loss": 4.5158, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.75353667508974e-05, | |
| "loss": 4.5221, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.752698080338688e-05, | |
| "loss": 4.5251, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.751859485587636e-05, | |
| "loss": 4.5122, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.751020890836584e-05, | |
| "loss": 4.5178, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7501839339659046e-05, | |
| "loss": 4.5151, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.481568813323975, | |
| "eval_runtime": 534.6878, | |
| "eval_samples_per_second": 713.671, | |
| "eval_steps_per_second": 22.303, | |
| "step": 152638 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7493453392148526e-05, | |
| "loss": 4.5286, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.748506744463801e-05, | |
| "loss": 4.524, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.747668149712749e-05, | |
| "loss": 4.5131, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.746829554961697e-05, | |
| "loss": 4.5121, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7459909602106446e-05, | |
| "loss": 4.5118, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7451523654595926e-05, | |
| "loss": 4.4943, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7443137707085406e-05, | |
| "loss": 4.5092, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7434751759574886e-05, | |
| "loss": 4.5075, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7426365812064366e-05, | |
| "loss": 4.4931, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7417979864553846e-05, | |
| "loss": 4.5115, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7409593917043326e-05, | |
| "loss": 4.5183, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7401207969532806e-05, | |
| "loss": 4.5013, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7392838400826015e-05, | |
| "loss": 4.4945, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7384452453315495e-05, | |
| "loss": 4.4981, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7376066505804975e-05, | |
| "loss": 4.5022, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.736768055829446e-05, | |
| "loss": 4.4961, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.735929461078394e-05, | |
| "loss": 4.4938, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.735090866327342e-05, | |
| "loss": 4.4925, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.73425227157629e-05, | |
| "loss": 4.4939, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.733413676825238e-05, | |
| "loss": 4.4959, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.732576719954559e-05, | |
| "loss": 4.496, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.731738125203507e-05, | |
| "loss": 4.4993, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730899530452455e-05, | |
| "loss": 4.4975, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730060935701403e-05, | |
| "loss": 4.4898, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.729222340950351e-05, | |
| "loss": 4.4924, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.728385384079672e-05, | |
| "loss": 4.4882, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.72754678932862e-05, | |
| "loss": 4.4922, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.726708194577568e-05, | |
| "loss": 4.4761, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.725869599826516e-05, | |
| "loss": 4.4876, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7250326429558375e-05, | |
| "loss": 4.4727, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7241940482047855e-05, | |
| "loss": 4.479, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7233554534537335e-05, | |
| "loss": 4.4911, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.722516858702681e-05, | |
| "loss": 4.4863, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.721678263951629e-05, | |
| "loss": 4.4757, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7208413070809504e-05, | |
| "loss": 4.4699, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7200027123298984e-05, | |
| "loss": 4.4908, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.719164117578846e-05, | |
| "loss": 4.472, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.718325522827794e-05, | |
| "loss": 4.4798, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.717486928076742e-05, | |
| "loss": 4.4762, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.71664833332569e-05, | |
| "loss": 4.4604, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7158097385746384e-05, | |
| "loss": 4.4584, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7149711438235864e-05, | |
| "loss": 4.4602, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714134186952907e-05, | |
| "loss": 4.4739, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.713295592201855e-05, | |
| "loss": 4.4638, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.712456997450803e-05, | |
| "loss": 4.4654, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.711620040580124e-05, | |
| "loss": 4.4579, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.710781445829072e-05, | |
| "loss": 4.4622, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.70994285107802e-05, | |
| "loss": 4.4778, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.709104256326968e-05, | |
| "loss": 4.4569, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.708265661575916e-05, | |
| "loss": 4.4575, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.707427066824864e-05, | |
| "loss": 4.4525, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.706590109954185e-05, | |
| "loss": 4.4724, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.705751515203134e-05, | |
| "loss": 4.4442, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704912920452082e-05, | |
| "loss": 4.4654, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.70407432570103e-05, | |
| "loss": 4.4453, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.703235730949978e-05, | |
| "loss": 4.4569, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.702397136198926e-05, | |
| "loss": 4.4589, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.701558541447874e-05, | |
| "loss": 4.4463, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7007215845771946e-05, | |
| "loss": 4.453, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6998829898261426e-05, | |
| "loss": 4.4456, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6990443950750906e-05, | |
| "loss": 4.4299, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6982058003240386e-05, | |
| "loss": 4.4503, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6973688434533595e-05, | |
| "loss": 4.447, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6965302487023075e-05, | |
| "loss": 4.4395, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6956916539512555e-05, | |
| "loss": 4.44, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6948530592002035e-05, | |
| "loss": 4.4398, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6940144644491515e-05, | |
| "loss": 4.4225, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6931758696980995e-05, | |
| "loss": 4.4486, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6923372749470475e-05, | |
| "loss": 4.4301, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.691500318076369e-05, | |
| "loss": 4.4349, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.690661723325317e-05, | |
| "loss": 4.4451, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6898231285742644e-05, | |
| "loss": 4.4332, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6889845338232124e-05, | |
| "loss": 4.4422, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6881459390721604e-05, | |
| "loss": 4.4515, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.687308982201481e-05, | |
| "loss": 4.4383, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.686470387450429e-05, | |
| "loss": 4.4457, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.685631792699377e-05, | |
| "loss": 4.426, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.684793197948325e-05, | |
| "loss": 4.4218, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683956241077647e-05, | |
| "loss": 4.4309, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683117646326595e-05, | |
| "loss": 4.4415, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.682279051575543e-05, | |
| "loss": 4.4291, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.681440456824491e-05, | |
| "loss": 4.4234, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.680601862073439e-05, | |
| "loss": 4.4144, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.679763267322387e-05, | |
| "loss": 4.4172, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678924672571335e-05, | |
| "loss": 4.4204, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678086077820283e-05, | |
| "loss": 4.4256, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.677247483069231e-05, | |
| "loss": 4.4337, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.676408888318179e-05, | |
| "loss": 4.4311, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.675570293567127e-05, | |
| "loss": 4.4213, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.674733336696448e-05, | |
| "loss": 4.4167, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.673894741945396e-05, | |
| "loss": 4.4112, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.673056147194344e-05, | |
| "loss": 4.4156, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6722175524432924e-05, | |
| "loss": 4.4159, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.671380595572613e-05, | |
| "loss": 4.4119, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.670542000821561e-05, | |
| "loss": 4.4094, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.669703406070509e-05, | |
| "loss": 4.4203, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.668864811319457e-05, | |
| "loss": 4.4191, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.668026216568405e-05, | |
| "loss": 4.4166, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.667187621817353e-05, | |
| "loss": 4.4169, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6663490270663006e-05, | |
| "loss": 4.4195, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6655104323152486e-05, | |
| "loss": 4.4143, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.66467347544457e-05, | |
| "loss": 4.4072, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6638348806935175e-05, | |
| "loss": 4.4234, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.662997923822839e-05, | |
| "loss": 4.4043, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.662159329071788e-05, | |
| "loss": 4.4241, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.661320734320736e-05, | |
| "loss": 4.3968, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.660482139569683e-05, | |
| "loss": 4.391, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.659643544818631e-05, | |
| "loss": 4.3983, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.658804950067579e-05, | |
| "loss": 4.4121, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657966355316527e-05, | |
| "loss": 4.4031, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657129398445848e-05, | |
| "loss": 4.3874, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6562924415751695e-05, | |
| "loss": 4.3895, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6554538468241175e-05, | |
| "loss": 4.3963, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.654615252073065e-05, | |
| "loss": 4.4109, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.653776657322013e-05, | |
| "loss": 4.3982, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6529397004513344e-05, | |
| "loss": 4.4072, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.652101105700283e-05, | |
| "loss": 4.4018, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6512625109492304e-05, | |
| "loss": 4.4, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6504239161981784e-05, | |
| "loss": 4.402, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6495853214471264e-05, | |
| "loss": 4.3896, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6487467266960744e-05, | |
| "loss": 4.391, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647909769825395e-05, | |
| "loss": 4.4013, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647071175074343e-05, | |
| "loss": 4.3849, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.646232580323291e-05, | |
| "loss": 4.3919, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.645393985572239e-05, | |
| "loss": 4.3944, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.64455702870156e-05, | |
| "loss": 4.3988, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.643718433950508e-05, | |
| "loss": 4.3828, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642879839199457e-05, | |
| "loss": 4.3812, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642041244448405e-05, | |
| "loss": 4.4068, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.641202649697353e-05, | |
| "loss": 4.385, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.640364054946301e-05, | |
| "loss": 4.3975, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.639525460195249e-05, | |
| "loss": 4.3889, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.638686865444197e-05, | |
| "loss": 4.3955, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637848270693145e-05, | |
| "loss": 4.385, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637009675942093e-05, | |
| "loss": 4.3822, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.636172719071414e-05, | |
| "loss": 4.3916, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.635334124320362e-05, | |
| "loss": 4.384, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.63449552956931e-05, | |
| "loss": 4.396, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.633656934818258e-05, | |
| "loss": 4.3905, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.632818340067206e-05, | |
| "loss": 4.3905, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.631979745316154e-05, | |
| "loss": 4.3852, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.631142788445475e-05, | |
| "loss": 4.3766, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.630304193694423e-05, | |
| "loss": 4.3845, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.629465598943371e-05, | |
| "loss": 4.3738, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.628627004192319e-05, | |
| "loss": 4.378, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6277884094412666e-05, | |
| "loss": 4.3789, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6269498146902146e-05, | |
| "loss": 4.3709, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6261112199391626e-05, | |
| "loss": 4.3752, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6252742630684835e-05, | |
| "loss": 4.374, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.347803592681885, | |
| "eval_runtime": 537.7927, | |
| "eval_samples_per_second": 709.55, | |
| "eval_steps_per_second": 22.174, | |
| "step": 228957 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6244356683174315e-05, | |
| "loss": 4.3823, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6235970735663795e-05, | |
| "loss": 4.3866, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6227584788153275e-05, | |
| "loss": 4.3735, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6219198840642755e-05, | |
| "loss": 4.3717, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6210812893132235e-05, | |
| "loss": 4.3742, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6202426945621715e-05, | |
| "loss": 4.3549, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.619405737691493e-05, | |
| "loss": 4.3695, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.618567142940441e-05, | |
| "loss": 4.3724, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.617728548189389e-05, | |
| "loss": 4.3604, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61689159131871e-05, | |
| "loss": 4.3743, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616052996567658e-05, | |
| "loss": 4.3822, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.615214401816606e-05, | |
| "loss": 4.362, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.614375807065554e-05, | |
| "loss": 4.3618, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.613537212314502e-05, | |
| "loss": 4.3639, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61269861756345e-05, | |
| "loss": 4.3686, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611860022812398e-05, | |
| "loss": 4.3596, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611021428061346e-05, | |
| "loss": 4.3583, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.610182833310294e-05, | |
| "loss": 4.3593, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.609344238559242e-05, | |
| "loss": 4.3625, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.60850564380819e-05, | |
| "loss": 4.3673, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.607667049057138e-05, | |
| "loss": 4.3629, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6068300921864595e-05, | |
| "loss": 4.3673, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6059931353157805e-05, | |
| "loss": 4.3658, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6051545405647285e-05, | |
| "loss": 4.3606, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6043175836940494e-05, | |
| "loss": 4.3571, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6034789889429974e-05, | |
| "loss": 4.3573, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6026403941919454e-05, | |
| "loss": 4.3649, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6018017994408933e-05, | |
| "loss": 4.3498, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6009632046898413e-05, | |
| "loss": 4.3596, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6001246099387893e-05, | |
| "loss": 4.3461, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.599286015187737e-05, | |
| "loss": 4.3492, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.598447420436685e-05, | |
| "loss": 4.3638, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.597610463566007e-05, | |
| "loss": 4.358, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.596771868814955e-05, | |
| "loss": 4.3514, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595933274063903e-05, | |
| "loss": 4.3472, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.59509467931285e-05, | |
| "loss": 4.36, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.594256084561798e-05, | |
| "loss": 4.3454, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.59341912769112e-05, | |
| "loss": 4.3565, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.592580532940067e-05, | |
| "loss": 4.3498, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.591741938189015e-05, | |
| "loss": 4.3385, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.590903343437963e-05, | |
| "loss": 4.3375, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.590064748686911e-05, | |
| "loss": 4.3388, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.589226153935859e-05, | |
| "loss": 4.3483, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.588387559184808e-05, | |
| "loss": 4.3403, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.587548964433756e-05, | |
| "loss": 4.3436, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.586712007563077e-05, | |
| "loss": 4.3391, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.585873412812025e-05, | |
| "loss": 4.342, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.585034818060973e-05, | |
| "loss": 4.3544, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.584196223309921e-05, | |
| "loss": 4.3362, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5833592664392416e-05, | |
| "loss": 4.3371, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5825206716881896e-05, | |
| "loss": 4.3383, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5816820769371376e-05, | |
| "loss": 4.3485, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5808434821860856e-05, | |
| "loss": 4.3278, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5800065253154065e-05, | |
| "loss": 4.3439, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5791679305643545e-05, | |
| "loss": 4.3272, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5783293358133025e-05, | |
| "loss": 4.3352, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.577492378942624e-05, | |
| "loss": 4.3411, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.576653784191572e-05, | |
| "loss": 4.3239, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.575816827320893e-05, | |
| "loss": 4.3414, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.574978232569841e-05, | |
| "loss": 4.3316, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.574139637818789e-05, | |
| "loss": 4.3154, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.573301043067737e-05, | |
| "loss": 4.3317, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.572462448316685e-05, | |
| "loss": 4.331, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.571625491446006e-05, | |
| "loss": 4.3296, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.570786896694954e-05, | |
| "loss": 4.3239, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.569948301943902e-05, | |
| "loss": 4.3265, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.56910970719285e-05, | |
| "loss": 4.3087, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.568271112441798e-05, | |
| "loss": 4.335, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5674325176907465e-05, | |
| "loss": 4.3171, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5665955608200674e-05, | |
| "loss": 4.322, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5657569660690154e-05, | |
| "loss": 4.3314, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5649183713179634e-05, | |
| "loss": 4.3208, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5640797765669114e-05, | |
| "loss": 4.3304, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5632411818158594e-05, | |
| "loss": 4.3356, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.562405862825553e-05, | |
| "loss": 4.3303, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.561567268074501e-05, | |
| "loss": 4.3351, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.560728673323449e-05, | |
| "loss": 4.3141, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.559890078572397e-05, | |
| "loss": 4.3132, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.559051483821345e-05, | |
| "loss": 4.3202, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.558212889070293e-05, | |
| "loss": 4.3333, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.557374294319242e-05, | |
| "loss": 4.3201, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.55653569956819e-05, | |
| "loss": 4.3127, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.555697104817138e-05, | |
| "loss": 4.3037, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554858510066086e-05, | |
| "loss": 4.3117, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554019915315033e-05, | |
| "loss": 4.3118, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.553181320563981e-05, | |
| "loss": 4.316, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.552344363693303e-05, | |
| "loss": 4.3228, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.551505768942251e-05, | |
| "loss": 4.3262, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.550668812071572e-05, | |
| "loss": 4.3136, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.54983021732052e-05, | |
| "loss": 4.3137, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5489932604498406e-05, | |
| "loss": 4.3026, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5481546656987886e-05, | |
| "loss": 4.3073, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.547316070947737e-05, | |
| "loss": 4.3132, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.546477476196685e-05, | |
| "loss": 4.3064, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.545638881445633e-05, | |
| "loss": 4.3016, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5448002866945806e-05, | |
| "loss": 4.3165, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5439616919435286e-05, | |
| "loss": 4.3109, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5431230971924765e-05, | |
| "loss": 4.3189, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5422845024414245e-05, | |
| "loss": 4.309, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5414475455707455e-05, | |
| "loss": 4.3157, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5406089508196934e-05, | |
| "loss": 4.3115, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5397703560686414e-05, | |
| "loss": 4.3053, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5389317613175894e-05, | |
| "loss": 4.3222, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5380931665665374e-05, | |
| "loss": 4.3018, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5372545718154854e-05, | |
| "loss": 4.3236, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.536415977064434e-05, | |
| "loss": 4.2935, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.535577382313382e-05, | |
| "loss": 4.2914, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.534740425442703e-05, | |
| "loss": 4.2968, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533901830691651e-05, | |
| "loss": 4.3108, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533064873820972e-05, | |
| "loss": 4.3032, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.53222627906992e-05, | |
| "loss": 4.2872, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.531387684318868e-05, | |
| "loss": 4.2888, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.530549089567816e-05, | |
| "loss": 4.2945, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.529710494816764e-05, | |
| "loss": 4.3131, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528871900065712e-05, | |
| "loss": 4.2975, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.52803330531466e-05, | |
| "loss": 4.3107, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.527194710563608e-05, | |
| "loss": 4.3018, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5263577536929295e-05, | |
| "loss": 4.304, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5255191589418775e-05, | |
| "loss": 4.3022, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5246805641908255e-05, | |
| "loss": 4.2934, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5238419694397735e-05, | |
| "loss": 4.294, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5230050125690944e-05, | |
| "loss": 4.3005, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5221664178180424e-05, | |
| "loss": 4.2898, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5213278230669904e-05, | |
| "loss": 4.295, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5204892283159384e-05, | |
| "loss": 4.2997, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.519652271445259e-05, | |
| "loss": 4.2973, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.518813676694207e-05, | |
| "loss": 4.2888, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517975081943155e-05, | |
| "loss": 4.2867, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517136487192103e-05, | |
| "loss": 4.3086, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.516299530321425e-05, | |
| "loss": 4.2887, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.515460935570373e-05, | |
| "loss": 4.3022, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.514622340819321e-05, | |
| "loss": 4.2963, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.513785383948642e-05, | |
| "loss": 4.299, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.51294678919759e-05, | |
| "loss": 4.2907, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512108194446538e-05, | |
| "loss": 4.2921, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.511269599695486e-05, | |
| "loss": 4.2944, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.510431004944434e-05, | |
| "loss": 4.29, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.509592410193382e-05, | |
| "loss": 4.3031, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.508753815442329e-05, | |
| "loss": 4.2931, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.507915220691277e-05, | |
| "loss": 4.2989, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5070782638205986e-05, | |
| "loss": 4.2968, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5062396690695466e-05, | |
| "loss": 4.2842, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5054010743184946e-05, | |
| "loss": 4.2919, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.504564117447816e-05, | |
| "loss": 4.2815, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.503725522696764e-05, | |
| "loss": 4.287, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5028869279457115e-05, | |
| "loss": 4.2849, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5020483331946595e-05, | |
| "loss": 4.2834, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5012097384436075e-05, | |
| "loss": 4.2782, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.500372781572929e-05, | |
| "loss": 4.2862, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.264499664306641, | |
| "eval_runtime": 538.3004, | |
| "eval_samples_per_second": 708.881, | |
| "eval_steps_per_second": 22.153, | |
| "step": 305276 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.49953582470225e-05, | |
| "loss": 4.2956, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.498697229951198e-05, | |
| "loss": 4.2931, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.497858635200146e-05, | |
| "loss": 4.2842, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.497020040449093e-05, | |
| "loss": 4.2781, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.496181445698042e-05, | |
| "loss": 4.2876, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.49534285094699e-05, | |
| "loss": 4.2654, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.494504256195938e-05, | |
| "loss": 4.2838, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.493665661444886e-05, | |
| "loss": 4.2791, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.492827066693834e-05, | |
| "loss": 4.2733, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491990109823155e-05, | |
| "loss": 4.2848, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491151515072103e-05, | |
| "loss": 4.2916, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.490312920321051e-05, | |
| "loss": 4.2721, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.489474325569999e-05, | |
| "loss": 4.2769, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.488635730818947e-05, | |
| "loss": 4.2753, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.487797136067895e-05, | |
| "loss": 4.2826, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486958541316843e-05, | |
| "loss": 4.2692, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486119946565791e-05, | |
| "loss": 4.2707, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.485281351814739e-05, | |
| "loss": 4.2708, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.484442757063687e-05, | |
| "loss": 4.2736, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.483604162312635e-05, | |
| "loss": 4.2841, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.482765567561583e-05, | |
| "loss": 4.2755, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.481926972810531e-05, | |
| "loss": 4.2828, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4810900159398524e-05, | |
| "loss": 4.2833, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.480253059069173e-05, | |
| "loss": 4.2746, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.479414464318121e-05, | |
| "loss": 4.2715, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.478575869567069e-05, | |
| "loss": 4.2717, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.477737274816017e-05, | |
| "loss": 4.2794, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.476898680064965e-05, | |
| "loss": 4.2666, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4760600853139126e-05, | |
| "loss": 4.2722, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4752214905628606e-05, | |
| "loss": 4.269, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4743828958118086e-05, | |
| "loss": 4.2605, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.473544301060757e-05, | |
| "loss": 4.277, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.472705706309705e-05, | |
| "loss": 4.2793, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.471867111558653e-05, | |
| "loss": 4.27, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.471030154687974e-05, | |
| "loss": 4.2643, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.470191559936922e-05, | |
| "loss": 4.2729, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.469354603066243e-05, | |
| "loss": 4.2702, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.468516008315191e-05, | |
| "loss": 4.2702, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.467677413564139e-05, | |
| "loss": 4.2683, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466838818813087e-05, | |
| "loss": 4.2555, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466000224062035e-05, | |
| "loss": 4.2551, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.465161629310983e-05, | |
| "loss": 4.2537, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.464323034559931e-05, | |
| "loss": 4.2678, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4634860776892526e-05, | |
| "loss": 4.2607, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4626474829382006e-05, | |
| "loss": 4.2611, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4618088881871486e-05, | |
| "loss": 4.2605, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4609702934360966e-05, | |
| "loss": 4.2588, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4601333365654175e-05, | |
| "loss": 4.2739, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4592947418143655e-05, | |
| "loss": 4.2569, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4584561470633135e-05, | |
| "loss": 4.2547, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4576175523122615e-05, | |
| "loss": 4.2614, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4567789575612095e-05, | |
| "loss": 4.2654, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4559420006905304e-05, | |
| "loss": 4.2506, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4551034059394784e-05, | |
| "loss": 4.2596, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4542648111884264e-05, | |
| "loss": 4.2515, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4534262164373744e-05, | |
| "loss": 4.257, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4525876216863224e-05, | |
| "loss": 4.2613, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4517490269352704e-05, | |
| "loss": 4.2451, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.450910432184219e-05, | |
| "loss": 4.2643, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4500718374331664e-05, | |
| "loss": 4.2502, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.449234880562488e-05, | |
| "loss": 4.2403, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.448396285811436e-05, | |
| "loss": 4.2505, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.447557691060384e-05, | |
| "loss": 4.2533, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.446719096309331e-05, | |
| "loss": 4.2562, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445880501558279e-05, | |
| "loss": 4.2465, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445045182567974e-05, | |
| "loss": 4.2511, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.444206587816922e-05, | |
| "loss": 4.2277, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.44336799306587e-05, | |
| "loss": 4.2594, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.442529398314818e-05, | |
| "loss": 4.2409, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.441690803563766e-05, | |
| "loss": 4.2449, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440852208812714e-05, | |
| "loss": 4.2577, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440013614061662e-05, | |
| "loss": 4.246, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.43917501931061e-05, | |
| "loss": 4.2513, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4383380624399314e-05, | |
| "loss": 4.2582, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.437499467688879e-05, | |
| "loss": 4.2553, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.436660872937827e-05, | |
| "loss": 4.2571, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4358222781867747e-05, | |
| "loss": 4.2425, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4349836834357227e-05, | |
| "loss": 4.237, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4341450886846707e-05, | |
| "loss": 4.2461, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4333064939336186e-05, | |
| "loss": 4.2568, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4324695370629396e-05, | |
| "loss": 4.2452, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.431630942311888e-05, | |
| "loss": 4.2408, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.430792347560836e-05, | |
| "loss": 4.2288, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.429953752809784e-05, | |
| "loss": 4.2398, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.429115158058732e-05, | |
| "loss": 4.2354, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.42827656330768e-05, | |
| "loss": 4.2453, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.427437968556628e-05, | |
| "loss": 4.2431, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.426601011685949e-05, | |
| "loss": 4.255, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.425762416934897e-05, | |
| "loss": 4.2385, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424923822183845e-05, | |
| "loss": 4.2418, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424085227432793e-05, | |
| "loss": 4.2286, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.423248270562114e-05, | |
| "loss": 4.2357, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.422409675811062e-05, | |
| "loss": 4.241, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.42157108106001e-05, | |
| "loss": 4.2309, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.420732486308958e-05, | |
| "loss": 4.2304, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419893891557907e-05, | |
| "loss": 4.2455, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419055296806855e-05, | |
| "loss": 4.234, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.418216702055803e-05, | |
| "loss": 4.2542, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.41737810730475e-05, | |
| "loss": 4.235, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.416539512553698e-05, | |
| "loss": 4.2429, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.415700917802646e-05, | |
| "loss": 4.2359, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.414862323051594e-05, | |
| "loss": 4.2338, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.414023728300542e-05, | |
| "loss": 4.2496, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4131884093102365e-05, | |
| "loss": 4.2317, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4123498145591845e-05, | |
| "loss": 4.2502, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.411511219808132e-05, | |
| "loss": 4.2271, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4106726250570805e-05, | |
| "loss": 4.2229, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4098340303060284e-05, | |
| "loss": 4.2233, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4089954355549764e-05, | |
| "loss": 4.238, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4081584786842974e-05, | |
| "loss": 4.2358, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4073198839332453e-05, | |
| "loss": 4.2216, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4064812891821933e-05, | |
| "loss": 4.2157, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4056426944311413e-05, | |
| "loss": 4.2255, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.404804099680089e-05, | |
| "loss": 4.2439, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403965504929037e-05, | |
| "loss": 4.2261, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403128548058358e-05, | |
| "loss": 4.2425, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.402289953307306e-05, | |
| "loss": 4.2318, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.401451358556254e-05, | |
| "loss": 4.2369, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.400612763805202e-05, | |
| "loss": 4.235, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.39977416905415e-05, | |
| "loss": 4.2242, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.398935574303098e-05, | |
| "loss": 4.2241, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.398096979552047e-05, | |
| "loss": 4.2314, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.397258384800995e-05, | |
| "loss": 4.2243, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.396421427930316e-05, | |
| "loss": 4.2272, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.395582833179264e-05, | |
| "loss": 4.2293, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.394744238428212e-05, | |
| "loss": 4.2298, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.39390564367716e-05, | |
| "loss": 4.2211, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.393068686806481e-05, | |
| "loss": 4.2202, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.392230092055429e-05, | |
| "loss": 4.2358, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.391391497304377e-05, | |
| "loss": 4.2203, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.390552902553325e-05, | |
| "loss": 4.2362, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3897159456826456e-05, | |
| "loss": 4.2296, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3888773509315936e-05, | |
| "loss": 4.2281, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.388038756180542e-05, | |
| "loss": 4.2267, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.38720016142949e-05, | |
| "loss": 4.2238, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.386363204558811e-05, | |
| "loss": 4.228, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.385524609807759e-05, | |
| "loss": 4.2227, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.384686015056707e-05, | |
| "loss": 4.2375, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.383849058186028e-05, | |
| "loss": 4.2304, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.383010463434976e-05, | |
| "loss": 4.2324, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.382171868683924e-05, | |
| "loss": 4.2303, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.381333273932872e-05, | |
| "loss": 4.2213, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.38049467918182e-05, | |
| "loss": 4.2243, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.379656084430768e-05, | |
| "loss": 4.2164, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.378817489679716e-05, | |
| "loss": 4.2196, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3779805328090376e-05, | |
| "loss": 4.2229, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3771419380579856e-05, | |
| "loss": 4.2184, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3763033433069336e-05, | |
| "loss": 4.21, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.375464748555881e-05, | |
| "loss": 4.2226, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.208243370056152, | |
| "eval_runtime": 539.7013, | |
| "eval_samples_per_second": 707.041, | |
| "eval_steps_per_second": 22.096, | |
| "step": 381595 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.374626153804829e-05, | |
| "loss": 4.228, | |
| "step": 381952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.373787559053777e-05, | |
| "loss": 4.2249, | |
| "step": 382464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372948964302725e-05, | |
| "loss": 4.2225, | |
| "step": 382976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372110369551673e-05, | |
| "loss": 4.2185, | |
| "step": 383488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.371273412680994e-05, | |
| "loss": 4.2164, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.370434817929942e-05, | |
| "loss": 4.2056, | |
| "step": 384512 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.36959622317889e-05, | |
| "loss": 4.2187, | |
| "step": 385024 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.368757628427838e-05, | |
| "loss": 4.2112, | |
| "step": 385536 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367919033676786e-05, | |
| "loss": 4.2165, | |
| "step": 386048 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3670820768061074e-05, | |
| "loss": 4.2206, | |
| "step": 386560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3662434820550554e-05, | |
| "loss": 4.2248, | |
| "step": 387072 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3654048873040034e-05, | |
| "loss": 4.2093, | |
| "step": 387584 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3645662925529514e-05, | |
| "loss": 4.2157, | |
| "step": 388096 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3637276978018994e-05, | |
| "loss": 4.2111, | |
| "step": 388608 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.36289074093122e-05, | |
| "loss": 4.2183, | |
| "step": 389120 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.362052146180168e-05, | |
| "loss": 4.2072, | |
| "step": 389632 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.361213551429116e-05, | |
| "loss": 4.2073, | |
| "step": 390144 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.360374956678064e-05, | |
| "loss": 4.2089, | |
| "step": 390656 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.359536361927012e-05, | |
| "loss": 4.2143, | |
| "step": 391168 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.35869776717596e-05, | |
| "loss": 4.2217, | |
| "step": 391680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357859172424908e-05, | |
| "loss": 4.2124, | |
| "step": 392192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357020577673856e-05, | |
| "loss": 4.2237, | |
| "step": 392704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.356183620803178e-05, | |
| "loss": 4.2168, | |
| "step": 393216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.355345026052126e-05, | |
| "loss": 4.2157, | |
| "step": 393728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.354506431301074e-05, | |
| "loss": 4.2085, | |
| "step": 394240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.353667836550022e-05, | |
| "loss": 4.2089, | |
| "step": 394752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.352830879679343e-05, | |
| "loss": 4.2204, | |
| "step": 395264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.351992284928291e-05, | |
| "loss": 4.2045, | |
| "step": 395776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.351153690177239e-05, | |
| "loss": 4.2128, | |
| "step": 396288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.350315095426187e-05, | |
| "loss": 4.2054, | |
| "step": 396800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3494781385555076e-05, | |
| "loss": 4.1978, | |
| "step": 397312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3486411816848285e-05, | |
| "loss": 4.2202, | |
| "step": 397824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3478025869337765e-05, | |
| "loss": 4.215, | |
| "step": 398336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.346963992182725e-05, | |
| "loss": 4.2127, | |
| "step": 398848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.346125397431673e-05, | |
| "loss": 4.2003, | |
| "step": 399360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.345286802680621e-05, | |
| "loss": 4.2157, | |
| "step": 399872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.344449845809942e-05, | |
| "loss": 4.2055, | |
| "step": 400384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.343612888939263e-05, | |
| "loss": 4.213, | |
| "step": 400896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.342774294188211e-05, | |
| "loss": 4.2074, | |
| "step": 401408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.341935699437159e-05, | |
| "loss": 4.1952, | |
| "step": 401920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.341097104686107e-05, | |
| "loss": 4.1956, | |
| "step": 402432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.340258509935055e-05, | |
| "loss": 4.1914, | |
| "step": 402944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.339419915184003e-05, | |
| "loss": 4.2113, | |
| "step": 403456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.338581320432951e-05, | |
| "loss": 4.2007, | |
| "step": 403968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.337742725681899e-05, | |
| "loss": 4.1962, | |
| "step": 404480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.336904130930847e-05, | |
| "loss": 4.2038, | |
| "step": 404992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3360671740601686e-05, | |
| "loss": 4.1994, | |
| "step": 405504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3352285793091166e-05, | |
| "loss": 4.2145, | |
| "step": 406016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.334389984558064e-05, | |
| "loss": 4.1954, | |
| "step": 406528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.333551389807012e-05, | |
| "loss": 4.1971, | |
| "step": 407040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.33271279505596e-05, | |
| "loss": 4.2042, | |
| "step": 407552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331874200304908e-05, | |
| "loss": 4.2038, | |
| "step": 408064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331035605553856e-05, | |
| "loss": 4.1928, | |
| "step": 408576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.330198648683177e-05, | |
| "loss": 4.1987, | |
| "step": 409088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.329360053932125e-05, | |
| "loss": 4.1947, | |
| "step": 409600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.328521459181073e-05, | |
| "loss": 4.1995, | |
| "step": 410112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.327682864430021e-05, | |
| "loss": 4.1989, | |
| "step": 410624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326844269678969e-05, | |
| "loss": 4.1926, | |
| "step": 411136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326005674927917e-05, | |
| "loss": 4.2054, | |
| "step": 411648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3251670801768654e-05, | |
| "loss": 4.1886, | |
| "step": 412160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3243284854258134e-05, | |
| "loss": 4.1866, | |
| "step": 412672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.323491528555134e-05, | |
| "loss": 4.1904, | |
| "step": 413184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.322652933804082e-05, | |
| "loss": 4.1896, | |
| "step": 413696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.321815976933403e-05, | |
| "loss": 4.2066, | |
| "step": 414208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320977382182351e-05, | |
| "loss": 4.1871, | |
| "step": 414720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320138787431299e-05, | |
| "loss": 4.1966, | |
| "step": 415232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.319300192680247e-05, | |
| "loss": 4.1755, | |
| "step": 415744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.318463235809568e-05, | |
| "loss": 4.1966, | |
| "step": 416256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.317624641058516e-05, | |
| "loss": 4.1843, | |
| "step": 416768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.316786046307464e-05, | |
| "loss": 4.1862, | |
| "step": 417280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315947451556412e-05, | |
| "loss": 4.1988, | |
| "step": 417792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315108856805361e-05, | |
| "loss": 4.1894, | |
| "step": 418304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.314270262054309e-05, | |
| "loss": 4.1947, | |
| "step": 418816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.313431667303257e-05, | |
| "loss": 4.2015, | |
| "step": 419328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.312593072552205e-05, | |
| "loss": 4.196, | |
| "step": 419840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.311756115681526e-05, | |
| "loss": 4.2038, | |
| "step": 420352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310917520930474e-05, | |
| "loss": 4.1884, | |
| "step": 420864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310078926179422e-05, | |
| "loss": 4.1801, | |
| "step": 421376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.30924033142837e-05, | |
| "loss": 4.1904, | |
| "step": 421888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.308401736677318e-05, | |
| "loss": 4.2001, | |
| "step": 422400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3075647798066386e-05, | |
| "loss": 4.1879, | |
| "step": 422912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3067261850555866e-05, | |
| "loss": 4.1871, | |
| "step": 423424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3058875903045346e-05, | |
| "loss": 4.1727, | |
| "step": 423936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3050489955534826e-05, | |
| "loss": 4.186, | |
| "step": 424448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.304212038682804e-05, | |
| "loss": 4.1776, | |
| "step": 424960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.303373443931752e-05, | |
| "loss": 4.1908, | |
| "step": 425472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3025348491807e-05, | |
| "loss": 4.1916, | |
| "step": 425984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.301697892310021e-05, | |
| "loss": 4.1959, | |
| "step": 426496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.300859297558969e-05, | |
| "loss": 4.1848, | |
| "step": 427008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.300020702807917e-05, | |
| "loss": 4.1884, | |
| "step": 427520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.299182108056865e-05, | |
| "loss": 4.1729, | |
| "step": 428032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2983435133058124e-05, | |
| "loss": 4.1807, | |
| "step": 428544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2975049185547604e-05, | |
| "loss": 4.1844, | |
| "step": 429056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2966663238037084e-05, | |
| "loss": 4.1791, | |
| "step": 429568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.29582936693303e-05, | |
| "loss": 4.1754, | |
| "step": 430080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294990772181978e-05, | |
| "loss": 4.1926, | |
| "step": 430592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294152177430926e-05, | |
| "loss": 4.1789, | |
| "step": 431104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.293313582679874e-05, | |
| "loss": 4.2014, | |
| "step": 431616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.292474987928822e-05, | |
| "loss": 4.1746, | |
| "step": 432128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.29163639317777e-05, | |
| "loss": 4.1929, | |
| "step": 432640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.290797798426718e-05, | |
| "loss": 4.1817, | |
| "step": 433152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289959203675666e-05, | |
| "loss": 4.1801, | |
| "step": 433664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289122246804987e-05, | |
| "loss": 4.1989, | |
| "step": 434176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.288283652053935e-05, | |
| "loss": 4.1751, | |
| "step": 434688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.287445057302883e-05, | |
| "loss": 4.1956, | |
| "step": 435200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.286606462551831e-05, | |
| "loss": 4.173, | |
| "step": 435712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.285769505681152e-05, | |
| "loss": 4.1707, | |
| "step": 436224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284932548810473e-05, | |
| "loss": 4.1698, | |
| "step": 436736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284093954059421e-05, | |
| "loss": 4.1827, | |
| "step": 437248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.283255359308369e-05, | |
| "loss": 4.1827, | |
| "step": 437760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.282416764557317e-05, | |
| "loss": 4.1728, | |
| "step": 438272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.281578169806265e-05, | |
| "loss": 4.16, | |
| "step": 438784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.280739575055213e-05, | |
| "loss": 4.1725, | |
| "step": 439296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.279902618184534e-05, | |
| "loss": 4.1866, | |
| "step": 439808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.279064023433482e-05, | |
| "loss": 4.1807, | |
| "step": 440320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.27822542868243e-05, | |
| "loss": 4.1848, | |
| "step": 440832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.277386833931378e-05, | |
| "loss": 4.1818, | |
| "step": 441344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.276548239180326e-05, | |
| "loss": 4.1811, | |
| "step": 441856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.275709644429274e-05, | |
| "loss": 4.1817, | |
| "step": 442368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.274871049678222e-05, | |
| "loss": 4.1745, | |
| "step": 442880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.27403245492717e-05, | |
| "loss": 4.1706, | |
| "step": 443392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.273195498056492e-05, | |
| "loss": 4.1788, | |
| "step": 443904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.27235690330544e-05, | |
| "loss": 4.172, | |
| "step": 444416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.271518308554388e-05, | |
| "loss": 4.1759, | |
| "step": 444928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.270679713803336e-05, | |
| "loss": 4.1775, | |
| "step": 445440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.269841119052284e-05, | |
| "loss": 4.1771, | |
| "step": 445952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2690041621816046e-05, | |
| "loss": 4.1734, | |
| "step": 446464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2681655674305526e-05, | |
| "loss": 4.1683, | |
| "step": 446976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2673269726795006e-05, | |
| "loss": 4.1839, | |
| "step": 447488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2664883779284486e-05, | |
| "loss": 4.1707, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2656514210577695e-05, | |
| "loss": 4.1821, | |
| "step": 448512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2648128263067175e-05, | |
| "loss": 4.1843, | |
| "step": 449024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2639742315556655e-05, | |
| "loss": 4.1759, | |
| "step": 449536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2631356368046135e-05, | |
| "loss": 4.1788, | |
| "step": 450048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.262298679933935e-05, | |
| "loss": 4.1732, | |
| "step": 450560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.261460085182883e-05, | |
| "loss": 4.1744, | |
| "step": 451072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.260621490431831e-05, | |
| "loss": 4.1676, | |
| "step": 451584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.259784533561152e-05, | |
| "loss": 4.1873, | |
| "step": 452096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2589459388101e-05, | |
| "loss": 4.1825, | |
| "step": 452608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.258107344059048e-05, | |
| "loss": 4.1783, | |
| "step": 453120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.257268749307996e-05, | |
| "loss": 4.1766, | |
| "step": 453632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.256430154556943e-05, | |
| "loss": 4.1733, | |
| "step": 454144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.255593197686265e-05, | |
| "loss": 4.1771, | |
| "step": 454656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.254754602935213e-05, | |
| "loss": 4.1689, | |
| "step": 455168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253916008184161e-05, | |
| "loss": 4.1644, | |
| "step": 455680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253077413433109e-05, | |
| "loss": 4.1752, | |
| "step": 456192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.252238818682057e-05, | |
| "loss": 4.1672, | |
| "step": 456704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.251400223931005e-05, | |
| "loss": 4.1592, | |
| "step": 457216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.250561629179953e-05, | |
| "loss": 4.1691, | |
| "step": 457728 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.166858196258545, | |
| "eval_runtime": 544.061, | |
| "eval_samples_per_second": 701.375, | |
| "eval_steps_per_second": 21.918, | |
| "step": 457914 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.249723034428901e-05, | |
| "loss": 4.1846, | |
| "step": 458240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.248884439677849e-05, | |
| "loss": 4.1739, | |
| "step": 458752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.248045844926797e-05, | |
| "loss": 4.1704, | |
| "step": 459264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.247207250175745e-05, | |
| "loss": 4.1694, | |
| "step": 459776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.246368655424693e-05, | |
| "loss": 4.1694, | |
| "step": 460288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.245530060673641e-05, | |
| "loss": 4.1586, | |
| "step": 460800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.244691465922589e-05, | |
| "loss": 4.1691, | |
| "step": 461312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243852871171537e-05, | |
| "loss": 4.1615, | |
| "step": 461824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243014276420485e-05, | |
| "loss": 4.1649, | |
| "step": 462336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.242175681669432e-05, | |
| "loss": 4.1709, | |
| "step": 462848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.241337086918381e-05, | |
| "loss": 4.1786, | |
| "step": 463360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.240498492167329e-05, | |
| "loss": 4.1558, | |
| "step": 463872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.239659897416277e-05, | |
| "loss": 4.1727, | |
| "step": 464384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.238821302665225e-05, | |
| "loss": 4.161, | |
| "step": 464896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.237982707914173e-05, | |
| "loss": 4.1724, | |
| "step": 465408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.237144113163121e-05, | |
| "loss": 4.1556, | |
| "step": 465920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.236305518412069e-05, | |
| "loss": 4.1575, | |
| "step": 466432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.235466923661017e-05, | |
| "loss": 4.1624, | |
| "step": 466944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.234628328909965e-05, | |
| "loss": 4.1621, | |
| "step": 467456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.233789734158913e-05, | |
| "loss": 4.1716, | |
| "step": 467968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.232952777288234e-05, | |
| "loss": 4.1667, | |
| "step": 468480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.232114182537182e-05, | |
| "loss": 4.1745, | |
| "step": 468992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2312772256665026e-05, | |
| "loss": 4.1709, | |
| "step": 469504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2304386309154506e-05, | |
| "loss": 4.1658, | |
| "step": 470016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2296000361643986e-05, | |
| "loss": 4.1597, | |
| "step": 470528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.228761441413347e-05, | |
| "loss": 4.1606, | |
| "step": 471040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.227922846662295e-05, | |
| "loss": 4.1708, | |
| "step": 471552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.227084251911243e-05, | |
| "loss": 4.1587, | |
| "step": 472064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.226245657160191e-05, | |
| "loss": 4.1623, | |
| "step": 472576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2254070624091386e-05, | |
| "loss": 4.1606, | |
| "step": 473088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.22457010553846e-05, | |
| "loss": 4.151, | |
| "step": 473600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.223731510787408e-05, | |
| "loss": 4.1681, | |
| "step": 474112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.222892916036356e-05, | |
| "loss": 4.169, | |
| "step": 474624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2220543212853035e-05, | |
| "loss": 4.1655, | |
| "step": 475136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.221217364414625e-05, | |
| "loss": 4.1531, | |
| "step": 475648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.220378769663573e-05, | |
| "loss": 4.1669, | |
| "step": 476160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.219541812792894e-05, | |
| "loss": 4.1599, | |
| "step": 476672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2187032180418426e-05, | |
| "loss": 4.1654, | |
| "step": 477184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2178646232907906e-05, | |
| "loss": 4.1621, | |
| "step": 477696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2170260285397386e-05, | |
| "loss": 4.1507, | |
| "step": 478208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.216187433788686e-05, | |
| "loss": 4.1455, | |
| "step": 478720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2153504769180075e-05, | |
| "loss": 4.1445, | |
| "step": 479232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2145118821669555e-05, | |
| "loss": 4.1688, | |
| "step": 479744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2136732874159035e-05, | |
| "loss": 4.1534, | |
| "step": 480256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.212834692664851e-05, | |
| "loss": 4.1495, | |
| "step": 480768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2119977357941724e-05, | |
| "loss": 4.16, | |
| "step": 481280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2111591410431204e-05, | |
| "loss": 4.1531, | |
| "step": 481792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.210320546292068e-05, | |
| "loss": 4.1649, | |
| "step": 482304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2094819515410164e-05, | |
| "loss": 4.1505, | |
| "step": 482816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2086433567899644e-05, | |
| "loss": 4.1533, | |
| "step": 483328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2078047620389124e-05, | |
| "loss": 4.1522, | |
| "step": 483840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2069661672878604e-05, | |
| "loss": 4.1608, | |
| "step": 484352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2061275725368084e-05, | |
| "loss": 4.1491, | |
| "step": 484864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.205290615666129e-05, | |
| "loss": 4.1478, | |
| "step": 485376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.204452020915077e-05, | |
| "loss": 4.1505, | |
| "step": 485888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.203613426164025e-05, | |
| "loss": 4.1572, | |
| "step": 486400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.202774831412973e-05, | |
| "loss": 4.1508, | |
| "step": 486912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201937874542294e-05, | |
| "loss": 4.1459, | |
| "step": 487424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201099279791242e-05, | |
| "loss": 4.1595, | |
| "step": 487936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.20026068504019e-05, | |
| "loss": 4.1447, | |
| "step": 488448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.199422090289138e-05, | |
| "loss": 4.1414, | |
| "step": 488960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.198583495538086e-05, | |
| "loss": 4.1452, | |
| "step": 489472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.197744900787035e-05, | |
| "loss": 4.1455, | |
| "step": 489984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.196906306035983e-05, | |
| "loss": 4.1621, | |
| "step": 490496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.196067711284931e-05, | |
| "loss": 4.1431, | |
| "step": 491008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.195230754414252e-05, | |
| "loss": 4.15, | |
| "step": 491520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1943921596632e-05, | |
| "loss": 4.1329, | |
| "step": 492032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.193553564912148e-05, | |
| "loss": 4.1494, | |
| "step": 492544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.192714970161096e-05, | |
| "loss": 4.1397, | |
| "step": 493056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1918780132904167e-05, | |
| "loss": 4.1425, | |
| "step": 493568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1910394185393647e-05, | |
| "loss": 4.1519, | |
| "step": 494080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1902008237883126e-05, | |
| "loss": 4.1458, | |
| "step": 494592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1893622290372606e-05, | |
| "loss": 4.1517, | |
| "step": 495104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1885252721665816e-05, | |
| "loss": 4.1578, | |
| "step": 495616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.18768667741553e-05, | |
| "loss": 4.152, | |
| "step": 496128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.186848082664478e-05, | |
| "loss": 4.1572, | |
| "step": 496640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.186009487913426e-05, | |
| "loss": 4.1502, | |
| "step": 497152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.185172531042747e-05, | |
| "loss": 4.1348, | |
| "step": 497664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.184335574172068e-05, | |
| "loss": 4.1455, | |
| "step": 498176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.183496979421016e-05, | |
| "loss": 4.152, | |
| "step": 498688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.182658384669964e-05, | |
| "loss": 4.1491, | |
| "step": 499200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.181819789918912e-05, | |
| "loss": 4.1461, | |
| "step": 499712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.18098119516786e-05, | |
| "loss": 4.1255, | |
| "step": 500224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.180142600416808e-05, | |
| "loss": 4.1447, | |
| "step": 500736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.179304005665756e-05, | |
| "loss": 4.1334, | |
| "step": 501248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.178467048795077e-05, | |
| "loss": 4.1484, | |
| "step": 501760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.177628454044025e-05, | |
| "loss": 4.1452, | |
| "step": 502272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1767898592929736e-05, | |
| "loss": 4.1539, | |
| "step": 502784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1759512645419216e-05, | |
| "loss": 4.1415, | |
| "step": 503296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1751126697908696e-05, | |
| "loss": 4.144, | |
| "step": 503808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.174274075039817e-05, | |
| "loss": 4.1334, | |
| "step": 504320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.173435480288765e-05, | |
| "loss": 4.1432, | |
| "step": 504832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.172596885537713e-05, | |
| "loss": 4.1374, | |
| "step": 505344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.171759928667034e-05, | |
| "loss": 4.1384, | |
| "step": 505856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.170921333915982e-05, | |
| "loss": 4.1311, | |
| "step": 506368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.17008273916493e-05, | |
| "loss": 4.1461, | |
| "step": 506880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.169244144413878e-05, | |
| "loss": 4.1362, | |
| "step": 507392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.168407187543199e-05, | |
| "loss": 4.1612, | |
| "step": 507904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1675685927921474e-05, | |
| "loss": 4.1343, | |
| "step": 508416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1667299980410954e-05, | |
| "loss": 4.1512, | |
| "step": 508928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1658914032900434e-05, | |
| "loss": 4.1398, | |
| "step": 509440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1650528085389914e-05, | |
| "loss": 4.1373, | |
| "step": 509952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1642142137879393e-05, | |
| "loss": 4.1536, | |
| "step": 510464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1633756190368873e-05, | |
| "loss": 4.1368, | |
| "step": 510976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1625370242858353e-05, | |
| "loss": 4.1508, | |
| "step": 511488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.161700067415156e-05, | |
| "loss": 4.1357, | |
| "step": 512000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160861472664104e-05, | |
| "loss": 4.1256, | |
| "step": 512512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160024515793425e-05, | |
| "loss": 4.1287, | |
| "step": 513024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.159185921042373e-05, | |
| "loss": 4.1408, | |
| "step": 513536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.158347326291321e-05, | |
| "loss": 4.1409, | |
| "step": 514048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.157508731540269e-05, | |
| "loss": 4.1352, | |
| "step": 514560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.156670136789217e-05, | |
| "loss": 4.1136, | |
| "step": 515072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.155833179918539e-05, | |
| "loss": 4.1319, | |
| "step": 515584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.154994585167487e-05, | |
| "loss": 4.1469, | |
| "step": 516096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.154155990416435e-05, | |
| "loss": 4.138, | |
| "step": 516608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.153317395665383e-05, | |
| "loss": 4.1443, | |
| "step": 517120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.152478800914331e-05, | |
| "loss": 4.1397, | |
| "step": 517632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.151640206163279e-05, | |
| "loss": 4.1409, | |
| "step": 518144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.150801611412227e-05, | |
| "loss": 4.1404, | |
| "step": 518656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.149963016661175e-05, | |
| "loss": 4.1358, | |
| "step": 519168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1491260597904956e-05, | |
| "loss": 4.1282, | |
| "step": 519680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1482874650394436e-05, | |
| "loss": 4.1365, | |
| "step": 520192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1474488702883916e-05, | |
| "loss": 4.1326, | |
| "step": 520704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1466102755373396e-05, | |
| "loss": 4.1352, | |
| "step": 521216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.145773318666661e-05, | |
| "loss": 4.1363, | |
| "step": 521728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.144934723915609e-05, | |
| "loss": 4.1369, | |
| "step": 522240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.144096129164557e-05, | |
| "loss": 4.1307, | |
| "step": 522752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.143257534413505e-05, | |
| "loss": 4.127, | |
| "step": 523264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.142418939662453e-05, | |
| "loss": 4.1418, | |
| "step": 523776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.141583620672147e-05, | |
| "loss": 4.128, | |
| "step": 524288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.140745025921095e-05, | |
| "loss": 4.1433, | |
| "step": 524800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.139906431170043e-05, | |
| "loss": 4.1446, | |
| "step": 525312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.139067836418991e-05, | |
| "loss": 4.1356, | |
| "step": 525824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.138229241667939e-05, | |
| "loss": 4.1392, | |
| "step": 526336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.137390646916887e-05, | |
| "loss": 4.1351, | |
| "step": 526848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.136552052165835e-05, | |
| "loss": 4.1333, | |
| "step": 527360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.135713457414783e-05, | |
| "loss": 4.1282, | |
| "step": 527872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1348765005441045e-05, | |
| "loss": 4.1455, | |
| "step": 528384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1340379057930525e-05, | |
| "loss": 4.1428, | |
| "step": 528896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1331993110420005e-05, | |
| "loss": 4.1398, | |
| "step": 529408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.132360716290948e-05, | |
| "loss": 4.1379, | |
| "step": 529920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1315237594202694e-05, | |
| "loss": 4.1309, | |
| "step": 530432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1306851646692174e-05, | |
| "loss": 4.1342, | |
| "step": 530944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1298482077985383e-05, | |
| "loss": 4.1293, | |
| "step": 531456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.129009613047486e-05, | |
| "loss": 4.1221, | |
| "step": 531968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.128171018296434e-05, | |
| "loss": 4.1384, | |
| "step": 532480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.127332423545382e-05, | |
| "loss": 4.1275, | |
| "step": 532992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.12649382879433e-05, | |
| "loss": 4.1203, | |
| "step": 533504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.125655234043278e-05, | |
| "loss": 4.1333, | |
| "step": 534016 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.136108875274658, | |
| "eval_runtime": 553.0927, | |
| "eval_samples_per_second": 689.922, | |
| "eval_steps_per_second": 21.561, | |
| "step": 534233 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1248182771726e-05, | |
| "loss": 4.1349, | |
| "step": 534528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.123979682421548e-05, | |
| "loss": 4.1327, | |
| "step": 535040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.123141087670495e-05, | |
| "loss": 4.1384, | |
| "step": 535552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.122302492919443e-05, | |
| "loss": 4.1294, | |
| "step": 536064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.121465536048765e-05, | |
| "loss": 4.1278, | |
| "step": 536576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.120626941297712e-05, | |
| "loss": 4.125, | |
| "step": 537088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.11978834654666e-05, | |
| "loss": 4.1284, | |
| "step": 537600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.118949751795608e-05, | |
| "loss": 4.1247, | |
| "step": 538112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.118111157044556e-05, | |
| "loss": 4.1257, | |
| "step": 538624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.117274200173877e-05, | |
| "loss": 4.1272, | |
| "step": 539136 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1164372433031986e-05, | |
| "loss": 4.1409, | |
| "step": 539648 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.115598648552147e-05, | |
| "loss": 4.1206, | |
| "step": 540160 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.114760053801095e-05, | |
| "loss": 4.1327, | |
| "step": 540672 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1139214590500426e-05, | |
| "loss": 4.1169, | |
| "step": 541184 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1130828642989906e-05, | |
| "loss": 4.136, | |
| "step": 541696 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1122442695479386e-05, | |
| "loss": 4.1169, | |
| "step": 542208 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1114056747968866e-05, | |
| "loss": 4.1207, | |
| "step": 542720 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1105670800458346e-05, | |
| "loss": 4.1235, | |
| "step": 543232 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1097284852947826e-05, | |
| "loss": 4.1233, | |
| "step": 543744 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1088898905437306e-05, | |
| "loss": 4.1333, | |
| "step": 544256 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1080512957926786e-05, | |
| "loss": 4.1285, | |
| "step": 544768 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1072143389219995e-05, | |
| "loss": 4.14, | |
| "step": 545280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1063757441709475e-05, | |
| "loss": 4.1285, | |
| "step": 545792 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1055371494198955e-05, | |
| "loss": 4.1276, | |
| "step": 546304 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1046985546688435e-05, | |
| "loss": 4.1252, | |
| "step": 546816 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.103859959917792e-05, | |
| "loss": 4.1235, | |
| "step": 547328 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.10302136516674e-05, | |
| "loss": 4.1294, | |
| "step": 547840 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.102182770415688e-05, | |
| "loss": 4.1226, | |
| "step": 548352 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.101344175664636e-05, | |
| "loss": 4.127, | |
| "step": 548864 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.100507218793957e-05, | |
| "loss": 4.1227, | |
| "step": 549376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.099668624042905e-05, | |
| "loss": 4.1147, | |
| "step": 549888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.098830029291853e-05, | |
| "loss": 4.1281, | |
| "step": 550400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.097991434540801e-05, | |
| "loss": 4.1319, | |
| "step": 550912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.097154477670122e-05, | |
| "loss": 4.1269, | |
| "step": 551424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.09631588291907e-05, | |
| "loss": 4.1164, | |
| "step": 551936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.095477288168018e-05, | |
| "loss": 4.1291, | |
| "step": 552448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.094640331297339e-05, | |
| "loss": 4.1232, | |
| "step": 552960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0938017365462875e-05, | |
| "loss": 4.1258, | |
| "step": 553472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0929631417952355e-05, | |
| "loss": 4.1248, | |
| "step": 553984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0921245470441835e-05, | |
| "loss": 4.1173, | |
| "step": 554496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.091285952293131e-05, | |
| "loss": 4.1104, | |
| "step": 555008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.090447357542079e-05, | |
| "loss": 4.1091, | |
| "step": 555520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.089608762791027e-05, | |
| "loss": 4.1276, | |
| "step": 556032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.088770168039975e-05, | |
| "loss": 4.1187, | |
| "step": 556544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.087931573288923e-05, | |
| "loss": 4.1121, | |
| "step": 557056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.087094616418244e-05, | |
| "loss": 4.124, | |
| "step": 557568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.086256021667192e-05, | |
| "loss": 4.1143, | |
| "step": 558080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.08541742691614e-05, | |
| "loss": 4.1282, | |
| "step": 558592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.084578832165088e-05, | |
| "loss": 4.113, | |
| "step": 559104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.083741875294409e-05, | |
| "loss": 4.1201, | |
| "step": 559616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.082903280543357e-05, | |
| "loss": 4.115, | |
| "step": 560128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.082064685792305e-05, | |
| "loss": 4.123, | |
| "step": 560640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.081226091041253e-05, | |
| "loss": 4.1137, | |
| "step": 561152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.080389134170574e-05, | |
| "loss": 4.1092, | |
| "step": 561664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.079550539419522e-05, | |
| "loss": 4.1171, | |
| "step": 562176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.07871194466847e-05, | |
| "loss": 4.1169, | |
| "step": 562688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.077873349917418e-05, | |
| "loss": 4.1153, | |
| "step": 563200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.077036393046739e-05, | |
| "loss": 4.1124, | |
| "step": 563712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.076197798295687e-05, | |
| "loss": 4.1244, | |
| "step": 564224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.075359203544635e-05, | |
| "loss": 4.1018, | |
| "step": 564736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.074520608793583e-05, | |
| "loss": 4.1107, | |
| "step": 565248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0736836519229046e-05, | |
| "loss": 4.1065, | |
| "step": 565760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0728450571718526e-05, | |
| "loss": 4.1129, | |
| "step": 566272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0720064624208006e-05, | |
| "loss": 4.1244, | |
| "step": 566784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0711678676697486e-05, | |
| "loss": 4.1073, | |
| "step": 567296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0703309107990695e-05, | |
| "loss": 4.1138, | |
| "step": 567808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0694923160480175e-05, | |
| "loss": 4.0993, | |
| "step": 568320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0686537212969655e-05, | |
| "loss": 4.1149, | |
| "step": 568832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0678151265459135e-05, | |
| "loss": 4.1039, | |
| "step": 569344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0669781696752344e-05, | |
| "loss": 4.1116, | |
| "step": 569856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0661395749241824e-05, | |
| "loss": 4.1157, | |
| "step": 570368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0653009801731304e-05, | |
| "loss": 4.1093, | |
| "step": 570880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0644623854220784e-05, | |
| "loss": 4.1124, | |
| "step": 571392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0636254285514e-05, | |
| "loss": 4.123, | |
| "step": 571904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.062788471680721e-05, | |
| "loss": 4.1149, | |
| "step": 572416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.061949876929669e-05, | |
| "loss": 4.1203, | |
| "step": 572928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.061111282178617e-05, | |
| "loss": 4.1145, | |
| "step": 573440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.060272687427565e-05, | |
| "loss": 4.1008, | |
| "step": 573952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.059434092676513e-05, | |
| "loss": 4.1097, | |
| "step": 574464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.058595497925461e-05, | |
| "loss": 4.1163, | |
| "step": 574976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.057756903174409e-05, | |
| "loss": 4.117, | |
| "step": 575488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.056918308423357e-05, | |
| "loss": 4.1064, | |
| "step": 576000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.056079713672305e-05, | |
| "loss": 4.0955, | |
| "step": 576512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.055242756801626e-05, | |
| "loss": 4.1043, | |
| "step": 577024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.054404162050574e-05, | |
| "loss": 4.1008, | |
| "step": 577536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.053565567299522e-05, | |
| "loss": 4.1136, | |
| "step": 578048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0527286104288434e-05, | |
| "loss": 4.1093, | |
| "step": 578560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0518900156777914e-05, | |
| "loss": 4.1176, | |
| "step": 579072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0510514209267394e-05, | |
| "loss": 4.1109, | |
| "step": 579584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0502128261756874e-05, | |
| "loss": 4.1046, | |
| "step": 580096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0493742314246353e-05, | |
| "loss": 4.1004, | |
| "step": 580608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.048537274553956e-05, | |
| "loss": 4.1073, | |
| "step": 581120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.047700317683277e-05, | |
| "loss": 4.101, | |
| "step": 581632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.046861722932225e-05, | |
| "loss": 4.1048, | |
| "step": 582144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.046023128181173e-05, | |
| "loss": 4.0996, | |
| "step": 582656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.045184533430121e-05, | |
| "loss": 4.11, | |
| "step": 583168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.044345938679069e-05, | |
| "loss": 4.1037, | |
| "step": 583680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.043507343928017e-05, | |
| "loss": 4.1267, | |
| "step": 584192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.042668749176965e-05, | |
| "loss": 4.0977, | |
| "step": 584704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.041830154425914e-05, | |
| "loss": 4.1146, | |
| "step": 585216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.040993197555235e-05, | |
| "loss": 4.1094, | |
| "step": 585728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.040154602804183e-05, | |
| "loss": 4.1023, | |
| "step": 586240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.039316008053131e-05, | |
| "loss": 4.1129, | |
| "step": 586752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0384790511824516e-05, | |
| "loss": 4.1087, | |
| "step": 587264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0376404564313996e-05, | |
| "loss": 4.115, | |
| "step": 587776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0368018616803476e-05, | |
| "loss": 4.1029, | |
| "step": 588288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0359632669292956e-05, | |
| "loss": 4.0911, | |
| "step": 588800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0351246721782436e-05, | |
| "loss": 4.0962, | |
| "step": 589312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.034286077427191e-05, | |
| "loss": 4.107, | |
| "step": 589824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0334491205565125e-05, | |
| "loss": 4.1068, | |
| "step": 590336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0326105258054605e-05, | |
| "loss": 4.0974, | |
| "step": 590848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.031771931054409e-05, | |
| "loss": 4.0816, | |
| "step": 591360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0309333363033565e-05, | |
| "loss": 4.0985, | |
| "step": 591872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.030096379432678e-05, | |
| "loss": 4.113, | |
| "step": 592384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.029257784681626e-05, | |
| "loss": 4.1052, | |
| "step": 592896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.028419189930574e-05, | |
| "loss": 4.114, | |
| "step": 593408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0275805951795214e-05, | |
| "loss": 4.1063, | |
| "step": 593920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0267420004284694e-05, | |
| "loss": 4.1027, | |
| "step": 594432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0259034056774174e-05, | |
| "loss": 4.1078, | |
| "step": 594944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0250648109263654e-05, | |
| "loss": 4.103, | |
| "step": 595456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0242262161753134e-05, | |
| "loss": 4.0956, | |
| "step": 595968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0233876214242614e-05, | |
| "loss": 4.1013, | |
| "step": 596480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0225490266732094e-05, | |
| "loss": 4.0998, | |
| "step": 596992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0217104319221574e-05, | |
| "loss": 4.1005, | |
| "step": 597504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.020873475051479e-05, | |
| "loss": 4.0978, | |
| "step": 598016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.020034880300427e-05, | |
| "loss": 4.108, | |
| "step": 598528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.019196285549375e-05, | |
| "loss": 4.0988, | |
| "step": 599040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.018357690798323e-05, | |
| "loss": 4.0927, | |
| "step": 599552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.017520733927644e-05, | |
| "loss": 4.1114, | |
| "step": 600064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.016682139176592e-05, | |
| "loss": 4.0922, | |
| "step": 600576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.01584354442554e-05, | |
| "loss": 4.1088, | |
| "step": 601088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.015004949674488e-05, | |
| "loss": 4.1144, | |
| "step": 601600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.014166354923436e-05, | |
| "loss": 4.1048, | |
| "step": 602112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.013327760172384e-05, | |
| "loss": 4.1022, | |
| "step": 602624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.012490803301705e-05, | |
| "loss": 4.1003, | |
| "step": 603136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.011652208550653e-05, | |
| "loss": 4.1003, | |
| "step": 603648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0108136137996014e-05, | |
| "loss": 4.0982, | |
| "step": 604160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0099750190485494e-05, | |
| "loss": 4.1146, | |
| "step": 604672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0091364242974974e-05, | |
| "loss": 4.1079, | |
| "step": 605184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0082978295464454e-05, | |
| "loss": 4.109, | |
| "step": 605696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.007459234795393e-05, | |
| "loss": 4.1023, | |
| "step": 606208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.006620640044341e-05, | |
| "loss": 4.0992, | |
| "step": 606720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.005783683173662e-05, | |
| "loss": 4.104, | |
| "step": 607232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.00494508842261e-05, | |
| "loss": 4.0952, | |
| "step": 607744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0041064936715576e-05, | |
| "loss": 4.0951, | |
| "step": 608256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0032678989205056e-05, | |
| "loss": 4.1048, | |
| "step": 608768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0024293041694536e-05, | |
| "loss": 4.0977, | |
| "step": 609280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.001592347298775e-05, | |
| "loss": 4.0849, | |
| "step": 609792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.000755390428097e-05, | |
| "loss": 4.1034, | |
| "step": 610304 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.111530303955078, | |
| "eval_runtime": 550.7994, | |
| "eval_samples_per_second": 692.795, | |
| "eval_steps_per_second": 21.65, | |
| "step": 610552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.999916795677045e-05, | |
| "loss": 4.1077, | |
| "step": 610816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.999078200925993e-05, | |
| "loss": 4.1081, | |
| "step": 611328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.99823960617494e-05, | |
| "loss": 4.105, | |
| "step": 611840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.997401011423888e-05, | |
| "loss": 4.0971, | |
| "step": 612352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.996562416672836e-05, | |
| "loss": 4.0965, | |
| "step": 612864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.995723821921784e-05, | |
| "loss": 4.0931, | |
| "step": 613376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.994885227170732e-05, | |
| "loss": 4.0959, | |
| "step": 613888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.99404663241968e-05, | |
| "loss": 4.0906, | |
| "step": 614400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.993208037668628e-05, | |
| "loss": 4.0925, | |
| "step": 614912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.992369442917576e-05, | |
| "loss": 4.0989, | |
| "step": 615424 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.991530848166524e-05, | |
| "loss": 4.107, | |
| "step": 615936 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.990692253415472e-05, | |
| "loss": 4.0898, | |
| "step": 616448 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.98985365866442e-05, | |
| "loss": 4.102, | |
| "step": 616960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.989015063913368e-05, | |
| "loss": 4.0848, | |
| "step": 617472 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.988176469162316e-05, | |
| "loss": 4.1082, | |
| "step": 617984 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.987337874411264e-05, | |
| "loss": 4.0835, | |
| "step": 618496 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.986499279660212e-05, | |
| "loss": 4.0911, | |
| "step": 619008 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.98566068490916e-05, | |
| "loss": 4.0933, | |
| "step": 619520 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.984822090158108e-05, | |
| "loss": 4.094, | |
| "step": 620032 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.983983495407056e-05, | |
| "loss": 4.0996, | |
| "step": 620544 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.983146538536377e-05, | |
| "loss": 4.0949, | |
| "step": 621056 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.982307943785325e-05, | |
| "loss": 4.1102, | |
| "step": 621568 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.981469349034273e-05, | |
| "loss": 4.0984, | |
| "step": 622080 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.980630754283221e-05, | |
| "loss": 4.0986, | |
| "step": 622592 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.979793797412542e-05, | |
| "loss": 4.0909, | |
| "step": 623104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.97895520266149e-05, | |
| "loss": 4.0979, | |
| "step": 623616 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9781166079104385e-05, | |
| "loss": 4.0966, | |
| "step": 624128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9772780131593865e-05, | |
| "loss": 4.0918, | |
| "step": 624640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9764410562887074e-05, | |
| "loss": 4.0933, | |
| "step": 625152 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9756024615376554e-05, | |
| "loss": 4.0932, | |
| "step": 625664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.974765504666976e-05, | |
| "loss": 4.0813, | |
| "step": 626176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.973926909915924e-05, | |
| "loss": 4.0993, | |
| "step": 626688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.973088315164872e-05, | |
| "loss": 4.1011, | |
| "step": 627200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.97224972041382e-05, | |
| "loss": 4.0949, | |
| "step": 627712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.971411125662768e-05, | |
| "loss": 4.0853, | |
| "step": 628224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.970572530911716e-05, | |
| "loss": 4.1002, | |
| "step": 628736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.969735574041037e-05, | |
| "loss": 4.0909, | |
| "step": 629248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.968896979289985e-05, | |
| "loss": 4.093, | |
| "step": 629760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.968058384538934e-05, | |
| "loss": 4.0961, | |
| "step": 630272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.967219789787882e-05, | |
| "loss": 4.085, | |
| "step": 630784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.966382832917203e-05, | |
| "loss": 4.0837, | |
| "step": 631296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.965544238166151e-05, | |
| "loss": 4.0766, | |
| "step": 631808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.964705643415099e-05, | |
| "loss": 4.0996, | |
| "step": 632320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.963867048664047e-05, | |
| "loss": 4.0885, | |
| "step": 632832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.963028453912995e-05, | |
| "loss": 4.0815, | |
| "step": 633344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.962189859161943e-05, | |
| "loss": 4.0966, | |
| "step": 633856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9613529022912636e-05, | |
| "loss": 4.0791, | |
| "step": 634368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9605143075402116e-05, | |
| "loss": 4.1, | |
| "step": 634880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9596757127891596e-05, | |
| "loss": 4.0851, | |
| "step": 635392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9588371180381076e-05, | |
| "loss": 4.0908, | |
| "step": 635904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9579985232870556e-05, | |
| "loss": 4.0863, | |
| "step": 636416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9571599285360036e-05, | |
| "loss": 4.0882, | |
| "step": 636928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.956321333784952e-05, | |
| "loss": 4.0906, | |
| "step": 637440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9554827390339e-05, | |
| "loss": 4.0753, | |
| "step": 637952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.954645782163221e-05, | |
| "loss": 4.0888, | |
| "step": 638464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.953807187412169e-05, | |
| "loss": 4.0849, | |
| "step": 638976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.952968592661117e-05, | |
| "loss": 4.0858, | |
| "step": 639488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.952129997910065e-05, | |
| "loss": 4.0838, | |
| "step": 640000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.951293041039386e-05, | |
| "loss": 4.097, | |
| "step": 640512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.950454446288334e-05, | |
| "loss": 4.0698, | |
| "step": 641024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.949615851537282e-05, | |
| "loss": 4.0828, | |
| "step": 641536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9487772567862294e-05, | |
| "loss": 4.0794, | |
| "step": 642048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.947940299915551e-05, | |
| "loss": 4.0829, | |
| "step": 642560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.947101705164499e-05, | |
| "loss": 4.0913, | |
| "step": 643072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.946263110413447e-05, | |
| "loss": 4.0812, | |
| "step": 643584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.945424515662395e-05, | |
| "loss": 4.0895, | |
| "step": 644096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9445891966720895e-05, | |
| "loss": 4.0635, | |
| "step": 644608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9437506019210375e-05, | |
| "loss": 4.0877, | |
| "step": 645120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9429120071699855e-05, | |
| "loss": 4.0733, | |
| "step": 645632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9420734124189335e-05, | |
| "loss": 4.0798, | |
| "step": 646144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9412348176678815e-05, | |
| "loss": 4.0838, | |
| "step": 646656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9403962229168295e-05, | |
| "loss": 4.0843, | |
| "step": 647168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.939557628165777e-05, | |
| "loss": 4.0827, | |
| "step": 647680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.938719033414725e-05, | |
| "loss": 4.098, | |
| "step": 648192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9378820765440464e-05, | |
| "loss": 4.0833, | |
| "step": 648704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9370434817929943e-05, | |
| "loss": 4.0911, | |
| "step": 649216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9362048870419423e-05, | |
| "loss": 4.0884, | |
| "step": 649728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9353662922908903e-05, | |
| "loss": 4.0738, | |
| "step": 650240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.934527697539838e-05, | |
| "loss": 4.0751, | |
| "step": 650752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.933689102788786e-05, | |
| "loss": 4.0864, | |
| "step": 651264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.932850508037734e-05, | |
| "loss": 4.0919, | |
| "step": 651776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.932011913286682e-05, | |
| "loss": 4.0852, | |
| "step": 652288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.93117331853563e-05, | |
| "loss": 4.0626, | |
| "step": 652800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.930334723784578e-05, | |
| "loss": 4.0739, | |
| "step": 653312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.929499404794272e-05, | |
| "loss": 4.0754, | |
| "step": 653824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.92866081004322e-05, | |
| "loss": 4.0826, | |
| "step": 654336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.927822215292168e-05, | |
| "loss": 4.0834, | |
| "step": 654848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.926983620541116e-05, | |
| "loss": 4.0881, | |
| "step": 655360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.926145025790065e-05, | |
| "loss": 4.0796, | |
| "step": 655872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.925306431039013e-05, | |
| "loss": 4.0778, | |
| "step": 656384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.924467836287961e-05, | |
| "loss": 4.0758, | |
| "step": 656896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.923629241536909e-05, | |
| "loss": 4.0766, | |
| "step": 657408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9227939225466026e-05, | |
| "loss": 4.067, | |
| "step": 657920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9219553277955506e-05, | |
| "loss": 4.0821, | |
| "step": 658432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9211167330444986e-05, | |
| "loss": 4.0725, | |
| "step": 658944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9202781382934466e-05, | |
| "loss": 4.079, | |
| "step": 659456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9194395435423946e-05, | |
| "loss": 4.0753, | |
| "step": 659968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9186009487913426e-05, | |
| "loss": 4.095, | |
| "step": 660480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9177623540402906e-05, | |
| "loss": 4.0732, | |
| "step": 660992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9169237592892386e-05, | |
| "loss": 4.0869, | |
| "step": 661504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.91608680241856e-05, | |
| "loss": 4.0858, | |
| "step": 662016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.915249845547881e-05, | |
| "loss": 4.0766, | |
| "step": 662528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.914411250796829e-05, | |
| "loss": 4.0821, | |
| "step": 663040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.913572656045777e-05, | |
| "loss": 4.0819, | |
| "step": 663552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.912734061294725e-05, | |
| "loss": 4.0866, | |
| "step": 664064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.911895466543673e-05, | |
| "loss": 4.0739, | |
| "step": 664576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.911056871792621e-05, | |
| "loss": 4.0656, | |
| "step": 665088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.910218277041569e-05, | |
| "loss": 4.0694, | |
| "step": 665600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.909379682290517e-05, | |
| "loss": 4.0774, | |
| "step": 666112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.908542725419838e-05, | |
| "loss": 4.0763, | |
| "step": 666624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.907704130668786e-05, | |
| "loss": 4.0752, | |
| "step": 667136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.906865535917734e-05, | |
| "loss": 4.052, | |
| "step": 667648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.906026941166682e-05, | |
| "loss": 4.0745, | |
| "step": 668160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9051899842960035e-05, | |
| "loss": 4.0808, | |
| "step": 668672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9043513895449515e-05, | |
| "loss": 4.0771, | |
| "step": 669184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9035144326742724e-05, | |
| "loss": 4.0858, | |
| "step": 669696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9026758379232204e-05, | |
| "loss": 4.0814, | |
| "step": 670208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9018372431721684e-05, | |
| "loss": 4.0718, | |
| "step": 670720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9009986484211164e-05, | |
| "loss": 4.0806, | |
| "step": 671232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9001600536700644e-05, | |
| "loss": 4.0786, | |
| "step": 671744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.899323096799385e-05, | |
| "loss": 4.0665, | |
| "step": 672256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.898484502048333e-05, | |
| "loss": 4.0763, | |
| "step": 672768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.897645907297281e-05, | |
| "loss": 4.0734, | |
| "step": 673280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.896807312546229e-05, | |
| "loss": 4.0742, | |
| "step": 673792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.895968717795177e-05, | |
| "loss": 4.0712, | |
| "step": 674304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.895130123044125e-05, | |
| "loss": 4.0778, | |
| "step": 674816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.894291528293073e-05, | |
| "loss": 4.0759, | |
| "step": 675328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.893454571422395e-05, | |
| "loss": 4.0641, | |
| "step": 675840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.892615976671343e-05, | |
| "loss": 4.0815, | |
| "step": 676352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.891779019800664e-05, | |
| "loss": 4.0668, | |
| "step": 676864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.890940425049612e-05, | |
| "loss": 4.0818, | |
| "step": 677376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.89010183029856e-05, | |
| "loss": 4.0877, | |
| "step": 677888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.889263235547508e-05, | |
| "loss": 4.0753, | |
| "step": 678400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.888424640796455e-05, | |
| "loss": 4.0772, | |
| "step": 678912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.887586046045403e-05, | |
| "loss": 4.0723, | |
| "step": 679424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.886747451294351e-05, | |
| "loss": 4.0741, | |
| "step": 679936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.885908856543299e-05, | |
| "loss": 4.0722, | |
| "step": 680448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.885070261792248e-05, | |
| "loss": 4.0852, | |
| "step": 680960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.884231667041196e-05, | |
| "loss": 4.0864, | |
| "step": 681472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.883393072290144e-05, | |
| "loss": 4.0778, | |
| "step": 681984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.882554477539092e-05, | |
| "loss": 4.0751, | |
| "step": 682496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.88171588278804e-05, | |
| "loss": 4.0782, | |
| "step": 683008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8808789259173606e-05, | |
| "loss": 4.0731, | |
| "step": 683520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8800403311663086e-05, | |
| "loss": 4.0706, | |
| "step": 684032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8792017364152566e-05, | |
| "loss": 4.0675, | |
| "step": 684544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8783647795445775e-05, | |
| "loss": 4.0765, | |
| "step": 685056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8775261847935255e-05, | |
| "loss": 4.0729, | |
| "step": 685568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8766875900424735e-05, | |
| "loss": 4.0609, | |
| "step": 686080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8758489952914215e-05, | |
| "loss": 4.0735, | |
| "step": 686592 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.0929107666015625, | |
| "eval_runtime": 530.2994, | |
| "eval_samples_per_second": 719.577, | |
| "eval_steps_per_second": 22.487, | |
| "step": 686871 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8750104005403695e-05, | |
| "loss": 4.0729, | |
| "step": 687104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.874173443669691e-05, | |
| "loss": 4.0814, | |
| "step": 687616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.873334848918639e-05, | |
| "loss": 4.0791, | |
| "step": 688128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.872496254167587e-05, | |
| "loss": 4.0662, | |
| "step": 688640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.871657659416535e-05, | |
| "loss": 4.0735, | |
| "step": 689152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.870819064665483e-05, | |
| "loss": 4.0688, | |
| "step": 689664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.869980469914431e-05, | |
| "loss": 4.0665, | |
| "step": 690176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.869143513043752e-05, | |
| "loss": 4.0667, | |
| "step": 690688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8683049182927e-05, | |
| "loss": 4.0642, | |
| "step": 691200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.867466323541648e-05, | |
| "loss": 4.0749, | |
| "step": 691712 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.866627728790596e-05, | |
| "loss": 4.0776, | |
| "step": 692224 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.865790771919917e-05, | |
| "loss": 4.065, | |
| "step": 692736 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.864952177168865e-05, | |
| "loss": 4.0764, | |
| "step": 693248 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.864113582417813e-05, | |
| "loss": 4.0592, | |
| "step": 693760 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.863274987666761e-05, | |
| "loss": 4.0767, | |
| "step": 694272 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.862436392915709e-05, | |
| "loss": 4.0553, | |
| "step": 694784 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.861597798164657e-05, | |
| "loss": 4.0668, | |
| "step": 695296 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.860759203413605e-05, | |
| "loss": 4.0695, | |
| "step": 695808 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.859920608662553e-05, | |
| "loss": 4.0668, | |
| "step": 696320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.859083651791874e-05, | |
| "loss": 4.0788, | |
| "step": 696832 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.858245057040822e-05, | |
| "loss": 4.0628, | |
| "step": 697344 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.85740646228977e-05, | |
| "loss": 4.0891, | |
| "step": 697856 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.856567867538718e-05, | |
| "loss": 4.0689, | |
| "step": 698368 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.855729272787666e-05, | |
| "loss": 4.0708, | |
| "step": 698880 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.854892315916987e-05, | |
| "loss": 4.0683, | |
| "step": 699392 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.854053721165935e-05, | |
| "loss": 4.073, | |
| "step": 699904 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8532151264148833e-05, | |
| "loss": 4.0699, | |
| "step": 700416 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.852376531663831e-05, | |
| "loss": 4.0635, | |
| "step": 700928 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.851539574793152e-05, | |
| "loss": 4.0682, | |
| "step": 701440 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.850702617922474e-05, | |
| "loss": 4.0707, | |
| "step": 701952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.849864023171421e-05, | |
| "loss": 4.0497, | |
| "step": 702464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.849025428420369e-05, | |
| "loss": 4.0745, | |
| "step": 702976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.848186833669317e-05, | |
| "loss": 4.0749, | |
| "step": 703488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.847348238918265e-05, | |
| "loss": 4.0692, | |
| "step": 704000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.846509644167213e-05, | |
| "loss": 4.0594, | |
| "step": 704512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.845672687296534e-05, | |
| "loss": 4.0734, | |
| "step": 705024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.844834092545482e-05, | |
| "loss": 4.0666, | |
| "step": 705536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.84399549779443e-05, | |
| "loss": 4.0676, | |
| "step": 706048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.843156903043379e-05, | |
| "loss": 4.0703, | |
| "step": 706560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.842318308292327e-05, | |
| "loss": 4.0563, | |
| "step": 707072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.841479713541275e-05, | |
| "loss": 4.0597, | |
| "step": 707584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.840641118790223e-05, | |
| "loss": 4.0494, | |
| "step": 708096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.839802524039171e-05, | |
| "loss": 4.0711, | |
| "step": 708608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8389655671684916e-05, | |
| "loss": 4.0626, | |
| "step": 709120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8381269724174396e-05, | |
| "loss": 4.0558, | |
| "step": 709632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8372900155467605e-05, | |
| "loss": 4.0697, | |
| "step": 710144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8364514207957085e-05, | |
| "loss": 4.0544, | |
| "step": 710656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8356128260446565e-05, | |
| "loss": 4.0715, | |
| "step": 711168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8347742312936045e-05, | |
| "loss": 4.0658, | |
| "step": 711680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8339356365425525e-05, | |
| "loss": 4.0635, | |
| "step": 712192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.833098679671874e-05, | |
| "loss": 4.0621, | |
| "step": 712704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.832260084920822e-05, | |
| "loss": 4.0584, | |
| "step": 713216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.83142149016977e-05, | |
| "loss": 4.067, | |
| "step": 713728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.830582895418718e-05, | |
| "loss": 4.0481, | |
| "step": 714240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.829744300667666e-05, | |
| "loss": 4.0659, | |
| "step": 714752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.828907343796987e-05, | |
| "loss": 4.0573, | |
| "step": 715264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.828068749045935e-05, | |
| "loss": 4.0615, | |
| "step": 715776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.827230154294883e-05, | |
| "loss": 4.0551, | |
| "step": 716288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.826391559543831e-05, | |
| "loss": 4.0714, | |
| "step": 716800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.825552964792779e-05, | |
| "loss": 4.052, | |
| "step": 717312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.824714370041727e-05, | |
| "loss": 4.0532, | |
| "step": 717824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.823875775290675e-05, | |
| "loss": 4.0593, | |
| "step": 718336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.823037180539622e-05, | |
| "loss": 4.0532, | |
| "step": 718848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.822200223668944e-05, | |
| "loss": 4.0623, | |
| "step": 719360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8213616289178925e-05, | |
| "loss": 4.0602, | |
| "step": 719872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.82052303416684e-05, | |
| "loss": 4.0638, | |
| "step": 720384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.819684439415788e-05, | |
| "loss": 4.0398, | |
| "step": 720896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.818845844664736e-05, | |
| "loss": 4.0617, | |
| "step": 721408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8180088877940574e-05, | |
| "loss": 4.0509, | |
| "step": 721920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.817170293043005e-05, | |
| "loss": 4.0528, | |
| "step": 722432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.816331698291953e-05, | |
| "loss": 4.0578, | |
| "step": 722944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.815493103540901e-05, | |
| "loss": 4.0575, | |
| "step": 723456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.814654508789849e-05, | |
| "loss": 4.0568, | |
| "step": 723968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.813815914038797e-05, | |
| "loss": 4.0734, | |
| "step": 724480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8129789571681176e-05, | |
| "loss": 4.0599, | |
| "step": 724992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.812140362417066e-05, | |
| "loss": 4.0661, | |
| "step": 725504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.811301767666014e-05, | |
| "loss": 4.0621, | |
| "step": 726016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.810463172914962e-05, | |
| "loss": 4.0523, | |
| "step": 726528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.80962457816391e-05, | |
| "loss": 4.0512, | |
| "step": 727040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.808785983412858e-05, | |
| "loss": 4.0598, | |
| "step": 727552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.807947388661806e-05, | |
| "loss": 4.0627, | |
| "step": 728064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.807108793910754e-05, | |
| "loss": 4.0663, | |
| "step": 728576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.806271837040075e-05, | |
| "loss": 4.0378, | |
| "step": 729088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.805433242289023e-05, | |
| "loss": 4.0455, | |
| "step": 729600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.804594647537971e-05, | |
| "loss": 4.0498, | |
| "step": 730112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.803756052786919e-05, | |
| "loss": 4.058, | |
| "step": 730624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.80291909591624e-05, | |
| "loss": 4.0561, | |
| "step": 731136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.802082139045561e-05, | |
| "loss": 4.0666, | |
| "step": 731648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8012435442945097e-05, | |
| "loss": 4.0567, | |
| "step": 732160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8004049495434577e-05, | |
| "loss": 4.0541, | |
| "step": 732672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7995663547924056e-05, | |
| "loss": 4.0505, | |
| "step": 733184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7987277600413536e-05, | |
| "loss": 4.0521, | |
| "step": 733696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7978908031706746e-05, | |
| "loss": 4.0428, | |
| "step": 734208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7970522084196225e-05, | |
| "loss": 4.0548, | |
| "step": 734720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7962136136685705e-05, | |
| "loss": 4.0512, | |
| "step": 735232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7953750189175185e-05, | |
| "loss": 4.051, | |
| "step": 735744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7945364241664665e-05, | |
| "loss": 4.0535, | |
| "step": 736256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7936978294154145e-05, | |
| "loss": 4.067, | |
| "step": 736768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7928608725447354e-05, | |
| "loss": 4.0504, | |
| "step": 737280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7920222777936834e-05, | |
| "loss": 4.0623, | |
| "step": 737792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7911836830426314e-05, | |
| "loss": 4.0597, | |
| "step": 738304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7903450882915794e-05, | |
| "loss": 4.0553, | |
| "step": 738816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.789506493540528e-05, | |
| "loss": 4.0511, | |
| "step": 739328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.788669536669849e-05, | |
| "loss": 4.0586, | |
| "step": 739840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.787830941918797e-05, | |
| "loss": 4.0624, | |
| "step": 740352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.786992347167745e-05, | |
| "loss": 4.0479, | |
| "step": 740864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.786153752416693e-05, | |
| "loss": 4.0462, | |
| "step": 741376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.785315157665641e-05, | |
| "loss": 4.0424, | |
| "step": 741888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.784476562914588e-05, | |
| "loss": 4.0488, | |
| "step": 742400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.78363960604391e-05, | |
| "loss": 4.055, | |
| "step": 742912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.782801011292858e-05, | |
| "loss": 4.0507, | |
| "step": 743424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.781962416541805e-05, | |
| "loss": 4.0338, | |
| "step": 743936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.781123821790753e-05, | |
| "loss": 4.0456, | |
| "step": 744448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.780286864920075e-05, | |
| "loss": 4.055, | |
| "step": 744960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7794482701690235e-05, | |
| "loss": 4.0543, | |
| "step": 745472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.778609675417971e-05, | |
| "loss": 4.0651, | |
| "step": 745984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.777771080666919e-05, | |
| "loss": 4.0556, | |
| "step": 746496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.776932485915867e-05, | |
| "loss": 4.0488, | |
| "step": 747008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.776093891164815e-05, | |
| "loss": 4.0533, | |
| "step": 747520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.775255296413763e-05, | |
| "loss": 4.0562, | |
| "step": 748032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.774416701662711e-05, | |
| "loss": 4.047, | |
| "step": 748544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.773578106911659e-05, | |
| "loss": 4.0484, | |
| "step": 749056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.77274115004098e-05, | |
| "loss": 4.0526, | |
| "step": 749568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.771902555289928e-05, | |
| "loss": 4.0431, | |
| "step": 750080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7710655984192486e-05, | |
| "loss": 4.0523, | |
| "step": 750592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.770227003668197e-05, | |
| "loss": 4.0519, | |
| "step": 751104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.769388408917145e-05, | |
| "loss": 4.0554, | |
| "step": 751616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.768549814166093e-05, | |
| "loss": 4.0409, | |
| "step": 752128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.767711219415041e-05, | |
| "loss": 4.055, | |
| "step": 752640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.766874262544362e-05, | |
| "loss": 4.045, | |
| "step": 753152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.76603566779331e-05, | |
| "loss": 4.0601, | |
| "step": 753664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.765197073042258e-05, | |
| "loss": 4.0651, | |
| "step": 754176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.764358478291206e-05, | |
| "loss": 4.0503, | |
| "step": 754688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.763519883540154e-05, | |
| "loss": 4.0515, | |
| "step": 755200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.762681288789102e-05, | |
| "loss": 4.0541, | |
| "step": 755712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.76184269403805e-05, | |
| "loss": 4.0483, | |
| "step": 756224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.761004099286998e-05, | |
| "loss": 4.048, | |
| "step": 756736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.760167142416319e-05, | |
| "loss": 4.062, | |
| "step": 757248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.759328547665267e-05, | |
| "loss": 4.0574, | |
| "step": 757760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.758489952914215e-05, | |
| "loss": 4.0567, | |
| "step": 758272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.757651358163164e-05, | |
| "loss": 4.0515, | |
| "step": 758784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.756812763412112e-05, | |
| "loss": 4.0573, | |
| "step": 759296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.75597416866106e-05, | |
| "loss": 4.0468, | |
| "step": 759808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7551388496707535e-05, | |
| "loss": 4.0453, | |
| "step": 760320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7543002549197015e-05, | |
| "loss": 4.0458, | |
| "step": 760832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7534616601686495e-05, | |
| "loss": 4.0533, | |
| "step": 761344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7526230654175975e-05, | |
| "loss": 4.0475, | |
| "step": 761856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7517844706665455e-05, | |
| "loss": 4.0319, | |
| "step": 762368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7509458759154935e-05, | |
| "loss": 4.051, | |
| "step": 762880 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.076761722564697, | |
| "eval_runtime": 530.228, | |
| "eval_samples_per_second": 719.673, | |
| "eval_steps_per_second": 22.49, | |
| "step": 763190 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7501072811644415e-05, | |
| "loss": 4.0675, | |
| "step": 763392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.749268686413389e-05, | |
| "loss": 4.0537, | |
| "step": 763904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7484300916623375e-05, | |
| "loss": 4.0572, | |
| "step": 764416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7475914969112855e-05, | |
| "loss": 4.0398, | |
| "step": 764928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7467529021602335e-05, | |
| "loss": 4.0542, | |
| "step": 765440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7459143074091815e-05, | |
| "loss": 4.046, | |
| "step": 765952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7450757126581294e-05, | |
| "loss": 4.043, | |
| "step": 766464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7442371179070774e-05, | |
| "loss": 4.042, | |
| "step": 766976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7433985231560254e-05, | |
| "loss": 4.0412, | |
| "step": 767488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7425599284049734e-05, | |
| "loss": 4.0534, | |
| "step": 768000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7417213336539214e-05, | |
| "loss": 4.0516, | |
| "step": 768512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7408827389028694e-05, | |
| "loss": 4.0447, | |
| "step": 769024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7400441441518174e-05, | |
| "loss": 4.0531, | |
| "step": 769536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7392055494007654e-05, | |
| "loss": 4.0347, | |
| "step": 770048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.738366954649713e-05, | |
| "loss": 4.0555, | |
| "step": 770560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.737528359898661e-05, | |
| "loss": 4.0346, | |
| "step": 771072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.736689765147609e-05, | |
| "loss": 4.0424, | |
| "step": 771584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7358511703965574e-05, | |
| "loss": 4.0462, | |
| "step": 772096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7350125756455054e-05, | |
| "loss": 4.0468, | |
| "step": 772608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7341739808944534e-05, | |
| "loss": 4.0515, | |
| "step": 773120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.733337024023774e-05, | |
| "loss": 4.0377, | |
| "step": 773632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.732498429272722e-05, | |
| "loss": 4.0647, | |
| "step": 774144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.73165983452167e-05, | |
| "loss": 4.0451, | |
| "step": 774656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.730821239770618e-05, | |
| "loss": 4.0515, | |
| "step": 775168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.729982645019566e-05, | |
| "loss": 4.0436, | |
| "step": 775680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.729145688148887e-05, | |
| "loss": 4.0533, | |
| "step": 776192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.728307093397835e-05, | |
| "loss": 4.0466, | |
| "step": 776704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.727468498646783e-05, | |
| "loss": 4.042, | |
| "step": 777216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.726629903895731e-05, | |
| "loss": 4.0487, | |
| "step": 777728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.725792947025053e-05, | |
| "loss": 4.0462, | |
| "step": 778240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.724955990154374e-05, | |
| "loss": 4.0295, | |
| "step": 778752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.724117395403322e-05, | |
| "loss": 4.0501, | |
| "step": 779264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.72327880065227e-05, | |
| "loss": 4.0496, | |
| "step": 779776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.722440205901218e-05, | |
| "loss": 4.046, | |
| "step": 780288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7216016111501657e-05, | |
| "loss": 4.0363, | |
| "step": 780800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7207630163991137e-05, | |
| "loss": 4.0568, | |
| "step": 781312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7199260595284346e-05, | |
| "loss": 4.0408, | |
| "step": 781824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7190874647773826e-05, | |
| "loss": 4.0501, | |
| "step": 782336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7182488700263306e-05, | |
| "loss": 4.0458, | |
| "step": 782848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7174102752752785e-05, | |
| "loss": 4.0337, | |
| "step": 783360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7165716805242265e-05, | |
| "loss": 4.039, | |
| "step": 783872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7157330857731745e-05, | |
| "loss": 4.0261, | |
| "step": 784384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7148944910221225e-05, | |
| "loss": 4.047, | |
| "step": 784896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7140558962710705e-05, | |
| "loss": 4.042, | |
| "step": 785408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.713218939400392e-05, | |
| "loss": 4.0324, | |
| "step": 785920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.71238034464934e-05, | |
| "loss": 4.0492, | |
| "step": 786432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.711541749898288e-05, | |
| "loss": 4.0342, | |
| "step": 786944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.710703155147236e-05, | |
| "loss": 4.0487, | |
| "step": 787456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.70986783615693e-05, | |
| "loss": 4.0432, | |
| "step": 787968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.709029241405878e-05, | |
| "loss": 4.0384, | |
| "step": 788480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.708190646654826e-05, | |
| "loss": 4.0417, | |
| "step": 788992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.707352051903774e-05, | |
| "loss": 4.0359, | |
| "step": 789504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.706513457152722e-05, | |
| "loss": 4.0423, | |
| "step": 790016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.70567486240167e-05, | |
| "loss": 4.0284, | |
| "step": 790528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7048379055309915e-05, | |
| "loss": 4.0422, | |
| "step": 791040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7039993107799395e-05, | |
| "loss": 4.0382, | |
| "step": 791552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7031607160288875e-05, | |
| "loss": 4.0389, | |
| "step": 792064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7023221212778355e-05, | |
| "loss": 4.0345, | |
| "step": 792576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7014835265267835e-05, | |
| "loss": 4.0466, | |
| "step": 793088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7006449317757315e-05, | |
| "loss": 4.0307, | |
| "step": 793600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6998063370246795e-05, | |
| "loss": 4.0304, | |
| "step": 794112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.698967742273627e-05, | |
| "loss": 4.036, | |
| "step": 794624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6981307854029484e-05, | |
| "loss": 4.03, | |
| "step": 795136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6972921906518964e-05, | |
| "loss": 4.0409, | |
| "step": 795648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.696453595900844e-05, | |
| "loss": 4.038, | |
| "step": 796160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.695615001149792e-05, | |
| "loss": 4.0406, | |
| "step": 796672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.694778044279113e-05, | |
| "loss": 4.0195, | |
| "step": 797184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.693939449528061e-05, | |
| "loss": 4.0403, | |
| "step": 797696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.693100854777009e-05, | |
| "loss": 4.0281, | |
| "step": 798208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.692263897906331e-05, | |
| "loss": 4.0313, | |
| "step": 798720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.691425303155279e-05, | |
| "loss": 4.0334, | |
| "step": 799232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.690586708404226e-05, | |
| "loss": 4.0394, | |
| "step": 799744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.689748113653174e-05, | |
| "loss": 4.0315, | |
| "step": 800256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.688909518902122e-05, | |
| "loss": 4.0515, | |
| "step": 800768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.68807092415107e-05, | |
| "loss": 4.0407, | |
| "step": 801280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.687232329400018e-05, | |
| "loss": 4.042, | |
| "step": 801792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.686395372529339e-05, | |
| "loss": 4.0412, | |
| "step": 802304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.685556777778287e-05, | |
| "loss": 4.0294, | |
| "step": 802816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.684718183027235e-05, | |
| "loss": 4.0322, | |
| "step": 803328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.683879588276184e-05, | |
| "loss": 4.0386, | |
| "step": 803840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6830426314055046e-05, | |
| "loss": 4.04, | |
| "step": 804352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6822040366544526e-05, | |
| "loss": 4.0448, | |
| "step": 804864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6813654419034006e-05, | |
| "loss": 4.0186, | |
| "step": 805376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6805268471523486e-05, | |
| "loss": 4.0234, | |
| "step": 805888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6796898902816695e-05, | |
| "loss": 4.025, | |
| "step": 806400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.678852933410991e-05, | |
| "loss": 4.0431, | |
| "step": 806912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6780143386599384e-05, | |
| "loss": 4.0342, | |
| "step": 807424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6771757439088864e-05, | |
| "loss": 4.0438, | |
| "step": 807936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6763371491578344e-05, | |
| "loss": 4.0335, | |
| "step": 808448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6754985544067824e-05, | |
| "loss": 4.0363, | |
| "step": 808960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6746599596557304e-05, | |
| "loss": 4.0241, | |
| "step": 809472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.673821364904679e-05, | |
| "loss": 4.0325, | |
| "step": 809984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.672982770153627e-05, | |
| "loss": 4.0247, | |
| "step": 810496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.672144175402575e-05, | |
| "loss": 4.0319, | |
| "step": 811008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.671305580651523e-05, | |
| "loss": 4.0289, | |
| "step": 811520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.670468623780844e-05, | |
| "loss": 4.032, | |
| "step": 812032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.669630029029792e-05, | |
| "loss": 4.0344, | |
| "step": 812544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.66879143427874e-05, | |
| "loss": 4.0437, | |
| "step": 813056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.667952839527688e-05, | |
| "loss": 4.0357, | |
| "step": 813568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.667114244776636e-05, | |
| "loss": 4.041, | |
| "step": 814080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.666277287905957e-05, | |
| "loss": 4.0363, | |
| "step": 814592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.665438693154905e-05, | |
| "loss": 4.0358, | |
| "step": 815104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.664600098403853e-05, | |
| "loss": 4.0338, | |
| "step": 815616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.663761503652801e-05, | |
| "loss": 4.0406, | |
| "step": 816128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.662922908901749e-05, | |
| "loss": 4.0398, | |
| "step": 816640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6620843141506975e-05, | |
| "loss": 4.028, | |
| "step": 817152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6612457193996455e-05, | |
| "loss": 4.0253, | |
| "step": 817664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.660407124648593e-05, | |
| "loss": 4.02, | |
| "step": 818176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6595701677779144e-05, | |
| "loss": 4.0312, | |
| "step": 818688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.658733210907235e-05, | |
| "loss": 4.0326, | |
| "step": 819200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.657894616156183e-05, | |
| "loss": 4.036, | |
| "step": 819712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.657056021405131e-05, | |
| "loss": 4.0111, | |
| "step": 820224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.656217426654079e-05, | |
| "loss": 4.0176, | |
| "step": 820736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.655378831903027e-05, | |
| "loss": 4.0376, | |
| "step": 821248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6545402371519746e-05, | |
| "loss": 4.0349, | |
| "step": 821760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6537016424009226e-05, | |
| "loss": 4.0446, | |
| "step": 822272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.652863047649871e-05, | |
| "loss": 4.0377, | |
| "step": 822784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.652024452898819e-05, | |
| "loss": 4.0256, | |
| "step": 823296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.65118749602814e-05, | |
| "loss": 4.0358, | |
| "step": 823808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.650348901277088e-05, | |
| "loss": 4.0331, | |
| "step": 824320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.649510306526036e-05, | |
| "loss": 4.0261, | |
| "step": 824832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.648673349655357e-05, | |
| "loss": 4.0274, | |
| "step": 825344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.647834754904305e-05, | |
| "loss": 4.0327, | |
| "step": 825856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.646996160153253e-05, | |
| "loss": 4.0283, | |
| "step": 826368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.646157565402201e-05, | |
| "loss": 4.0307, | |
| "step": 826880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.645318970651149e-05, | |
| "loss": 4.0295, | |
| "step": 827392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.644480375900097e-05, | |
| "loss": 4.0383, | |
| "step": 827904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.643641781149045e-05, | |
| "loss": 4.0237, | |
| "step": 828416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.642803186397993e-05, | |
| "loss": 4.0297, | |
| "step": 828928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.641966229527315e-05, | |
| "loss": 4.029, | |
| "step": 829440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.641127634776263e-05, | |
| "loss": 4.0395, | |
| "step": 829952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6402890400252107e-05, | |
| "loss": 4.0411, | |
| "step": 830464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6394504452741587e-05, | |
| "loss": 4.0323, | |
| "step": 830976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6386134884034796e-05, | |
| "loss": 4.0308, | |
| "step": 831488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6377748936524276e-05, | |
| "loss": 4.0362, | |
| "step": 832000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6369362989013756e-05, | |
| "loss": 4.0287, | |
| "step": 832512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6360977041503236e-05, | |
| "loss": 4.0275, | |
| "step": 833024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6352607472796445e-05, | |
| "loss": 4.0413, | |
| "step": 833536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6344221525285925e-05, | |
| "loss": 4.038, | |
| "step": 834048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6335835577775405e-05, | |
| "loss": 4.0401, | |
| "step": 834560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6327449630264884e-05, | |
| "loss": 4.0334, | |
| "step": 835072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.63190800615581e-05, | |
| "loss": 4.0392, | |
| "step": 835584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.631069411404758e-05, | |
| "loss": 4.0231, | |
| "step": 836096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.630230816653706e-05, | |
| "loss": 4.0271, | |
| "step": 836608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.629393859783027e-05, | |
| "loss": 4.0277, | |
| "step": 837120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.628555265031975e-05, | |
| "loss": 4.0321, | |
| "step": 837632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.627716670280923e-05, | |
| "loss": 4.0281, | |
| "step": 838144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.626878075529871e-05, | |
| "loss": 4.0155, | |
| "step": 838656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.626041118659192e-05, | |
| "loss": 4.0334, | |
| "step": 839168 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.0648016929626465, | |
| "eval_runtime": 543.7091, | |
| "eval_samples_per_second": 701.829, | |
| "eval_steps_per_second": 21.933, | |
| "step": 839509 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.62520252390814e-05, | |
| "loss": 4.0312, | |
| "step": 839680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.624363929157088e-05, | |
| "loss": 4.0365, | |
| "step": 840192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.623525334406036e-05, | |
| "loss": 4.0358, | |
| "step": 840704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.622686739654984e-05, | |
| "loss": 4.0243, | |
| "step": 841216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.621848144903932e-05, | |
| "loss": 4.0331, | |
| "step": 841728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.62100955015288e-05, | |
| "loss": 4.0277, | |
| "step": 842240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6201709554018285e-05, | |
| "loss": 4.0211, | |
| "step": 842752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.619332360650776e-05, | |
| "loss": 4.0242, | |
| "step": 843264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.618493765899724e-05, | |
| "loss": 4.0274, | |
| "step": 843776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.617655171148672e-05, | |
| "loss": 4.0309, | |
| "step": 844288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6168182142779934e-05, | |
| "loss": 4.0287, | |
| "step": 844800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.615979619526941e-05, | |
| "loss": 4.026, | |
| "step": 845312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.615141024775889e-05, | |
| "loss": 4.0333, | |
| "step": 845824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.614302430024837e-05, | |
| "loss": 4.0215, | |
| "step": 846336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.613463835273785e-05, | |
| "loss": 4.0306, | |
| "step": 846848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.612625240522733e-05, | |
| "loss": 4.0168, | |
| "step": 847360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.611786645771681e-05, | |
| "loss": 4.0212, | |
| "step": 847872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.610948051020629e-05, | |
| "loss": 4.0288, | |
| "step": 848384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.61011109414995e-05, | |
| "loss": 4.0255, | |
| "step": 848896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.609272499398898e-05, | |
| "loss": 4.0305, | |
| "step": 849408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.608433904647846e-05, | |
| "loss": 4.0218, | |
| "step": 849920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.607595309896794e-05, | |
| "loss": 4.0468, | |
| "step": 850432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.606758353026115e-05, | |
| "loss": 4.0251, | |
| "step": 850944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.605919758275063e-05, | |
| "loss": 4.0289, | |
| "step": 851456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.605081163524011e-05, | |
| "loss": 4.0273, | |
| "step": 851968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.604242568772959e-05, | |
| "loss": 4.0365, | |
| "step": 852480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.60340561190228e-05, | |
| "loss": 4.0261, | |
| "step": 852992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.602567017151228e-05, | |
| "loss": 4.02, | |
| "step": 853504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.601728422400176e-05, | |
| "loss": 4.0281, | |
| "step": 854016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.600889827649124e-05, | |
| "loss": 4.0303, | |
| "step": 854528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.600051232898072e-05, | |
| "loss": 4.0121, | |
| "step": 855040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.599212638147021e-05, | |
| "loss": 4.0261, | |
| "step": 855552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.598374043395969e-05, | |
| "loss": 4.0337, | |
| "step": 856064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.597535448644917e-05, | |
| "loss": 4.0294, | |
| "step": 856576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.596696853893865e-05, | |
| "loss": 4.0168, | |
| "step": 857088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5958598970231856e-05, | |
| "loss": 4.0315, | |
| "step": 857600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5950229401525065e-05, | |
| "loss": 4.026, | |
| "step": 858112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5941843454014545e-05, | |
| "loss": 4.028, | |
| "step": 858624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5933457506504025e-05, | |
| "loss": 4.0254, | |
| "step": 859136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5925071558993505e-05, | |
| "loss": 4.0182, | |
| "step": 859648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5916685611482985e-05, | |
| "loss": 4.021, | |
| "step": 860160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5908316042776194e-05, | |
| "loss": 4.0081, | |
| "step": 860672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5899930095265674e-05, | |
| "loss": 4.0255, | |
| "step": 861184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5891544147755154e-05, | |
| "loss": 4.0251, | |
| "step": 861696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.588315820024464e-05, | |
| "loss": 4.0165, | |
| "step": 862208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.587477225273412e-05, | |
| "loss": 4.0257, | |
| "step": 862720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5866386305223594e-05, | |
| "loss": 4.0154, | |
| "step": 863232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5858000357713074e-05, | |
| "loss": 4.0332, | |
| "step": 863744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5849614410202554e-05, | |
| "loss": 4.0263, | |
| "step": 864256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5841228462692034e-05, | |
| "loss": 4.013, | |
| "step": 864768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.583285889398524e-05, | |
| "loss": 4.0212, | |
| "step": 865280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.582447294647472e-05, | |
| "loss": 4.0172, | |
| "step": 865792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.58160869989642e-05, | |
| "loss": 4.0292, | |
| "step": 866304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.580770105145368e-05, | |
| "loss": 4.0068, | |
| "step": 866816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.579933148274689e-05, | |
| "loss": 4.0269, | |
| "step": 867328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.579094553523638e-05, | |
| "loss": 4.0183, | |
| "step": 867840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.578255958772586e-05, | |
| "loss": 4.0211, | |
| "step": 868352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.577417364021534e-05, | |
| "loss": 4.0186, | |
| "step": 868864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.576580407150855e-05, | |
| "loss": 4.0288, | |
| "step": 869376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.575741812399803e-05, | |
| "loss": 4.0078, | |
| "step": 869888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.574903217648751e-05, | |
| "loss": 4.0127, | |
| "step": 870400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.574064622897699e-05, | |
| "loss": 4.0182, | |
| "step": 870912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.573226028146647e-05, | |
| "loss": 4.0124, | |
| "step": 871424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5723890712759676e-05, | |
| "loss": 4.0213, | |
| "step": 871936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5715504765249156e-05, | |
| "loss": 4.02, | |
| "step": 872448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5707118817738636e-05, | |
| "loss": 4.0263, | |
| "step": 872960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5698732870228116e-05, | |
| "loss": 3.9986, | |
| "step": 873472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.569036330152133e-05, | |
| "loss": 4.0198, | |
| "step": 873984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.568197735401081e-05, | |
| "loss": 4.0147, | |
| "step": 874496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.567359140650029e-05, | |
| "loss": 4.0115, | |
| "step": 875008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.566520545898977e-05, | |
| "loss": 4.0159, | |
| "step": 875520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.565683589028298e-05, | |
| "loss": 4.0236, | |
| "step": 876032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.564844994277246e-05, | |
| "loss": 4.0092, | |
| "step": 876544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.564006399526194e-05, | |
| "loss": 4.0338, | |
| "step": 877056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.563167804775142e-05, | |
| "loss": 4.0217, | |
| "step": 877568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.562330847904463e-05, | |
| "loss": 4.0254, | |
| "step": 878080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.561492253153411e-05, | |
| "loss": 4.0229, | |
| "step": 878592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.560653658402359e-05, | |
| "loss": 4.0127, | |
| "step": 879104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.559815063651307e-05, | |
| "loss": 4.0133, | |
| "step": 879616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.558976468900255e-05, | |
| "loss": 4.0213, | |
| "step": 880128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5581411499099495e-05, | |
| "loss": 4.0247, | |
| "step": 880640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5573025551588975e-05, | |
| "loss": 4.022, | |
| "step": 881152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5564639604078455e-05, | |
| "loss": 4.002, | |
| "step": 881664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5556253656567935e-05, | |
| "loss": 4.0076, | |
| "step": 882176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5547867709057415e-05, | |
| "loss": 4.0037, | |
| "step": 882688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5539498140350624e-05, | |
| "loss": 4.0251, | |
| "step": 883200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5531112192840104e-05, | |
| "loss": 4.0132, | |
| "step": 883712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5522726245329584e-05, | |
| "loss": 4.0263, | |
| "step": 884224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5514340297819064e-05, | |
| "loss": 4.0177, | |
| "step": 884736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5505954350308544e-05, | |
| "loss": 4.0143, | |
| "step": 885248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.549758478160175e-05, | |
| "loss": 4.0087, | |
| "step": 885760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.548919883409124e-05, | |
| "loss": 4.0128, | |
| "step": 886272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.548081288658072e-05, | |
| "loss": 4.011, | |
| "step": 886784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.54724269390702e-05, | |
| "loss": 4.011, | |
| "step": 887296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.546404099155968e-05, | |
| "loss": 4.0089, | |
| "step": 887808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.545567142285289e-05, | |
| "loss": 4.017, | |
| "step": 888320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.54473018541461e-05, | |
| "loss": 4.0169, | |
| "step": 888832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.543891590663558e-05, | |
| "loss": 4.023, | |
| "step": 889344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.543052995912506e-05, | |
| "loss": 4.0198, | |
| "step": 889856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.542214401161454e-05, | |
| "loss": 4.0206, | |
| "step": 890368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.541375806410402e-05, | |
| "loss": 4.0171, | |
| "step": 890880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.54053721165935e-05, | |
| "loss": 4.0213, | |
| "step": 891392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.539698616908298e-05, | |
| "loss": 4.015, | |
| "step": 891904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.538860022157246e-05, | |
| "loss": 4.0222, | |
| "step": 892416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.538021427406194e-05, | |
| "loss": 4.0164, | |
| "step": 892928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5371828326551424e-05, | |
| "loss": 4.0168, | |
| "step": 893440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5363442379040904e-05, | |
| "loss": 4.0097, | |
| "step": 893952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.535505643153038e-05, | |
| "loss": 3.9967, | |
| "step": 894464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.534668686282359e-05, | |
| "loss": 4.0141, | |
| "step": 894976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.533833367292053e-05, | |
| "loss": 4.0179, | |
| "step": 895488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.532994772541001e-05, | |
| "loss": 4.0139, | |
| "step": 896000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.532156177789949e-05, | |
| "loss": 3.9941, | |
| "step": 896512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.531317583038897e-05, | |
| "loss": 4.0011, | |
| "step": 897024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.530478988287845e-05, | |
| "loss": 4.0189, | |
| "step": 897536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.529640393536793e-05, | |
| "loss": 4.0193, | |
| "step": 898048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.528801798785741e-05, | |
| "loss": 4.0247, | |
| "step": 898560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.527963204034689e-05, | |
| "loss": 4.0197, | |
| "step": 899072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.527124609283637e-05, | |
| "loss": 4.0136, | |
| "step": 899584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.526286014532585e-05, | |
| "loss": 4.0152, | |
| "step": 900096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.525447419781533e-05, | |
| "loss": 4.0155, | |
| "step": 900608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.524608825030481e-05, | |
| "loss": 4.0052, | |
| "step": 901120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.523770230279429e-05, | |
| "loss": 4.0076, | |
| "step": 901632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.52293327340875e-05, | |
| "loss": 4.0177, | |
| "step": 902144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.522094678657698e-05, | |
| "loss": 4.0047, | |
| "step": 902656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.521256083906646e-05, | |
| "loss": 4.015, | |
| "step": 903168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.520417489155594e-05, | |
| "loss": 4.0093, | |
| "step": 903680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5195821701652885e-05, | |
| "loss": 4.0213, | |
| "step": 904192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5187435754142365e-05, | |
| "loss": 4.0043, | |
| "step": 904704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5179049806631844e-05, | |
| "loss": 4.0155, | |
| "step": 905216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5170663859121324e-05, | |
| "loss": 4.0098, | |
| "step": 905728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5162277911610804e-05, | |
| "loss": 4.0217, | |
| "step": 906240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5153891964100284e-05, | |
| "loss": 4.0223, | |
| "step": 906752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5145506016589764e-05, | |
| "loss": 4.0169, | |
| "step": 907264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5137120069079244e-05, | |
| "loss": 4.0103, | |
| "step": 907776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.512875050037245e-05, | |
| "loss": 4.0147, | |
| "step": 908288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.512036455286193e-05, | |
| "loss": 4.0124, | |
| "step": 908800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.511197860535141e-05, | |
| "loss": 4.0104, | |
| "step": 909312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.510359265784089e-05, | |
| "loss": 4.0232, | |
| "step": 909824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.509520671033037e-05, | |
| "loss": 4.0185, | |
| "step": 910336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.508682076281985e-05, | |
| "loss": 4.0221, | |
| "step": 910848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.507843481530933e-05, | |
| "loss": 4.0186, | |
| "step": 911360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.507004886779881e-05, | |
| "loss": 4.0183, | |
| "step": 911872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.506167929909203e-05, | |
| "loss": 4.0063, | |
| "step": 912384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.505329335158151e-05, | |
| "loss": 4.0095, | |
| "step": 912896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.504490740407099e-05, | |
| "loss": 4.0104, | |
| "step": 913408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.503652145656047e-05, | |
| "loss": 4.0159, | |
| "step": 913920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.502815188785368e-05, | |
| "loss": 4.0134, | |
| "step": 914432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.501976594034316e-05, | |
| "loss": 3.9974, | |
| "step": 914944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.501137999283264e-05, | |
| "loss": 4.0153, | |
| "step": 915456 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.053233623504639, | |
| "eval_runtime": 542.9356, | |
| "eval_samples_per_second": 702.829, | |
| "eval_steps_per_second": 21.964, | |
| "step": 915828 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 3.708640544457162e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |