| { | |
| "best_metric": 4.172706604003906, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl/lstm/1/checkpoints/checkpoint-457914", | |
| "epoch": 0.025000278439663435, | |
| "eval_steps": 10, | |
| "global_step": 457914, | |
| "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": 10.8173, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 7.5621, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 7.0552, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 6.9778, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 6.9512, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 6.8598, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 6.7142, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994129836742636e-05, | |
| "loss": 6.6064, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993291241991584e-05, | |
| "loss": 6.5223, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992452647240532e-05, | |
| "loss": 6.4391, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99161405248948e-05, | |
| "loss": 6.405, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990775457738428e-05, | |
| "loss": 6.347, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989936862987376e-05, | |
| "loss": 6.2854, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989098268236324e-05, | |
| "loss": 6.2171, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988259673485272e-05, | |
| "loss": 6.1624, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.98742107873422e-05, | |
| "loss": 6.1169, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986582483983168e-05, | |
| "loss": 6.0706, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985743889232116e-05, | |
| "loss": 6.0137, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984905294481064e-05, | |
| "loss": 5.9735, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984066699730012e-05, | |
| "loss": 5.9362, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.8977, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.982391148108281e-05, | |
| "loss": 5.8699, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.981552553357229e-05, | |
| "loss": 5.8394, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9807155964865506e-05, | |
| "loss": 5.807, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9798770017354986e-05, | |
| "loss": 5.7783, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9790384069844466e-05, | |
| "loss": 5.7604, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9781998122333946e-05, | |
| "loss": 5.7276, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773612174823426e-05, | |
| "loss": 5.7055, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9765242606116635e-05, | |
| "loss": 5.6716, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9756856658606115e-05, | |
| "loss": 5.667, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748470711095595e-05, | |
| "loss": 5.6342, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.974008476358507e-05, | |
| "loss": 5.6179, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.973169881607455e-05, | |
| "loss": 5.6075, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.972331286856403e-05, | |
| "loss": 5.589, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714926921053515e-05, | |
| "loss": 5.5594, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9706540973542995e-05, | |
| "loss": 5.5456, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9698171404836204e-05, | |
| "loss": 5.5373, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9689785457325684e-05, | |
| "loss": 5.5153, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9681399509815164e-05, | |
| "loss": 5.5032, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9673013562304644e-05, | |
| "loss": 5.4829, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966464399359785e-05, | |
| "loss": 5.4577, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965625804608733e-05, | |
| "loss": 5.4464, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964787209857681e-05, | |
| "loss": 5.4322, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963948615106629e-05, | |
| "loss": 5.4251, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96311165823595e-05, | |
| "loss": 5.4132, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962273063484898e-05, | |
| "loss": 5.4073, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.961434468733847e-05, | |
| "loss": 5.3832, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.960595873982795e-05, | |
| "loss": 5.3828, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.959757279231743e-05, | |
| "loss": 5.3878, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958920322361064e-05, | |
| "loss": 5.3593, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958081727610012e-05, | |
| "loss": 5.3489, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95724313285896e-05, | |
| "loss": 5.3364, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956404538107908e-05, | |
| "loss": 5.3456, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955565943356856e-05, | |
| "loss": 5.3071, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.954732262246923e-05, | |
| "loss": 5.3256, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953893667495871e-05, | |
| "loss": 5.2948, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953055072744819e-05, | |
| "loss": 5.2928, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.952216477993767e-05, | |
| "loss": 5.2861, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951377883242715e-05, | |
| "loss": 5.2759, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.950539288491663e-05, | |
| "loss": 5.2628, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949700693740611e-05, | |
| "loss": 5.2518, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948862098989559e-05, | |
| "loss": 5.2376, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948023504238507e-05, | |
| "loss": 5.2442, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947186547367828e-05, | |
| "loss": 5.2314, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946347952616776e-05, | |
| "loss": 5.2213, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945509357865724e-05, | |
| "loss": 5.2138, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944670763114672e-05, | |
| "loss": 5.2024, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94383216836362e-05, | |
| "loss": 5.1858, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942993573612568e-05, | |
| "loss": 5.203, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942154978861516e-05, | |
| "loss": 5.1795, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.941316384110464e-05, | |
| "loss": 5.1736, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.940477789359412e-05, | |
| "loss": 5.1824, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.93963919460836e-05, | |
| "loss": 5.1575, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388022377376816e-05, | |
| "loss": 5.1701, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379636429866296e-05, | |
| "loss": 5.1621, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371250482355776e-05, | |
| "loss": 5.1505, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362864534845256e-05, | |
| "loss": 5.1492, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.935447858733473e-05, | |
| "loss": 5.1231, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.934609263982421e-05, | |
| "loss": 5.1214, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.933770669231369e-05, | |
| "loss": 5.1192, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329337123606905e-05, | |
| "loss": 5.1184, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932095117609638e-05, | |
| "loss": 5.1077, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.931256522858586e-05, | |
| "loss": 5.0946, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.930417928107534e-05, | |
| "loss": 5.0808, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9295793333564825e-05, | |
| "loss": 5.0878, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9287407386054305e-05, | |
| "loss": 5.0868, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9279037817347514e-05, | |
| "loss": 5.0809, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9270651869836994e-05, | |
| "loss": 5.0794, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9262265922326474e-05, | |
| "loss": 5.0745, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9253879974815954e-05, | |
| "loss": 5.0617, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9245494027305433e-05, | |
| "loss": 5.0432, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923712445859864e-05, | |
| "loss": 5.0402, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922873851108812e-05, | |
| "loss": 5.0459, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.92203525635776e-05, | |
| "loss": 5.0425, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921196661606708e-05, | |
| "loss": 5.0189, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920359704736029e-05, | |
| "loss": 5.0237, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919521109984978e-05, | |
| "loss": 5.0224, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918682515233926e-05, | |
| "loss": 5.0212, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917843920482874e-05, | |
| "loss": 5.0117, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917005325731822e-05, | |
| "loss": 5.0127, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.916168368861143e-05, | |
| "loss": 5.0045, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.915329774110091e-05, | |
| "loss": 4.9936, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.914491179359039e-05, | |
| "loss": 4.9952, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.913652584607987e-05, | |
| "loss": 4.9946, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.912813989856935e-05, | |
| "loss": 4.9785, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9119770329862556e-05, | |
| "loss": 4.9824, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9111384382352036e-05, | |
| "loss": 4.9603, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9102998434841516e-05, | |
| "loss": 4.9551, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9094612487330996e-05, | |
| "loss": 4.9542, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9086226539820476e-05, | |
| "loss": 4.9587, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907784059230996e-05, | |
| "loss": 4.9513, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906945464479944e-05, | |
| "loss": 4.9289, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906106869728892e-05, | |
| "loss": 4.9339, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905269912858213e-05, | |
| "loss": 4.9377, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904431318107161e-05, | |
| "loss": 4.9436, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903592723356109e-05, | |
| "loss": 4.9289, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90275576648543e-05, | |
| "loss": 4.9288, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901918809614751e-05, | |
| "loss": 4.9254, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901080214863699e-05, | |
| "loss": 4.9262, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.900241620112647e-05, | |
| "loss": 4.9102, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899403025361595e-05, | |
| "loss": 4.9102, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.898564430610543e-05, | |
| "loss": 4.9012, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.897725835859491e-05, | |
| "loss": 4.905, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968872411084396e-05, | |
| "loss": 4.8878, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.896048646357387e-05, | |
| "loss": 4.8999, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.895210051606335e-05, | |
| "loss": 4.8903, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8943730947356565e-05, | |
| "loss": 4.8972, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.893534499984604e-05, | |
| "loss": 4.8688, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8926975431139254e-05, | |
| "loss": 4.8801, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918589483628734e-05, | |
| "loss": 4.8895, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8910203536118214e-05, | |
| "loss": 4.8614, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890181758860769e-05, | |
| "loss": 4.882, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889343164109717e-05, | |
| "loss": 4.8646, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888504569358665e-05, | |
| "loss": 4.868, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8876659746076134e-05, | |
| "loss": 4.8582, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8868273798565614e-05, | |
| "loss": 4.8538, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8859887851055094e-05, | |
| "loss": 4.8555, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8851501903544574e-05, | |
| "loss": 4.8526, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884313233483778e-05, | |
| "loss": 4.8526, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883474638732726e-05, | |
| "loss": 4.8449, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882636043981674e-05, | |
| "loss": 4.8474, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881797449230622e-05, | |
| "loss": 4.8465, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88095885447957e-05, | |
| "loss": 4.8267, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880120259728518e-05, | |
| "loss": 4.8271, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879283302857839e-05, | |
| "loss": 4.818, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.878444708106787e-05, | |
| "loss": 4.8204, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877606113355735e-05, | |
| "loss": 4.8187, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.876767518604683e-05, | |
| "loss": 4.8119, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875930561734005e-05, | |
| "loss": 4.8099, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875091966982953e-05, | |
| "loss": 4.8102, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.7706427574157715, | |
| "eval_runtime": 545.2364, | |
| "eval_samples_per_second": 699.863, | |
| "eval_steps_per_second": 21.871, | |
| "step": 76319 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.874253372231901e-05, | |
| "loss": 4.819, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.873414777480849e-05, | |
| "loss": 4.8106, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.872576182729797e-05, | |
| "loss": 4.7982, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.871737587978745e-05, | |
| "loss": 4.8001, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.870898993227693e-05, | |
| "loss": 4.7934, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.87006039847664e-05, | |
| "loss": 4.7726, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.869221803725588e-05, | |
| "loss": 4.785, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.868383208974536e-05, | |
| "loss": 4.7882, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.867544614223484e-05, | |
| "loss": 4.7692, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.866706019472432e-05, | |
| "loss": 4.785, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.86586742472138e-05, | |
| "loss": 4.7886, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.865028829970329e-05, | |
| "loss": 4.7729, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.864190235219277e-05, | |
| "loss": 4.7658, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.863351640468225e-05, | |
| "loss": 4.7677, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.862513045717173e-05, | |
| "loss": 4.767, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.861674450966121e-05, | |
| "loss": 4.7616, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.860835856215069e-05, | |
| "loss": 4.7565, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.859997261464017e-05, | |
| "loss": 4.756, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.859158666712964e-05, | |
| "loss": 4.7509, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8583233477226585e-05, | |
| "loss": 4.7549, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8574847529716065e-05, | |
| "loss": 4.7551, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8566461582205545e-05, | |
| "loss": 4.7543, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8558075634695025e-05, | |
| "loss": 4.7496, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8549689687184505e-05, | |
| "loss": 4.7474, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.854132011847772e-05, | |
| "loss": 4.7419, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.85329341709672e-05, | |
| "loss": 4.7387, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.852454822345668e-05, | |
| "loss": 4.7425, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.851616227594616e-05, | |
| "loss": 4.723, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.850777632843564e-05, | |
| "loss": 4.7354, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8499390380925114e-05, | |
| "loss": 4.7177, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8491004433414594e-05, | |
| "loss": 4.722, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8482618485904074e-05, | |
| "loss": 4.7319, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.847424891719729e-05, | |
| "loss": 4.7296, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.846586296968676e-05, | |
| "loss": 4.715, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.845747702217624e-05, | |
| "loss": 4.707, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.844910745346946e-05, | |
| "loss": 4.7239, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.844072150595894e-05, | |
| "loss": 4.7108, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.843233555844842e-05, | |
| "loss": 4.7093, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.84239496109379e-05, | |
| "loss": 4.7055, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.841556366342738e-05, | |
| "loss": 4.6932, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.840717771591686e-05, | |
| "loss": 4.6894, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.839879176840634e-05, | |
| "loss": 4.6913, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.839040582089582e-05, | |
| "loss": 4.6954, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.83820198733853e-05, | |
| "loss": 4.6896, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.837363392587478e-05, | |
| "loss": 4.6924, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.836524797836426e-05, | |
| "loss": 4.6844, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.835686203085374e-05, | |
| "loss": 4.6879, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.834849246214695e-05, | |
| "loss": 4.6993, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.834010651463643e-05, | |
| "loss": 4.6794, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.833172056712591e-05, | |
| "loss": 4.6801, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.832333461961539e-05, | |
| "loss": 4.6698, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8314948672104874e-05, | |
| "loss": 4.6902, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.830657910339808e-05, | |
| "loss": 4.6616, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.829819315588756e-05, | |
| "loss": 4.6838, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.828982358718077e-05, | |
| "loss": 4.6604, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.828143763967025e-05, | |
| "loss": 4.6699, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.827305169215973e-05, | |
| "loss": 4.6682, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.826466574464921e-05, | |
| "loss": 4.6641, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.825627979713869e-05, | |
| "loss": 4.6559, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.824789384962817e-05, | |
| "loss": 4.6559, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.823950790211765e-05, | |
| "loss": 4.6452, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8231121954607125e-05, | |
| "loss": 4.6602, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.822273600709661e-05, | |
| "loss": 4.6511, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.821435005958609e-05, | |
| "loss": 4.6454, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.82059804908793e-05, | |
| "loss": 4.6438, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.819759454336878e-05, | |
| "loss": 4.6443, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.818920859585826e-05, | |
| "loss": 4.6261, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.818082264834774e-05, | |
| "loss": 4.65, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.817245307964095e-05, | |
| "loss": 4.6354, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.816406713213043e-05, | |
| "loss": 4.6349, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.815568118461991e-05, | |
| "loss": 4.6435, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.814729523710939e-05, | |
| "loss": 4.6292, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.813890928959887e-05, | |
| "loss": 4.6431, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.813053972089208e-05, | |
| "loss": 4.6458, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8122153773381565e-05, | |
| "loss": 4.6363, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8113767825871045e-05, | |
| "loss": 4.6383, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8105381878360525e-05, | |
| "loss": 4.6205, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8096995930850005e-05, | |
| "loss": 4.6156, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8088609983339485e-05, | |
| "loss": 4.6269, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8080240414632694e-05, | |
| "loss": 4.6279, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8071854467122174e-05, | |
| "loss": 4.6204, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8063468519611654e-05, | |
| "loss": 4.6159, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8055082572101134e-05, | |
| "loss": 4.5988, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8046696624590614e-05, | |
| "loss": 4.61, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8038310677080094e-05, | |
| "loss": 4.6138, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8029924729569574e-05, | |
| "loss": 4.6127, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8021538782059054e-05, | |
| "loss": 4.6194, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8013152834548534e-05, | |
| "loss": 4.6174, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.800478326584174e-05, | |
| "loss": 4.607, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.799639731833123e-05, | |
| "loss": 4.5994, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.798802774962444e-05, | |
| "loss": 4.5926, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.797964180211392e-05, | |
| "loss": 4.599, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.79712558546034e-05, | |
| "loss": 4.6017, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.796286990709288e-05, | |
| "loss": 4.5893, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.795448395958236e-05, | |
| "loss": 4.5903, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.794611439087557e-05, | |
| "loss": 4.5981, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.793772844336505e-05, | |
| "loss": 4.6013, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792934249585453e-05, | |
| "loss": 4.5899, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792095654834401e-05, | |
| "loss": 4.596, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.791257060083349e-05, | |
| "loss": 4.5978, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.790418465332297e-05, | |
| "loss": 4.5883, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.789579870581245e-05, | |
| "loss": 4.585, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.788741275830193e-05, | |
| "loss": 4.5951, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7879043189595136e-05, | |
| "loss": 4.5802, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7870657242084616e-05, | |
| "loss": 4.5931, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.786228767337783e-05, | |
| "loss": 4.573, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.785390172586731e-05, | |
| "loss": 4.5631, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7845515778356785e-05, | |
| "loss": 4.5704, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7837129830846265e-05, | |
| "loss": 4.5787, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.782876026213948e-05, | |
| "loss": 4.572, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.782037431462896e-05, | |
| "loss": 4.5589, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7811988367118434e-05, | |
| "loss": 4.559, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.780360241960792e-05, | |
| "loss": 4.5672, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.77952164720974e-05, | |
| "loss": 4.5783, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.778683052458688e-05, | |
| "loss": 4.565, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.777844457707636e-05, | |
| "loss": 4.5736, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.777005862956584e-05, | |
| "loss": 4.5694, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.776168906085905e-05, | |
| "loss": 4.5658, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.775331949215226e-05, | |
| "loss": 4.5613, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.774493354464174e-05, | |
| "loss": 4.5578, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.773654759713122e-05, | |
| "loss": 4.5541, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.77281616496207e-05, | |
| "loss": 4.5645, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.771977570211018e-05, | |
| "loss": 4.5491, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.771138975459966e-05, | |
| "loss": 4.5512, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.770300380708914e-05, | |
| "loss": 4.5553, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.769461785957862e-05, | |
| "loss": 4.5625, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7686248290871835e-05, | |
| "loss": 4.5395, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7677862343361315e-05, | |
| "loss": 4.543, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7669476395850795e-05, | |
| "loss": 4.5655, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7661090448340274e-05, | |
| "loss": 4.5387, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7652720879633484e-05, | |
| "loss": 4.5605, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.764435131092669e-05, | |
| "loss": 4.5427, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.763596536341617e-05, | |
| "loss": 4.5539, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.762757941590565e-05, | |
| "loss": 4.5414, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.761919346839513e-05, | |
| "loss": 4.537, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.761080752088461e-05, | |
| "loss": 4.5451, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.760242157337409e-05, | |
| "loss": 4.5409, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.759405200466731e-05, | |
| "loss": 4.5498, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.758566605715679e-05, | |
| "loss": 4.5441, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.757728010964627e-05, | |
| "loss": 4.5433, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.756889416213575e-05, | |
| "loss": 4.5408, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.756052459342896e-05, | |
| "loss": 4.5256, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.755213864591844e-05, | |
| "loss": 4.536, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.754375269840792e-05, | |
| "loss": 4.5226, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.75353667508974e-05, | |
| "loss": 4.5275, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7526997182190606e-05, | |
| "loss": 4.5307, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7518611234680086e-05, | |
| "loss": 4.518, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7510225287169566e-05, | |
| "loss": 4.5247, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7501839339659046e-05, | |
| "loss": 4.5218, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.488372802734375, | |
| "eval_runtime": 546.2477, | |
| "eval_samples_per_second": 698.568, | |
| "eval_steps_per_second": 21.831, | |
| "step": 152638 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7493453392148526e-05, | |
| "loss": 4.5335, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.748508382344174e-05, | |
| "loss": 4.5308, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.747669787593122e-05, | |
| "loss": 4.5211, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.74683119284207e-05, | |
| "loss": 4.5176, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.745992598091018e-05, | |
| "loss": 4.5188, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.745155641220339e-05, | |
| "loss": 4.5028, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.744317046469287e-05, | |
| "loss": 4.5165, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.743480089598608e-05, | |
| "loss": 4.5148, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.742641494847556e-05, | |
| "loss": 4.4999, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.741802900096504e-05, | |
| "loss": 4.5196, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.740964305345452e-05, | |
| "loss": 4.5243, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7401257105944e-05, | |
| "loss": 4.5079, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.739287115843348e-05, | |
| "loss": 4.5009, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7384485210922966e-05, | |
| "loss": 4.5046, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7376099263412446e-05, | |
| "loss": 4.508, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7367729694705656e-05, | |
| "loss": 4.5042, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7359343747195135e-05, | |
| "loss": 4.4989, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7350957799684615e-05, | |
| "loss": 4.4989, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7342571852174095e-05, | |
| "loss": 4.4999, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.733418590466357e-05, | |
| "loss": 4.5027, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7325816335956784e-05, | |
| "loss": 4.5034, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7317430388446264e-05, | |
| "loss": 4.5055, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.730904444093574e-05, | |
| "loss": 4.5033, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.730065849342522e-05, | |
| "loss": 4.4952, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7292288924718433e-05, | |
| "loss": 4.5005, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.728390297720791e-05, | |
| "loss": 4.4946, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.727551702969739e-05, | |
| "loss": 4.4982, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.726713108218687e-05, | |
| "loss": 4.4837, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.725876151348009e-05, | |
| "loss": 4.4946, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.725037556596957e-05, | |
| "loss": 4.4793, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.724198961845904e-05, | |
| "loss": 4.4858, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.723360367094852e-05, | |
| "loss": 4.4955, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7225217723438e-05, | |
| "loss": 4.4939, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.721684815473121e-05, | |
| "loss": 4.4818, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.720846220722069e-05, | |
| "loss": 4.476, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.720007625971017e-05, | |
| "loss": 4.498, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.719169031219965e-05, | |
| "loss": 4.4778, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.718330436468914e-05, | |
| "loss": 4.4861, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.717491841717862e-05, | |
| "loss": 4.4812, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.71665324696681e-05, | |
| "loss": 4.4669, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.715814652215758e-05, | |
| "loss": 4.4645, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.714977695345079e-05, | |
| "loss": 4.4655, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.714139100594027e-05, | |
| "loss": 4.4806, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7133021437233476e-05, | |
| "loss": 4.4716, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7124635489722956e-05, | |
| "loss": 4.4708, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7116249542212436e-05, | |
| "loss": 4.4651, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7107863594701916e-05, | |
| "loss": 4.4686, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7099477647191396e-05, | |
| "loss": 4.4834, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7091091699680876e-05, | |
| "loss": 4.4632, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.708272213097409e-05, | |
| "loss": 4.4633, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.707433618346357e-05, | |
| "loss": 4.459, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.706595023595305e-05, | |
| "loss": 4.4801, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.705756428844253e-05, | |
| "loss": 4.4513, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.704917834093201e-05, | |
| "loss": 4.4739, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.704079239342149e-05, | |
| "loss": 4.4501, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.70324228247147e-05, | |
| "loss": 4.4635, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.702403687720418e-05, | |
| "loss": 4.4644, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.701565092969366e-05, | |
| "loss": 4.453, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.700726498218314e-05, | |
| "loss": 4.4572, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.699887903467262e-05, | |
| "loss": 4.4511, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.69904930871621e-05, | |
| "loss": 4.4369, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6982107139651573e-05, | |
| "loss": 4.4557, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.697373757094479e-05, | |
| "loss": 4.4539, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6965351623434276e-05, | |
| "loss": 4.4444, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6956965675923756e-05, | |
| "loss": 4.4472, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.694857972841323e-05, | |
| "loss": 4.4476, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.694019378090271e-05, | |
| "loss": 4.429, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.693180783339219e-05, | |
| "loss": 4.4563, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.692342188588167e-05, | |
| "loss": 4.437, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.691503593837115e-05, | |
| "loss": 4.441, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.690666636966436e-05, | |
| "loss": 4.4512, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.689828042215384e-05, | |
| "loss": 4.4379, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.688989447464332e-05, | |
| "loss": 4.4489, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.68815085271328e-05, | |
| "loss": 4.4582, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.687312257962228e-05, | |
| "loss": 4.4445, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.686473663211176e-05, | |
| "loss": 4.4521, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6856367063404974e-05, | |
| "loss": 4.4343, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6847981115894454e-05, | |
| "loss": 4.4262, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6839595168383934e-05, | |
| "loss": 4.4387, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6831209220873414e-05, | |
| "loss": 4.4461, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6822823273362894e-05, | |
| "loss": 4.437, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6814437325852373e-05, | |
| "loss": 4.4288, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6806051378341853e-05, | |
| "loss": 4.4202, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.679768180963506e-05, | |
| "loss": 4.4244, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.678929586212454e-05, | |
| "loss": 4.4274, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.678090991461402e-05, | |
| "loss": 4.431, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.67725239671035e-05, | |
| "loss": 4.4403, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.676415439839671e-05, | |
| "loss": 4.4375, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.675576845088619e-05, | |
| "loss": 4.4276, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.674738250337568e-05, | |
| "loss": 4.423, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.673899655586516e-05, | |
| "loss": 4.4165, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.673061060835464e-05, | |
| "loss": 4.4228, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.672222466084412e-05, | |
| "loss": 4.4209, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.671383871333359e-05, | |
| "loss": 4.4166, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.670545276582307e-05, | |
| "loss": 4.4163, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.669706681831255e-05, | |
| "loss": 4.426, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.668868087080203e-05, | |
| "loss": 4.424, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.668031130209524e-05, | |
| "loss": 4.4243, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.667192535458472e-05, | |
| "loss": 4.4239, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.66635394070742e-05, | |
| "loss": 4.4267, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.665515345956368e-05, | |
| "loss": 4.4198, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6646783890856896e-05, | |
| "loss": 4.4136, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6638397943346376e-05, | |
| "loss": 4.4292, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.663002837463959e-05, | |
| "loss": 4.4085, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6621642427129065e-05, | |
| "loss": 4.4311, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6613256479618545e-05, | |
| "loss": 4.4025, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6604870532108025e-05, | |
| "loss": 4.3963, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6596484584597505e-05, | |
| "loss": 4.4052, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6588098637086985e-05, | |
| "loss": 4.4194, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6579712689576465e-05, | |
| "loss": 4.4091, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6571326742065945e-05, | |
| "loss": 4.3951, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6562940794555425e-05, | |
| "loss": 4.3962, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6554554847044905e-05, | |
| "loss": 4.4031, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6546168899534385e-05, | |
| "loss": 4.418, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6537782952023865e-05, | |
| "loss": 4.4033, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.652942976212081e-05, | |
| "loss": 4.4126, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.652104381461029e-05, | |
| "loss": 4.4087, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.651265786709977e-05, | |
| "loss": 4.406, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.650427191958925e-05, | |
| "loss": 4.4093, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.649588597207873e-05, | |
| "loss": 4.3976, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.648750002456821e-05, | |
| "loss": 4.3992, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.647911407705769e-05, | |
| "loss": 4.4079, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.647072812954717e-05, | |
| "loss": 4.3909, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.646234218203665e-05, | |
| "loss": 4.3972, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.645397261332986e-05, | |
| "loss": 4.401, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.644558666581934e-05, | |
| "loss": 4.4045, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.643720071830882e-05, | |
| "loss": 4.3891, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6428831149602034e-05, | |
| "loss": 4.3887, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6420445202091514e-05, | |
| "loss": 4.4135, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6412059254580994e-05, | |
| "loss": 4.3912, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6403673307070474e-05, | |
| "loss": 4.4041, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.639528735955995e-05, | |
| "loss": 4.3949, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.638690141204943e-05, | |
| "loss": 4.4027, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.637851546453891e-05, | |
| "loss": 4.3912, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.637012951702839e-05, | |
| "loss": 4.3886, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6361759948321596e-05, | |
| "loss": 4.3994, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6353374000811076e-05, | |
| "loss": 4.3889, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6344988053300556e-05, | |
| "loss": 4.4018, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.633661848459377e-05, | |
| "loss": 4.399, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.632823253708325e-05, | |
| "loss": 4.3967, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.631984658957273e-05, | |
| "loss": 4.3914, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.631146064206221e-05, | |
| "loss": 4.3834, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.630307469455169e-05, | |
| "loss": 4.3895, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.629468874704117e-05, | |
| "loss": 4.3791, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.628630279953065e-05, | |
| "loss": 4.3838, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.627791685202013e-05, | |
| "loss": 4.3844, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.626953090450961e-05, | |
| "loss": 4.3776, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.626114495699909e-05, | |
| "loss": 4.3796, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.62527753882923e-05, | |
| "loss": 4.3799, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.354189872741699, | |
| "eval_runtime": 546.5547, | |
| "eval_samples_per_second": 698.175, | |
| "eval_steps_per_second": 21.818, | |
| "step": 228957 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.624438944078178e-05, | |
| "loss": 4.3894, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.623600349327126e-05, | |
| "loss": 4.3938, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.622761754576074e-05, | |
| "loss": 4.3809, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.621923159825022e-05, | |
| "loss": 4.3769, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.62108456507397e-05, | |
| "loss": 4.3806, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.620245970322919e-05, | |
| "loss": 4.362, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.619407375571867e-05, | |
| "loss": 4.3777, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.618568780820814e-05, | |
| "loss": 4.379, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.617730186069762e-05, | |
| "loss": 4.3667, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6168932291990836e-05, | |
| "loss": 4.3811, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.616054634448031e-05, | |
| "loss": 4.3879, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.615216039696979e-05, | |
| "loss": 4.3686, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.614377444945927e-05, | |
| "loss": 4.3683, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.613538850194875e-05, | |
| "loss": 4.3707, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.612700255443823e-05, | |
| "loss": 4.3748, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.611861660692771e-05, | |
| "loss": 4.3688, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.611023065941719e-05, | |
| "loss": 4.3635, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.610184471190667e-05, | |
| "loss": 4.3654, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6093475143199885e-05, | |
| "loss": 4.3667, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6085089195689365e-05, | |
| "loss": 4.3738, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6076703248178845e-05, | |
| "loss": 4.3706, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6068317300668325e-05, | |
| "loss": 4.3753, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6059931353157805e-05, | |
| "loss": 4.3721, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6051545405647285e-05, | |
| "loss": 4.3666, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6043159458136764e-05, | |
| "loss": 4.3652, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6034773510626244e-05, | |
| "loss": 4.3653, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6026387563115724e-05, | |
| "loss": 4.3701, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.60180016156052e-05, | |
| "loss": 4.3575, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6009632046898413e-05, | |
| "loss": 4.3661, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6001246099387893e-05, | |
| "loss": 4.3523, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.599286015187737e-05, | |
| "loss": 4.3565, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.598449058317059e-05, | |
| "loss": 4.3703, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.597610463566007e-05, | |
| "loss": 4.3659, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.596771868814955e-05, | |
| "loss": 4.3581, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.595933274063903e-05, | |
| "loss": 4.352, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.59509467931285e-05, | |
| "loss": 4.3663, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.594256084561798e-05, | |
| "loss": 4.3538, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.593417489810746e-05, | |
| "loss": 4.3646, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.592578895059694e-05, | |
| "loss": 4.3548, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.591740300308642e-05, | |
| "loss": 4.3442, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.59090170555759e-05, | |
| "loss": 4.3437, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.590064748686911e-05, | |
| "loss": 4.3455, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.589226153935859e-05, | |
| "loss": 4.3555, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.588387559184808e-05, | |
| "loss": 4.3468, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.587548964433756e-05, | |
| "loss": 4.3488, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.586712007563077e-05, | |
| "loss": 4.3452, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.585873412812025e-05, | |
| "loss": 4.3494, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.585034818060973e-05, | |
| "loss": 4.3605, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.584196223309921e-05, | |
| "loss": 4.342, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.583357628558869e-05, | |
| "loss": 4.3423, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.582519033807817e-05, | |
| "loss": 4.3463, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.581680439056765e-05, | |
| "loss": 4.3567, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5808418443057127e-05, | |
| "loss": 4.3344, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5800048874350336e-05, | |
| "loss": 4.3516, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5791662926839816e-05, | |
| "loss": 4.3329, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5783276979329296e-05, | |
| "loss": 4.3415, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5774891031818776e-05, | |
| "loss": 4.3468, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.576653784191572e-05, | |
| "loss": 4.3334, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.57581518944052e-05, | |
| "loss": 4.347, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.574976594689468e-05, | |
| "loss": 4.3389, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.574137999938416e-05, | |
| "loss": 4.3219, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.573299405187364e-05, | |
| "loss": 4.3363, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.572462448316685e-05, | |
| "loss": 4.337, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.571623853565633e-05, | |
| "loss": 4.3356, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.570785258814581e-05, | |
| "loss": 4.3327, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.569946664063529e-05, | |
| "loss": 4.3332, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.569108069312477e-05, | |
| "loss": 4.3157, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.568269474561425e-05, | |
| "loss": 4.3427, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5674325176907465e-05, | |
| "loss": 4.3236, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5665939229396945e-05, | |
| "loss": 4.3279, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5657569660690154e-05, | |
| "loss": 4.3391, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5649183713179634e-05, | |
| "loss": 4.3266, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5640797765669114e-05, | |
| "loss": 4.3361, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5632411818158594e-05, | |
| "loss": 4.344, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5624025870648074e-05, | |
| "loss": 4.3349, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.561565630194128e-05, | |
| "loss": 4.3427, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.560727035443076e-05, | |
| "loss": 4.3235, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.559888440692024e-05, | |
| "loss": 4.3186, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.559049845940972e-05, | |
| "loss": 4.3268, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.55821125118992e-05, | |
| "loss": 4.3385, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.557372656438868e-05, | |
| "loss": 4.3271, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.556534061687816e-05, | |
| "loss": 4.3198, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.555695466936764e-05, | |
| "loss": 4.3105, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.554858510066086e-05, | |
| "loss": 4.3204, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.554019915315033e-05, | |
| "loss": 4.3183, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.553181320563981e-05, | |
| "loss": 4.3221, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.552342725812929e-05, | |
| "loss": 4.3288, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.551504131061877e-05, | |
| "loss": 4.334, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.550665536310825e-05, | |
| "loss": 4.3216, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.549826941559773e-05, | |
| "loss": 4.3195, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.548989984689094e-05, | |
| "loss": 4.309, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.548151389938042e-05, | |
| "loss": 4.3138, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.54731279518699e-05, | |
| "loss": 4.3189, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.546474200435939e-05, | |
| "loss": 4.3126, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.545635605684887e-05, | |
| "loss": 4.3085, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.544797010933835e-05, | |
| "loss": 4.3228, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.543958416182783e-05, | |
| "loss": 4.3159, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.543119821431731e-05, | |
| "loss": 4.3262, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5422828645610516e-05, | |
| "loss": 4.3164, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5414442698099996e-05, | |
| "loss": 4.3242, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5406073129393205e-05, | |
| "loss": 4.3154, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5397687181882685e-05, | |
| "loss": 4.31, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5389301234372165e-05, | |
| "loss": 4.328, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5380915286861645e-05, | |
| "loss": 4.3068, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5372529339351125e-05, | |
| "loss": 4.3303, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5364143391840605e-05, | |
| "loss": 4.3016, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5355757444330085e-05, | |
| "loss": 4.2976, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.534737149681957e-05, | |
| "loss": 4.3033, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.533900192811278e-05, | |
| "loss": 4.3188, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.533063235940599e-05, | |
| "loss": 4.3097, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.532224641189547e-05, | |
| "loss": 4.2946, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.531386046438495e-05, | |
| "loss": 4.2941, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.530547451687443e-05, | |
| "loss": 4.3027, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.529710494816764e-05, | |
| "loss": 4.3206, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.528871900065712e-05, | |
| "loss": 4.3046, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.52803330531466e-05, | |
| "loss": 4.3176, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.527194710563608e-05, | |
| "loss": 4.3086, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.526356115812556e-05, | |
| "loss": 4.3094, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.525517521061504e-05, | |
| "loss": 4.3105, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.524678926310452e-05, | |
| "loss": 4.3001, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5238403315594e-05, | |
| "loss": 4.3013, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5230033746887215e-05, | |
| "loss": 4.3081, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5221647799376694e-05, | |
| "loss": 4.2952, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5213278230669904e-05, | |
| "loss": 4.3016, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5204892283159384e-05, | |
| "loss": 4.3066, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.519652271445259e-05, | |
| "loss": 4.3043, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.518813676694207e-05, | |
| "loss": 4.2946, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.517975081943155e-05, | |
| "loss": 4.2942, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.517136487192103e-05, | |
| "loss": 4.3154, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.516297892441051e-05, | |
| "loss": 4.2976, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.515459297689999e-05, | |
| "loss": 4.3084, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.514620702938947e-05, | |
| "loss": 4.3015, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.513782108187895e-05, | |
| "loss": 4.3068, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.512943513436843e-05, | |
| "loss": 4.2982, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.512104918685791e-05, | |
| "loss": 4.2981, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.511266323934739e-05, | |
| "loss": 4.3029, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.510427729183687e-05, | |
| "loss": 4.2965, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.509590772313008e-05, | |
| "loss": 4.3116, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.508752177561956e-05, | |
| "loss": 4.3008, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.507913582810904e-05, | |
| "loss": 4.3056, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.507076625940225e-05, | |
| "loss": 4.302, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.506238031189173e-05, | |
| "loss": 4.2914, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.505399436438121e-05, | |
| "loss": 4.2963, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.50456084168707e-05, | |
| "loss": 4.2877, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.503722246936018e-05, | |
| "loss": 4.293, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.502883652184966e-05, | |
| "loss": 4.2916, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.502045057433914e-05, | |
| "loss": 4.291, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.501206462682862e-05, | |
| "loss": 4.2846, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5003695058121826e-05, | |
| "loss": 4.2923, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.270674228668213, | |
| "eval_runtime": 545.1308, | |
| "eval_samples_per_second": 699.999, | |
| "eval_steps_per_second": 21.875, | |
| "step": 305276 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4995309110611306e-05, | |
| "loss": 4.2996, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4986939541904515e-05, | |
| "loss": 4.3004, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4978553594393995e-05, | |
| "loss": 4.2917, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4970167646883475e-05, | |
| "loss": 4.2845, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4961781699372955e-05, | |
| "loss": 4.2946, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4953395751862435e-05, | |
| "loss": 4.2724, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4945009804351915e-05, | |
| "loss": 4.2906, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4936623856841395e-05, | |
| "loss": 4.2865, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.492823790933088e-05, | |
| "loss": 4.28, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.491986834062409e-05, | |
| "loss": 4.2908, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.491148239311357e-05, | |
| "loss": 4.2999, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.490309644560305e-05, | |
| "loss": 4.2791, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.489471049809253e-05, | |
| "loss": 4.2842, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4886324550582003e-05, | |
| "loss": 4.2821, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4877938603071483e-05, | |
| "loss": 4.2884, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.486955265556096e-05, | |
| "loss": 4.2777, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.486116670805044e-05, | |
| "loss": 4.2773, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.485279713934365e-05, | |
| "loss": 4.2781, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.484441119183313e-05, | |
| "loss": 4.2789, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.483602524432262e-05, | |
| "loss": 4.2904, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.48276392968121e-05, | |
| "loss": 4.2828, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.481925334930158e-05, | |
| "loss": 4.289, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.481086740179106e-05, | |
| "loss": 4.2888, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.480249783308427e-05, | |
| "loss": 4.2807, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.479411188557375e-05, | |
| "loss": 4.2772, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.478574231686696e-05, | |
| "loss": 4.2789, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.477735636935644e-05, | |
| "loss": 4.2853, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.476897042184592e-05, | |
| "loss": 4.2729, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.47605844743354e-05, | |
| "loss": 4.2781, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.475219852682488e-05, | |
| "loss": 4.2747, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.474381257931436e-05, | |
| "loss": 4.2666, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.473542663180384e-05, | |
| "loss": 4.2829, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.472705706309705e-05, | |
| "loss": 4.2869, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.471867111558653e-05, | |
| "loss": 4.2762, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.471028516807601e-05, | |
| "loss": 4.2683, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.470189922056549e-05, | |
| "loss": 4.2774, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.46935296518587e-05, | |
| "loss": 4.2767, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.468514370434818e-05, | |
| "loss": 4.2769, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.467675775683766e-05, | |
| "loss": 4.2707, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.466837180932714e-05, | |
| "loss": 4.2622, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.465998586181662e-05, | |
| "loss": 4.2614, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.46515999143061e-05, | |
| "loss": 4.2622, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.464323034559931e-05, | |
| "loss": 4.2743, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.463484439808879e-05, | |
| "loss": 4.2677, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.462645845057827e-05, | |
| "loss": 4.2671, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.461807250306775e-05, | |
| "loss": 4.2647, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.460968655555724e-05, | |
| "loss": 4.2664, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.460130060804672e-05, | |
| "loss": 4.2827, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.459291466053619e-05, | |
| "loss": 4.2628, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4584545091829406e-05, | |
| "loss": 4.2615, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4576159144318886e-05, | |
| "loss": 4.2679, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4567773196808366e-05, | |
| "loss": 4.2713, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.455938724929784e-05, | |
| "loss": 4.2589, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.455100130178732e-05, | |
| "loss": 4.2677, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.45426153542768e-05, | |
| "loss": 4.2589, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.453422940676628e-05, | |
| "loss": 4.262, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.452585983805949e-05, | |
| "loss": 4.2673, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4517473890548975e-05, | |
| "loss": 4.2523, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4509087943038455e-05, | |
| "loss": 4.2702, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4500701995527935e-05, | |
| "loss": 4.2569, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4492316048017415e-05, | |
| "loss": 4.2456, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4483930100506895e-05, | |
| "loss": 4.2535, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4475544152996375e-05, | |
| "loss": 4.2603, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4467158205485855e-05, | |
| "loss": 4.2632, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4458788636779064e-05, | |
| "loss": 4.2538, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4450402689268544e-05, | |
| "loss": 4.2588, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4442016741758024e-05, | |
| "loss": 4.2344, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.443364717305123e-05, | |
| "loss": 4.2659, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.442526122554071e-05, | |
| "loss": 4.2481, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.441687527803019e-05, | |
| "loss": 4.2494, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.440848933051967e-05, | |
| "loss": 4.2633, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.440010338300916e-05, | |
| "loss": 4.2516, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.439171743549864e-05, | |
| "loss": 4.2566, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.438334786679185e-05, | |
| "loss": 4.2659, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.437496191928133e-05, | |
| "loss": 4.2594, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.436657597177081e-05, | |
| "loss": 4.2633, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.435819002426029e-05, | |
| "loss": 4.2492, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.434980407674977e-05, | |
| "loss": 4.2418, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.434141812923925e-05, | |
| "loss": 4.2535, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.433303218172873e-05, | |
| "loss": 4.261, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.43246462342182e-05, | |
| "loss": 4.2512, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.431626028670768e-05, | |
| "loss": 4.247, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.43078907180009e-05, | |
| "loss": 4.2344, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.429950477049038e-05, | |
| "loss": 4.2475, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.429111882297986e-05, | |
| "loss": 4.2416, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.428273287546934e-05, | |
| "loss": 4.2519, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.427434692795882e-05, | |
| "loss": 4.2501, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.426599373805576e-05, | |
| "loss": 4.2618, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.425760779054524e-05, | |
| "loss": 4.2449, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.424922184303472e-05, | |
| "loss": 4.2483, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.42408358955242e-05, | |
| "loss": 4.2342, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4232449948013675e-05, | |
| "loss": 4.2417, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4224064000503155e-05, | |
| "loss": 4.2453, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4215678052992635e-05, | |
| "loss": 4.2382, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4207292105482115e-05, | |
| "loss": 4.238, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.419893891557907e-05, | |
| "loss": 4.2513, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.419055296806855e-05, | |
| "loss": 4.2391, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.418216702055803e-05, | |
| "loss": 4.2594, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.41737810730475e-05, | |
| "loss": 4.2419, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4165411504340716e-05, | |
| "loss": 4.2506, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4157025556830196e-05, | |
| "loss": 4.2414, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4148639609319676e-05, | |
| "loss": 4.2389, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.414025366180915e-05, | |
| "loss": 4.2556, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4131884093102365e-05, | |
| "loss": 4.2361, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4123498145591845e-05, | |
| "loss": 4.2576, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.411511219808132e-05, | |
| "loss": 4.2332, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4106726250570805e-05, | |
| "loss": 4.2303, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4098340303060284e-05, | |
| "loss": 4.2295, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4089954355549764e-05, | |
| "loss": 4.2463, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.408160116564671e-05, | |
| "loss": 4.2427, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.407321521813619e-05, | |
| "loss": 4.2296, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.406482927062567e-05, | |
| "loss": 4.2225, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.405644332311514e-05, | |
| "loss": 4.2337, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.404805737560462e-05, | |
| "loss": 4.2507, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.40396714280941e-05, | |
| "loss": 4.2335, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.403130185938732e-05, | |
| "loss": 4.248, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.402291591187679e-05, | |
| "loss": 4.2387, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.401452996436627e-05, | |
| "loss": 4.2415, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.400614401685576e-05, | |
| "loss": 4.2428, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.399775806934524e-05, | |
| "loss": 4.2301, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.398938850063845e-05, | |
| "loss": 4.2313, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.398100255312793e-05, | |
| "loss": 4.2374, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.397261660561741e-05, | |
| "loss": 4.229, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.396423065810689e-05, | |
| "loss": 4.2329, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.395584471059637e-05, | |
| "loss": 4.2372, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.394745876308585e-05, | |
| "loss": 4.2362, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.393907281557533e-05, | |
| "loss": 4.2272, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.393068686806481e-05, | |
| "loss": 4.2286, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.392230092055429e-05, | |
| "loss": 4.2431, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.391391497304377e-05, | |
| "loss": 4.2277, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.390552902553325e-05, | |
| "loss": 4.2414, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.389714307802273e-05, | |
| "loss": 4.2358, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3888773509315936e-05, | |
| "loss": 4.2373, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.388040394060915e-05, | |
| "loss": 4.2323, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.387201799309863e-05, | |
| "loss": 4.2289, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.386363204558811e-05, | |
| "loss": 4.2358, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.385524609807759e-05, | |
| "loss": 4.2288, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.384686015056707e-05, | |
| "loss": 4.244, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.383847420305655e-05, | |
| "loss": 4.2376, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.383008825554603e-05, | |
| "loss": 4.2386, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3821702308035505e-05, | |
| "loss": 4.2361, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.381333273932872e-05, | |
| "loss": 4.2262, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.38049467918182e-05, | |
| "loss": 4.2285, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.379656084430768e-05, | |
| "loss": 4.2224, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.378817489679716e-05, | |
| "loss": 4.2261, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.377978894928664e-05, | |
| "loss": 4.2266, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.377140300177612e-05, | |
| "loss": 4.2252, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.37630170542656e-05, | |
| "loss": 4.2148, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.375463110675508e-05, | |
| "loss": 4.2284, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.213342666625977, | |
| "eval_runtime": 543.7311, | |
| "eval_samples_per_second": 701.801, | |
| "eval_steps_per_second": 21.932, | |
| "step": 381595 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.374626153804829e-05, | |
| "loss": 4.24, | |
| "step": 381952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.373787559053777e-05, | |
| "loss": 4.2331, | |
| "step": 382464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.372948964302725e-05, | |
| "loss": 4.2285, | |
| "step": 382976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.372110369551673e-05, | |
| "loss": 4.2234, | |
| "step": 383488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.371271774800621e-05, | |
| "loss": 4.2237, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.370433180049569e-05, | |
| "loss": 4.2124, | |
| "step": 384512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.369594585298517e-05, | |
| "loss": 4.2258, | |
| "step": 385024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.368755990547465e-05, | |
| "loss": 4.2155, | |
| "step": 385536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.367917395796413e-05, | |
| "loss": 4.2218, | |
| "step": 386048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.367078801045361e-05, | |
| "loss": 4.2282, | |
| "step": 386560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.366240206294309e-05, | |
| "loss": 4.2312, | |
| "step": 387072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.3654016115432576e-05, | |
| "loss": 4.2147, | |
| "step": 387584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.364563016792205e-05, | |
| "loss": 4.2231, | |
| "step": 388096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.363724422041153e-05, | |
| "loss": 4.2184, | |
| "step": 388608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.362885827290101e-05, | |
| "loss": 4.2233, | |
| "step": 389120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.3620488704194225e-05, | |
| "loss": 4.215, | |
| "step": 389632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.36121027566837e-05, | |
| "loss": 4.2129, | |
| "step": 390144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.360371680917318e-05, | |
| "loss": 4.2159, | |
| "step": 390656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.359533086166266e-05, | |
| "loss": 4.22, | |
| "step": 391168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.358694491415214e-05, | |
| "loss": 4.2275, | |
| "step": 391680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.357855896664162e-05, | |
| "loss": 4.2192, | |
| "step": 392192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.35701730191311e-05, | |
| "loss": 4.2302, | |
| "step": 392704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.356178707162058e-05, | |
| "loss": 4.2226, | |
| "step": 393216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.355341750291379e-05, | |
| "loss": 4.2216, | |
| "step": 393728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.354503155540327e-05, | |
| "loss": 4.2149, | |
| "step": 394240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.353664560789275e-05, | |
| "loss": 4.2163, | |
| "step": 394752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.352825966038223e-05, | |
| "loss": 4.2256, | |
| "step": 395264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.351989009167544e-05, | |
| "loss": 4.2125, | |
| "step": 395776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.351150414416492e-05, | |
| "loss": 4.2184, | |
| "step": 396288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.35031181966544e-05, | |
| "loss": 4.2121, | |
| "step": 396800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.349474862794761e-05, | |
| "loss": 4.2043, | |
| "step": 397312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.348636268043709e-05, | |
| "loss": 4.227, | |
| "step": 397824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.347797673292657e-05, | |
| "loss": 4.2204, | |
| "step": 398336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.346959078541605e-05, | |
| "loss": 4.2202, | |
| "step": 398848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.346120483790553e-05, | |
| "loss": 4.2049, | |
| "step": 399360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.345281889039501e-05, | |
| "loss": 4.2214, | |
| "step": 399872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.344444932168823e-05, | |
| "loss": 4.2121, | |
| "step": 400384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.343606337417771e-05, | |
| "loss": 4.2191, | |
| "step": 400896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.342767742666719e-05, | |
| "loss": 4.212, | |
| "step": 401408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.341929147915667e-05, | |
| "loss": 4.202, | |
| "step": 401920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3410921910449876e-05, | |
| "loss": 4.202, | |
| "step": 402432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3402535962939356e-05, | |
| "loss": 4.1998, | |
| "step": 402944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3394150015428836e-05, | |
| "loss": 4.2196, | |
| "step": 403456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3385764067918316e-05, | |
| "loss": 4.2074, | |
| "step": 403968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3377378120407796e-05, | |
| "loss": 4.2034, | |
| "step": 404480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3368992172897276e-05, | |
| "loss": 4.2104, | |
| "step": 404992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3360606225386756e-05, | |
| "loss": 4.2084, | |
| "step": 405504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.335222027787623e-05, | |
| "loss": 4.2211, | |
| "step": 406016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3343850709169445e-05, | |
| "loss": 4.201, | |
| "step": 406528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.333546476165893e-05, | |
| "loss": 4.2051, | |
| "step": 407040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.332707881414841e-05, | |
| "loss": 4.2094, | |
| "step": 407552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3318692866637885e-05, | |
| "loss": 4.2109, | |
| "step": 408064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.33103232979311e-05, | |
| "loss": 4.1996, | |
| "step": 408576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.330193735042058e-05, | |
| "loss": 4.2071, | |
| "step": 409088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.329356778171379e-05, | |
| "loss": 4.2019, | |
| "step": 409600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.328518183420327e-05, | |
| "loss": 4.2057, | |
| "step": 410112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.327679588669275e-05, | |
| "loss": 4.2045, | |
| "step": 410624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.326840993918223e-05, | |
| "loss": 4.2012, | |
| "step": 411136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.32600239916717e-05, | |
| "loss": 4.2125, | |
| "step": 411648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.325165442296492e-05, | |
| "loss": 4.1965, | |
| "step": 412160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.32432684754544e-05, | |
| "loss": 4.1929, | |
| "step": 412672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3234882527943885e-05, | |
| "loss": 4.1955, | |
| "step": 413184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.322649658043336e-05, | |
| "loss": 4.1959, | |
| "step": 413696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.321811063292284e-05, | |
| "loss": 4.2146, | |
| "step": 414208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3209741064216054e-05, | |
| "loss": 4.1951, | |
| "step": 414720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.320135511670553e-05, | |
| "loss": 4.2024, | |
| "step": 415232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.319296916919501e-05, | |
| "loss": 4.1818, | |
| "step": 415744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.318458322168449e-05, | |
| "loss": 4.2041, | |
| "step": 416256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.317619727417397e-05, | |
| "loss": 4.1917, | |
| "step": 416768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3167827705467176e-05, | |
| "loss": 4.1921, | |
| "step": 417280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3159441757956656e-05, | |
| "loss": 4.2067, | |
| "step": 417792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3151055810446136e-05, | |
| "loss": 4.1965, | |
| "step": 418304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.314266986293562e-05, | |
| "loss": 4.2019, | |
| "step": 418816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.31342839154251e-05, | |
| "loss": 4.2084, | |
| "step": 419328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.312589796791458e-05, | |
| "loss": 4.2016, | |
| "step": 419840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.311751202040406e-05, | |
| "loss": 4.2113, | |
| "step": 420352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.310912607289354e-05, | |
| "loss": 4.1952, | |
| "step": 420864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.310074012538302e-05, | |
| "loss": 4.1859, | |
| "step": 421376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.309237055667623e-05, | |
| "loss": 4.1971, | |
| "step": 421888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.308398460916571e-05, | |
| "loss": 4.2061, | |
| "step": 422400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.307559866165519e-05, | |
| "loss": 4.1949, | |
| "step": 422912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.306721271414467e-05, | |
| "loss": 4.1935, | |
| "step": 423424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.305884314543788e-05, | |
| "loss": 4.1796, | |
| "step": 423936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.305045719792736e-05, | |
| "loss": 4.1922, | |
| "step": 424448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.304207125041684e-05, | |
| "loss": 4.1847, | |
| "step": 424960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.303368530290632e-05, | |
| "loss": 4.1969, | |
| "step": 425472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3025315734199536e-05, | |
| "loss": 4.198, | |
| "step": 425984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3016929786689016e-05, | |
| "loss": 4.2035, | |
| "step": 426496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3008560217982226e-05, | |
| "loss": 4.1935, | |
| "step": 427008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3000174270471705e-05, | |
| "loss": 4.1946, | |
| "step": 427520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2991788322961185e-05, | |
| "loss": 4.1812, | |
| "step": 428032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2983402375450665e-05, | |
| "loss": 4.1865, | |
| "step": 428544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2975016427940145e-05, | |
| "loss": 4.1905, | |
| "step": 429056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2966646859233354e-05, | |
| "loss": 4.1873, | |
| "step": 429568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2958260911722834e-05, | |
| "loss": 4.184, | |
| "step": 430080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2949874964212314e-05, | |
| "loss": 4.1984, | |
| "step": 430592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2941489016701794e-05, | |
| "loss": 4.1849, | |
| "step": 431104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2933103069191274e-05, | |
| "loss": 4.2072, | |
| "step": 431616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2924717121680754e-05, | |
| "loss": 4.1826, | |
| "step": 432128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.291633117417024e-05, | |
| "loss": 4.1992, | |
| "step": 432640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2907945226659714e-05, | |
| "loss": 4.1884, | |
| "step": 433152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.289957565795293e-05, | |
| "loss": 4.1881, | |
| "step": 433664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.289118971044241e-05, | |
| "loss": 4.2054, | |
| "step": 434176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.288282014173562e-05, | |
| "loss": 4.1816, | |
| "step": 434688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.28744341942251e-05, | |
| "loss": 4.2024, | |
| "step": 435200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.286604824671458e-05, | |
| "loss": 4.1794, | |
| "step": 435712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.285766229920406e-05, | |
| "loss": 4.1779, | |
| "step": 436224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.284927635169354e-05, | |
| "loss": 4.1783, | |
| "step": 436736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.284090678298675e-05, | |
| "loss": 4.1908, | |
| "step": 437248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.283252083547623e-05, | |
| "loss": 4.1889, | |
| "step": 437760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.282413488796571e-05, | |
| "loss": 4.1801, | |
| "step": 438272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.281574894045519e-05, | |
| "loss": 4.1682, | |
| "step": 438784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.280736299294467e-05, | |
| "loss": 4.1809, | |
| "step": 439296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.279897704543415e-05, | |
| "loss": 4.1951, | |
| "step": 439808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.279059109792363e-05, | |
| "loss": 4.1879, | |
| "step": 440320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.278220515041311e-05, | |
| "loss": 4.1925, | |
| "step": 440832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.277383558170632e-05, | |
| "loss": 4.1901, | |
| "step": 441344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.27654496341958e-05, | |
| "loss": 4.189, | |
| "step": 441856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.275706368668528e-05, | |
| "loss": 4.1905, | |
| "step": 442368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.274867773917476e-05, | |
| "loss": 4.1819, | |
| "step": 442880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.27403245492717e-05, | |
| "loss": 4.1783, | |
| "step": 443392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.273193860176118e-05, | |
| "loss": 4.1858, | |
| "step": 443904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.272355265425066e-05, | |
| "loss": 4.1777, | |
| "step": 444416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.271516670674014e-05, | |
| "loss": 4.184, | |
| "step": 444928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.270678075922962e-05, | |
| "loss": 4.1842, | |
| "step": 445440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.26983948117191e-05, | |
| "loss": 4.1842, | |
| "step": 445952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.269000886420858e-05, | |
| "loss": 4.1806, | |
| "step": 446464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.268162291669806e-05, | |
| "loss": 4.1746, | |
| "step": 446976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.267323696918754e-05, | |
| "loss": 4.192, | |
| "step": 447488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.266486740048075e-05, | |
| "loss": 4.1778, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.265648145297023e-05, | |
| "loss": 4.1893, | |
| "step": 448512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.264809550545971e-05, | |
| "loss": 4.1905, | |
| "step": 449024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.263970955794919e-05, | |
| "loss": 4.1848, | |
| "step": 449536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.26313399892424e-05, | |
| "loss": 4.1857, | |
| "step": 450048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2622954041731886e-05, | |
| "loss": 4.1797, | |
| "step": 450560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2614568094221366e-05, | |
| "loss": 4.1811, | |
| "step": 451072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2606182146710846e-05, | |
| "loss": 4.1761, | |
| "step": 451584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2597828956807784e-05, | |
| "loss": 4.1948, | |
| "step": 452096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2589443009297264e-05, | |
| "loss": 4.1917, | |
| "step": 452608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2581057061786744e-05, | |
| "loss": 4.1861, | |
| "step": 453120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2572671114276224e-05, | |
| "loss": 4.1844, | |
| "step": 453632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2564285166765704e-05, | |
| "loss": 4.1792, | |
| "step": 454144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.255591559805891e-05, | |
| "loss": 4.1836, | |
| "step": 454656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.254752965054839e-05, | |
| "loss": 4.1756, | |
| "step": 455168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.253916008184161e-05, | |
| "loss": 4.1711, | |
| "step": 455680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.253077413433109e-05, | |
| "loss": 4.1824, | |
| "step": 456192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.252238818682057e-05, | |
| "loss": 4.1765, | |
| "step": 456704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.251400223931005e-05, | |
| "loss": 4.1653, | |
| "step": 457216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.250561629179953e-05, | |
| "loss": 4.1771, | |
| "step": 457728 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.172706604003906, | |
| "eval_runtime": 535.8869, | |
| "eval_samples_per_second": 712.074, | |
| "eval_steps_per_second": 22.253, | |
| "step": 457914 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 1.854320272228581e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |