| { | |
| "best_metric": 4.08057165145874, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-sent-neg/lstm/3/checkpoints/checkpoint-686880", | |
| "epoch": 0.025000606015738065, | |
| "eval_steps": 10, | |
| "global_step": 686880, | |
| "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": 10.8187, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 7.5487, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 7.0597, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 6.9956, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 6.9491, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 6.9236, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 6.7673, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994129836742636e-05, | |
| "loss": 6.6605, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993291241991584e-05, | |
| "loss": 6.5615, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992454285120906e-05, | |
| "loss": 6.4811, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.991615690369854e-05, | |
| "loss": 6.4166, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990777095618801e-05, | |
| "loss": 6.3434, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989938500867749e-05, | |
| "loss": 6.2739, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989099906116697e-05, | |
| "loss": 6.1955, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988261311365645e-05, | |
| "loss": 6.1487, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987422716614593e-05, | |
| "loss": 6.0869, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986584121863541e-05, | |
| "loss": 6.044, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985745527112489e-05, | |
| "loss": 5.9981, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984906932361437e-05, | |
| "loss": 5.9589, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984068337610385e-05, | |
| "loss": 5.917, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.8838, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.982391148108281e-05, | |
| "loss": 5.8415, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.981552553357229e-05, | |
| "loss": 5.8082, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9807155964865506e-05, | |
| "loss": 5.7764, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9798770017354986e-05, | |
| "loss": 5.7561, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9790384069844466e-05, | |
| "loss": 5.7185, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9781998122333946e-05, | |
| "loss": 5.6943, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773612174823426e-05, | |
| "loss": 5.6664, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9765242606116635e-05, | |
| "loss": 5.6364, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9756856658606115e-05, | |
| "loss": 5.6261, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748470711095595e-05, | |
| "loss": 5.6005, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9740101142388804e-05, | |
| "loss": 5.5925, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9731715194878284e-05, | |
| "loss": 5.556, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9723329247367764e-05, | |
| "loss": 5.5449, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714943299857244e-05, | |
| "loss": 5.5334, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9706557352346724e-05, | |
| "loss": 5.5162, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9698171404836204e-05, | |
| "loss": 5.4948, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9689785457325684e-05, | |
| "loss": 5.46, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9681399509815164e-05, | |
| "loss": 5.4632, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9673013562304644e-05, | |
| "loss": 5.429, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966464399359785e-05, | |
| "loss": 5.4315, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965625804608733e-05, | |
| "loss": 5.4151, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964787209857681e-05, | |
| "loss": 5.4045, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963948615106629e-05, | |
| "loss": 5.3846, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96311165823595e-05, | |
| "loss": 5.3785, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962274701365272e-05, | |
| "loss": 5.3782, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96143610661422e-05, | |
| "loss": 5.3608, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.960597511863168e-05, | |
| "loss": 5.3561, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.959758917112116e-05, | |
| "loss": 5.3276, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958920322361064e-05, | |
| "loss": 5.3134, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958081727610012e-05, | |
| "loss": 5.3225, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.957244770739333e-05, | |
| "loss": 5.3091, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956406175988281e-05, | |
| "loss": 5.3008, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955567581237229e-05, | |
| "loss": 5.2703, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.954728986486177e-05, | |
| "loss": 5.286, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953890391735125e-05, | |
| "loss": 5.2603, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9530517969840727e-05, | |
| "loss": 5.2814, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9522132022330207e-05, | |
| "loss": 5.2324, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9513746074819686e-05, | |
| "loss": 5.2438, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9505360127309166e-05, | |
| "loss": 5.2241, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9496974179798646e-05, | |
| "loss": 5.2122, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948858823228813e-05, | |
| "loss": 5.2144, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948021866358134e-05, | |
| "loss": 5.2016, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947183271607082e-05, | |
| "loss": 5.1798, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946346314736403e-05, | |
| "loss": 5.1807, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945509357865724e-05, | |
| "loss": 5.2013, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944672400995045e-05, | |
| "loss": 5.1745, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.943833806243993e-05, | |
| "loss": 5.1634, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942995211492941e-05, | |
| "loss": 5.152, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942156616741889e-05, | |
| "loss": 5.1378, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.941318021990837e-05, | |
| "loss": 5.1515, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9404794272397856e-05, | |
| "loss": 5.1429, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9396408324887336e-05, | |
| "loss": 5.1357, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388022377376816e-05, | |
| "loss": 5.1418, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379636429866296e-05, | |
| "loss": 5.1269, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371250482355776e-05, | |
| "loss": 5.1211, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362864534845256e-05, | |
| "loss": 5.0985, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.935447858733473e-05, | |
| "loss": 5.099, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.934609263982421e-05, | |
| "loss": 5.0898, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9337723071117425e-05, | |
| "loss": 5.0805, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329337123606905e-05, | |
| "loss": 5.0804, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932095117609638e-05, | |
| "loss": 5.0785, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.931256522858586e-05, | |
| "loss": 5.0806, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.930417928107534e-05, | |
| "loss": 5.0692, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9295793333564825e-05, | |
| "loss": 5.0437, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9287423764858034e-05, | |
| "loss": 5.0544, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9279037817347514e-05, | |
| "loss": 5.0547, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9270651869836994e-05, | |
| "loss": 5.0554, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9262265922326474e-05, | |
| "loss": 5.0337, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9253879974815954e-05, | |
| "loss": 5.037, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.924551040610916e-05, | |
| "loss": 5.0288, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923712445859864e-05, | |
| "loss": 5.0207, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922873851108812e-05, | |
| "loss": 5.0154, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.92203525635776e-05, | |
| "loss": 5.0115, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921198299487081e-05, | |
| "loss": 4.9989, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920361342616403e-05, | |
| "loss": 5.0067, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919522747865351e-05, | |
| "loss": 4.9898, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918684153114299e-05, | |
| "loss": 4.9809, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917845558363247e-05, | |
| "loss": 4.9775, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9170086014925676e-05, | |
| "loss": 4.9819, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9161700067415156e-05, | |
| "loss": 4.9759, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9153314119904636e-05, | |
| "loss": 4.9611, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9144928172394116e-05, | |
| "loss": 4.9623, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9136542224883596e-05, | |
| "loss": 4.9493, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9128156277373076e-05, | |
| "loss": 4.9535, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9119770329862556e-05, | |
| "loss": 4.9288, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9111384382352036e-05, | |
| "loss": 4.9392, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9102998434841516e-05, | |
| "loss": 4.9317, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9094612487330996e-05, | |
| "loss": 4.9285, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.908624291862421e-05, | |
| "loss": 4.9313, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907787334991742e-05, | |
| "loss": 4.9255, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90694874024069e-05, | |
| "loss": 4.904, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906110145489638e-05, | |
| "loss": 4.913, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905273188618959e-05, | |
| "loss": 4.9153, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904434593867907e-05, | |
| "loss": 4.9072, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903595999116855e-05, | |
| "loss": 4.907, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.902757404365803e-05, | |
| "loss": 4.8983, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901918809614751e-05, | |
| "loss": 4.8945, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901080214863699e-05, | |
| "loss": 4.8826, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.900241620112647e-05, | |
| "loss": 4.8892, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899403025361595e-05, | |
| "loss": 4.8845, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.898564430610543e-05, | |
| "loss": 4.8695, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.897725835859491e-05, | |
| "loss": 4.8716, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968872411084396e-05, | |
| "loss": 4.8665, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.896048646357387e-05, | |
| "loss": 4.8736, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.895210051606335e-05, | |
| "loss": 4.8501, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.894371456855283e-05, | |
| "loss": 4.8564, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.893532862104231e-05, | |
| "loss": 4.8429, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.892695905233552e-05, | |
| "loss": 4.851, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918589483628734e-05, | |
| "loss": 4.8394, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8910203536118214e-05, | |
| "loss": 4.8539, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890181758860769e-05, | |
| "loss": 4.8413, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88934480199009e-05, | |
| "loss": 4.8344, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888506207239038e-05, | |
| "loss": 4.833, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.887667612487986e-05, | |
| "loss": 4.8279, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.886829017736934e-05, | |
| "loss": 4.8365, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.885990422985882e-05, | |
| "loss": 4.8333, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88515182823483e-05, | |
| "loss": 4.8244, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884313233483778e-05, | |
| "loss": 4.8187, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883474638732726e-05, | |
| "loss": 4.8202, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882637681862047e-05, | |
| "loss": 4.8096, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881799087110995e-05, | |
| "loss": 4.7973, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880960492359943e-05, | |
| "loss": 4.8054, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880121897608891e-05, | |
| "loss": 4.7966, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879284940738212e-05, | |
| "loss": 4.7888, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.87844634598716e-05, | |
| "loss": 4.79, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877607751236109e-05, | |
| "loss": 4.7877, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.876769156485057e-05, | |
| "loss": 4.7912, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875932199614378e-05, | |
| "loss": 4.7911, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875093604863326e-05, | |
| "loss": 4.7787, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.7453789710998535, | |
| "eval_runtime": 301.8396, | |
| "eval_samples_per_second": 1264.218, | |
| "eval_steps_per_second": 39.508, | |
| "step": 76320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.874255010112274e-05, | |
| "loss": 4.769, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.873416415361222e-05, | |
| "loss": 4.7668, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.87257782061017e-05, | |
| "loss": 4.7781, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.871739225859118e-05, | |
| "loss": 4.7697, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8709006311080657e-05, | |
| "loss": 4.7717, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8700620363570137e-05, | |
| "loss": 4.7534, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8692234416059616e-05, | |
| "loss": 4.7553, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8683848468549096e-05, | |
| "loss": 4.7501, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8675478899842306e-05, | |
| "loss": 4.7553, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8667092952331785e-05, | |
| "loss": 4.754, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865870700482127e-05, | |
| "loss": 4.7511, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865032105731075e-05, | |
| "loss": 4.7574, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.864195148860396e-05, | |
| "loss": 4.7257, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.863356554109344e-05, | |
| "loss": 4.7374, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.862517959358292e-05, | |
| "loss": 4.7316, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.86167936460724e-05, | |
| "loss": 4.7222, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8608407698561874e-05, | |
| "loss": 4.7355, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8600021751051354e-05, | |
| "loss": 4.7259, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8591635803540834e-05, | |
| "loss": 4.7222, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.858326623483405e-05, | |
| "loss": 4.7441, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.857489666612726e-05, | |
| "loss": 4.7242, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.856651071861674e-05, | |
| "loss": 4.7179, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8558141149909955e-05, | |
| "loss": 4.7171, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8549755202399435e-05, | |
| "loss": 4.7316, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8541369254888915e-05, | |
| "loss": 4.7105, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8532983307378395e-05, | |
| "loss": 4.7052, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8524597359867875e-05, | |
| "loss": 4.7058, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.851621141235735e-05, | |
| "loss": 4.7019, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.850782546484683e-05, | |
| "loss": 4.7012, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.849943951733631e-05, | |
| "loss": 4.6945, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8491069948629524e-05, | |
| "loss": 4.7137, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8482684001119e-05, | |
| "loss": 4.6941, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.847429805360848e-05, | |
| "loss": 4.6929, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8465912106097964e-05, | |
| "loss": 4.6948, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.845754253739117e-05, | |
| "loss": 4.6935, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.844915658988065e-05, | |
| "loss": 4.6874, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.844077064237013e-05, | |
| "loss": 4.6686, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.843238469485961e-05, | |
| "loss": 4.6766, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.842399874734909e-05, | |
| "loss": 4.6618, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.841561279983857e-05, | |
| "loss": 4.6796, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.840722685232805e-05, | |
| "loss": 4.6675, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839884090481753e-05, | |
| "loss": 4.6728, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839047133611074e-05, | |
| "loss": 4.6631, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.838208538860022e-05, | |
| "loss": 4.6676, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.83736994410897e-05, | |
| "loss": 4.6705, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.836531349357918e-05, | |
| "loss": 4.663, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8356960303676126e-05, | |
| "loss": 4.6663, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8348574356165606e-05, | |
| "loss": 4.6531, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834020478745882e-05, | |
| "loss": 4.6455, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8331818839948295e-05, | |
| "loss": 4.6571, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8323432892437775e-05, | |
| "loss": 4.6515, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8315046944927255e-05, | |
| "loss": 4.6565, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8306660997416735e-05, | |
| "loss": 4.6289, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8298275049906215e-05, | |
| "loss": 4.6521, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8289905481199424e-05, | |
| "loss": 4.6297, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8281519533688904e-05, | |
| "loss": 4.6618, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8273133586178384e-05, | |
| "loss": 4.6242, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.826474763866787e-05, | |
| "loss": 4.6398, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.825636169115735e-05, | |
| "loss": 4.6291, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.824797574364683e-05, | |
| "loss": 4.6216, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823960617494004e-05, | |
| "loss": 4.6257, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823122022742952e-05, | |
| "loss": 4.6241, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8222834279919e-05, | |
| "loss": 4.6067, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.821444833240848e-05, | |
| "loss": 4.6132, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.820606238489796e-05, | |
| "loss": 4.6288, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.819767643738744e-05, | |
| "loss": 4.6155, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.818929048987692e-05, | |
| "loss": 4.6117, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.81809045423664e-05, | |
| "loss": 4.6075, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.817251859485588e-05, | |
| "loss": 4.5979, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.816413264734535e-05, | |
| "loss": 4.6164, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.815574669983483e-05, | |
| "loss": 4.6106, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.814736075232432e-05, | |
| "loss": 4.6116, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8138991183617535e-05, | |
| "loss": 4.6166, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.813060523610701e-05, | |
| "loss": 4.6133, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.812221928859649e-05, | |
| "loss": 4.6115, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8113849719889704e-05, | |
| "loss": 4.5901, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8105463772379184e-05, | |
| "loss": 4.6017, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.809707782486866e-05, | |
| "loss": 4.5959, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808869187735814e-05, | |
| "loss": 4.5843, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808032230865135e-05, | |
| "loss": 4.5946, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8071936361140827e-05, | |
| "loss": 4.5944, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8063550413630307e-05, | |
| "loss": 4.6033, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8055164466119786e-05, | |
| "loss": 4.5956, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8046794897413e-05, | |
| "loss": 4.5747, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803840894990248e-05, | |
| "loss": 4.5844, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803002300239196e-05, | |
| "loss": 4.5929, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.802163705488144e-05, | |
| "loss": 4.5948, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.801325110737092e-05, | |
| "loss": 4.5843, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.80048651598604e-05, | |
| "loss": 4.5896, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.799647921234988e-05, | |
| "loss": 4.5851, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.798810964364309e-05, | |
| "loss": 4.5725, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797972369613257e-05, | |
| "loss": 4.5782, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797133774862205e-05, | |
| "loss": 4.5791, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.796295180111153e-05, | |
| "loss": 4.5686, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.795456585360101e-05, | |
| "loss": 4.5834, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.794617990609049e-05, | |
| "loss": 4.5649, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.793781033738371e-05, | |
| "loss": 4.5621, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.792942438987319e-05, | |
| "loss": 4.5651, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.792103844236267e-05, | |
| "loss": 4.5721, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.791265249485215e-05, | |
| "loss": 4.5693, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.790426654734163e-05, | |
| "loss": 4.5579, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7895880599831107e-05, | |
| "loss": 4.5561, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7887494652320587e-05, | |
| "loss": 4.5512, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7879108704810067e-05, | |
| "loss": 4.5619, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7870739136103276e-05, | |
| "loss": 4.5383, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7862369567396485e-05, | |
| "loss": 4.5467, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7853983619885965e-05, | |
| "loss": 4.5543, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7845597672375445e-05, | |
| "loss": 4.5489, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7837211724864925e-05, | |
| "loss": 4.5484, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782882577735441e-05, | |
| "loss": 4.5472, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782045620864762e-05, | |
| "loss": 4.5365, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.78120702611371e-05, | |
| "loss": 4.5444, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.780368431362658e-05, | |
| "loss": 4.5444, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.779529836611606e-05, | |
| "loss": 4.5434, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.778691241860554e-05, | |
| "loss": 4.5474, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777852647109502e-05, | |
| "loss": 4.5466, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777015690238823e-05, | |
| "loss": 4.5378, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.776177095487771e-05, | |
| "loss": 4.5318, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.775338500736719e-05, | |
| "loss": 4.5394, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.774499905985666e-05, | |
| "loss": 4.5344, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.773662949114988e-05, | |
| "loss": 4.5235, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.772824354363936e-05, | |
| "loss": 4.5318, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7719857596128845e-05, | |
| "loss": 4.5267, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771147164861832e-05, | |
| "loss": 4.5355, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.77030857011078e-05, | |
| "loss": 4.518, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.769469975359728e-05, | |
| "loss": 4.5272, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.768631380608676e-05, | |
| "loss": 4.5064, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.767792785857624e-05, | |
| "loss": 4.5284, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766955828986945e-05, | |
| "loss": 4.5149, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766117234235893e-05, | |
| "loss": 4.532, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.765278639484841e-05, | |
| "loss": 4.5243, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.764440044733789e-05, | |
| "loss": 4.5194, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7636030878631096e-05, | |
| "loss": 4.5112, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.762764493112058e-05, | |
| "loss": 4.513, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761925898361006e-05, | |
| "loss": 4.5269, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761087303609954e-05, | |
| "loss": 4.5295, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.760251984619649e-05, | |
| "loss": 4.5181, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.759413389868596e-05, | |
| "loss": 4.5144, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.758574795117544e-05, | |
| "loss": 4.5117, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.757736200366492e-05, | |
| "loss": 4.5085, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75689760561544e-05, | |
| "loss": 4.5035, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756059010864388e-05, | |
| "loss": 4.5093, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.755220416113336e-05, | |
| "loss": 4.5035, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.754381821362284e-05, | |
| "loss": 4.4966, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.753543226611232e-05, | |
| "loss": 4.4994, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75270463186018e-05, | |
| "loss": 4.5001, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.751866037109128e-05, | |
| "loss": 4.5022, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7510290802384496e-05, | |
| "loss": 4.5016, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7501904854873976e-05, | |
| "loss": 4.4985, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.4658379554748535, | |
| "eval_runtime": 298.8499, | |
| "eval_samples_per_second": 1276.865, | |
| "eval_steps_per_second": 39.903, | |
| "step": 152640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7493518907363456e-05, | |
| "loss": 4.4838, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7485132959852936e-05, | |
| "loss": 4.4883, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7476747012342416e-05, | |
| "loss": 4.4969, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7468361064831896e-05, | |
| "loss": 4.4946, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7459975117321376e-05, | |
| "loss": 4.4977, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745158916981085e-05, | |
| "loss": 4.4775, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.744320322230033e-05, | |
| "loss": 4.4889, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.743481727478981e-05, | |
| "loss": 4.4785, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.742643132727929e-05, | |
| "loss": 4.4889, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.741804537976877e-05, | |
| "loss": 4.4876, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.740965943225825e-05, | |
| "loss": 4.4873, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7401273484747736e-05, | |
| "loss": 4.4901, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7392887537237216e-05, | |
| "loss": 4.4648, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7384501589726696e-05, | |
| "loss": 4.4752, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7376115642216176e-05, | |
| "loss": 4.4756, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7367729694705656e-05, | |
| "loss": 4.467, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7359343747195135e-05, | |
| "loss": 4.4763, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7350957799684615e-05, | |
| "loss": 4.47, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7342571852174095e-05, | |
| "loss": 4.4703, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7334202283467305e-05, | |
| "loss": 4.4981, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7325816335956784e-05, | |
| "loss": 4.4722, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7317430388446264e-05, | |
| "loss": 4.4733, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730904444093574e-05, | |
| "loss": 4.4692, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7300674872228953e-05, | |
| "loss": 4.4853, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7292288924718433e-05, | |
| "loss": 4.4646, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.728390297720791e-05, | |
| "loss": 4.4623, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.727551702969739e-05, | |
| "loss": 4.4663, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.726714746099061e-05, | |
| "loss": 4.4662, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.725876151348009e-05, | |
| "loss": 4.4581, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.725037556596957e-05, | |
| "loss": 4.4605, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.724200599726278e-05, | |
| "loss": 4.4758, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.723362004975226e-05, | |
| "loss": 4.4616, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.722523410224174e-05, | |
| "loss": 4.4582, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.721684815473121e-05, | |
| "loss": 4.4654, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720846220722069e-05, | |
| "loss": 4.4608, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720007625971017e-05, | |
| "loss": 4.457, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.719170669100339e-05, | |
| "loss": 4.4395, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.718332074349287e-05, | |
| "loss": 4.4556, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.717493479598235e-05, | |
| "loss": 4.4386, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.716654884847183e-05, | |
| "loss": 4.4521, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.715816290096131e-05, | |
| "loss": 4.4452, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714977695345079e-05, | |
| "loss": 4.4538, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714139100594027e-05, | |
| "loss": 4.4448, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.713300505842975e-05, | |
| "loss": 4.4455, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.712466824733042e-05, | |
| "loss": 4.4497, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.71162822998199e-05, | |
| "loss": 4.4482, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.710789635230938e-05, | |
| "loss": 4.4512, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.709951040479886e-05, | |
| "loss": 4.4372, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.709112445728834e-05, | |
| "loss": 4.4283, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.708273850977782e-05, | |
| "loss": 4.4476, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.70743525622673e-05, | |
| "loss": 4.4383, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.706596661475678e-05, | |
| "loss": 4.4424, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.705758066724626e-05, | |
| "loss": 4.4218, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704921109853947e-05, | |
| "loss": 4.4458, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704082515102895e-05, | |
| "loss": 4.4213, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.703243920351843e-05, | |
| "loss": 4.4509, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.702405325600791e-05, | |
| "loss": 4.4141, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.701566730849739e-05, | |
| "loss": 4.4368, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.700728136098687e-05, | |
| "loss": 4.4252, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699889541347635e-05, | |
| "loss": 4.4238, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699050946596583e-05, | |
| "loss": 4.4233, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6982139897259045e-05, | |
| "loss": 4.4242, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6973770328552254e-05, | |
| "loss": 4.4066, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6965384381041734e-05, | |
| "loss": 4.4146, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6956998433531214e-05, | |
| "loss": 4.426, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6948612486020694e-05, | |
| "loss": 4.4234, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6940226538510174e-05, | |
| "loss": 4.4131, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6931840590999654e-05, | |
| "loss": 4.4139, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6923454643489134e-05, | |
| "loss": 4.4102, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6915068695978614e-05, | |
| "loss": 4.4169, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.690671550607555e-05, | |
| "loss": 4.4198, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.689832955856503e-05, | |
| "loss": 4.419, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688994361105451e-05, | |
| "loss": 4.4277, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6881557663544e-05, | |
| "loss": 4.4275, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.687317171603348e-05, | |
| "loss": 4.4221, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.686478576852296e-05, | |
| "loss": 4.4032, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.685641619981617e-05, | |
| "loss": 4.4131, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.684803025230565e-05, | |
| "loss": 4.4168, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683966068359886e-05, | |
| "loss": 4.3985, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683127473608834e-05, | |
| "loss": 4.4122, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.682288878857782e-05, | |
| "loss": 4.4083, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68145028410673e-05, | |
| "loss": 4.4191, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.680611689355678e-05, | |
| "loss": 4.4124, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.679773094604626e-05, | |
| "loss": 4.3936, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678934499853574e-05, | |
| "loss": 4.4071, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678095905102522e-05, | |
| "loss": 4.4109, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6772573103514697e-05, | |
| "loss": 4.4145, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6764187156004177e-05, | |
| "loss": 4.4071, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6755801208493657e-05, | |
| "loss": 4.4117, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.674743163978687e-05, | |
| "loss": 4.4093, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6739045692276346e-05, | |
| "loss": 4.3943, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6730659744765826e-05, | |
| "loss": 4.4063, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6722273797255305e-05, | |
| "loss": 4.4017, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6713887849744785e-05, | |
| "loss": 4.3941, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6705501902234265e-05, | |
| "loss": 4.4133, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6697132333527474e-05, | |
| "loss": 4.3952, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6688746386016954e-05, | |
| "loss": 4.3861, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6680360438506434e-05, | |
| "loss": 4.395, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.667197449099592e-05, | |
| "loss": 4.4033, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.66635885434854e-05, | |
| "loss": 4.3996, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.665520259597488e-05, | |
| "loss": 4.3901, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.664683302726809e-05, | |
| "loss": 4.3853, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.66384634585613e-05, | |
| "loss": 4.3837, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663007751105078e-05, | |
| "loss": 4.395, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.662169156354026e-05, | |
| "loss": 4.3729, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.661330561602974e-05, | |
| "loss": 4.3846, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.660491966851922e-05, | |
| "loss": 4.3892, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.65965337210087e-05, | |
| "loss": 4.3859, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.658814777349818e-05, | |
| "loss": 4.3841, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657976182598766e-05, | |
| "loss": 4.3872, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657139225728087e-05, | |
| "loss": 4.3774, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6563006309770355e-05, | |
| "loss": 4.383, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6554620362259835e-05, | |
| "loss": 4.3828, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6546234414749315e-05, | |
| "loss": 4.381, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6537848467238795e-05, | |
| "loss": 4.3888, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6529462519728275e-05, | |
| "loss": 4.3883, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6521092951021484e-05, | |
| "loss": 4.3857, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6512707003510964e-05, | |
| "loss": 4.3715, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6504321056000444e-05, | |
| "loss": 4.3788, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6495935108489924e-05, | |
| "loss": 4.377, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6487549160979403e-05, | |
| "loss": 4.3668, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6479163213468883e-05, | |
| "loss": 4.3815, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647077726595836e-05, | |
| "loss": 4.3682, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6462391318447837e-05, | |
| "loss": 4.3817, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.645402174974105e-05, | |
| "loss": 4.3651, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.644563580223053e-05, | |
| "loss": 4.3752, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.643724985472001e-05, | |
| "loss": 4.3565, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642888028601323e-05, | |
| "loss": 4.3757, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642049433850271e-05, | |
| "loss": 4.3608, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.641210839099218e-05, | |
| "loss": 4.3817, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.640372244348166e-05, | |
| "loss": 4.3793, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.639533649597114e-05, | |
| "loss": 4.3692, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.638695054846062e-05, | |
| "loss": 4.3615, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637858097975383e-05, | |
| "loss": 4.3657, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637019503224331e-05, | |
| "loss": 4.3773, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.636180908473279e-05, | |
| "loss": 4.3835, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.635342313722228e-05, | |
| "loss": 4.3683, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.634503718971176e-05, | |
| "loss": 4.3697, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6336667621004966e-05, | |
| "loss": 4.3675, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6328281673494446e-05, | |
| "loss": 4.3656, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6319895725983926e-05, | |
| "loss": 4.3574, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6311509778473406e-05, | |
| "loss": 4.3613, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6303123830962886e-05, | |
| "loss": 4.3629, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6294737883452366e-05, | |
| "loss": 4.3542, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6286368314745575e-05, | |
| "loss": 4.3553, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6277982367235055e-05, | |
| "loss": 4.3608, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6269596419724535e-05, | |
| "loss": 4.3543, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6261210472214015e-05, | |
| "loss": 4.3613, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.625284090350723e-05, | |
| "loss": 4.3554, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.332054615020752, | |
| "eval_runtime": 293.744, | |
| "eval_samples_per_second": 1299.06, | |
| "eval_steps_per_second": 40.597, | |
| "step": 228960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.624445495599671e-05, | |
| "loss": 4.3516, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.623606900848619e-05, | |
| "loss": 4.3474, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.622768306097567e-05, | |
| "loss": 4.3614, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.621929711346515e-05, | |
| "loss": 4.3507, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.621091116595463e-05, | |
| "loss": 4.3623, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.620252521844411e-05, | |
| "loss": 4.3395, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.619415564973732e-05, | |
| "loss": 4.3517, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61857697022268e-05, | |
| "loss": 4.3396, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.617738375471628e-05, | |
| "loss": 4.3559, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616899780720576e-05, | |
| "loss": 4.3505, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616061185969524e-05, | |
| "loss": 4.3525, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.615222591218472e-05, | |
| "loss": 4.3555, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61438399646742e-05, | |
| "loss": 4.3321, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.613545401716368e-05, | |
| "loss": 4.3415, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.612706806965316e-05, | |
| "loss": 4.3408, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611868212214264e-05, | |
| "loss": 4.3385, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611029617463212e-05, | |
| "loss": 4.34, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61019102271216e-05, | |
| "loss": 4.344, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.609354065841481e-05, | |
| "loss": 4.3352, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.608515471090429e-05, | |
| "loss": 4.3694, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.607676876339377e-05, | |
| "loss": 4.3391, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606838281588325e-05, | |
| "loss": 4.3419, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606001324717646e-05, | |
| "loss": 4.341, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.605162729966594e-05, | |
| "loss": 4.3519, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.604324135215542e-05, | |
| "loss": 4.3335, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.603487178344863e-05, | |
| "loss": 4.3399, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.602648583593811e-05, | |
| "loss": 4.3341, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.601809988842759e-05, | |
| "loss": 4.3372, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.600971394091707e-05, | |
| "loss": 4.3267, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.600132799340655e-05, | |
| "loss": 4.3366, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.599294204589603e-05, | |
| "loss": 4.3461, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.598455609838551e-05, | |
| "loss": 4.3384, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.597617015087499e-05, | |
| "loss": 4.3365, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.59678005821682e-05, | |
| "loss": 4.3379, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595943101346141e-05, | |
| "loss": 4.3328, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595104506595089e-05, | |
| "loss": 4.3369, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.594265911844037e-05, | |
| "loss": 4.3157, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.593427317092985e-05, | |
| "loss": 4.3322, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.592588722341933e-05, | |
| "loss": 4.3127, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5917517654712546e-05, | |
| "loss": 4.3311, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5909131707202026e-05, | |
| "loss": 4.3241, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5900745759691506e-05, | |
| "loss": 4.3317, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5892359812180986e-05, | |
| "loss": 4.3218, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5883973864670466e-05, | |
| "loss": 4.3224, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5875604295963675e-05, | |
| "loss": 4.3307, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5867218348453155e-05, | |
| "loss": 4.3271, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5858832400942635e-05, | |
| "loss": 4.3273, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5850446453432115e-05, | |
| "loss": 4.3218, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5842060505921595e-05, | |
| "loss": 4.3066, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5833690937214804e-05, | |
| "loss": 4.3287, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5825304989704284e-05, | |
| "loss": 4.3196, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.581691904219377e-05, | |
| "loss": 4.3233, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.580853309468325e-05, | |
| "loss": 4.308, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.580014714717273e-05, | |
| "loss": 4.3311, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5791761199662204e-05, | |
| "loss": 4.2995, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5783375252151684e-05, | |
| "loss": 4.3349, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5774989304641164e-05, | |
| "loss": 4.2988, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.576661973593438e-05, | |
| "loss": 4.3159, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.575823378842385e-05, | |
| "loss": 4.3113, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.574984784091333e-05, | |
| "loss": 4.3091, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.574147827220655e-05, | |
| "loss": 4.3043, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.573309232469602e-05, | |
| "loss": 4.3162, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.572470637718551e-05, | |
| "loss": 4.2906, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.571632042967499e-05, | |
| "loss": 4.2985, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.570793448216447e-05, | |
| "loss": 4.314, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.569954853465395e-05, | |
| "loss": 4.309, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.569116258714343e-05, | |
| "loss": 4.3022, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.568277663963291e-05, | |
| "loss": 4.302, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.567440707092612e-05, | |
| "loss": 4.2917, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.566603750221933e-05, | |
| "loss": 4.3053, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.565765155470881e-05, | |
| "loss": 4.3103, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.564926560719829e-05, | |
| "loss": 4.3085, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5640879659687767e-05, | |
| "loss": 4.3184, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5632493712177247e-05, | |
| "loss": 4.3155, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.562412414347046e-05, | |
| "loss": 4.3132, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.561573819595994e-05, | |
| "loss": 4.2934, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.560735224844942e-05, | |
| "loss": 4.3012, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.55989663009389e-05, | |
| "loss": 4.308, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.559059673223211e-05, | |
| "loss": 4.2893, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.558221078472159e-05, | |
| "loss": 4.3059, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.557382483721107e-05, | |
| "loss": 4.2988, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.556543888970055e-05, | |
| "loss": 4.3078, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.555705294219003e-05, | |
| "loss": 4.305, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554866699467951e-05, | |
| "loss": 4.2876, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554028104716899e-05, | |
| "loss": 4.3023, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.553189509965847e-05, | |
| "loss": 4.3007, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.552352553095168e-05, | |
| "loss": 4.31, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.551513958344116e-05, | |
| "loss": 4.303, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.550675363593064e-05, | |
| "loss": 4.3031, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.549836768842013e-05, | |
| "loss": 4.3067, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5489998119713336e-05, | |
| "loss": 4.2881, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5481612172202816e-05, | |
| "loss": 4.3027, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5473226224692296e-05, | |
| "loss": 4.2943, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5464840277181776e-05, | |
| "loss": 4.2949, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5456470708474985e-05, | |
| "loss": 4.3076, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5448084760964465e-05, | |
| "loss": 4.2953, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5439698813453945e-05, | |
| "loss": 4.2811, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5431312865943425e-05, | |
| "loss": 4.2919, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5422943297236634e-05, | |
| "loss": 4.3013, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5414557349726114e-05, | |
| "loss": 4.2982, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5406171402215594e-05, | |
| "loss": 4.2886, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.539778545470508e-05, | |
| "loss": 4.2836, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.538941588599829e-05, | |
| "loss": 4.2838, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.538102993848777e-05, | |
| "loss": 4.2967, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.537264399097725e-05, | |
| "loss": 4.2692, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.536425804346673e-05, | |
| "loss": 4.2847, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.535588847475994e-05, | |
| "loss": 4.2884, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.534750252724942e-05, | |
| "loss": 4.2878, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.53391165797389e-05, | |
| "loss": 4.2807, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533073063222838e-05, | |
| "loss": 4.2902, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.532237744232532e-05, | |
| "loss": 4.2808, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.53139914948148e-05, | |
| "loss": 4.2809, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.530560554730428e-05, | |
| "loss": 4.2865, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.529721959979376e-05, | |
| "loss": 4.2815, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528883365228324e-05, | |
| "loss": 4.2957, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528044770477272e-05, | |
| "loss": 4.2868, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.527207813606593e-05, | |
| "loss": 4.2882, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.526369218855541e-05, | |
| "loss": 4.273, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.525530624104489e-05, | |
| "loss": 4.2827, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.524692029353437e-05, | |
| "loss": 4.2818, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.523853434602385e-05, | |
| "loss": 4.2704, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.523014839851333e-05, | |
| "loss": 4.2836, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5221762451002805e-05, | |
| "loss": 4.2738, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5213376503492285e-05, | |
| "loss": 4.2853, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.520499055598177e-05, | |
| "loss": 4.2703, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.519660460847125e-05, | |
| "loss": 4.2802, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.518821866096073e-05, | |
| "loss": 4.256, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517983271345021e-05, | |
| "loss": 4.2838, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517146314474342e-05, | |
| "loss": 4.2687, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.51630771972329e-05, | |
| "loss": 4.2836, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.515470762852611e-05, | |
| "loss": 4.282, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5146338059819326e-05, | |
| "loss": 4.281, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5137952112308806e-05, | |
| "loss": 4.2646, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512956616479828e-05, | |
| "loss": 4.2762, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512118021728776e-05, | |
| "loss": 4.2815, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.511279426977724e-05, | |
| "loss": 4.2868, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5104408322266726e-05, | |
| "loss": 4.2777, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5096038753559935e-05, | |
| "loss": 4.2759, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5087652806049415e-05, | |
| "loss": 4.2756, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5079266858538895e-05, | |
| "loss": 4.2789, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5070880911028375e-05, | |
| "loss": 4.2614, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5062494963517855e-05, | |
| "loss": 4.2702, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5054109016007334e-05, | |
| "loss": 4.2695, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5045723068496814e-05, | |
| "loss": 4.2649, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5037337120986294e-05, | |
| "loss": 4.2626, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5028951173475774e-05, | |
| "loss": 4.2722, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5020581604768983e-05, | |
| "loss": 4.2609, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.501219565725846e-05, | |
| "loss": 4.269, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.500380970974794e-05, | |
| "loss": 4.2729, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.2502336502075195, | |
| "eval_runtime": 294.515, | |
| "eval_samples_per_second": 1295.659, | |
| "eval_steps_per_second": 40.49, | |
| "step": 305280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.499542376223742e-05, | |
| "loss": 4.2551, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.498703781472691e-05, | |
| "loss": 4.2585, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.497865186721639e-05, | |
| "loss": 4.27, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.497026591970587e-05, | |
| "loss": 4.2617, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.496187997219535e-05, | |
| "loss": 4.2749, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.495349402468482e-05, | |
| "loss": 4.2494, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.49451080771743e-05, | |
| "loss": 4.2602, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.493672212966378e-05, | |
| "loss": 4.2522, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.492833618215326e-05, | |
| "loss": 4.2662, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491995023464274e-05, | |
| "loss": 4.2655, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491156428713222e-05, | |
| "loss": 4.2568, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.49031783396217e-05, | |
| "loss": 4.2727, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.489479239211118e-05, | |
| "loss": 4.241, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.488640644460066e-05, | |
| "loss": 4.2544, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.487803687589387e-05, | |
| "loss": 4.2584, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486965092838336e-05, | |
| "loss": 4.248, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486126498087284e-05, | |
| "loss": 4.2534, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.485287903336232e-05, | |
| "loss": 4.2564, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.48444930858518e-05, | |
| "loss": 4.2515, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.483610713834128e-05, | |
| "loss": 4.2806, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.482772119083076e-05, | |
| "loss": 4.255, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.481933524332024e-05, | |
| "loss": 4.2536, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.481094929580971e-05, | |
| "loss": 4.2611, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.480257972710293e-05, | |
| "loss": 4.268, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.479419377959241e-05, | |
| "loss": 4.2483, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.478580783208188e-05, | |
| "loss": 4.2584, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.477742188457136e-05, | |
| "loss": 4.2497, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.476903593706084e-05, | |
| "loss": 4.2527, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.476064998955033e-05, | |
| "loss": 4.2447, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.475226404203981e-05, | |
| "loss": 4.2532, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.474387809452929e-05, | |
| "loss": 4.2643, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4735508525822496e-05, | |
| "loss": 4.2538, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4727122578311976e-05, | |
| "loss": 4.2546, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4718736630801456e-05, | |
| "loss": 4.2565, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4710367062094665e-05, | |
| "loss": 4.251, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4701981114584145e-05, | |
| "loss": 4.2559, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4693595167073625e-05, | |
| "loss": 4.2345, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4685209219563105e-05, | |
| "loss": 4.2476, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4676823272052585e-05, | |
| "loss": 4.2368, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4668437324542065e-05, | |
| "loss": 4.2433, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466006775583528e-05, | |
| "loss": 4.2454, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.465168180832476e-05, | |
| "loss": 4.2521, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.464329586081424e-05, | |
| "loss": 4.2408, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.463490991330372e-05, | |
| "loss": 4.2437, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.46265239657932e-05, | |
| "loss": 4.248, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.461813801828268e-05, | |
| "loss": 4.2463, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.460976844957589e-05, | |
| "loss": 4.247, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.460138250206537e-05, | |
| "loss": 4.2424, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.459299655455485e-05, | |
| "loss": 4.2262, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.458461060704433e-05, | |
| "loss": 4.2501, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.457624103833754e-05, | |
| "loss": 4.2367, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.456785509082702e-05, | |
| "loss": 4.2475, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.45594691433165e-05, | |
| "loss": 4.2329, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.455108319580598e-05, | |
| "loss": 4.2469, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4542697248295465e-05, | |
| "loss": 4.2234, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4534311300784945e-05, | |
| "loss": 4.254, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4525925353274425e-05, | |
| "loss": 4.221, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.45175394057639e-05, | |
| "loss": 4.2317, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.450918621586084e-05, | |
| "loss": 4.2383, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.450080026835032e-05, | |
| "loss": 4.2299, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.44924143208398e-05, | |
| "loss": 4.2315, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.448402837332928e-05, | |
| "loss": 4.2351, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.447564242581876e-05, | |
| "loss": 4.2225, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.446725647830824e-05, | |
| "loss": 4.2193, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445888690960145e-05, | |
| "loss": 4.2332, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445050096209093e-05, | |
| "loss": 4.2354, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.444211501458041e-05, | |
| "loss": 4.2235, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.44337290670699e-05, | |
| "loss": 4.2319, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.442535949836311e-05, | |
| "loss": 4.2138, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.441697355085259e-05, | |
| "loss": 4.23, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440858760334207e-05, | |
| "loss": 4.2337, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440020165583154e-05, | |
| "loss": 4.2309, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.439181570832102e-05, | |
| "loss": 4.2407, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.43834297608105e-05, | |
| "loss": 4.2382, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.437504381329998e-05, | |
| "loss": 4.2439, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4366690623396926e-05, | |
| "loss": 4.2186, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4358304675886406e-05, | |
| "loss": 4.2269, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4349918728375886e-05, | |
| "loss": 4.2338, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4341532780865366e-05, | |
| "loss": 4.2121, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4333146833354846e-05, | |
| "loss": 4.2347, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4324760885844326e-05, | |
| "loss": 4.2265, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.431639131713754e-05, | |
| "loss": 4.2331, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4308005369627015e-05, | |
| "loss": 4.2329, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4299619422116495e-05, | |
| "loss": 4.2141, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4291233474605975e-05, | |
| "loss": 4.2292, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4282847527095455e-05, | |
| "loss": 4.2255, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4274461579584935e-05, | |
| "loss": 4.2336, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4266075632074415e-05, | |
| "loss": 4.2281, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4257689684563894e-05, | |
| "loss": 4.2346, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4249320115857104e-05, | |
| "loss": 4.2304, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424095054715032e-05, | |
| "loss": 4.2159, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.42325645996398e-05, | |
| "loss": 4.2286, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.422417865212928e-05, | |
| "loss": 4.2192, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.421579270461876e-05, | |
| "loss": 4.2269, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.420740675710824e-05, | |
| "loss": 4.2348, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419902080959772e-05, | |
| "loss": 4.2185, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.41906348620872e-05, | |
| "loss": 4.214, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.418224891457668e-05, | |
| "loss": 4.2183, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.417386296706616e-05, | |
| "loss": 4.2276, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.416549339835937e-05, | |
| "loss": 4.2316, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.415710745084885e-05, | |
| "loss": 4.2156, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.414872150333833e-05, | |
| "loss": 4.215, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.414033555582781e-05, | |
| "loss": 4.2112, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4131965987121024e-05, | |
| "loss": 4.2263, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4123580039610504e-05, | |
| "loss": 4.2002, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4115194092099984e-05, | |
| "loss": 4.2138, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4106808144589464e-05, | |
| "loss": 4.2188, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4098422197078944e-05, | |
| "loss": 4.2143, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4090036249568424e-05, | |
| "loss": 4.2121, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4081650302057904e-05, | |
| "loss": 4.2233, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.407326435454738e-05, | |
| "loss": 4.2112, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.406487840703686e-05, | |
| "loss": 4.2076, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.405650883833007e-05, | |
| "loss": 4.2194, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.404812289081955e-05, | |
| "loss": 4.2152, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4039736943309026e-05, | |
| "loss": 4.2248, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403135099579851e-05, | |
| "loss": 4.2198, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.402298142709173e-05, | |
| "loss": 4.2184, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.40145954795812e-05, | |
| "loss": 4.2034, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.400620953207068e-05, | |
| "loss": 4.2134, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.399782358456016e-05, | |
| "loss": 4.213, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.398943763704964e-05, | |
| "loss": 4.2003, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.398105168953912e-05, | |
| "loss": 4.2201, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.397268212083233e-05, | |
| "loss": 4.2026, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.396429617332181e-05, | |
| "loss": 4.2169, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.395591022581129e-05, | |
| "loss": 4.2065, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.394752427830077e-05, | |
| "loss": 4.2118, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.393915470959398e-05, | |
| "loss": 4.191, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3930768762083466e-05, | |
| "loss": 4.2107, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3922382814572946e-05, | |
| "loss": 4.2067, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3913996867062426e-05, | |
| "loss": 4.2118, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3905610919551906e-05, | |
| "loss": 4.2141, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3897224972041386e-05, | |
| "loss": 4.2192, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3888839024530866e-05, | |
| "loss": 4.1961, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3880469455824075e-05, | |
| "loss": 4.2117, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3872083508313555e-05, | |
| "loss": 4.2176, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3863697560803035e-05, | |
| "loss": 4.2172, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3855311613292515e-05, | |
| "loss": 4.2123, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3846925665781995e-05, | |
| "loss": 4.214, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3838539718271475e-05, | |
| "loss": 4.2082, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3830153770760955e-05, | |
| "loss": 4.2125, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3821767823250435e-05, | |
| "loss": 4.1936, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3813398254543644e-05, | |
| "loss": 4.2071, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.380501230703313e-05, | |
| "loss": 4.2013, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.379662635952261e-05, | |
| "loss": 4.2031, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.378825679081582e-05, | |
| "loss": 4.2013, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.37798708433053e-05, | |
| "loss": 4.2026, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.377148489579478e-05, | |
| "loss": 4.2021, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.376309894828426e-05, | |
| "loss": 4.2003, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.375471300077374e-05, | |
| "loss": 4.2103, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.194133281707764, | |
| "eval_runtime": 312.5876, | |
| "eval_samples_per_second": 1220.749, | |
| "eval_steps_per_second": 38.149, | |
| "step": 381600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.374634343206695e-05, | |
| "loss": 4.1983, | |
| "step": 381952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.373795748455643e-05, | |
| "loss": 4.1925, | |
| "step": 382464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372957153704591e-05, | |
| "loss": 4.2049, | |
| "step": 382976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372120196833912e-05, | |
| "loss": 4.2017, | |
| "step": 383488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.37128160208286e-05, | |
| "loss": 4.2041, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3704430073318084e-05, | |
| "loss": 4.1937, | |
| "step": 384512 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3696044125807564e-05, | |
| "loss": 4.1923, | |
| "step": 385024 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.368765817829704e-05, | |
| "loss": 4.193, | |
| "step": 385536 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367928860959025e-05, | |
| "loss": 4.1993, | |
| "step": 386048 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367090266207973e-05, | |
| "loss": 4.2076, | |
| "step": 386560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.366251671456921e-05, | |
| "loss": 4.1943, | |
| "step": 387072 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3654130767058686e-05, | |
| "loss": 4.2074, | |
| "step": 387584 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3645744819548166e-05, | |
| "loss": 4.1854, | |
| "step": 388096 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.363737525084138e-05, | |
| "loss": 4.1914, | |
| "step": 388608 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.362898930333086e-05, | |
| "loss": 4.1985, | |
| "step": 389120 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3620603355820335e-05, | |
| "loss": 4.1824, | |
| "step": 389632 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.361221740830982e-05, | |
| "loss": 4.1919, | |
| "step": 390144 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.36038314607993e-05, | |
| "loss": 4.1953, | |
| "step": 390656 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.359544551328878e-05, | |
| "loss": 4.1868, | |
| "step": 391168 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.358705956577826e-05, | |
| "loss": 4.2157, | |
| "step": 391680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357867361826774e-05, | |
| "loss": 4.1984, | |
| "step": 392192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357030404956095e-05, | |
| "loss": 4.1943, | |
| "step": 392704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.356191810205043e-05, | |
| "loss": 4.1968, | |
| "step": 393216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.355354853334364e-05, | |
| "loss": 4.2077, | |
| "step": 393728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.354516258583312e-05, | |
| "loss": 4.1861, | |
| "step": 394240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.35367766383226e-05, | |
| "loss": 4.1953, | |
| "step": 394752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.352839069081208e-05, | |
| "loss": 4.193, | |
| "step": 395264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.352002112210529e-05, | |
| "loss": 4.1881, | |
| "step": 395776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3511635174594776e-05, | |
| "loss": 4.181, | |
| "step": 396288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3503249227084256e-05, | |
| "loss": 4.196, | |
| "step": 396800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3494863279573736e-05, | |
| "loss": 4.204, | |
| "step": 397312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3486477332063216e-05, | |
| "loss": 4.1944, | |
| "step": 397824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3478091384552696e-05, | |
| "loss": 4.198, | |
| "step": 398336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3469705437042175e-05, | |
| "loss": 4.1927, | |
| "step": 398848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3461335868335385e-05, | |
| "loss": 4.1912, | |
| "step": 399360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3452949920824865e-05, | |
| "loss": 4.1943, | |
| "step": 399872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3444580352118074e-05, | |
| "loss": 4.1811, | |
| "step": 400384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3436194404607554e-05, | |
| "loss": 4.1834, | |
| "step": 400896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3427808457097034e-05, | |
| "loss": 4.1783, | |
| "step": 401408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3419422509586514e-05, | |
| "loss": 4.182, | |
| "step": 401920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3411036562075993e-05, | |
| "loss": 4.1885, | |
| "step": 402432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3402650614565473e-05, | |
| "loss": 4.1908, | |
| "step": 402944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.339426466705496e-05, | |
| "loss": 4.1819, | |
| "step": 403456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.338587871954444e-05, | |
| "loss": 4.1862, | |
| "step": 403968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.337749277203392e-05, | |
| "loss": 4.1914, | |
| "step": 404480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.33691068245234e-05, | |
| "loss": 4.1873, | |
| "step": 404992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.336072087701287e-05, | |
| "loss": 4.1897, | |
| "step": 405504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.335233492950235e-05, | |
| "loss": 4.1842, | |
| "step": 406016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.334396536079557e-05, | |
| "loss": 4.169, | |
| "step": 406528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.333559579208878e-05, | |
| "loss": 4.1936, | |
| "step": 407040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.332720984457826e-05, | |
| "loss": 4.1784, | |
| "step": 407552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331882389706774e-05, | |
| "loss": 4.1888, | |
| "step": 408064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331043794955722e-05, | |
| "loss": 4.1751, | |
| "step": 408576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.330206838085043e-05, | |
| "loss": 4.1867, | |
| "step": 409088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3293682433339914e-05, | |
| "loss": 4.1663, | |
| "step": 409600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3285296485829394e-05, | |
| "loss": 4.1907, | |
| "step": 410112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3276910538318874e-05, | |
| "loss": 4.1652, | |
| "step": 410624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326852459080835e-05, | |
| "loss": 4.1774, | |
| "step": 411136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326013864329783e-05, | |
| "loss": 4.1834, | |
| "step": 411648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.325175269578731e-05, | |
| "loss": 4.1716, | |
| "step": 412160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.324336674827679e-05, | |
| "loss": 4.1712, | |
| "step": 412672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.323498080076627e-05, | |
| "loss": 4.1831, | |
| "step": 413184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3226611232059476e-05, | |
| "loss": 4.1617, | |
| "step": 413696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3218225284548956e-05, | |
| "loss": 4.1614, | |
| "step": 414208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3209839337038436e-05, | |
| "loss": 4.1763, | |
| "step": 414720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3201453389527916e-05, | |
| "loss": 4.1803, | |
| "step": 415232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3193067442017396e-05, | |
| "loss": 4.1648, | |
| "step": 415744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.318469787331061e-05, | |
| "loss": 4.1787, | |
| "step": 416256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.317631192580009e-05, | |
| "loss": 4.1543, | |
| "step": 416768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.316792597828957e-05, | |
| "loss": 4.1708, | |
| "step": 417280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315954003077905e-05, | |
| "loss": 4.1787, | |
| "step": 417792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315115408326853e-05, | |
| "loss": 4.1773, | |
| "step": 418304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.314276813575801e-05, | |
| "loss": 4.1805, | |
| "step": 418816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.313438218824749e-05, | |
| "loss": 4.1828, | |
| "step": 419328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.312599624073697e-05, | |
| "loss": 4.191, | |
| "step": 419840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.311764305083391e-05, | |
| "loss": 4.1606, | |
| "step": 420352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310925710332339e-05, | |
| "loss": 4.1763, | |
| "step": 420864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310087115581287e-05, | |
| "loss": 4.1795, | |
| "step": 421376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.309248520830235e-05, | |
| "loss": 4.1537, | |
| "step": 421888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3084115639595565e-05, | |
| "loss": 4.178, | |
| "step": 422400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3075729692085045e-05, | |
| "loss": 4.1722, | |
| "step": 422912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3067343744574525e-05, | |
| "loss": 4.1771, | |
| "step": 423424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3058957797064005e-05, | |
| "loss": 4.1781, | |
| "step": 423936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3050571849553485e-05, | |
| "loss": 4.1627, | |
| "step": 424448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3042202280846694e-05, | |
| "loss": 4.1696, | |
| "step": 424960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3033816333336174e-05, | |
| "loss": 4.1715, | |
| "step": 425472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3025430385825654e-05, | |
| "loss": 4.1815, | |
| "step": 425984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3017044438315134e-05, | |
| "loss": 4.1695, | |
| "step": 426496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3008658490804614e-05, | |
| "loss": 4.1805, | |
| "step": 427008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3000272543294094e-05, | |
| "loss": 4.1805, | |
| "step": 427520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2991886595783574e-05, | |
| "loss": 4.1647, | |
| "step": 428032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2983500648273054e-05, | |
| "loss": 4.1728, | |
| "step": 428544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.297513107956627e-05, | |
| "loss": 4.1632, | |
| "step": 429056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.296674513205575e-05, | |
| "loss": 4.1746, | |
| "step": 429568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.295835918454523e-05, | |
| "loss": 4.184, | |
| "step": 430080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294997323703471e-05, | |
| "loss": 4.1608, | |
| "step": 430592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294162004713165e-05, | |
| "loss": 4.1649, | |
| "step": 431104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.293323409962113e-05, | |
| "loss": 4.1597, | |
| "step": 431616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.292484815211061e-05, | |
| "loss": 4.173, | |
| "step": 432128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.291646220460009e-05, | |
| "loss": 4.1803, | |
| "step": 432640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.290807625708957e-05, | |
| "loss": 4.1646, | |
| "step": 433152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289969030957905e-05, | |
| "loss": 4.16, | |
| "step": 433664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289130436206853e-05, | |
| "loss": 4.1532, | |
| "step": 434176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.288291841455801e-05, | |
| "loss": 4.1759, | |
| "step": 434688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.287454884585122e-05, | |
| "loss": 4.1461, | |
| "step": 435200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.28661628983407e-05, | |
| "loss": 4.1645, | |
| "step": 435712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.285779332963391e-05, | |
| "loss": 4.1622, | |
| "step": 436224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284940738212339e-05, | |
| "loss": 4.1654, | |
| "step": 436736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.28410378134166e-05, | |
| "loss": 4.161, | |
| "step": 437248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.283265186590608e-05, | |
| "loss": 4.1673, | |
| "step": 437760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.282426591839556e-05, | |
| "loss": 4.162, | |
| "step": 438272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.281587997088504e-05, | |
| "loss": 4.1559, | |
| "step": 438784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.280749402337452e-05, | |
| "loss": 4.1657, | |
| "step": 439296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2799108075864e-05, | |
| "loss": 4.165, | |
| "step": 439808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2790722128353474e-05, | |
| "loss": 4.1699, | |
| "step": 440320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.278233618084296e-05, | |
| "loss": 4.1716, | |
| "step": 440832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.277396661213618e-05, | |
| "loss": 4.163, | |
| "step": 441344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.276558066462566e-05, | |
| "loss": 4.1526, | |
| "step": 441856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.275719471711513e-05, | |
| "loss": 4.1609, | |
| "step": 442368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.274880876960461e-05, | |
| "loss": 4.1652, | |
| "step": 442880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.274042282209409e-05, | |
| "loss": 4.1518, | |
| "step": 443392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.273203687458357e-05, | |
| "loss": 4.1652, | |
| "step": 443904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.272365092707305e-05, | |
| "loss": 4.1507, | |
| "step": 444416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.271526497956253e-05, | |
| "loss": 4.1654, | |
| "step": 444928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.270689541085574e-05, | |
| "loss": 4.1558, | |
| "step": 445440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.269850946334522e-05, | |
| "loss": 4.1589, | |
| "step": 445952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.26901235158347e-05, | |
| "loss": 4.1463, | |
| "step": 446464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.268173756832418e-05, | |
| "loss": 4.157, | |
| "step": 446976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2673367999617395e-05, | |
| "loss": 4.1589, | |
| "step": 447488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2664982052106875e-05, | |
| "loss": 4.1638, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2656596104596355e-05, | |
| "loss": 4.159, | |
| "step": 448512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2648210157085835e-05, | |
| "loss": 4.1662, | |
| "step": 449024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2639824209575315e-05, | |
| "loss": 4.1483, | |
| "step": 449536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2631454640868524e-05, | |
| "loss": 4.16, | |
| "step": 450048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2623068693358004e-05, | |
| "loss": 4.1675, | |
| "step": 450560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2614682745847484e-05, | |
| "loss": 4.1654, | |
| "step": 451072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2606296798336964e-05, | |
| "loss": 4.1632, | |
| "step": 451584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2597910850826443e-05, | |
| "loss": 4.1674, | |
| "step": 452096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2589524903315923e-05, | |
| "loss": 4.1557, | |
| "step": 452608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2581138955805403e-05, | |
| "loss": 4.1648, | |
| "step": 453120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.257275300829488e-05, | |
| "loss": 4.1454, | |
| "step": 453632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.256436706078436e-05, | |
| "loss": 4.1567, | |
| "step": 454144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.255599749207758e-05, | |
| "loss": 4.1511, | |
| "step": 454656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.254762792337079e-05, | |
| "loss": 4.153, | |
| "step": 455168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253924197586027e-05, | |
| "loss": 4.1541, | |
| "step": 455680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253085602834975e-05, | |
| "loss": 4.1486, | |
| "step": 456192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.252247008083923e-05, | |
| "loss": 4.1597, | |
| "step": 456704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.251408413332871e-05, | |
| "loss": 4.1447, | |
| "step": 457216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.250569818581819e-05, | |
| "loss": 4.1659, | |
| "step": 457728 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.1535773277282715, | |
| "eval_runtime": 302.8427, | |
| "eval_samples_per_second": 1260.03, | |
| "eval_steps_per_second": 39.377, | |
| "step": 457920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.249731223830766e-05, | |
| "loss": 4.1472, | |
| "step": 458240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.248892629079714e-05, | |
| "loss": 4.1437, | |
| "step": 458752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.248054034328662e-05, | |
| "loss": 4.1562, | |
| "step": 459264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.24721543957761e-05, | |
| "loss": 4.1526, | |
| "step": 459776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.246376844826558e-05, | |
| "loss": 4.1555, | |
| "step": 460288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.245538250075506e-05, | |
| "loss": 4.1467, | |
| "step": 460800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.244699655324455e-05, | |
| "loss": 4.1367, | |
| "step": 461312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243861060573403e-05, | |
| "loss": 4.1481, | |
| "step": 461824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243022465822351e-05, | |
| "loss": 4.1502, | |
| "step": 462336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.242185508951672e-05, | |
| "loss": 4.1578, | |
| "step": 462848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.24134691420062e-05, | |
| "loss": 4.1502, | |
| "step": 463360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.240508319449568e-05, | |
| "loss": 4.1543, | |
| "step": 463872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.239669724698516e-05, | |
| "loss": 4.1401, | |
| "step": 464384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2388311299474637e-05, | |
| "loss": 4.1429, | |
| "step": 464896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2379925351964117e-05, | |
| "loss": 4.1496, | |
| "step": 465408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2371539404453596e-05, | |
| "loss": 4.1348, | |
| "step": 465920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2363153456943076e-05, | |
| "loss": 4.1439, | |
| "step": 466432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.235476750943255e-05, | |
| "loss": 4.1492, | |
| "step": 466944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.234638156192203e-05, | |
| "loss": 4.1417, | |
| "step": 467456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2337995614411516e-05, | |
| "loss": 4.1656, | |
| "step": 467968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2329609666900996e-05, | |
| "loss": 4.1554, | |
| "step": 468480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2321223719390476e-05, | |
| "loss": 4.1436, | |
| "step": 468992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2312837771879956e-05, | |
| "loss": 4.1498, | |
| "step": 469504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2304468203173165e-05, | |
| "loss": 4.1629, | |
| "step": 470016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2296082255662645e-05, | |
| "loss": 4.1391, | |
| "step": 470528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2287696308152125e-05, | |
| "loss": 4.1507, | |
| "step": 471040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2279310360641605e-05, | |
| "loss": 4.1426, | |
| "step": 471552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2270924413131085e-05, | |
| "loss": 4.1423, | |
| "step": 472064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2262538465620565e-05, | |
| "loss": 4.1371, | |
| "step": 472576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2254152518110045e-05, | |
| "loss": 4.1478, | |
| "step": 473088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2245766570599525e-05, | |
| "loss": 4.1553, | |
| "step": 473600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2237397001892734e-05, | |
| "loss": 4.1466, | |
| "step": 474112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2229011054382214e-05, | |
| "loss": 4.1534, | |
| "step": 474624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.22206251068717e-05, | |
| "loss": 4.1496, | |
| "step": 475136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.221225553816491e-05, | |
| "loss": 4.1426, | |
| "step": 475648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.220386959065439e-05, | |
| "loss": 4.1475, | |
| "step": 476160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.219548364314387e-05, | |
| "loss": 4.1377, | |
| "step": 476672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.218711407443708e-05, | |
| "loss": 4.1358, | |
| "step": 477184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.217872812692656e-05, | |
| "loss": 4.1343, | |
| "step": 477696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.217034217941604e-05, | |
| "loss": 4.1343, | |
| "step": 478208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.216195623190552e-05, | |
| "loss": 4.1412, | |
| "step": 478720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2153570284395e-05, | |
| "loss": 4.1468, | |
| "step": 479232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.214518433688448e-05, | |
| "loss": 4.1361, | |
| "step": 479744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.213679838937396e-05, | |
| "loss": 4.1372, | |
| "step": 480256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.212841244186344e-05, | |
| "loss": 4.148, | |
| "step": 480768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.212002649435292e-05, | |
| "loss": 4.1409, | |
| "step": 481280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.21116405468424e-05, | |
| "loss": 4.1465, | |
| "step": 481792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.210325459933188e-05, | |
| "loss": 4.1389, | |
| "step": 482304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.209486865182136e-05, | |
| "loss": 4.1201, | |
| "step": 482816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.208649908311457e-05, | |
| "loss": 4.1515, | |
| "step": 483328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.207812951440778e-05, | |
| "loss": 4.1314, | |
| "step": 483840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.206974356689726e-05, | |
| "loss": 4.1451, | |
| "step": 484352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2061357619386736e-05, | |
| "loss": 4.134, | |
| "step": 484864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2052971671876216e-05, | |
| "loss": 4.1357, | |
| "step": 485376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.204460210316943e-05, | |
| "loss": 4.1289, | |
| "step": 485888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.203621615565891e-05, | |
| "loss": 4.1424, | |
| "step": 486400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2027830208148385e-05, | |
| "loss": 4.1245, | |
| "step": 486912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201944426063787e-05, | |
| "loss": 4.1331, | |
| "step": 487424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201105831312735e-05, | |
| "loss": 4.1429, | |
| "step": 487936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.200267236561683e-05, | |
| "loss": 4.1234, | |
| "step": 488448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.199428641810631e-05, | |
| "loss": 4.1269, | |
| "step": 488960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.198590047059579e-05, | |
| "loss": 4.1443, | |
| "step": 489472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.197751452308527e-05, | |
| "loss": 4.1143, | |
| "step": 489984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.196912857557475e-05, | |
| "loss": 4.1184, | |
| "step": 490496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.196074262806423e-05, | |
| "loss": 4.1328, | |
| "step": 491008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.195235668055371e-05, | |
| "loss": 4.1364, | |
| "step": 491520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.194398711184692e-05, | |
| "loss": 4.1246, | |
| "step": 492032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.193561754314013e-05, | |
| "loss": 4.1364, | |
| "step": 492544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.192723159562961e-05, | |
| "loss": 4.1147, | |
| "step": 493056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.191884564811909e-05, | |
| "loss": 4.1259, | |
| "step": 493568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.191045970060857e-05, | |
| "loss": 4.1367, | |
| "step": 494080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1902073753098057e-05, | |
| "loss": 4.1336, | |
| "step": 494592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1893704184391266e-05, | |
| "loss": 4.1376, | |
| "step": 495104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1885318236880746e-05, | |
| "loss": 4.1378, | |
| "step": 495616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1876932289370226e-05, | |
| "loss": 4.1461, | |
| "step": 496128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1868562720663435e-05, | |
| "loss": 4.1192, | |
| "step": 496640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1860176773152915e-05, | |
| "loss": 4.1307, | |
| "step": 497152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1851790825642395e-05, | |
| "loss": 4.1363, | |
| "step": 497664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1843404878131875e-05, | |
| "loss": 4.1131, | |
| "step": 498176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1835018930621355e-05, | |
| "loss": 4.1338, | |
| "step": 498688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1826632983110834e-05, | |
| "loss": 4.1305, | |
| "step": 499200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1818247035600314e-05, | |
| "loss": 4.1354, | |
| "step": 499712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1809861088089794e-05, | |
| "loss": 4.1347, | |
| "step": 500224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1801475140579274e-05, | |
| "loss": 4.1227, | |
| "step": 500736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1793089193068754e-05, | |
| "loss": 4.1257, | |
| "step": 501248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1784703245558234e-05, | |
| "loss": 4.1299, | |
| "step": 501760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1776317298047714e-05, | |
| "loss": 4.1333, | |
| "step": 502272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.176794772934093e-05, | |
| "loss": 4.131, | |
| "step": 502784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.17595617818304e-05, | |
| "loss": 4.138, | |
| "step": 503296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.175119221312362e-05, | |
| "loss": 4.1372, | |
| "step": 503808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.17428062656131e-05, | |
| "loss": 4.1251, | |
| "step": 504320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.173442031810257e-05, | |
| "loss": 4.13, | |
| "step": 504832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.172603437059205e-05, | |
| "loss": 4.1196, | |
| "step": 505344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.171764842308153e-05, | |
| "loss": 4.1384, | |
| "step": 505856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.170926247557101e-05, | |
| "loss": 4.1376, | |
| "step": 506368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.170087652806049e-05, | |
| "loss": 4.1232, | |
| "step": 506880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.169249058054998e-05, | |
| "loss": 4.1217, | |
| "step": 507392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.168412101184319e-05, | |
| "loss": 4.1168, | |
| "step": 507904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.167573506433267e-05, | |
| "loss": 4.1331, | |
| "step": 508416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.166734911682215e-05, | |
| "loss": 4.1385, | |
| "step": 508928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165897954811536e-05, | |
| "loss": 4.1214, | |
| "step": 509440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165059360060484e-05, | |
| "loss": 4.1216, | |
| "step": 509952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.164220765309432e-05, | |
| "loss": 4.1165, | |
| "step": 510464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.16338217055838e-05, | |
| "loss": 4.1286, | |
| "step": 510976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.162543575807328e-05, | |
| "loss": 4.1096, | |
| "step": 511488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.161704981056276e-05, | |
| "loss": 4.1227, | |
| "step": 512000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160866386305224e-05, | |
| "loss": 4.1146, | |
| "step": 512512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160027791554172e-05, | |
| "loss": 4.128, | |
| "step": 513024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1591908346834926e-05, | |
| "loss": 4.1187, | |
| "step": 513536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.158353877812814e-05, | |
| "loss": 4.1262, | |
| "step": 514048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.157515283061762e-05, | |
| "loss": 4.1208, | |
| "step": 514560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.15667668831071e-05, | |
| "loss": 4.113, | |
| "step": 515072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.155838093559658e-05, | |
| "loss": 4.1269, | |
| "step": 515584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.154999498808606e-05, | |
| "loss": 4.1244, | |
| "step": 516096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.154162541937927e-05, | |
| "loss": 4.1295, | |
| "step": 516608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.153323947186875e-05, | |
| "loss": 4.1313, | |
| "step": 517120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.152485352435823e-05, | |
| "loss": 4.1185, | |
| "step": 517632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.151646757684771e-05, | |
| "loss": 4.1215, | |
| "step": 518144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.150808162933719e-05, | |
| "loss": 4.1144, | |
| "step": 518656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.149969568182667e-05, | |
| "loss": 4.1259, | |
| "step": 519168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.149130973431615e-05, | |
| "loss": 4.1186, | |
| "step": 519680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.148292378680563e-05, | |
| "loss": 4.1206, | |
| "step": 520192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1474554218098846e-05, | |
| "loss": 4.1073, | |
| "step": 520704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1466184649392055e-05, | |
| "loss": 4.1238, | |
| "step": 521216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1457815080685264e-05, | |
| "loss": 4.1138, | |
| "step": 521728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1449429133174744e-05, | |
| "loss": 4.1206, | |
| "step": 522240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1441043185664224e-05, | |
| "loss": 4.1038, | |
| "step": 522752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1432657238153704e-05, | |
| "loss": 4.1206, | |
| "step": 523264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1424271290643184e-05, | |
| "loss": 4.1154, | |
| "step": 523776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1415885343132664e-05, | |
| "loss": 4.1294, | |
| "step": 524288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1407499395622144e-05, | |
| "loss": 4.1181, | |
| "step": 524800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1399113448111624e-05, | |
| "loss": 4.125, | |
| "step": 525312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1390727500601104e-05, | |
| "loss": 4.1083, | |
| "step": 525824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1382341553090584e-05, | |
| "loss": 4.1235, | |
| "step": 526336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1373955605580064e-05, | |
| "loss": 4.1275, | |
| "step": 526848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1365569658069544e-05, | |
| "loss": 4.1249, | |
| "step": 527360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.135720008936276e-05, | |
| "loss": 4.1238, | |
| "step": 527872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.134881414185223e-05, | |
| "loss": 4.1277, | |
| "step": 528384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.134042819434171e-05, | |
| "loss": 4.1157, | |
| "step": 528896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.133204224683119e-05, | |
| "loss": 4.1267, | |
| "step": 529408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.132365629932067e-05, | |
| "loss": 4.107, | |
| "step": 529920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.131528673061388e-05, | |
| "loss": 4.1181, | |
| "step": 530432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.130690078310336e-05, | |
| "loss": 4.1118, | |
| "step": 530944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.129853121439658e-05, | |
| "loss": 4.1145, | |
| "step": 531456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.129014526688606e-05, | |
| "loss": 4.1105, | |
| "step": 531968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.128175931937554e-05, | |
| "loss": 4.1088, | |
| "step": 532480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.127337337186502e-05, | |
| "loss": 4.1195, | |
| "step": 532992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.126500380315823e-05, | |
| "loss": 4.1104, | |
| "step": 533504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1256617855647707e-05, | |
| "loss": 4.1289, | |
| "step": 534016 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.122705459594727, | |
| "eval_runtime": 295.5193, | |
| "eval_samples_per_second": 1291.256, | |
| "eval_steps_per_second": 40.353, | |
| "step": 534240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1248231908137186e-05, | |
| "loss": 4.1089, | |
| "step": 534528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1239845960626666e-05, | |
| "loss": 4.1035, | |
| "step": 535040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1231460013116146e-05, | |
| "loss": 4.1183, | |
| "step": 535552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1223074065605626e-05, | |
| "loss": 4.1133, | |
| "step": 536064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1214704496898835e-05, | |
| "loss": 4.1119, | |
| "step": 536576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1206318549388315e-05, | |
| "loss": 4.113, | |
| "step": 537088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1197932601877795e-05, | |
| "loss": 4.1042, | |
| "step": 537600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1189546654367275e-05, | |
| "loss": 4.1071, | |
| "step": 538112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1181160706856755e-05, | |
| "loss": 4.1134, | |
| "step": 538624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.117277475934624e-05, | |
| "loss": 4.1207, | |
| "step": 539136 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.116438881183572e-05, | |
| "loss": 4.1122, | |
| "step": 539648 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.115601924312893e-05, | |
| "loss": 4.1114, | |
| "step": 540160 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.114763329561841e-05, | |
| "loss": 4.1078, | |
| "step": 540672 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.113924734810789e-05, | |
| "loss": 4.1046, | |
| "step": 541184 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.113086140059737e-05, | |
| "loss": 4.1086, | |
| "step": 541696 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.112247545308685e-05, | |
| "loss": 4.0964, | |
| "step": 542208 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.111410588438006e-05, | |
| "loss": 4.1082, | |
| "step": 542720 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.110571993686954e-05, | |
| "loss": 4.1121, | |
| "step": 543232 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.109735036816275e-05, | |
| "loss": 4.1023, | |
| "step": 543744 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.108896442065223e-05, | |
| "loss": 4.1257, | |
| "step": 544256 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.108057847314171e-05, | |
| "loss": 4.1214, | |
| "step": 544768 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1072192525631196e-05, | |
| "loss": 4.1064, | |
| "step": 545280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1063806578120676e-05, | |
| "loss": 4.1121, | |
| "step": 545792 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1055420630610156e-05, | |
| "loss": 4.123, | |
| "step": 546304 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1047034683099636e-05, | |
| "loss": 4.1015, | |
| "step": 546816 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1038648735589115e-05, | |
| "loss": 4.1138, | |
| "step": 547328 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1030279166882325e-05, | |
| "loss": 4.1069, | |
| "step": 547840 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1021893219371805e-05, | |
| "loss": 4.1047, | |
| "step": 548352 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1013507271861285e-05, | |
| "loss": 4.0987, | |
| "step": 548864 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1005121324350764e-05, | |
| "loss": 4.1137, | |
| "step": 549376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0996735376840244e-05, | |
| "loss": 4.1167, | |
| "step": 549888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0988365808133454e-05, | |
| "loss": 4.1138, | |
| "step": 550400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0979979860622933e-05, | |
| "loss": 4.116, | |
| "step": 550912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0971593913112413e-05, | |
| "loss": 4.107, | |
| "step": 551424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.096320796560189e-05, | |
| "loss": 4.1053, | |
| "step": 551936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.095483839689511e-05, | |
| "loss": 4.1126, | |
| "step": 552448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.094645244938459e-05, | |
| "loss": 4.1032, | |
| "step": 552960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.09380828806778e-05, | |
| "loss": 4.0947, | |
| "step": 553472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.092969693316728e-05, | |
| "loss": 4.0977, | |
| "step": 553984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.092131098565676e-05, | |
| "loss": 4.0984, | |
| "step": 554496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.091292503814624e-05, | |
| "loss": 4.1027, | |
| "step": 555008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.090453909063572e-05, | |
| "loss": 4.1103, | |
| "step": 555520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.089615314312519e-05, | |
| "loss": 4.1032, | |
| "step": 556032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.088776719561467e-05, | |
| "loss": 4.0982, | |
| "step": 556544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.087938124810415e-05, | |
| "loss": 4.1121, | |
| "step": 557056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.087101167939737e-05, | |
| "loss": 4.1076, | |
| "step": 557568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.086262573188685e-05, | |
| "loss": 4.108, | |
| "step": 558080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.085423978437633e-05, | |
| "loss": 4.1047, | |
| "step": 558592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.084585383686581e-05, | |
| "loss": 4.0793, | |
| "step": 559104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.083746788935529e-05, | |
| "loss": 4.1193, | |
| "step": 559616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.082908194184477e-05, | |
| "loss": 4.0957, | |
| "step": 560128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.082069599433425e-05, | |
| "loss": 4.1106, | |
| "step": 560640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.081231004682373e-05, | |
| "loss": 4.0963, | |
| "step": 561152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.080392409931321e-05, | |
| "loss": 4.1006, | |
| "step": 561664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0795554530606416e-05, | |
| "loss": 4.0924, | |
| "step": 562176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0787168583095896e-05, | |
| "loss": 4.104, | |
| "step": 562688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0778799014389105e-05, | |
| "loss": 4.0917, | |
| "step": 563200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0770413066878585e-05, | |
| "loss": 4.0959, | |
| "step": 563712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0762027119368065e-05, | |
| "loss": 4.109, | |
| "step": 564224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.075364117185755e-05, | |
| "loss": 4.0896, | |
| "step": 564736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.074525522434703e-05, | |
| "loss": 4.0877, | |
| "step": 565248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.073686927683651e-05, | |
| "loss": 4.1073, | |
| "step": 565760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.072848332932599e-05, | |
| "loss": 4.083, | |
| "step": 566272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.072009738181547e-05, | |
| "loss": 4.0815, | |
| "step": 566784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.071172781310868e-05, | |
| "loss": 4.0983, | |
| "step": 567296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.070334186559816e-05, | |
| "loss": 4.0997, | |
| "step": 567808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.069495591808764e-05, | |
| "loss": 4.0908, | |
| "step": 568320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.068656997057712e-05, | |
| "loss": 4.1007, | |
| "step": 568832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.067820040187033e-05, | |
| "loss": 4.0773, | |
| "step": 569344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.066981445435981e-05, | |
| "loss": 4.0919, | |
| "step": 569856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.066142850684929e-05, | |
| "loss": 4.1013, | |
| "step": 570368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.065304255933877e-05, | |
| "loss": 4.0959, | |
| "step": 570880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0644672990631985e-05, | |
| "loss": 4.1058, | |
| "step": 571392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0636287043121465e-05, | |
| "loss": 4.1024, | |
| "step": 571904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0627901095610945e-05, | |
| "loss": 4.1085, | |
| "step": 572416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0619531526904154e-05, | |
| "loss": 4.0834, | |
| "step": 572928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0611145579393634e-05, | |
| "loss": 4.0989, | |
| "step": 573440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0602759631883114e-05, | |
| "loss": 4.0995, | |
| "step": 573952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0594373684372594e-05, | |
| "loss": 4.0815, | |
| "step": 574464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0585987736862074e-05, | |
| "loss": 4.0977, | |
| "step": 574976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.057761816815528e-05, | |
| "loss": 4.0974, | |
| "step": 575488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.056923222064476e-05, | |
| "loss": 4.1008, | |
| "step": 576000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.056084627313424e-05, | |
| "loss": 4.1044, | |
| "step": 576512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.055246032562372e-05, | |
| "loss": 4.0833, | |
| "step": 577024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.054409075691694e-05, | |
| "loss": 4.0905, | |
| "step": 577536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.053570480940642e-05, | |
| "loss": 4.0961, | |
| "step": 578048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.052733524069963e-05, | |
| "loss": 4.1003, | |
| "step": 578560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.051894929318911e-05, | |
| "loss": 4.0991, | |
| "step": 579072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.051056334567859e-05, | |
| "loss": 4.0999, | |
| "step": 579584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.050217739816807e-05, | |
| "loss": 4.1023, | |
| "step": 580096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.049379145065755e-05, | |
| "loss": 4.0947, | |
| "step": 580608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.048540550314702e-05, | |
| "loss": 4.0946, | |
| "step": 581120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.04770195556365e-05, | |
| "loss": 4.0884, | |
| "step": 581632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.046863360812598e-05, | |
| "loss": 4.1015, | |
| "step": 582144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.046024766061546e-05, | |
| "loss": 4.1013, | |
| "step": 582656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.045186171310494e-05, | |
| "loss": 4.0929, | |
| "step": 583168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0443492144398157e-05, | |
| "loss": 4.0851, | |
| "step": 583680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0435106196887637e-05, | |
| "loss": 4.0837, | |
| "step": 584192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0426720249377116e-05, | |
| "loss": 4.0975, | |
| "step": 584704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0418350680670326e-05, | |
| "loss": 4.1071, | |
| "step": 585216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0409964733159806e-05, | |
| "loss": 4.0887, | |
| "step": 585728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0401578785649285e-05, | |
| "loss": 4.0883, | |
| "step": 586240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0393192838138765e-05, | |
| "loss": 4.0869, | |
| "step": 586752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0384806890628245e-05, | |
| "loss": 4.0912, | |
| "step": 587264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0376420943117725e-05, | |
| "loss": 4.0786, | |
| "step": 587776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0368034995607205e-05, | |
| "loss": 4.0916, | |
| "step": 588288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0359649048096685e-05, | |
| "loss": 4.0786, | |
| "step": 588800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0351263100586165e-05, | |
| "loss": 4.0926, | |
| "step": 589312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0342877153075645e-05, | |
| "loss": 4.0895, | |
| "step": 589824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0334491205565125e-05, | |
| "loss": 4.0902, | |
| "step": 590336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0326105258054605e-05, | |
| "loss": 4.0863, | |
| "step": 590848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.031773568934782e-05, | |
| "loss": 4.086, | |
| "step": 591360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.030936612064103e-05, | |
| "loss": 4.0903, | |
| "step": 591872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.030098017313051e-05, | |
| "loss": 4.0914, | |
| "step": 592384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.029259422561999e-05, | |
| "loss": 4.0953, | |
| "step": 592896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.028420827810947e-05, | |
| "loss": 4.1023, | |
| "step": 593408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.027582233059895e-05, | |
| "loss": 4.0864, | |
| "step": 593920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.026743638308843e-05, | |
| "loss": 4.0893, | |
| "step": 594432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.025905043557791e-05, | |
| "loss": 4.083, | |
| "step": 594944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.025068086687112e-05, | |
| "loss": 4.0898, | |
| "step": 595456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.02422949193606e-05, | |
| "loss": 4.0879, | |
| "step": 595968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.023390897185008e-05, | |
| "loss": 4.0874, | |
| "step": 596480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.022552302433956e-05, | |
| "loss": 4.0759, | |
| "step": 596992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.021713707682904e-05, | |
| "loss": 4.0885, | |
| "step": 597504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0208767508122255e-05, | |
| "loss": 4.0865, | |
| "step": 598016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0200381560611735e-05, | |
| "loss": 4.0859, | |
| "step": 598528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0191995613101214e-05, | |
| "loss": 4.0679, | |
| "step": 599040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.018360966559069e-05, | |
| "loss": 4.0934, | |
| "step": 599552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.017522371808017e-05, | |
| "loss": 4.0811, | |
| "step": 600064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.016683777056965e-05, | |
| "loss": 4.0953, | |
| "step": 600576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.015846820186286e-05, | |
| "loss": 4.0879, | |
| "step": 601088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.015008225435234e-05, | |
| "loss": 4.0923, | |
| "step": 601600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0141696306841817e-05, | |
| "loss": 4.0757, | |
| "step": 602112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0133310359331297e-05, | |
| "loss": 4.0933, | |
| "step": 602624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.012492441182078e-05, | |
| "loss": 4.095, | |
| "step": 603136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.011653846431026e-05, | |
| "loss": 4.0927, | |
| "step": 603648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.010815251679974e-05, | |
| "loss": 4.0929, | |
| "step": 604160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.009976656928922e-05, | |
| "loss": 4.0931, | |
| "step": 604672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.009139700058243e-05, | |
| "loss": 4.0844, | |
| "step": 605184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.008301105307191e-05, | |
| "loss": 4.0942, | |
| "step": 605696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.007462510556139e-05, | |
| "loss": 4.0774, | |
| "step": 606208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.006623915805087e-05, | |
| "loss": 4.089, | |
| "step": 606720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.005786958934408e-05, | |
| "loss": 4.0804, | |
| "step": 607232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.004948364183356e-05, | |
| "loss": 4.0884, | |
| "step": 607744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.004111407312677e-05, | |
| "loss": 4.0747, | |
| "step": 608256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.003272812561625e-05, | |
| "loss": 4.0786, | |
| "step": 608768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.002434217810574e-05, | |
| "loss": 4.0837, | |
| "step": 609280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.001595623059522e-05, | |
| "loss": 4.0832, | |
| "step": 609792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.00075702830847e-05, | |
| "loss": 4.0969, | |
| "step": 610304 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.099428653717041, | |
| "eval_runtime": 294.2439, | |
| "eval_samples_per_second": 1296.853, | |
| "eval_steps_per_second": 40.528, | |
| "step": 610560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.999918433557418e-05, | |
| "loss": 4.0751, | |
| "step": 610816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.999079838806366e-05, | |
| "loss": 4.0768, | |
| "step": 611328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.998241244055314e-05, | |
| "loss": 4.08, | |
| "step": 611840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.997402649304262e-05, | |
| "loss": 4.0823, | |
| "step": 612352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.99656405455321e-05, | |
| "loss": 4.0854, | |
| "step": 612864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9957254598021577e-05, | |
| "loss": 4.078, | |
| "step": 613376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.994886865051105e-05, | |
| "loss": 4.0829, | |
| "step": 613888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.994048270300053e-05, | |
| "loss": 4.0761, | |
| "step": 614400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.993209675549001e-05, | |
| "loss": 4.0778, | |
| "step": 614912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.992371080797949e-05, | |
| "loss": 4.0932, | |
| "step": 615424 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.991532486046897e-05, | |
| "loss": 4.0784, | |
| "step": 615936 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.990693891295845e-05, | |
| "loss": 4.0814, | |
| "step": 616448 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.989855296544793e-05, | |
| "loss": 4.0765, | |
| "step": 616960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9890167017937416e-05, | |
| "loss": 4.0699, | |
| "step": 617472 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9881781070426896e-05, | |
| "loss": 4.0824, | |
| "step": 617984 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9873395122916376e-05, | |
| "loss": 4.0667, | |
| "step": 618496 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9865009175405856e-05, | |
| "loss": 4.0787, | |
| "step": 619008 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9856623227895336e-05, | |
| "loss": 4.0826, | |
| "step": 619520 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9848237280384816e-05, | |
| "loss": 4.0702, | |
| "step": 620032 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.983985133287429e-05, | |
| "loss": 4.0927, | |
| "step": 620544 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9831481764167505e-05, | |
| "loss": 4.0935, | |
| "step": 621056 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9823095816656985e-05, | |
| "loss": 4.0788, | |
| "step": 621568 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.981470986914646e-05, | |
| "loss": 4.0788, | |
| "step": 622080 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.980632392163594e-05, | |
| "loss": 4.0911, | |
| "step": 622592 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9797954352929154e-05, | |
| "loss": 4.0777, | |
| "step": 623104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9789568405418634e-05, | |
| "loss": 4.0773, | |
| "step": 623616 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9781182457908114e-05, | |
| "loss": 4.077, | |
| "step": 624128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9772796510397594e-05, | |
| "loss": 4.0747, | |
| "step": 624640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.976442694169081e-05, | |
| "loss": 4.0762, | |
| "step": 625152 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.975604099418029e-05, | |
| "loss": 4.0779, | |
| "step": 625664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.974765504666976e-05, | |
| "loss": 4.082, | |
| "step": 626176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.973928547796298e-05, | |
| "loss": 4.0857, | |
| "step": 626688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.973089953045246e-05, | |
| "loss": 4.0862, | |
| "step": 627200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.972251358294193e-05, | |
| "loss": 4.079, | |
| "step": 627712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.971412763543141e-05, | |
| "loss": 4.0741, | |
| "step": 628224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.970574168792089e-05, | |
| "loss": 4.0844, | |
| "step": 628736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.969737211921411e-05, | |
| "loss": 4.0747, | |
| "step": 629248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.968898617170359e-05, | |
| "loss": 4.066, | |
| "step": 629760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.968060022419307e-05, | |
| "loss": 4.0697, | |
| "step": 630272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.967221427668255e-05, | |
| "loss": 4.0669, | |
| "step": 630784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.966382832917203e-05, | |
| "loss": 4.0732, | |
| "step": 631296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.965544238166151e-05, | |
| "loss": 4.0785, | |
| "step": 631808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.964705643415099e-05, | |
| "loss": 4.0764, | |
| "step": 632320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.963867048664047e-05, | |
| "loss": 4.0735, | |
| "step": 632832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.963028453912995e-05, | |
| "loss": 4.0772, | |
| "step": 633344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9621914970423156e-05, | |
| "loss": 4.0789, | |
| "step": 633856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9613545401716366e-05, | |
| "loss": 4.0744, | |
| "step": 634368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9605159454205845e-05, | |
| "loss": 4.0774, | |
| "step": 634880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9596773506695325e-05, | |
| "loss": 4.0509, | |
| "step": 635392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9588387559184805e-05, | |
| "loss": 4.0891, | |
| "step": 635904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.958000161167429e-05, | |
| "loss": 4.0667, | |
| "step": 636416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.957161566416377e-05, | |
| "loss": 4.084, | |
| "step": 636928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.956322971665325e-05, | |
| "loss": 4.0655, | |
| "step": 637440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.955484376914273e-05, | |
| "loss": 4.0693, | |
| "step": 637952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.954647420043594e-05, | |
| "loss": 4.0665, | |
| "step": 638464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.953808825292542e-05, | |
| "loss": 4.0733, | |
| "step": 638976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.95297023054149e-05, | |
| "loss": 4.0644, | |
| "step": 639488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.952131635790438e-05, | |
| "loss": 4.0671, | |
| "step": 640000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.951294678919759e-05, | |
| "loss": 4.0782, | |
| "step": 640512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.95045772204908e-05, | |
| "loss": 4.0649, | |
| "step": 641024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.949619127298028e-05, | |
| "loss": 4.0563, | |
| "step": 641536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.948780532546976e-05, | |
| "loss": 4.0807, | |
| "step": 642048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9479419377959246e-05, | |
| "loss": 4.054, | |
| "step": 642560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9471033430448726e-05, | |
| "loss": 4.0511, | |
| "step": 643072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9462647482938206e-05, | |
| "loss": 4.0704, | |
| "step": 643584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9454261535427686e-05, | |
| "loss": 4.0723, | |
| "step": 644096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9445875587917166e-05, | |
| "loss": 4.0648, | |
| "step": 644608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9437506019210375e-05, | |
| "loss": 4.069, | |
| "step": 645120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9429120071699855e-05, | |
| "loss": 4.0509, | |
| "step": 645632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9420734124189335e-05, | |
| "loss": 4.0615, | |
| "step": 646144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9412348176678815e-05, | |
| "loss": 4.0736, | |
| "step": 646656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9403962229168295e-05, | |
| "loss": 4.0669, | |
| "step": 647168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9395592660461504e-05, | |
| "loss": 4.0789, | |
| "step": 647680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9387206712950984e-05, | |
| "loss": 4.0693, | |
| "step": 648192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9378820765440464e-05, | |
| "loss": 4.0823, | |
| "step": 648704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.937045119673368e-05, | |
| "loss": 4.0558, | |
| "step": 649216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.936206524922316e-05, | |
| "loss": 4.0705, | |
| "step": 649728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.935367930171264e-05, | |
| "loss": 4.067, | |
| "step": 650240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.934529335420212e-05, | |
| "loss": 4.0583, | |
| "step": 650752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.933690740669159e-05, | |
| "loss": 4.068, | |
| "step": 651264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.932852145918107e-05, | |
| "loss": 4.0717, | |
| "step": 651776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.932013551167055e-05, | |
| "loss": 4.0699, | |
| "step": 652288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.931176594296377e-05, | |
| "loss": 4.0776, | |
| "step": 652800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.930337999545324e-05, | |
| "loss": 4.0536, | |
| "step": 653312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.929499404794272e-05, | |
| "loss": 4.0591, | |
| "step": 653824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.92866081004322e-05, | |
| "loss": 4.0718, | |
| "step": 654336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.927822215292168e-05, | |
| "loss": 4.0697, | |
| "step": 654848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.926983620541116e-05, | |
| "loss": 4.0707, | |
| "step": 655360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.926145025790065e-05, | |
| "loss": 4.0739, | |
| "step": 655872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.925308068919386e-05, | |
| "loss": 4.0695, | |
| "step": 656384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.924469474168334e-05, | |
| "loss": 4.0691, | |
| "step": 656896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.923630879417282e-05, | |
| "loss": 4.0658, | |
| "step": 657408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.92279228466623e-05, | |
| "loss": 4.0592, | |
| "step": 657920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9219553277955506e-05, | |
| "loss": 4.0773, | |
| "step": 658432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9211167330444986e-05, | |
| "loss": 4.0694, | |
| "step": 658944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9202781382934466e-05, | |
| "loss": 4.0668, | |
| "step": 659456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9194395435423946e-05, | |
| "loss": 4.0614, | |
| "step": 659968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9186009487913426e-05, | |
| "loss": 4.0539, | |
| "step": 660480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9177639919206635e-05, | |
| "loss": 4.0673, | |
| "step": 660992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9169253971696115e-05, | |
| "loss": 4.0769, | |
| "step": 661504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.91608680241856e-05, | |
| "loss": 4.0621, | |
| "step": 662016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.915248207667508e-05, | |
| "loss": 4.0615, | |
| "step": 662528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.914409612916456e-05, | |
| "loss": 4.0589, | |
| "step": 663040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.913572656045777e-05, | |
| "loss": 4.0606, | |
| "step": 663552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.912734061294725e-05, | |
| "loss": 4.0582, | |
| "step": 664064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.911895466543673e-05, | |
| "loss": 4.0575, | |
| "step": 664576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.911056871792621e-05, | |
| "loss": 4.0505, | |
| "step": 665088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.910218277041569e-05, | |
| "loss": 4.068, | |
| "step": 665600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.90938132017089e-05, | |
| "loss": 4.0642, | |
| "step": 666112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.908542725419838e-05, | |
| "loss": 4.0623, | |
| "step": 666624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.907704130668786e-05, | |
| "loss": 4.0539, | |
| "step": 667136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.906865535917734e-05, | |
| "loss": 4.0629, | |
| "step": 667648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.906026941166682e-05, | |
| "loss": 4.0636, | |
| "step": 668160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.90518834641563e-05, | |
| "loss": 4.063, | |
| "step": 668672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9043497516645786e-05, | |
| "loss": 4.0687, | |
| "step": 669184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9035127947938995e-05, | |
| "loss": 4.0728, | |
| "step": 669696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9026758379232204e-05, | |
| "loss": 4.0574, | |
| "step": 670208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.901838881052541e-05, | |
| "loss": 4.0655, | |
| "step": 670720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.901000286301489e-05, | |
| "loss": 4.0501, | |
| "step": 671232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.900161691550437e-05, | |
| "loss": 4.0635, | |
| "step": 671744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.899323096799385e-05, | |
| "loss": 4.0614, | |
| "step": 672256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.898484502048333e-05, | |
| "loss": 4.0581, | |
| "step": 672768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.897645907297281e-05, | |
| "loss": 4.0532, | |
| "step": 673280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.896807312546229e-05, | |
| "loss": 4.0619, | |
| "step": 673792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.895968717795177e-05, | |
| "loss": 4.0612, | |
| "step": 674304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.895131760924499e-05, | |
| "loss": 4.0583, | |
| "step": 674816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.894293166173447e-05, | |
| "loss": 4.0433, | |
| "step": 675328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.893454571422395e-05, | |
| "loss": 4.0626, | |
| "step": 675840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.892615976671343e-05, | |
| "loss": 4.0543, | |
| "step": 676352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.89177738192029e-05, | |
| "loss": 4.0678, | |
| "step": 676864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.890938787169238e-05, | |
| "loss": 4.0629, | |
| "step": 677376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.890100192418186e-05, | |
| "loss": 4.0607, | |
| "step": 677888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.889261597667134e-05, | |
| "loss": 4.0514, | |
| "step": 678400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.888424640796455e-05, | |
| "loss": 4.0671, | |
| "step": 678912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.887587683925777e-05, | |
| "loss": 4.0676, | |
| "step": 679424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.886749089174725e-05, | |
| "loss": 4.0668, | |
| "step": 679936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.885910494423673e-05, | |
| "loss": 4.0672, | |
| "step": 680448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.885071899672621e-05, | |
| "loss": 4.0655, | |
| "step": 680960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.884233304921569e-05, | |
| "loss": 4.0594, | |
| "step": 681472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8833947101705167e-05, | |
| "loss": 4.066, | |
| "step": 681984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8825561154194647e-05, | |
| "loss": 4.0494, | |
| "step": 682496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8817175206684127e-05, | |
| "loss": 4.0617, | |
| "step": 683008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8808805637977336e-05, | |
| "loss": 4.0554, | |
| "step": 683520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8800419690466816e-05, | |
| "loss": 4.0581, | |
| "step": 684032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8792033742956296e-05, | |
| "loss": 4.0489, | |
| "step": 684544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8783664174249505e-05, | |
| "loss": 4.0517, | |
| "step": 685056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8775278226738985e-05, | |
| "loss": 4.0579, | |
| "step": 685568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.87669086580322e-05, | |
| "loss": 4.0589, | |
| "step": 686080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.875852271052168e-05, | |
| "loss": 4.0686, | |
| "step": 686592 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.08057165145874, | |
| "eval_runtime": 294.6161, | |
| "eval_samples_per_second": 1295.214, | |
| "eval_steps_per_second": 40.476, | |
| "step": 686880 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 2.8194795582850138e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |