| { | |
| "best_metric": 4.0643205642700195, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/superlative-quantifier/lstm/4/checkpoints/checkpoint-915840", | |
| "epoch": 1.0250006060157382, | |
| "eval_steps": 10, | |
| "global_step": 915840, | |
| "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": 10.8207, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 7.5574, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 7.0552, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 6.9906, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 6.946, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 6.8958, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 6.7246, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994129836742636e-05, | |
| "loss": 6.6133, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993291241991584e-05, | |
| "loss": 6.5249, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992452647240532e-05, | |
| "loss": 6.4571, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99161405248948e-05, | |
| "loss": 6.3993, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990777095618801e-05, | |
| "loss": 6.3338, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989938500867749e-05, | |
| "loss": 6.2675, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989099906116697e-05, | |
| "loss": 6.2045, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988261311365645e-05, | |
| "loss": 6.1463, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987422716614593e-05, | |
| "loss": 6.0968, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986584121863541e-05, | |
| "loss": 6.049, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985745527112489e-05, | |
| "loss": 6.0115, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984906932361437e-05, | |
| "loss": 5.9629, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984068337610385e-05, | |
| "loss": 5.9354, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.8991, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.982391148108281e-05, | |
| "loss": 5.8529, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.981552553357229e-05, | |
| "loss": 5.8222, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.980713958606178e-05, | |
| "loss": 5.7872, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.979875363855125e-05, | |
| "loss": 5.767, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.979036769104073e-05, | |
| "loss": 5.7332, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9781998122333946e-05, | |
| "loss": 5.7032, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773612174823426e-05, | |
| "loss": 5.6743, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.97652262273129e-05, | |
| "loss": 5.6674, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.975684027980238e-05, | |
| "loss": 5.6273, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.974845433229186e-05, | |
| "loss": 5.6144, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.974008476358507e-05, | |
| "loss": 5.6057, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.973169881607455e-05, | |
| "loss": 5.5758, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.972331286856403e-05, | |
| "loss": 5.5575, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714926921053515e-05, | |
| "loss": 5.5534, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9706557352346724e-05, | |
| "loss": 5.5314, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9698171404836204e-05, | |
| "loss": 5.5222, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9689785457325684e-05, | |
| "loss": 5.483, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9681399509815164e-05, | |
| "loss": 5.4775, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9673013562304644e-05, | |
| "loss": 5.4499, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966464399359785e-05, | |
| "loss": 5.447, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965625804608733e-05, | |
| "loss": 5.4338, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964787209857681e-05, | |
| "loss": 5.4286, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963948615106629e-05, | |
| "loss": 5.3972, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963110020355577e-05, | |
| "loss": 5.3972, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962273063484898e-05, | |
| "loss": 5.3838, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.961434468733847e-05, | |
| "loss": 5.3851, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.960595873982795e-05, | |
| "loss": 5.3776, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.959757279231743e-05, | |
| "loss": 5.3492, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958920322361064e-05, | |
| "loss": 5.3342, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958081727610012e-05, | |
| "loss": 5.3436, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95724313285896e-05, | |
| "loss": 5.3374, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956404538107908e-05, | |
| "loss": 5.3125, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955565943356856e-05, | |
| "loss": 5.3033, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.954728986486177e-05, | |
| "loss": 5.3002, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953890391735125e-05, | |
| "loss": 5.2821, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9530517969840727e-05, | |
| "loss": 5.2895, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9522132022330207e-05, | |
| "loss": 5.2555, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9513746074819686e-05, | |
| "loss": 5.2601, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95053765061129e-05, | |
| "loss": 5.25, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949699055860238e-05, | |
| "loss": 5.2477, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948860461109186e-05, | |
| "loss": 5.2243, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948021866358134e-05, | |
| "loss": 5.2321, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947183271607082e-05, | |
| "loss": 5.2042, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94634467685603e-05, | |
| "loss": 5.2114, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945507719985351e-05, | |
| "loss": 5.2097, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944669125234299e-05, | |
| "loss": 5.2035, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.943830530483247e-05, | |
| "loss": 5.1832, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942991935732195e-05, | |
| "loss": 5.1654, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942154978861516e-05, | |
| "loss": 5.1598, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.941316384110464e-05, | |
| "loss": 5.175, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.940477789359412e-05, | |
| "loss": 5.1631, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.93963919460836e-05, | |
| "loss": 5.1481, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388038756180545e-05, | |
| "loss": 5.1584, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379652808670025e-05, | |
| "loss": 5.1567, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371266861159505e-05, | |
| "loss": 5.1405, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362880913648985e-05, | |
| "loss": 5.1299, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9354494966138465e-05, | |
| "loss": 5.1196, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9346109018627945e-05, | |
| "loss": 5.1157, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9337723071117425e-05, | |
| "loss": 5.1023, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329337123606905e-05, | |
| "loss": 5.1036, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932095117609638e-05, | |
| "loss": 5.0931, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9312581607389594e-05, | |
| "loss": 5.1048, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9304195659879074e-05, | |
| "loss": 5.0914, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9295809712368554e-05, | |
| "loss": 5.0688, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9287423764858034e-05, | |
| "loss": 5.0707, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927905419615125e-05, | |
| "loss": 5.07, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927066824864073e-05, | |
| "loss": 5.0695, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.92622823011302e-05, | |
| "loss": 5.0631, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.925389635361968e-05, | |
| "loss": 5.0464, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.924554316371663e-05, | |
| "loss": 5.044, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923715721620611e-05, | |
| "loss": 5.0398, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922877126869559e-05, | |
| "loss": 5.0253, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922038532118507e-05, | |
| "loss": 5.0173, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921199937367455e-05, | |
| "loss": 5.0157, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920361342616403e-05, | |
| "loss": 5.0239, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919524385745724e-05, | |
| "loss": 5.0111, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918685790994672e-05, | |
| "loss": 4.9996, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.91784719624362e-05, | |
| "loss": 4.9906, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9170086014925676e-05, | |
| "loss": 4.9931, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9161700067415156e-05, | |
| "loss": 4.9954, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9153314119904636e-05, | |
| "loss": 4.9883, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9144928172394116e-05, | |
| "loss": 4.9821, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9136558603687325e-05, | |
| "loss": 4.9697, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9128172656176805e-05, | |
| "loss": 4.9795, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9119786708666285e-05, | |
| "loss": 4.9571, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9111400761155765e-05, | |
| "loss": 4.9607, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9103014813645245e-05, | |
| "loss": 4.9518, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.909462886613473e-05, | |
| "loss": 4.9462, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.908624291862421e-05, | |
| "loss": 4.9414, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907785697111369e-05, | |
| "loss": 4.946, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906947102360317e-05, | |
| "loss": 4.9338, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906110145489638e-05, | |
| "loss": 4.9318, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905271550738586e-05, | |
| "loss": 4.9294, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904432955987534e-05, | |
| "loss": 4.9234, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903594361236482e-05, | |
| "loss": 4.9276, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90275576648543e-05, | |
| "loss": 4.9204, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901918809614751e-05, | |
| "loss": 4.9169, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901080214863699e-05, | |
| "loss": 4.9141, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.900241620112647e-05, | |
| "loss": 4.9064, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899403025361595e-05, | |
| "loss": 4.9058, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.898564430610543e-05, | |
| "loss": 4.8941, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8977274737398646e-05, | |
| "loss": 4.8901, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968888789888125e-05, | |
| "loss": 4.888, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8960502842377605e-05, | |
| "loss": 4.8866, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8952133273670815e-05, | |
| "loss": 4.8816, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8943747326160294e-05, | |
| "loss": 4.8786, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8935361378649774e-05, | |
| "loss": 4.8764, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8926975431139254e-05, | |
| "loss": 4.8636, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918589483628734e-05, | |
| "loss": 4.8734, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8910203536118214e-05, | |
| "loss": 4.8709, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890181758860769e-05, | |
| "loss": 4.8526, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889343164109717e-05, | |
| "loss": 4.8398, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888506207239038e-05, | |
| "loss": 4.8469, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.887667612487986e-05, | |
| "loss": 4.8546, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.886829017736934e-05, | |
| "loss": 4.861, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.885990422985882e-05, | |
| "loss": 4.8445, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88515182823483e-05, | |
| "loss": 4.8428, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884316509244525e-05, | |
| "loss": 4.842, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883477914493473e-05, | |
| "loss": 4.8422, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882639319742421e-05, | |
| "loss": 4.829, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881800724991369e-05, | |
| "loss": 4.8211, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88096376812069e-05, | |
| "loss": 4.8289, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880125173369638e-05, | |
| "loss": 4.8184, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879286578618586e-05, | |
| "loss": 4.8181, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.878447983867534e-05, | |
| "loss": 4.8206, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877609389116482e-05, | |
| "loss": 4.8042, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.87677079436543e-05, | |
| "loss": 4.8157, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875932199614378e-05, | |
| "loss": 4.8113, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875093604863326e-05, | |
| "loss": 4.7903, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.776516914367676, | |
| "eval_runtime": 292.6321, | |
| "eval_samples_per_second": 1303.996, | |
| "eval_steps_per_second": 40.751, | |
| "step": 76320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8742566479926466e-05, | |
| "loss": 4.8033, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.873419691121968e-05, | |
| "loss": 4.794, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.872581096370916e-05, | |
| "loss": 4.8041, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8717425016198635e-05, | |
| "loss": 4.7906, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8709039068688115e-05, | |
| "loss": 4.7918, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8700653121177595e-05, | |
| "loss": 4.7772, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8692267173667075e-05, | |
| "loss": 4.7778, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8683881226156555e-05, | |
| "loss": 4.7584, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.867549527864604e-05, | |
| "loss": 4.7853, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.866710933113552e-05, | |
| "loss": 4.7773, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865873976242873e-05, | |
| "loss": 4.7728, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865035381491821e-05, | |
| "loss": 4.7767, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.864198424621142e-05, | |
| "loss": 4.7598, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.86335982987009e-05, | |
| "loss": 4.7594, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.862521235119038e-05, | |
| "loss": 4.7544, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.861682640367986e-05, | |
| "loss": 4.7521, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.860844045616934e-05, | |
| "loss": 4.753, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.860007088746255e-05, | |
| "loss": 4.7452, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.859168493995203e-05, | |
| "loss": 4.7542, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.858329899244151e-05, | |
| "loss": 4.7572, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8574913044930995e-05, | |
| "loss": 4.7434, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8566527097420475e-05, | |
| "loss": 4.7367, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8558141149909955e-05, | |
| "loss": 4.734, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8549771581203164e-05, | |
| "loss": 4.7401, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8541385633692644e-05, | |
| "loss": 4.7311, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8532999686182124e-05, | |
| "loss": 4.7187, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8524613738671604e-05, | |
| "loss": 4.7188, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8516227791161084e-05, | |
| "loss": 4.7325, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8507841843650564e-05, | |
| "loss": 4.7052, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8499455896140044e-05, | |
| "loss": 4.7121, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8491069948629524e-05, | |
| "loss": 4.7216, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8482684001119e-05, | |
| "loss": 4.7172, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.847433081121595e-05, | |
| "loss": 4.7167, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.846594486370543e-05, | |
| "loss": 4.715, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.845755891619491e-05, | |
| "loss": 4.712, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.844917296868439e-05, | |
| "loss": 4.7131, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.844078702117387e-05, | |
| "loss": 4.6901, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.843240107366335e-05, | |
| "loss": 4.699, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.842401512615282e-05, | |
| "loss": 4.6822, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.84156291786423e-05, | |
| "loss": 4.6882, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.840724323113178e-05, | |
| "loss": 4.6896, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839885728362126e-05, | |
| "loss": 4.6874, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839048771491447e-05, | |
| "loss": 4.6832, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.838210176740395e-05, | |
| "loss": 4.6832, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.837371581989343e-05, | |
| "loss": 4.6822, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.836532987238292e-05, | |
| "loss": 4.6858, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.83569439248724e-05, | |
| "loss": 4.6883, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834855797736188e-05, | |
| "loss": 4.6697, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834017202985136e-05, | |
| "loss": 4.662, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.833178608234084e-05, | |
| "loss": 4.6782, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8323416513634046e-05, | |
| "loss": 4.6823, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8315030566123526e-05, | |
| "loss": 4.6651, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8306644618613006e-05, | |
| "loss": 4.662, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8298258671102486e-05, | |
| "loss": 4.6617, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8289889102395695e-05, | |
| "loss": 4.6559, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8281503154885175e-05, | |
| "loss": 4.6652, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8273117207374655e-05, | |
| "loss": 4.6451, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8264731259864135e-05, | |
| "loss": 4.6572, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8256345312353615e-05, | |
| "loss": 4.6529, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.824797574364683e-05, | |
| "loss": 4.654, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823958979613631e-05, | |
| "loss": 4.6349, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823120384862579e-05, | |
| "loss": 4.6561, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.822281790111527e-05, | |
| "loss": 4.6299, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.821443195360475e-05, | |
| "loss": 4.645, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.820604600609423e-05, | |
| "loss": 4.6447, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.819767643738744e-05, | |
| "loss": 4.6458, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.818929048987692e-05, | |
| "loss": 4.6293, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.81809045423664e-05, | |
| "loss": 4.6246, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.817251859485588e-05, | |
| "loss": 4.623, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.816413264734535e-05, | |
| "loss": 4.6347, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.815574669983483e-05, | |
| "loss": 4.6347, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.814736075232432e-05, | |
| "loss": 4.6272, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.81389748048138e-05, | |
| "loss": 4.6337, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.813060523610701e-05, | |
| "loss": 4.6387, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.812221928859649e-05, | |
| "loss": 4.635, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.811383334108597e-05, | |
| "loss": 4.6181, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.810544739357545e-05, | |
| "loss": 4.6217, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.809707782486866e-05, | |
| "loss": 4.6231, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808869187735814e-05, | |
| "loss": 4.6027, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808030592984762e-05, | |
| "loss": 4.6243, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.80719199823371e-05, | |
| "loss": 4.606, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.806353403482658e-05, | |
| "loss": 4.6272, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.805514808731606e-05, | |
| "loss": 4.6163, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.804677851860927e-05, | |
| "loss": 4.5943, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803839257109875e-05, | |
| "loss": 4.6105, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803000662358823e-05, | |
| "loss": 4.5993, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.802162067607771e-05, | |
| "loss": 4.6137, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.801325110737092e-05, | |
| "loss": 4.6062, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.800488153866413e-05, | |
| "loss": 4.6037, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.799649559115361e-05, | |
| "loss": 4.6002, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.798812602244683e-05, | |
| "loss": 4.5935, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.79797400749363e-05, | |
| "loss": 4.5893, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797135412742578e-05, | |
| "loss": 4.5834, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.796296817991526e-05, | |
| "loss": 4.5852, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.795458223240474e-05, | |
| "loss": 4.5972, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7946212663697956e-05, | |
| "loss": 4.5869, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7937826716187436e-05, | |
| "loss": 4.5812, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7929440768676916e-05, | |
| "loss": 4.5766, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7921054821166396e-05, | |
| "loss": 4.5839, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7912668873655876e-05, | |
| "loss": 4.5887, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7904282926145356e-05, | |
| "loss": 4.582, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7895896978634836e-05, | |
| "loss": 4.5777, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7887511031124316e-05, | |
| "loss": 4.5763, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7879125083613796e-05, | |
| "loss": 4.5817, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7870739136103276e-05, | |
| "loss": 4.5626, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7862353188592756e-05, | |
| "loss": 4.5719, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7853967241082236e-05, | |
| "loss": 4.5696, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7845581293571715e-05, | |
| "loss": 4.5638, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7837211724864925e-05, | |
| "loss": 4.5637, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782882577735441e-05, | |
| "loss": 4.5671, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782043982984389e-05, | |
| "loss": 4.5628, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.781205388233337e-05, | |
| "loss": 4.5604, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7803667934822844e-05, | |
| "loss": 4.5609, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.779529836611606e-05, | |
| "loss": 4.5576, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.778691241860554e-05, | |
| "loss": 4.5689, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777852647109502e-05, | |
| "loss": 4.563, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777014052358449e-05, | |
| "loss": 4.5598, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.776175457607397e-05, | |
| "loss": 4.5604, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.775336862856345e-05, | |
| "loss": 4.5582, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.774498268105293e-05, | |
| "loss": 4.5565, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.773659673354241e-05, | |
| "loss": 4.5474, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.772822716483563e-05, | |
| "loss": 4.5458, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771984121732511e-05, | |
| "loss": 4.5502, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771145526981459e-05, | |
| "loss": 4.5512, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.770306932230407e-05, | |
| "loss": 4.5495, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.769469975359728e-05, | |
| "loss": 4.5481, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.768631380608676e-05, | |
| "loss": 4.542, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.767794423737997e-05, | |
| "loss": 4.5436, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766955828986945e-05, | |
| "loss": 4.5426, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766117234235893e-05, | |
| "loss": 4.551, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.765278639484841e-05, | |
| "loss": 4.5318, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.764440044733789e-05, | |
| "loss": 4.5237, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.763601449982737e-05, | |
| "loss": 4.5292, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.762764493112058e-05, | |
| "loss": 4.542, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761925898361006e-05, | |
| "loss": 4.5461, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761087303609954e-05, | |
| "loss": 4.5364, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.760248708858902e-05, | |
| "loss": 4.5355, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75941011410785e-05, | |
| "loss": 4.5376, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.758571519356798e-05, | |
| "loss": 4.5372, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.757734562486119e-05, | |
| "loss": 4.5283, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756895967735067e-05, | |
| "loss": 4.5244, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756057372984015e-05, | |
| "loss": 4.5295, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.755218778232963e-05, | |
| "loss": 4.5261, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.754383459242657e-05, | |
| "loss": 4.5192, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.753544864491605e-05, | |
| "loss": 4.5314, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7527062697405536e-05, | |
| "loss": 4.515, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7518676749895016e-05, | |
| "loss": 4.525, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7510290802384496e-05, | |
| "loss": 4.5232, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7501904854873976e-05, | |
| "loss": 4.5105, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.495769500732422, | |
| "eval_runtime": 291.918, | |
| "eval_samples_per_second": 1307.186, | |
| "eval_steps_per_second": 40.851, | |
| "step": 152640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7493518907363456e-05, | |
| "loss": 4.5214, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7485132959852936e-05, | |
| "loss": 4.5117, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7476747012342416e-05, | |
| "loss": 4.5216, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7468361064831896e-05, | |
| "loss": 4.5149, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7459975117321376e-05, | |
| "loss": 4.5181, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745158916981085e-05, | |
| "loss": 4.5047, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.744320322230033e-05, | |
| "loss": 4.5042, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.743481727478981e-05, | |
| "loss": 4.4898, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.742643132727929e-05, | |
| "loss": 4.5199, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.741804537976877e-05, | |
| "loss": 4.511, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.740965943225825e-05, | |
| "loss": 4.5028, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7401273484747736e-05, | |
| "loss": 4.5177, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7392887537237216e-05, | |
| "loss": 4.4953, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7384501589726696e-05, | |
| "loss": 4.502, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7376115642216176e-05, | |
| "loss": 4.4931, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7367729694705656e-05, | |
| "loss": 4.4899, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7359360125998865e-05, | |
| "loss": 4.5008, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7350974178488345e-05, | |
| "loss": 4.4872, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7342588230977825e-05, | |
| "loss": 4.501, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7334202283467305e-05, | |
| "loss": 4.5052, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7325816335956784e-05, | |
| "loss": 4.4955, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7317430388446264e-05, | |
| "loss": 4.4906, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730904444093574e-05, | |
| "loss": 4.489, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7300674872228953e-05, | |
| "loss": 4.4939, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7292288924718433e-05, | |
| "loss": 4.4816, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.728390297720791e-05, | |
| "loss": 4.4834, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.727551702969739e-05, | |
| "loss": 4.4717, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.726713108218687e-05, | |
| "loss": 4.4936, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.725874513467635e-05, | |
| "loss": 4.4653, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.725035918716583e-05, | |
| "loss": 4.4797, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.724197323965531e-05, | |
| "loss": 4.4828, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.723360367094852e-05, | |
| "loss": 4.4839, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.722523410224174e-05, | |
| "loss": 4.4809, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.721684815473121e-05, | |
| "loss": 4.4819, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720847858602443e-05, | |
| "loss": 4.4794, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720009263851391e-05, | |
| "loss": 4.4828, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.719170669100339e-05, | |
| "loss": 4.466, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.718332074349287e-05, | |
| "loss": 4.4711, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.717493479598235e-05, | |
| "loss": 4.455, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.716654884847183e-05, | |
| "loss": 4.4652, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.715816290096131e-05, | |
| "loss": 4.4634, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714977695345079e-05, | |
| "loss": 4.4702, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714139100594027e-05, | |
| "loss": 4.4615, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.713300505842975e-05, | |
| "loss": 4.4635, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.712461911091923e-05, | |
| "loss": 4.4642, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.711623316340871e-05, | |
| "loss": 4.4678, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7107863594701916e-05, | |
| "loss": 4.4699, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7099494025995125e-05, | |
| "loss": 4.456, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7091108078484605e-05, | |
| "loss": 4.4444, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.708272213097409e-05, | |
| "loss": 4.4687, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.707433618346357e-05, | |
| "loss": 4.4655, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.706595023595305e-05, | |
| "loss": 4.4567, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.705756428844253e-05, | |
| "loss": 4.4552, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704917834093201e-05, | |
| "loss": 4.4483, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704079239342149e-05, | |
| "loss": 4.4509, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.70324228247147e-05, | |
| "loss": 4.4568, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.702403687720418e-05, | |
| "loss": 4.4372, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.701565092969366e-05, | |
| "loss": 4.4514, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.700726498218314e-05, | |
| "loss": 4.4481, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699889541347635e-05, | |
| "loss": 4.4471, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699050946596583e-05, | |
| "loss": 4.4326, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.698212351845531e-05, | |
| "loss": 4.456, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6973753949748525e-05, | |
| "loss": 4.433, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6965368002238005e-05, | |
| "loss": 4.4457, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6956982054727485e-05, | |
| "loss": 4.4421, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6948596107216965e-05, | |
| "loss": 4.4501, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6940226538510174e-05, | |
| "loss": 4.4313, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6931840590999654e-05, | |
| "loss": 4.4337, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6923454643489134e-05, | |
| "loss": 4.4226, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6915068695978614e-05, | |
| "loss": 4.4421, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6906682748468094e-05, | |
| "loss": 4.4416, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6898296800957574e-05, | |
| "loss": 4.4362, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688991085344705e-05, | |
| "loss": 4.4398, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688152490593653e-05, | |
| "loss": 4.4479, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6873138958426014e-05, | |
| "loss": 4.446, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.686476938971923e-05, | |
| "loss": 4.426, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.685639982101244e-05, | |
| "loss": 4.4333, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.684801387350192e-05, | |
| "loss": 4.4408, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68396279259914e-05, | |
| "loss": 4.4181, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683124197848087e-05, | |
| "loss": 4.4355, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.682285603097035e-05, | |
| "loss": 4.419, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.681447008345983e-05, | |
| "loss": 4.445, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.680608413594931e-05, | |
| "loss": 4.4346, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.679769818843879e-05, | |
| "loss": 4.4117, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6789328619732e-05, | |
| "loss": 4.4256, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678094267222148e-05, | |
| "loss": 4.4229, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.677255672471097e-05, | |
| "loss": 4.4328, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.676417077720045e-05, | |
| "loss": 4.428, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6755801208493657e-05, | |
| "loss": 4.425, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6747415260983136e-05, | |
| "loss": 4.4209, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6739029313472616e-05, | |
| "loss": 4.4159, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6730643365962096e-05, | |
| "loss": 4.4136, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6722257418451576e-05, | |
| "loss": 4.4052, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6713871470941056e-05, | |
| "loss": 4.414, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6705501902234265e-05, | |
| "loss": 4.4223, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6697115954723745e-05, | |
| "loss": 4.4144, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6688730007213225e-05, | |
| "loss": 4.4093, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6680344059702705e-05, | |
| "loss": 4.4069, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6671958112192185e-05, | |
| "loss": 4.4104, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6663572164681665e-05, | |
| "loss": 4.4207, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6655186217171145e-05, | |
| "loss": 4.4108, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.664680026966063e-05, | |
| "loss": 4.4108, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663843070095384e-05, | |
| "loss": 4.4019, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663004475344332e-05, | |
| "loss": 4.4148, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.662167518473653e-05, | |
| "loss": 4.3939, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.661328923722601e-05, | |
| "loss": 4.4071, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.660490328971549e-05, | |
| "loss": 4.3999, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.659651734220497e-05, | |
| "loss": 4.4018, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.658813139469445e-05, | |
| "loss": 4.3967, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657974544718393e-05, | |
| "loss": 4.4026, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657135949967341e-05, | |
| "loss": 4.4003, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.656297355216288e-05, | |
| "loss": 4.3957, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.65546039834561e-05, | |
| "loss": 4.4037, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6546234414749315e-05, | |
| "loss": 4.3956, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6537848467238795e-05, | |
| "loss": 4.4055, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6529462519728275e-05, | |
| "loss": 4.4044, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6521076572217755e-05, | |
| "loss": 4.4014, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6512690624707234e-05, | |
| "loss": 4.4017, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.650430467719671e-05, | |
| "loss": 4.398, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.649591872968619e-05, | |
| "loss": 4.3938, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6487549160979403e-05, | |
| "loss": 4.3933, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6479163213468883e-05, | |
| "loss": 4.3929, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647077726595836e-05, | |
| "loss": 4.39, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6462391318447837e-05, | |
| "loss": 4.4001, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.645400537093732e-05, | |
| "loss": 4.3914, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.64456194234268e-05, | |
| "loss": 4.3951, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.643723347591628e-05, | |
| "loss": 4.3905, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642886390720949e-05, | |
| "loss": 4.3891, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642047795969897e-05, | |
| "loss": 4.387, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.641209201218845e-05, | |
| "loss": 4.4012, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.640370606467793e-05, | |
| "loss": 4.379, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.639532011716741e-05, | |
| "loss": 4.3769, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.638693416965689e-05, | |
| "loss": 4.3742, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637854822214637e-05, | |
| "loss": 4.3873, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637016227463585e-05, | |
| "loss": 4.3944, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.636179270592906e-05, | |
| "loss": 4.3919, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.635340675841854e-05, | |
| "loss": 4.3837, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.634502081090802e-05, | |
| "loss": 4.3908, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.633663486339751e-05, | |
| "loss": 4.3849, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.632826529469072e-05, | |
| "loss": 4.3854, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.63198793471802e-05, | |
| "loss": 4.3764, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.631149339966968e-05, | |
| "loss": 4.3856, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.630310745215916e-05, | |
| "loss": 4.3815, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6294737883452366e-05, | |
| "loss": 4.3706, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6286351935941846e-05, | |
| "loss": 4.387, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6277965988431326e-05, | |
| "loss": 4.3751, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6269596419724535e-05, | |
| "loss": 4.3772, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6261210472214015e-05, | |
| "loss": 4.3786, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6252824524703495e-05, | |
| "loss": 4.3672, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.3605637550354, | |
| "eval_runtime": 316.5441, | |
| "eval_samples_per_second": 1205.491, | |
| "eval_steps_per_second": 37.672, | |
| "step": 228960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6244438577192975e-05, | |
| "loss": 4.3802, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.623605262968246e-05, | |
| "loss": 4.3693, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.62276994397794e-05, | |
| "loss": 4.3784, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.621931349226888e-05, | |
| "loss": 4.3708, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.621092754475836e-05, | |
| "loss": 4.3825, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.620254159724784e-05, | |
| "loss": 4.3659, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.619415564973732e-05, | |
| "loss": 4.3619, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61857697022268e-05, | |
| "loss": 4.3538, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.617738375471628e-05, | |
| "loss": 4.3782, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616899780720576e-05, | |
| "loss": 4.3774, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616061185969524e-05, | |
| "loss": 4.3619, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.615222591218472e-05, | |
| "loss": 4.3776, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.614385634347793e-05, | |
| "loss": 4.3606, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6135470395967415e-05, | |
| "loss": 4.3643, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6127084448456895e-05, | |
| "loss": 4.3642, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611869850094637e-05, | |
| "loss": 4.3546, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611031255343585e-05, | |
| "loss": 4.3638, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.610192660592533e-05, | |
| "loss": 4.3554, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.609354065841481e-05, | |
| "loss": 4.367, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.608515471090429e-05, | |
| "loss": 4.3686, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.60767851421975e-05, | |
| "loss": 4.3669, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606839919468698e-05, | |
| "loss": 4.3587, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606001324717646e-05, | |
| "loss": 4.3523, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.605162729966594e-05, | |
| "loss": 4.3689, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.604325773095915e-05, | |
| "loss": 4.3477, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.60349045410561e-05, | |
| "loss": 4.3542, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.602651859354558e-05, | |
| "loss": 4.3428, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.601813264603506e-05, | |
| "loss": 4.3628, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.600974669852454e-05, | |
| "loss": 4.3394, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.600136075101402e-05, | |
| "loss": 4.348, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.599297480350349e-05, | |
| "loss": 4.3563, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.598458885599297e-05, | |
| "loss": 4.3523, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.597620290848245e-05, | |
| "loss": 4.3535, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.596781696097193e-05, | |
| "loss": 4.3545, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595944739226514e-05, | |
| "loss": 4.3546, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595106144475462e-05, | |
| "loss": 4.3554, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.59426754972441e-05, | |
| "loss": 4.3457, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5934289549733587e-05, | |
| "loss": 4.3388, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5925919981026796e-05, | |
| "loss": 4.3332, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5917534033516276e-05, | |
| "loss": 4.34, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5909148086005756e-05, | |
| "loss": 4.3391, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5900762138495235e-05, | |
| "loss": 4.3491, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5892392569788445e-05, | |
| "loss": 4.337, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5884006622277925e-05, | |
| "loss": 4.3408, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5875620674767404e-05, | |
| "loss": 4.3461, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5867234727256884e-05, | |
| "loss": 4.3394, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5858865158550094e-05, | |
| "loss": 4.3528, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5850479211039573e-05, | |
| "loss": 4.3341, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5842093263529053e-05, | |
| "loss": 4.3198, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.583370731601854e-05, | |
| "loss": 4.3507, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.582532136850802e-05, | |
| "loss": 4.3423, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.58169354209975e-05, | |
| "loss": 4.3413, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.580854947348698e-05, | |
| "loss": 4.3341, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.580016352597646e-05, | |
| "loss": 4.3301, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.57918103360734e-05, | |
| "loss": 4.3309, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.578344076736661e-05, | |
| "loss": 4.3365, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.577505481985609e-05, | |
| "loss": 4.3227, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.576666887234557e-05, | |
| "loss": 4.3283, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.575828292483505e-05, | |
| "loss": 4.336, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.574989697732453e-05, | |
| "loss": 4.3279, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.574151102981401e-05, | |
| "loss": 4.3152, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5733125082303494e-05, | |
| "loss": 4.3415, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5724739134792974e-05, | |
| "loss": 4.3151, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.571636956608618e-05, | |
| "loss": 4.325, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.570798361857566e-05, | |
| "loss": 4.3325, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.569959767106514e-05, | |
| "loss": 4.3343, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.569121172355462e-05, | |
| "loss": 4.3172, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.56828257760441e-05, | |
| "loss": 4.3215, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.567443982853358e-05, | |
| "loss": 4.3054, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.566605388102306e-05, | |
| "loss": 4.3323, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.565766793351254e-05, | |
| "loss": 4.326, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.564928198600202e-05, | |
| "loss": 4.3252, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5640896038491496e-05, | |
| "loss": 4.3228, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5632510090980976e-05, | |
| "loss": 4.3345, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.562414052227419e-05, | |
| "loss": 4.337, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.561575457476368e-05, | |
| "loss": 4.3145, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.560736862725315e-05, | |
| "loss": 4.3232, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.559898267974263e-05, | |
| "loss": 4.3308, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.559059673223211e-05, | |
| "loss": 4.3069, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.558222716352533e-05, | |
| "loss": 4.3266, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.55738412160148e-05, | |
| "loss": 4.3121, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.556545526850428e-05, | |
| "loss": 4.329, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.555706932099376e-05, | |
| "loss": 4.3269, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554868337348324e-05, | |
| "loss": 4.3048, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554029742597272e-05, | |
| "loss": 4.3177, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.55319114784622e-05, | |
| "loss": 4.3127, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5523541909755416e-05, | |
| "loss": 4.3255, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5515155962244896e-05, | |
| "loss": 4.3143, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5506786393538105e-05, | |
| "loss": 4.3144, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5498400446027585e-05, | |
| "loss": 4.3127, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5490014498517065e-05, | |
| "loss": 4.311, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5481628551006545e-05, | |
| "loss": 4.3083, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5473258982299754e-05, | |
| "loss": 4.298, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5464873034789234e-05, | |
| "loss": 4.3075, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5456487087278714e-05, | |
| "loss": 4.3158, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5448101139768194e-05, | |
| "loss": 4.3159, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5439715192257674e-05, | |
| "loss": 4.2966, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5431329244747154e-05, | |
| "loss": 4.3034, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5422943297236634e-05, | |
| "loss": 4.3066, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5414557349726114e-05, | |
| "loss": 4.3165, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5406171402215594e-05, | |
| "loss": 4.306, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.539780183350881e-05, | |
| "loss": 4.3067, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.538941588599829e-05, | |
| "loss": 4.2983, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.538102993848777e-05, | |
| "loss": 4.3146, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.537264399097725e-05, | |
| "loss": 4.2893, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.536425804346673e-05, | |
| "loss": 4.3055, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.535588847475994e-05, | |
| "loss": 4.2964, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.534750252724942e-05, | |
| "loss": 4.3009, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.53391165797389e-05, | |
| "loss": 4.2945, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533073063222838e-05, | |
| "loss": 4.3065, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.532234468471786e-05, | |
| "loss": 4.2937, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.531395873720733e-05, | |
| "loss": 4.2953, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.530557278969682e-05, | |
| "loss": 4.3004, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.52971868421863e-05, | |
| "loss": 4.2963, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528883365228324e-05, | |
| "loss": 4.3062, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528044770477272e-05, | |
| "loss": 4.3051, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.52720617572622e-05, | |
| "loss": 4.3008, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.526367580975168e-05, | |
| "loss": 4.3055, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.525528986224116e-05, | |
| "loss": 4.2982, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5246903914730636e-05, | |
| "loss": 4.2921, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5238517967220116e-05, | |
| "loss": 4.2933, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.523014839851333e-05, | |
| "loss": 4.2946, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5221762451002805e-05, | |
| "loss": 4.295, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5213376503492285e-05, | |
| "loss": 4.3009, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.520499055598177e-05, | |
| "loss": 4.2919, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.519660460847125e-05, | |
| "loss": 4.2984, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.518821866096073e-05, | |
| "loss": 4.2884, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517983271345021e-05, | |
| "loss": 4.2921, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517144676593969e-05, | |
| "loss": 4.2905, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.51630771972329e-05, | |
| "loss": 4.3043, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.515469124972238e-05, | |
| "loss": 4.2869, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.514630530221186e-05, | |
| "loss": 4.2809, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.513793573350507e-05, | |
| "loss": 4.2752, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512954978599455e-05, | |
| "loss": 4.2922, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512116383848403e-05, | |
| "loss": 4.2975, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.511277789097351e-05, | |
| "loss": 4.2972, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.510439194346299e-05, | |
| "loss": 4.2917, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.509600599595247e-05, | |
| "loss": 4.2965, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5087620048441956e-05, | |
| "loss": 4.2912, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5079250479735165e-05, | |
| "loss": 4.2903, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5070864532224645e-05, | |
| "loss": 4.2843, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5062478584714125e-05, | |
| "loss": 4.2906, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5054092637203605e-05, | |
| "loss": 4.2887, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5045706689693085e-05, | |
| "loss": 4.2763, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5037337120986294e-05, | |
| "loss": 4.294, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5028951173475774e-05, | |
| "loss": 4.2856, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5020565225965254e-05, | |
| "loss": 4.2812, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5012179278454734e-05, | |
| "loss": 4.2843, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.500380970974794e-05, | |
| "loss": 4.282, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.276997089385986, | |
| "eval_runtime": 289.737, | |
| "eval_samples_per_second": 1317.026, | |
| "eval_steps_per_second": 41.158, | |
| "step": 305280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.499542376223742e-05, | |
| "loss": 4.2866, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.498703781472691e-05, | |
| "loss": 4.2803, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.497865186721639e-05, | |
| "loss": 4.2901, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.497026591970587e-05, | |
| "loss": 4.2787, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.496187997219535e-05, | |
| "loss": 4.2915, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.495349402468482e-05, | |
| "loss": 4.28, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.49451080771743e-05, | |
| "loss": 4.2694, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.493672212966378e-05, | |
| "loss": 4.2689, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.492833618215326e-05, | |
| "loss": 4.2835, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491996661344647e-05, | |
| "loss": 4.2875, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491158066593595e-05, | |
| "loss": 4.2762, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.490319471842543e-05, | |
| "loss": 4.2852, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.489480877091491e-05, | |
| "loss": 4.2737, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.488642282340439e-05, | |
| "loss": 4.2773, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.487803687589387e-05, | |
| "loss": 4.2725, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486965092838336e-05, | |
| "loss": 4.2673, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486126498087284e-05, | |
| "loss": 4.2728, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.485287903336232e-05, | |
| "loss": 4.2676, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.48444930858518e-05, | |
| "loss": 4.2809, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.483610713834128e-05, | |
| "loss": 4.281, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.482772119083076e-05, | |
| "loss": 4.2816, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.481933524332024e-05, | |
| "loss": 4.2744, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.481094929580971e-05, | |
| "loss": 4.2673, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.480256334829919e-05, | |
| "loss": 4.2789, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.479417740078867e-05, | |
| "loss": 4.2667, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.478580783208188e-05, | |
| "loss": 4.2675, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4777438263375096e-05, | |
| "loss": 4.2564, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4769052315864576e-05, | |
| "loss": 4.2725, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4760666368354056e-05, | |
| "loss": 4.2596, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4752280420843536e-05, | |
| "loss": 4.258, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4743894473333016e-05, | |
| "loss": 4.2736, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.473552490462623e-05, | |
| "loss": 4.267, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.472713895711571e-05, | |
| "loss": 4.2733, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4718753009605185e-05, | |
| "loss": 4.273, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4710367062094665e-05, | |
| "loss": 4.2642, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.470199749338788e-05, | |
| "loss": 4.2742, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4693611545877354e-05, | |
| "loss": 4.2655, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4685225598366834e-05, | |
| "loss": 4.2539, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4676839650856314e-05, | |
| "loss": 4.2503, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4668453703345794e-05, | |
| "loss": 4.256, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466006775583528e-05, | |
| "loss": 4.2573, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.465168180832476e-05, | |
| "loss": 4.2681, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.464329586081424e-05, | |
| "loss": 4.2547, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.463492629210745e-05, | |
| "loss": 4.2572, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.462654034459693e-05, | |
| "loss": 4.2654, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.461815439708641e-05, | |
| "loss": 4.2611, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.460976844957589e-05, | |
| "loss": 4.2667, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.460138250206537e-05, | |
| "loss": 4.2506, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.459299655455485e-05, | |
| "loss": 4.2357, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.458461060704433e-05, | |
| "loss": 4.2686, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.457622465953381e-05, | |
| "loss": 4.2673, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.456787146963075e-05, | |
| "loss": 4.2592, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4559485522120234e-05, | |
| "loss": 4.2527, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4551099574609714e-05, | |
| "loss": 4.2495, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4542730005902923e-05, | |
| "loss": 4.2502, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.453436043719613e-05, | |
| "loss": 4.2548, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.452597448968561e-05, | |
| "loss": 4.2439, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.451758854217509e-05, | |
| "loss": 4.249, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.450920259466457e-05, | |
| "loss": 4.2584, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.450081664715405e-05, | |
| "loss": 4.2522, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.449243069964353e-05, | |
| "loss": 4.2373, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.448404475213301e-05, | |
| "loss": 4.2626, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.447565880462249e-05, | |
| "loss": 4.2382, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.446727285711197e-05, | |
| "loss": 4.2408, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445888690960145e-05, | |
| "loss": 4.2561, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445050096209093e-05, | |
| "loss": 4.2592, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.444211501458041e-05, | |
| "loss": 4.2356, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.44337290670699e-05, | |
| "loss": 4.245, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.442534311955937e-05, | |
| "loss": 4.2278, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.441697355085259e-05, | |
| "loss": 4.2552, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440858760334207e-05, | |
| "loss": 4.2531, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440020165583154e-05, | |
| "loss": 4.2437, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.439181570832102e-05, | |
| "loss": 4.2481, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.43834297608105e-05, | |
| "loss": 4.2561, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.437504381329998e-05, | |
| "loss": 4.265, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.436667424459319e-05, | |
| "loss": 4.237, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.435828829708267e-05, | |
| "loss": 4.2481, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.434990234957216e-05, | |
| "loss": 4.2536, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.434151640206164e-05, | |
| "loss": 4.2301, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4333130454551117e-05, | |
| "loss": 4.2527, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4324744507040597e-05, | |
| "loss": 4.2373, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4316358559530076e-05, | |
| "loss": 4.254, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4307972612019556e-05, | |
| "loss": 4.2522, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4299586664509036e-05, | |
| "loss": 4.233, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4291217095802245e-05, | |
| "loss": 4.2377, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4282831148291725e-05, | |
| "loss": 4.2398, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4274445200781205e-05, | |
| "loss": 4.2547, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4266059253270685e-05, | |
| "loss": 4.2409, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4257689684563894e-05, | |
| "loss": 4.2409, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4249320115857104e-05, | |
| "loss": 4.236, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424093416834659e-05, | |
| "loss": 4.2425, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.423254822083607e-05, | |
| "loss": 4.2357, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.422416227332555e-05, | |
| "loss": 4.2221, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.421577632581503e-05, | |
| "loss": 4.2366, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.420739037830451e-05, | |
| "loss": 4.2406, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419900443079399e-05, | |
| "loss": 4.2426, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419061848328347e-05, | |
| "loss": 4.2244, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.418224891457668e-05, | |
| "loss": 4.2312, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.417386296706616e-05, | |
| "loss": 4.2365, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.416547701955564e-05, | |
| "loss": 4.2405, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.415709107204512e-05, | |
| "loss": 4.2402, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.41487051245346e-05, | |
| "loss": 4.2336, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.414033555582781e-05, | |
| "loss": 4.2292, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.413194960831729e-05, | |
| "loss": 4.241, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.412359641841423e-05, | |
| "loss": 4.2176, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.411521047090371e-05, | |
| "loss": 4.2343, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.410684090219692e-05, | |
| "loss": 4.223, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.40984549546864e-05, | |
| "loss": 4.2338, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.409006900717588e-05, | |
| "loss": 4.2203, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.408168305966536e-05, | |
| "loss": 4.2354, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.407329711215484e-05, | |
| "loss": 4.2268, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.406491116464432e-05, | |
| "loss": 4.2241, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.40565252171338e-05, | |
| "loss": 4.2287, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.404813926962328e-05, | |
| "loss": 4.2271, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403975332211276e-05, | |
| "loss": 4.2349, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403136737460224e-05, | |
| "loss": 4.2393, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.402299780589546e-05, | |
| "loss": 4.2323, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.401462823718867e-05, | |
| "loss": 4.2335, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4006242289678147e-05, | |
| "loss": 4.2283, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3997856342167627e-05, | |
| "loss": 4.2244, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3989470394657106e-05, | |
| "loss": 4.2213, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3981084447146586e-05, | |
| "loss": 4.2316, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3972698499636066e-05, | |
| "loss": 4.2212, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3964312552125546e-05, | |
| "loss": 4.2299, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3955942983418755e-05, | |
| "loss": 4.2243, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3947557035908235e-05, | |
| "loss": 4.2309, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3939171088397715e-05, | |
| "loss": 4.2217, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3930785140887195e-05, | |
| "loss": 4.2255, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3922399193376675e-05, | |
| "loss": 4.2203, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3914013245866155e-05, | |
| "loss": 4.2334, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3905627298355635e-05, | |
| "loss": 4.2226, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3897241350845115e-05, | |
| "loss": 4.2142, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3888855403334595e-05, | |
| "loss": 4.2068, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3880485834627804e-05, | |
| "loss": 4.2253, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3872099887117284e-05, | |
| "loss": 4.2295, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3863713939606764e-05, | |
| "loss": 4.2296, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3855327992096244e-05, | |
| "loss": 4.2278, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3846942044585724e-05, | |
| "loss": 4.2274, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3838556097075204e-05, | |
| "loss": 4.2236, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3830170149564684e-05, | |
| "loss": 4.228, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3821784202054164e-05, | |
| "loss": 4.2155, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3813398254543644e-05, | |
| "loss": 4.224, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.380501230703313e-05, | |
| "loss": 4.2203, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.379662635952261e-05, | |
| "loss": 4.2155, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.378824041201209e-05, | |
| "loss": 4.2281, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.37798708433053e-05, | |
| "loss": 4.2196, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.377148489579478e-05, | |
| "loss": 4.2173, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.376309894828426e-05, | |
| "loss": 4.2188, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.375471300077374e-05, | |
| "loss": 4.2196, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.220510005950928, | |
| "eval_runtime": 294.3224, | |
| "eval_samples_per_second": 1296.507, | |
| "eval_steps_per_second": 40.517, | |
| "step": 381600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.374632705326321e-05, | |
| "loss": 4.2171, | |
| "step": 381952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.373795748455643e-05, | |
| "loss": 4.2172, | |
| "step": 382464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372957153704591e-05, | |
| "loss": 4.2232, | |
| "step": 382976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372118558953539e-05, | |
| "loss": 4.2136, | |
| "step": 383488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.371279964202487e-05, | |
| "loss": 4.2267, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.370446283092554e-05, | |
| "loss": 4.217, | |
| "step": 384512 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.369607688341502e-05, | |
| "loss": 4.2056, | |
| "step": 385024 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.36876909359045e-05, | |
| "loss": 4.2043, | |
| "step": 385536 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367930498839398e-05, | |
| "loss": 4.2201, | |
| "step": 386048 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367091904088346e-05, | |
| "loss": 4.2225, | |
| "step": 386560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.366253309337294e-05, | |
| "loss": 4.2148, | |
| "step": 387072 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.365414714586242e-05, | |
| "loss": 4.2192, | |
| "step": 387584 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.364577757715563e-05, | |
| "loss": 4.2112, | |
| "step": 388096 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.363739162964511e-05, | |
| "loss": 4.2144, | |
| "step": 388608 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.362900568213459e-05, | |
| "loss": 4.2092, | |
| "step": 389120 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.362061973462407e-05, | |
| "loss": 4.2005, | |
| "step": 389632 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.361223378711355e-05, | |
| "loss": 4.2122, | |
| "step": 390144 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.360384783960304e-05, | |
| "loss": 4.2033, | |
| "step": 390656 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.359546189209251e-05, | |
| "loss": 4.2174, | |
| "step": 391168 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.358707594458199e-05, | |
| "loss": 4.2188, | |
| "step": 391680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357868999707147e-05, | |
| "loss": 4.2215, | |
| "step": 392192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357030404956095e-05, | |
| "loss": 4.211, | |
| "step": 392704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.356191810205043e-05, | |
| "loss": 4.2077, | |
| "step": 393216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.355353215453991e-05, | |
| "loss": 4.2139, | |
| "step": 393728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.354516258583312e-05, | |
| "loss": 4.2028, | |
| "step": 394240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.35367766383226e-05, | |
| "loss": 4.2081, | |
| "step": 394752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.352839069081208e-05, | |
| "loss": 4.1919, | |
| "step": 395264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.352000474330156e-05, | |
| "loss": 4.2143, | |
| "step": 395776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3511635174594776e-05, | |
| "loss": 4.2017, | |
| "step": 396288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3503249227084256e-05, | |
| "loss": 4.197, | |
| "step": 396800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3494863279573736e-05, | |
| "loss": 4.207, | |
| "step": 397312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3486477332063216e-05, | |
| "loss": 4.2111, | |
| "step": 397824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3478091384552696e-05, | |
| "loss": 4.2137, | |
| "step": 398336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3469721815845905e-05, | |
| "loss": 4.2089, | |
| "step": 398848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3461335868335385e-05, | |
| "loss": 4.2043, | |
| "step": 399360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3452966299628594e-05, | |
| "loss": 4.2122, | |
| "step": 399872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3444580352118074e-05, | |
| "loss": 4.2072, | |
| "step": 400384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3436194404607554e-05, | |
| "loss": 4.192, | |
| "step": 400896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3427808457097034e-05, | |
| "loss": 4.1919, | |
| "step": 401408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3419422509586514e-05, | |
| "loss": 4.1934, | |
| "step": 401920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3411036562075993e-05, | |
| "loss": 4.1982, | |
| "step": 402432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.340266699336921e-05, | |
| "loss": 4.2056, | |
| "step": 402944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.339428104585869e-05, | |
| "loss": 4.1973, | |
| "step": 403456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.338589509834817e-05, | |
| "loss": 4.1943, | |
| "step": 403968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.337750915083765e-05, | |
| "loss": 4.2071, | |
| "step": 404480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.336912320332713e-05, | |
| "loss": 4.1985, | |
| "step": 404992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.336073725581661e-05, | |
| "loss": 4.2132, | |
| "step": 405504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.335235130830609e-05, | |
| "loss": 4.1928, | |
| "step": 406016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.334396536079557e-05, | |
| "loss": 4.1737, | |
| "step": 406528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.333559579208878e-05, | |
| "loss": 4.2109, | |
| "step": 407040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.332720984457826e-05, | |
| "loss": 4.205, | |
| "step": 407552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331882389706774e-05, | |
| "loss": 4.203, | |
| "step": 408064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331043794955722e-05, | |
| "loss": 4.1974, | |
| "step": 408576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.330206838085043e-05, | |
| "loss": 4.1926, | |
| "step": 409088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3293682433339914e-05, | |
| "loss": 4.1906, | |
| "step": 409600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3285296485829394e-05, | |
| "loss": 4.1944, | |
| "step": 410112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3276910538318874e-05, | |
| "loss": 4.1885, | |
| "step": 410624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326854096961208e-05, | |
| "loss": 4.1897, | |
| "step": 411136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326015502210156e-05, | |
| "loss": 4.1977, | |
| "step": 411648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.325178545339477e-05, | |
| "loss": 4.1925, | |
| "step": 412160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.324339950588425e-05, | |
| "loss": 4.1827, | |
| "step": 412672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.323501355837373e-05, | |
| "loss": 4.2019, | |
| "step": 413184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.322664398966694e-05, | |
| "loss": 4.1837, | |
| "step": 413696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.321825804215642e-05, | |
| "loss": 4.1837, | |
| "step": 414208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320988847344964e-05, | |
| "loss": 4.1964, | |
| "step": 414720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320150252593912e-05, | |
| "loss": 4.199, | |
| "step": 415232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.31931165784286e-05, | |
| "loss": 4.1796, | |
| "step": 415744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3184730630918077e-05, | |
| "loss": 4.1893, | |
| "step": 416256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3176344683407557e-05, | |
| "loss": 4.1719, | |
| "step": 416768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3167958735897036e-05, | |
| "loss": 4.1948, | |
| "step": 417280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3159572788386516e-05, | |
| "loss": 4.1949, | |
| "step": 417792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3151186840875996e-05, | |
| "loss": 4.1927, | |
| "step": 418304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.314280089336547e-05, | |
| "loss": 4.1908, | |
| "step": 418816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.313441494585495e-05, | |
| "loss": 4.2001, | |
| "step": 419328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.312602899834443e-05, | |
| "loss": 4.207, | |
| "step": 419840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.311764305083391e-05, | |
| "loss": 4.18, | |
| "step": 420352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310925710332339e-05, | |
| "loss": 4.1955, | |
| "step": 420864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310087115581287e-05, | |
| "loss": 4.1975, | |
| "step": 421376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3092501587106085e-05, | |
| "loss": 4.1727, | |
| "step": 421888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3084115639595565e-05, | |
| "loss": 4.1961, | |
| "step": 422400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3075729692085045e-05, | |
| "loss": 4.1802, | |
| "step": 422912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3067343744574525e-05, | |
| "loss": 4.197, | |
| "step": 423424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3058957797064005e-05, | |
| "loss": 4.1961, | |
| "step": 423936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3050571849553485e-05, | |
| "loss": 4.1801, | |
| "step": 424448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3042202280846694e-05, | |
| "loss": 4.1796, | |
| "step": 424960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3033816333336174e-05, | |
| "loss": 4.1847, | |
| "step": 425472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3025430385825654e-05, | |
| "loss": 4.2001, | |
| "step": 425984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3017044438315134e-05, | |
| "loss": 4.1849, | |
| "step": 426496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.300869124841207e-05, | |
| "loss": 4.1874, | |
| "step": 427008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.300030530090155e-05, | |
| "loss": 4.1813, | |
| "step": 427520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.299191935339104e-05, | |
| "loss": 4.1877, | |
| "step": 428032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.298353340588052e-05, | |
| "loss": 4.1827, | |
| "step": 428544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.297514745837e-05, | |
| "loss": 4.1653, | |
| "step": 429056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.296676151085948e-05, | |
| "loss": 4.1818, | |
| "step": 429568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.295837556334896e-05, | |
| "loss": 4.1898, | |
| "step": 430080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294998961583844e-05, | |
| "loss": 4.1891, | |
| "step": 430592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294162004713165e-05, | |
| "loss": 4.1721, | |
| "step": 431104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.293325047842486e-05, | |
| "loss": 4.1759, | |
| "step": 431616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.292486453091434e-05, | |
| "loss": 4.1812, | |
| "step": 432128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.291647858340382e-05, | |
| "loss": 4.1839, | |
| "step": 432640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.29080926358933e-05, | |
| "loss": 4.1911, | |
| "step": 433152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289970668838278e-05, | |
| "loss": 4.1743, | |
| "step": 433664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289132074087226e-05, | |
| "loss": 4.1807, | |
| "step": 434176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2882934793361737e-05, | |
| "loss": 4.1873, | |
| "step": 434688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.287454884585122e-05, | |
| "loss": 4.1677, | |
| "step": 435200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.28661628983407e-05, | |
| "loss": 4.1786, | |
| "step": 435712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.285779332963391e-05, | |
| "loss": 4.1687, | |
| "step": 436224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284940738212339e-05, | |
| "loss": 4.1827, | |
| "step": 436736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.28410378134166e-05, | |
| "loss": 4.1686, | |
| "step": 437248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.283265186590608e-05, | |
| "loss": 4.184, | |
| "step": 437760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.282426591839556e-05, | |
| "loss": 4.1726, | |
| "step": 438272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.281587997088504e-05, | |
| "loss": 4.1742, | |
| "step": 438784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.280749402337452e-05, | |
| "loss": 4.1767, | |
| "step": 439296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2799108075864e-05, | |
| "loss": 4.1769, | |
| "step": 439808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2790722128353474e-05, | |
| "loss": 4.1762, | |
| "step": 440320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.278233618084296e-05, | |
| "loss": 4.1903, | |
| "step": 440832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.277396661213618e-05, | |
| "loss": 4.1805, | |
| "step": 441344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.276558066462566e-05, | |
| "loss": 4.1827, | |
| "step": 441856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.275719471711513e-05, | |
| "loss": 4.1722, | |
| "step": 442368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.274880876960461e-05, | |
| "loss": 4.1769, | |
| "step": 442880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.274042282209409e-05, | |
| "loss": 4.171, | |
| "step": 443392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.273203687458357e-05, | |
| "loss": 4.1759, | |
| "step": 443904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.272365092707305e-05, | |
| "loss": 4.1732, | |
| "step": 444416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.271528135836626e-05, | |
| "loss": 4.1763, | |
| "step": 444928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.270689541085574e-05, | |
| "loss": 4.1767, | |
| "step": 445440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.269850946334522e-05, | |
| "loss": 4.1756, | |
| "step": 445952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.26901235158347e-05, | |
| "loss": 4.1704, | |
| "step": 446464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2681753947127915e-05, | |
| "loss": 4.1781, | |
| "step": 446976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2673367999617395e-05, | |
| "loss": 4.1707, | |
| "step": 447488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2664982052106875e-05, | |
| "loss": 4.1785, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2656596104596355e-05, | |
| "loss": 4.1716, | |
| "step": 448512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2648226535889564e-05, | |
| "loss": 4.1675, | |
| "step": 449024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.263985696718277e-05, | |
| "loss": 4.1526, | |
| "step": 449536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.263147101967225e-05, | |
| "loss": 4.1793, | |
| "step": 450048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.262310145096547e-05, | |
| "loss": 4.1798, | |
| "step": 450560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.261471550345495e-05, | |
| "loss": 4.1787, | |
| "step": 451072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.260632955594442e-05, | |
| "loss": 4.1751, | |
| "step": 451584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.25979436084339e-05, | |
| "loss": 4.1765, | |
| "step": 452096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.258955766092338e-05, | |
| "loss": 4.1699, | |
| "step": 452608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.258117171341287e-05, | |
| "loss": 4.183, | |
| "step": 453120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.257278576590235e-05, | |
| "loss": 4.164, | |
| "step": 453632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.256439981839183e-05, | |
| "loss": 4.1721, | |
| "step": 454144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.255601387088131e-05, | |
| "loss": 4.1691, | |
| "step": 454656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.254762792337079e-05, | |
| "loss": 4.1678, | |
| "step": 455168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253924197586027e-05, | |
| "loss": 4.1775, | |
| "step": 455680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253085602834975e-05, | |
| "loss": 4.1655, | |
| "step": 456192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.252248645964296e-05, | |
| "loss": 4.1748, | |
| "step": 456704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.251410051213244e-05, | |
| "loss": 4.165, | |
| "step": 457216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.250571456462192e-05, | |
| "loss": 4.1715, | |
| "step": 457728 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.179671764373779, | |
| "eval_runtime": 297.5058, | |
| "eval_samples_per_second": 1282.634, | |
| "eval_steps_per_second": 40.083, | |
| "step": 457920 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.24973286171114e-05, | |
| "loss": 4.1706, | |
| "step": 458240 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.248894266960088e-05, | |
| "loss": 4.1666, | |
| "step": 458752 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.248055672209036e-05, | |
| "loss": 4.1742, | |
| "step": 459264 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.247217077457984e-05, | |
| "loss": 4.1638, | |
| "step": 459776 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.246378482706932e-05, | |
| "loss": 4.1788, | |
| "step": 460288 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.24553988795588e-05, | |
| "loss": 4.1672, | |
| "step": 460800 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.244704568965574e-05, | |
| "loss": 4.1606, | |
| "step": 461312 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.243865974214522e-05, | |
| "loss": 4.1544, | |
| "step": 461824 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.24302737946347e-05, | |
| "loss": 4.1723, | |
| "step": 462336 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.242188784712418e-05, | |
| "loss": 4.1757, | |
| "step": 462848 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.241350189961366e-05, | |
| "loss": 4.166, | |
| "step": 463360 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.2405115952103135e-05, | |
| "loss": 4.1712, | |
| "step": 463872 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.239674638339635e-05, | |
| "loss": 4.1647, | |
| "step": 464384 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.238836043588583e-05, | |
| "loss": 4.1639, | |
| "step": 464896 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.237997448837531e-05, | |
| "loss": 4.1585, | |
| "step": 465408 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.2371588540864784e-05, | |
| "loss": 4.1553, | |
| "step": 465920 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.236320259335427e-05, | |
| "loss": 4.1639, | |
| "step": 466432 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.235481664584375e-05, | |
| "loss": 4.1599, | |
| "step": 466944 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.234643069833323e-05, | |
| "loss": 4.1666, | |
| "step": 467456 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.233806112962644e-05, | |
| "loss": 4.1709, | |
| "step": 467968 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.232967518211592e-05, | |
| "loss": 4.1748, | |
| "step": 468480 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.23212892346054e-05, | |
| "loss": 4.1656, | |
| "step": 468992 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.231290328709488e-05, | |
| "loss": 4.161, | |
| "step": 469504 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.230451733958436e-05, | |
| "loss": 4.1642, | |
| "step": 470016 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.229613139207384e-05, | |
| "loss": 4.159, | |
| "step": 470528 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.228774544456332e-05, | |
| "loss": 4.1608, | |
| "step": 471040 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.22793594970528e-05, | |
| "loss": 4.1416, | |
| "step": 471552 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.227098992834601e-05, | |
| "loss": 4.1693, | |
| "step": 472064 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.226260398083549e-05, | |
| "loss": 4.1579, | |
| "step": 472576 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.2254234412128704e-05, | |
| "loss": 4.1487, | |
| "step": 473088 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.2245848464618184e-05, | |
| "loss": 4.1584, | |
| "step": 473600 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.2237462517107664e-05, | |
| "loss": 4.1663, | |
| "step": 474112 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.2229076569597144e-05, | |
| "loss": 4.1614, | |
| "step": 474624 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.2220690622086624e-05, | |
| "loss": 4.1662, | |
| "step": 475136 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.221232105337983e-05, | |
| "loss": 4.1552, | |
| "step": 475648 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.220393510586931e-05, | |
| "loss": 4.1672, | |
| "step": 476160 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.219554915835879e-05, | |
| "loss": 4.1606, | |
| "step": 476672 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.218716321084827e-05, | |
| "loss": 4.1445, | |
| "step": 477184 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.217877726333775e-05, | |
| "loss": 4.1466, | |
| "step": 477696 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.217039131582723e-05, | |
| "loss": 4.1466, | |
| "step": 478208 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.216200536831671e-05, | |
| "loss": 4.1532, | |
| "step": 478720 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.215361942080619e-05, | |
| "loss": 4.162, | |
| "step": 479232 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.214524985209941e-05, | |
| "loss": 4.149, | |
| "step": 479744 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.213688028339262e-05, | |
| "loss": 4.1517, | |
| "step": 480256 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.21284943358821e-05, | |
| "loss": 4.1563, | |
| "step": 480768 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.212010838837158e-05, | |
| "loss": 4.1509, | |
| "step": 481280 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.211172244086106e-05, | |
| "loss": 4.167, | |
| "step": 481792 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.210335287215427e-05, | |
| "loss": 4.1489, | |
| "step": 482304 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.209496692464375e-05, | |
| "loss": 4.1242, | |
| "step": 482816 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.208658097713323e-05, | |
| "loss": 4.1663, | |
| "step": 483328 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.207819502962271e-05, | |
| "loss": 4.1598, | |
| "step": 483840 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.206980908211219e-05, | |
| "loss": 4.1606, | |
| "step": 484352 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.2061439513405396e-05, | |
| "loss": 4.1513, | |
| "step": 484864 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.2053053565894876e-05, | |
| "loss": 4.1472, | |
| "step": 485376 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.204466761838436e-05, | |
| "loss": 4.1457, | |
| "step": 485888 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.203628167087384e-05, | |
| "loss": 4.153, | |
| "step": 486400 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.202789572336332e-05, | |
| "loss": 4.1451, | |
| "step": 486912 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.201952615465653e-05, | |
| "loss": 4.1426, | |
| "step": 487424 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.201114020714601e-05, | |
| "loss": 4.1532, | |
| "step": 487936 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.200275425963549e-05, | |
| "loss": 4.1498, | |
| "step": 488448 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.199436831212497e-05, | |
| "loss": 4.1358, | |
| "step": 488960 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.1985982364614444e-05, | |
| "loss": 4.1575, | |
| "step": 489472 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.197761279590766e-05, | |
| "loss": 4.1361, | |
| "step": 489984 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.196922684839714e-05, | |
| "loss": 4.1409, | |
| "step": 490496 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.196084090088662e-05, | |
| "loss": 4.1538, | |
| "step": 491008 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.19524549533761e-05, | |
| "loss": 4.1536, | |
| "step": 491520 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.194406900586558e-05, | |
| "loss": 4.1338, | |
| "step": 492032 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.193568305835506e-05, | |
| "loss": 4.145, | |
| "step": 492544 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.192729711084454e-05, | |
| "loss": 4.1269, | |
| "step": 493056 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.191891116333402e-05, | |
| "loss": 4.1509, | |
| "step": 493568 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.19105252158235e-05, | |
| "loss": 4.1501, | |
| "step": 494080 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.190215564711671e-05, | |
| "loss": 4.1503, | |
| "step": 494592 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.189376969960619e-05, | |
| "loss": 4.1448, | |
| "step": 495104 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.188538375209567e-05, | |
| "loss": 4.1538, | |
| "step": 495616 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.187701418338888e-05, | |
| "loss": 4.1617, | |
| "step": 496128 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.186862823587836e-05, | |
| "loss": 4.1376, | |
| "step": 496640 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.186024228836784e-05, | |
| "loss": 4.1497, | |
| "step": 497152 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.185185634085732e-05, | |
| "loss": 4.1535, | |
| "step": 497664 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.18434703933468e-05, | |
| "loss": 4.1336, | |
| "step": 498176 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.183508444583628e-05, | |
| "loss": 4.1489, | |
| "step": 498688 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.1826714877129494e-05, | |
| "loss": 4.144, | |
| "step": 499200 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.1818328929618974e-05, | |
| "loss": 4.1504, | |
| "step": 499712 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.1809942982108454e-05, | |
| "loss": 4.1536, | |
| "step": 500224 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.1801557034597934e-05, | |
| "loss": 4.1385, | |
| "step": 500736 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.1793171087087414e-05, | |
| "loss": 4.137, | |
| "step": 501248 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.1784785139576894e-05, | |
| "loss": 4.1438, | |
| "step": 501760 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.1776399192066373e-05, | |
| "loss": 4.1492, | |
| "step": 502272 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.1768013244555853e-05, | |
| "loss": 4.1441, | |
| "step": 502784 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.175966005465279e-05, | |
| "loss": 4.1427, | |
| "step": 503296 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.175127410714227e-05, | |
| "loss": 4.1407, | |
| "step": 503808 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.174288815963175e-05, | |
| "loss": 4.1458, | |
| "step": 504320 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.173450221212123e-05, | |
| "loss": 4.1391, | |
| "step": 504832 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.172611626461072e-05, | |
| "loss": 4.1232, | |
| "step": 505344 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.17177303171002e-05, | |
| "loss": 4.1405, | |
| "step": 505856 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.170936074839341e-05, | |
| "loss": 4.1435, | |
| "step": 506368 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.170097480088289e-05, | |
| "loss": 4.1455, | |
| "step": 506880 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.169258885337237e-05, | |
| "loss": 4.1316, | |
| "step": 507392 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.168420290586185e-05, | |
| "loss": 4.1321, | |
| "step": 507904 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.167581695835133e-05, | |
| "loss": 4.1404, | |
| "step": 508416 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1667447389644536e-05, | |
| "loss": 4.1408, | |
| "step": 508928 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1659061442134016e-05, | |
| "loss": 4.1497, | |
| "step": 509440 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1650675494623496e-05, | |
| "loss": 4.1335, | |
| "step": 509952 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1642289547112976e-05, | |
| "loss": 4.1378, | |
| "step": 510464 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1633903599602456e-05, | |
| "loss": 4.1441, | |
| "step": 510976 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1625517652091936e-05, | |
| "loss": 4.1251, | |
| "step": 511488 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1617131704581416e-05, | |
| "loss": 4.1381, | |
| "step": 512000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.160876213587463e-05, | |
| "loss": 4.1222, | |
| "step": 512512 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1600376188364105e-05, | |
| "loss": 4.1447, | |
| "step": 513024 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1591990240853585e-05, | |
| "loss": 4.125, | |
| "step": 513536 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1583604293343065e-05, | |
| "loss": 4.1413, | |
| "step": 514048 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1575218345832545e-05, | |
| "loss": 4.1311, | |
| "step": 514560 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1566832398322025e-05, | |
| "loss": 4.1306, | |
| "step": 515072 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1558446450811505e-05, | |
| "loss": 4.1349, | |
| "step": 515584 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1550076882104714e-05, | |
| "loss": 4.1384, | |
| "step": 516096 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1541690934594194e-05, | |
| "loss": 4.1346, | |
| "step": 516608 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1533304987083674e-05, | |
| "loss": 4.1484, | |
| "step": 517120 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.152493541837689e-05, | |
| "loss": 4.1337, | |
| "step": 517632 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.151654947086637e-05, | |
| "loss": 4.1492, | |
| "step": 518144 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.150816352335585e-05, | |
| "loss": 4.1308, | |
| "step": 518656 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.149977757584533e-05, | |
| "loss": 4.1327, | |
| "step": 519168 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.149139162833481e-05, | |
| "loss": 4.1335, | |
| "step": 519680 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.148300568082429e-05, | |
| "loss": 4.1311, | |
| "step": 520192 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.147461973331377e-05, | |
| "loss": 4.1302, | |
| "step": 520704 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.146623378580325e-05, | |
| "loss": 4.1366, | |
| "step": 521216 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.145784783829273e-05, | |
| "loss": 4.1321, | |
| "step": 521728 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.144946189078221e-05, | |
| "loss": 4.1372, | |
| "step": 522240 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.144107594327169e-05, | |
| "loss": 4.1323, | |
| "step": 522752 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.14327063745649e-05, | |
| "loss": 4.1379, | |
| "step": 523264 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.142432042705438e-05, | |
| "loss": 4.1318, | |
| "step": 523776 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.141593447954386e-05, | |
| "loss": 4.1336, | |
| "step": 524288 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.140754853203334e-05, | |
| "loss": 4.13, | |
| "step": 524800 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.139916258452282e-05, | |
| "loss": 4.1302, | |
| "step": 525312 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1390793015816034e-05, | |
| "loss": 4.1128, | |
| "step": 525824 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1382407068305514e-05, | |
| "loss": 4.1385, | |
| "step": 526336 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1374021120794994e-05, | |
| "loss": 4.1379, | |
| "step": 526848 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.136563517328447e-05, | |
| "loss": 4.1389, | |
| "step": 527360 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.135724922577395e-05, | |
| "loss": 4.1312, | |
| "step": 527872 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.134886327826343e-05, | |
| "loss": 4.1394, | |
| "step": 528384 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.134047733075291e-05, | |
| "loss": 4.1325, | |
| "step": 528896 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.133209138324239e-05, | |
| "loss": 4.1419, | |
| "step": 529408 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1323721814535596e-05, | |
| "loss": 4.1219, | |
| "step": 529920 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1315335867025076e-05, | |
| "loss": 4.1282, | |
| "step": 530432 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.1306949919514556e-05, | |
| "loss": 4.1313, | |
| "step": 530944 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.129856397200404e-05, | |
| "loss": 4.1307, | |
| "step": 531456 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.129019440329725e-05, | |
| "loss": 4.134, | |
| "step": 531968 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.128182483459047e-05, | |
| "loss": 4.1287, | |
| "step": 532480 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.127343888707994e-05, | |
| "loss": 4.1334, | |
| "step": 532992 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.126505293956942e-05, | |
| "loss": 4.1258, | |
| "step": 533504 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.12566669920589e-05, | |
| "loss": 4.1356, | |
| "step": 534016 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "eval_loss": 4.148338317871094, | |
| "eval_runtime": 296.5235, | |
| "eval_samples_per_second": 1286.883, | |
| "eval_steps_per_second": 40.216, | |
| "step": 534240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.124828104454838e-05, | |
| "loss": 4.1339, | |
| "step": 534528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.123989509703786e-05, | |
| "loss": 4.1319, | |
| "step": 535040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.123150914952734e-05, | |
| "loss": 4.1295, | |
| "step": 535552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.122312320201682e-05, | |
| "loss": 4.1242, | |
| "step": 536064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.12147372545063e-05, | |
| "loss": 4.1371, | |
| "step": 536576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.120635130699578e-05, | |
| "loss": 4.1311, | |
| "step": 537088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.119796535948526e-05, | |
| "loss": 4.1226, | |
| "step": 537600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.118957941197474e-05, | |
| "loss": 4.1189, | |
| "step": 538112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.118119346446423e-05, | |
| "loss": 4.1268, | |
| "step": 538624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.117280751695371e-05, | |
| "loss": 4.1389, | |
| "step": 539136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.116442156944318e-05, | |
| "loss": 4.1245, | |
| "step": 539648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.115603562193266e-05, | |
| "loss": 4.1312, | |
| "step": 540160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1147666053225876e-05, | |
| "loss": 4.131, | |
| "step": 540672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1139280105715356e-05, | |
| "loss": 4.1208, | |
| "step": 541184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.113089415820483e-05, | |
| "loss": 4.1233, | |
| "step": 541696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.112250821069431e-05, | |
| "loss": 4.1159, | |
| "step": 542208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.111412226318379e-05, | |
| "loss": 4.124, | |
| "step": 542720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.110573631567327e-05, | |
| "loss": 4.1198, | |
| "step": 543232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.109735036816275e-05, | |
| "loss": 4.1236, | |
| "step": 543744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.108896442065223e-05, | |
| "loss": 4.1351, | |
| "step": 544256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.108057847314171e-05, | |
| "loss": 4.1359, | |
| "step": 544768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1072192525631196e-05, | |
| "loss": 4.1268, | |
| "step": 545280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1063806578120676e-05, | |
| "loss": 4.1203, | |
| "step": 545792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1055420630610156e-05, | |
| "loss": 4.1223, | |
| "step": 546304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1047051061903365e-05, | |
| "loss": 4.123, | |
| "step": 546816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1038681493196574e-05, | |
| "loss": 4.1217, | |
| "step": 547328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1030295545686054e-05, | |
| "loss": 4.1064, | |
| "step": 547840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1021909598175534e-05, | |
| "loss": 4.1288, | |
| "step": 548352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1013523650665014e-05, | |
| "loss": 4.1221, | |
| "step": 548864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1005137703154494e-05, | |
| "loss": 4.114, | |
| "step": 549376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0996751755643974e-05, | |
| "loss": 4.1162, | |
| "step": 549888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0988365808133454e-05, | |
| "loss": 4.131, | |
| "step": 550400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0979979860622933e-05, | |
| "loss": 4.1256, | |
| "step": 550912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0971593913112413e-05, | |
| "loss": 4.1249, | |
| "step": 551424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.096322434440563e-05, | |
| "loss": 4.1183, | |
| "step": 551936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.095485477569884e-05, | |
| "loss": 4.125, | |
| "step": 552448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.094646882818832e-05, | |
| "loss": 4.1254, | |
| "step": 552960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.09380828806778e-05, | |
| "loss": 4.108, | |
| "step": 553472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.092969693316728e-05, | |
| "loss": 4.1087, | |
| "step": 553984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.092131098565676e-05, | |
| "loss": 4.1088, | |
| "step": 554496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.091292503814624e-05, | |
| "loss": 4.1147, | |
| "step": 555008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.090453909063572e-05, | |
| "loss": 4.1213, | |
| "step": 555520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.089616952192893e-05, | |
| "loss": 4.1151, | |
| "step": 556032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.088778357441841e-05, | |
| "loss": 4.1133, | |
| "step": 556544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.087939762690789e-05, | |
| "loss": 4.1195, | |
| "step": 557056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.087101167939737e-05, | |
| "loss": 4.1144, | |
| "step": 557568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.086264211069058e-05, | |
| "loss": 4.1264, | |
| "step": 558080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.085425616318006e-05, | |
| "loss": 4.118, | |
| "step": 558592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.084588659447327e-05, | |
| "loss": 4.089, | |
| "step": 559104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.083750064696275e-05, | |
| "loss": 4.1257, | |
| "step": 559616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.082911469945223e-05, | |
| "loss": 4.1216, | |
| "step": 560128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.082072875194171e-05, | |
| "loss": 4.1264, | |
| "step": 560640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.081234280443119e-05, | |
| "loss": 4.1131, | |
| "step": 561152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0803956856920665e-05, | |
| "loss": 4.1089, | |
| "step": 561664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.079558728821388e-05, | |
| "loss": 4.1119, | |
| "step": 562176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.078720134070336e-05, | |
| "loss": 4.1137, | |
| "step": 562688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0778815393192834e-05, | |
| "loss": 4.1129, | |
| "step": 563200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.077042944568232e-05, | |
| "loss": 4.1017, | |
| "step": 563712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.07620434981718e-05, | |
| "loss": 4.1187, | |
| "step": 564224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.075365755066128e-05, | |
| "loss": 4.116, | |
| "step": 564736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.074527160315076e-05, | |
| "loss": 4.0988, | |
| "step": 565248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.073690203444397e-05, | |
| "loss": 4.1218, | |
| "step": 565760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.072851608693345e-05, | |
| "loss": 4.1003, | |
| "step": 566272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.072013013942293e-05, | |
| "loss": 4.1031, | |
| "step": 566784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.071174419191241e-05, | |
| "loss": 4.1174, | |
| "step": 567296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.070335824440189e-05, | |
| "loss": 4.1172, | |
| "step": 567808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.069497229689137e-05, | |
| "loss": 4.1012, | |
| "step": 568320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.068658634938085e-05, | |
| "loss": 4.1086, | |
| "step": 568832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.067820040187033e-05, | |
| "loss": 4.0902, | |
| "step": 569344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.066983083316354e-05, | |
| "loss": 4.1144, | |
| "step": 569856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.066144488565302e-05, | |
| "loss": 4.1139, | |
| "step": 570368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0653075316946234e-05, | |
| "loss": 4.1156, | |
| "step": 570880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0644689369435714e-05, | |
| "loss": 4.1073, | |
| "step": 571392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0636303421925194e-05, | |
| "loss": 4.1166, | |
| "step": 571904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0627917474414674e-05, | |
| "loss": 4.13, | |
| "step": 572416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0619531526904154e-05, | |
| "loss": 4.0994, | |
| "step": 572928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0611145579393634e-05, | |
| "loss": 4.1151, | |
| "step": 573440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0602759631883114e-05, | |
| "loss": 4.1179, | |
| "step": 573952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0594373684372594e-05, | |
| "loss": 4.1003, | |
| "step": 574464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0585987736862074e-05, | |
| "loss": 4.1129, | |
| "step": 574976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.057761816815528e-05, | |
| "loss": 4.11, | |
| "step": 575488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.056923222064476e-05, | |
| "loss": 4.1134, | |
| "step": 576000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.056084627313424e-05, | |
| "loss": 4.1216, | |
| "step": 576512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.055246032562372e-05, | |
| "loss": 4.0985, | |
| "step": 577024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.054410713572067e-05, | |
| "loss": 4.1042, | |
| "step": 577536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.053572118821015e-05, | |
| "loss": 4.1074, | |
| "step": 578048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.052733524069963e-05, | |
| "loss": 4.1159, | |
| "step": 578560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.051894929318911e-05, | |
| "loss": 4.1064, | |
| "step": 579072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.051057972448232e-05, | |
| "loss": 4.1125, | |
| "step": 579584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.05021937769718e-05, | |
| "loss": 4.1085, | |
| "step": 580096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.049380782946128e-05, | |
| "loss": 4.1058, | |
| "step": 580608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.048542188195076e-05, | |
| "loss": 4.1074, | |
| "step": 581120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.047703593444024e-05, | |
| "loss": 4.0871, | |
| "step": 581632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0468666365733446e-05, | |
| "loss": 4.1039, | |
| "step": 582144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0460280418222926e-05, | |
| "loss": 4.1095, | |
| "step": 582656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.045189447071241e-05, | |
| "loss": 4.1126, | |
| "step": 583168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.044350852320189e-05, | |
| "loss": 4.0984, | |
| "step": 583680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.043512257569137e-05, | |
| "loss": 4.0955, | |
| "step": 584192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.042675300698458e-05, | |
| "loss": 4.1049, | |
| "step": 584704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.041836705947406e-05, | |
| "loss": 4.1068, | |
| "step": 585216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.040998111196354e-05, | |
| "loss": 4.1128, | |
| "step": 585728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.040159516445302e-05, | |
| "loss": 4.1043, | |
| "step": 586240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0393209216942495e-05, | |
| "loss": 4.1063, | |
| "step": 586752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0384823269431975e-05, | |
| "loss": 4.104, | |
| "step": 587264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0376437321921454e-05, | |
| "loss": 4.0966, | |
| "step": 587776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0368051374410934e-05, | |
| "loss": 4.099, | |
| "step": 588288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.035968180570415e-05, | |
| "loss": 4.0899, | |
| "step": 588800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.035129585819363e-05, | |
| "loss": 4.1118, | |
| "step": 589312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0342926289486846e-05, | |
| "loss": 4.091, | |
| "step": 589824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0334540341976326e-05, | |
| "loss": 4.1068, | |
| "step": 590336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.03261543944658e-05, | |
| "loss": 4.0982, | |
| "step": 590848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.031776844695528e-05, | |
| "loss": 4.1014, | |
| "step": 591360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.030938249944476e-05, | |
| "loss": 4.099, | |
| "step": 591872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.030099655193424e-05, | |
| "loss": 4.1027, | |
| "step": 592384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.029261060442372e-05, | |
| "loss": 4.1011, | |
| "step": 592896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.02842246569132e-05, | |
| "loss": 4.1161, | |
| "step": 593408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.027585508820641e-05, | |
| "loss": 4.0989, | |
| "step": 593920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.026748551949962e-05, | |
| "loss": 4.1145, | |
| "step": 594432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0259099571989104e-05, | |
| "loss": 4.0958, | |
| "step": 594944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0250713624478584e-05, | |
| "loss": 4.0994, | |
| "step": 595456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0242327676968064e-05, | |
| "loss": 4.1002, | |
| "step": 595968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0233941729457544e-05, | |
| "loss": 4.1017, | |
| "step": 596480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0225555781947024e-05, | |
| "loss": 4.093, | |
| "step": 596992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0217169834436504e-05, | |
| "loss": 4.1038, | |
| "step": 597504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0208783886925984e-05, | |
| "loss": 4.0958, | |
| "step": 598016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0200397939415464e-05, | |
| "loss": 4.1036, | |
| "step": 598528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0192011991904944e-05, | |
| "loss": 4.1027, | |
| "step": 599040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.018364242319815e-05, | |
| "loss": 4.0979, | |
| "step": 599552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.017525647568763e-05, | |
| "loss": 4.1015, | |
| "step": 600064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.016687052817711e-05, | |
| "loss": 4.1066, | |
| "step": 600576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.015848458066659e-05, | |
| "loss": 4.0934, | |
| "step": 601088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.015013139076354e-05, | |
| "loss": 4.0985, | |
| "step": 601600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.014174544325302e-05, | |
| "loss": 4.0779, | |
| "step": 602112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.01333594957425e-05, | |
| "loss": 4.1052, | |
| "step": 602624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.012497354823198e-05, | |
| "loss": 4.1058, | |
| "step": 603136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.011658760072146e-05, | |
| "loss": 4.1074, | |
| "step": 603648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.010820165321094e-05, | |
| "loss": 4.0996, | |
| "step": 604160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0099832084504146e-05, | |
| "loss": 4.1054, | |
| "step": 604672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0091446136993626e-05, | |
| "loss": 4.1002, | |
| "step": 605184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0083060189483106e-05, | |
| "loss": 4.1069, | |
| "step": 605696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0074674241972586e-05, | |
| "loss": 4.0898, | |
| "step": 606208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0066288294462066e-05, | |
| "loss": 4.0984, | |
| "step": 606720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0057902346951546e-05, | |
| "loss": 4.0963, | |
| "step": 607232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0049516399441026e-05, | |
| "loss": 4.1004, | |
| "step": 607744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.0041146830734235e-05, | |
| "loss": 4.1017, | |
| "step": 608256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.003276088322372e-05, | |
| "loss": 4.0923, | |
| "step": 608768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.00243749357132e-05, | |
| "loss": 4.1032, | |
| "step": 609280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.001598898820268e-05, | |
| "loss": 4.0923, | |
| "step": 609792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.000760304069216e-05, | |
| "loss": 4.1053, | |
| "step": 610304 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.123390197753906, | |
| "eval_runtime": 301.8228, | |
| "eval_samples_per_second": 1264.288, | |
| "eval_steps_per_second": 39.51, | |
| "step": 610560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9999217093181635e-05, | |
| "loss": 4.0888, | |
| "step": 610816 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9990831145671115e-05, | |
| "loss": 4.0986, | |
| "step": 611328 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9982445198160595e-05, | |
| "loss": 4.0979, | |
| "step": 611840 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9974075629453804e-05, | |
| "loss": 4.095, | |
| "step": 612352 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9965689681943284e-05, | |
| "loss": 4.1072, | |
| "step": 612864 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.99573201132365e-05, | |
| "loss": 4.0991, | |
| "step": 613376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.994893416572598e-05, | |
| "loss": 4.0886, | |
| "step": 613888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.994054821821546e-05, | |
| "loss": 4.0852, | |
| "step": 614400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.993216227070494e-05, | |
| "loss": 4.094, | |
| "step": 614912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.992377632319442e-05, | |
| "loss": 4.1068, | |
| "step": 615424 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.99153903756839e-05, | |
| "loss": 4.0974, | |
| "step": 615936 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.990702080697711e-05, | |
| "loss": 4.096, | |
| "step": 616448 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9898651238270325e-05, | |
| "loss": 4.0979, | |
| "step": 616960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9890265290759805e-05, | |
| "loss": 4.0889, | |
| "step": 617472 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.988187934324928e-05, | |
| "loss": 4.0875, | |
| "step": 617984 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.987349339573876e-05, | |
| "loss": 4.084, | |
| "step": 618496 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.986510744822824e-05, | |
| "loss": 4.0945, | |
| "step": 619008 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.985672150071772e-05, | |
| "loss": 4.0881, | |
| "step": 619520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.98483355532072e-05, | |
| "loss": 4.0947, | |
| "step": 620032 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.983994960569668e-05, | |
| "loss": 4.1029, | |
| "step": 620544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.983156365818616e-05, | |
| "loss": 4.1034, | |
| "step": 621056 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9823177710675644e-05, | |
| "loss": 4.0928, | |
| "step": 621568 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.9814791763165124e-05, | |
| "loss": 4.0934, | |
| "step": 622080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.980642219445833e-05, | |
| "loss": 4.0905, | |
| "step": 622592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.979803624694781e-05, | |
| "loss": 4.0893, | |
| "step": 623104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.978965029943729e-05, | |
| "loss": 4.0902, | |
| "step": 623616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.97812807307305e-05, | |
| "loss": 4.0777, | |
| "step": 624128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.977289478321998e-05, | |
| "loss": 4.0994, | |
| "step": 624640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.976450883570946e-05, | |
| "loss": 4.0909, | |
| "step": 625152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.975612288819894e-05, | |
| "loss": 4.0786, | |
| "step": 625664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.974773694068842e-05, | |
| "loss": 4.0844, | |
| "step": 626176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.97393509931779e-05, | |
| "loss": 4.0965, | |
| "step": 626688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.973096504566738e-05, | |
| "loss": 4.0985, | |
| "step": 627200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.972257909815686e-05, | |
| "loss": 4.0939, | |
| "step": 627712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.971420952945008e-05, | |
| "loss": 4.0885, | |
| "step": 628224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.970582358193956e-05, | |
| "loss": 4.0878, | |
| "step": 628736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.969743763442904e-05, | |
| "loss": 4.0998, | |
| "step": 629248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.968905168691852e-05, | |
| "loss": 4.0775, | |
| "step": 629760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.968068211821173e-05, | |
| "loss": 4.0764, | |
| "step": 630272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.967229617070121e-05, | |
| "loss": 4.0811, | |
| "step": 630784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.966391022319069e-05, | |
| "loss": 4.0823, | |
| "step": 631296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.965552427568017e-05, | |
| "loss": 4.0905, | |
| "step": 631808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.964713832816964e-05, | |
| "loss": 4.0797, | |
| "step": 632320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.963875238065912e-05, | |
| "loss": 4.0906, | |
| "step": 632832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.96303664331486e-05, | |
| "loss": 4.0853, | |
| "step": 633344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.962198048563808e-05, | |
| "loss": 4.0823, | |
| "step": 633856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9613610916931296e-05, | |
| "loss": 4.0998, | |
| "step": 634368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9605224969420776e-05, | |
| "loss": 4.0853, | |
| "step": 634880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9596839021910256e-05, | |
| "loss": 4.0625, | |
| "step": 635392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9588453074399736e-05, | |
| "loss": 4.0935, | |
| "step": 635904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9580083505692945e-05, | |
| "loss": 4.0877, | |
| "step": 636416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.957174669459362e-05, | |
| "loss": 4.0966, | |
| "step": 636928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.95633607470831e-05, | |
| "loss": 4.0841, | |
| "step": 637440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.955497479957258e-05, | |
| "loss": 4.0784, | |
| "step": 637952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.954658885206206e-05, | |
| "loss": 4.0809, | |
| "step": 638464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.953820290455154e-05, | |
| "loss": 4.0803, | |
| "step": 638976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.952981695704102e-05, | |
| "loss": 4.0865, | |
| "step": 639488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.95214310095305e-05, | |
| "loss": 4.0693, | |
| "step": 640000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9513045062019985e-05, | |
| "loss": 4.0914, | |
| "step": 640512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9504659114509465e-05, | |
| "loss": 4.0852, | |
| "step": 641024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.949627316699894e-05, | |
| "loss": 4.0703, | |
| "step": 641536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.948788721948842e-05, | |
| "loss": 4.0911, | |
| "step": 642048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.94795012719779e-05, | |
| "loss": 4.0703, | |
| "step": 642560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.947111532446738e-05, | |
| "loss": 4.0728, | |
| "step": 643072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.946274575576059e-05, | |
| "loss": 4.0851, | |
| "step": 643584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.945435980825007e-05, | |
| "loss": 4.0907, | |
| "step": 644096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.944597386073955e-05, | |
| "loss": 4.067, | |
| "step": 644608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.943758791322903e-05, | |
| "loss": 4.0828, | |
| "step": 645120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.942920196571851e-05, | |
| "loss": 4.0626, | |
| "step": 645632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.942081601820799e-05, | |
| "loss": 4.0851, | |
| "step": 646144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.941243007069747e-05, | |
| "loss": 4.0851, | |
| "step": 646656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9404044123186954e-05, | |
| "loss": 4.0773, | |
| "step": 647168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9395658175676434e-05, | |
| "loss": 4.084, | |
| "step": 647680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.938728860696964e-05, | |
| "loss": 4.0846, | |
| "step": 648192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.937890265945912e-05, | |
| "loss": 4.1034, | |
| "step": 648704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.93705167119486e-05, | |
| "loss": 4.0679, | |
| "step": 649216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.936213076443808e-05, | |
| "loss": 4.0859, | |
| "step": 649728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.935376119573129e-05, | |
| "loss": 4.084, | |
| "step": 650240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.934537524822077e-05, | |
| "loss": 4.0761, | |
| "step": 650752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.933700567951398e-05, | |
| "loss": 4.0821, | |
| "step": 651264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.932861973200346e-05, | |
| "loss": 4.0785, | |
| "step": 651776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.932023378449294e-05, | |
| "loss": 4.0842, | |
| "step": 652288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.931184783698242e-05, | |
| "loss": 4.0935, | |
| "step": 652800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.930346188947191e-05, | |
| "loss": 4.0694, | |
| "step": 653312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.929507594196139e-05, | |
| "loss": 4.075, | |
| "step": 653824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.928668999445087e-05, | |
| "loss": 4.0771, | |
| "step": 654336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9278320425744076e-05, | |
| "loss": 4.0894, | |
| "step": 654848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9269934478233556e-05, | |
| "loss": 4.073, | |
| "step": 655360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9261564909526766e-05, | |
| "loss": 4.0824, | |
| "step": 655872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9253178962016245e-05, | |
| "loss": 4.0818, | |
| "step": 656384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9244793014505725e-05, | |
| "loss": 4.0772, | |
| "step": 656896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9236407066995205e-05, | |
| "loss": 4.077, | |
| "step": 657408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9228021119484685e-05, | |
| "loss": 4.0594, | |
| "step": 657920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9219635171974165e-05, | |
| "loss": 4.0743, | |
| "step": 658432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9211249224463645e-05, | |
| "loss": 4.0791, | |
| "step": 658944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9202863276953125e-05, | |
| "loss": 4.0844, | |
| "step": 659456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9194477329442605e-05, | |
| "loss": 4.072, | |
| "step": 659968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9186091381932085e-05, | |
| "loss": 4.0656, | |
| "step": 660480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.91777218132253e-05, | |
| "loss": 4.0735, | |
| "step": 660992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9169335865714774e-05, | |
| "loss": 4.0798, | |
| "step": 661504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9160949918204254e-05, | |
| "loss": 4.0842, | |
| "step": 662016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9152563970693734e-05, | |
| "loss": 4.0749, | |
| "step": 662528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9144178023183214e-05, | |
| "loss": 4.0799, | |
| "step": 663040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.913580845447642e-05, | |
| "loss": 4.0728, | |
| "step": 663552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.91274225069659e-05, | |
| "loss": 4.0735, | |
| "step": 664064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.911903655945538e-05, | |
| "loss": 4.0661, | |
| "step": 664576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.91106669907486e-05, | |
| "loss": 4.0681, | |
| "step": 665088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.910228104323808e-05, | |
| "loss": 4.0846, | |
| "step": 665600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.909389509572756e-05, | |
| "loss": 4.0627, | |
| "step": 666112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.908550914821704e-05, | |
| "loss": 4.0761, | |
| "step": 666624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.907712320070652e-05, | |
| "loss": 4.0656, | |
| "step": 667136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9068737253196e-05, | |
| "loss": 4.0745, | |
| "step": 667648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.906035130568548e-05, | |
| "loss": 4.0714, | |
| "step": 668160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.905196535817496e-05, | |
| "loss": 4.074, | |
| "step": 668672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.904357941066444e-05, | |
| "loss": 4.0731, | |
| "step": 669184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.903519346315392e-05, | |
| "loss": 4.0863, | |
| "step": 669696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.90268075156434e-05, | |
| "loss": 4.0737, | |
| "step": 670208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.901842156813288e-05, | |
| "loss": 4.0827, | |
| "step": 670720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.901005199942609e-05, | |
| "loss": 4.0701, | |
| "step": 671232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.900166605191557e-05, | |
| "loss": 4.0733, | |
| "step": 671744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.899328010440505e-05, | |
| "loss": 4.0747, | |
| "step": 672256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.898489415689453e-05, | |
| "loss": 4.0742, | |
| "step": 672768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.897652458818774e-05, | |
| "loss": 4.0654, | |
| "step": 673280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.896813864067722e-05, | |
| "loss": 4.0748, | |
| "step": 673792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.89597526931667e-05, | |
| "loss": 4.0691, | |
| "step": 674304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.895136674565618e-05, | |
| "loss": 4.0766, | |
| "step": 674816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.894298079814566e-05, | |
| "loss": 4.072, | |
| "step": 675328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.893461122943887e-05, | |
| "loss": 4.0777, | |
| "step": 675840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.892622528192835e-05, | |
| "loss": 4.0712, | |
| "step": 676352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.891783933441783e-05, | |
| "loss": 4.081, | |
| "step": 676864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.890945338690731e-05, | |
| "loss": 4.0635, | |
| "step": 677376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.890108381820052e-05, | |
| "loss": 4.0723, | |
| "step": 677888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.889269787069e-05, | |
| "loss": 4.0468, | |
| "step": 678400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.888431192317948e-05, | |
| "loss": 4.0801, | |
| "step": 678912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.887592597566896e-05, | |
| "loss": 4.0785, | |
| "step": 679424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.886755640696218e-05, | |
| "loss": 4.0814, | |
| "step": 679936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.885917045945166e-05, | |
| "loss": 4.0693, | |
| "step": 680448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.885078451194114e-05, | |
| "loss": 4.0809, | |
| "step": 680960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8842414943234346e-05, | |
| "loss": 4.071, | |
| "step": 681472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8834028995723826e-05, | |
| "loss": 4.084, | |
| "step": 681984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8825643048213306e-05, | |
| "loss": 4.0639, | |
| "step": 682496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8817257100702786e-05, | |
| "loss": 4.0689, | |
| "step": 683008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.880887115319226e-05, | |
| "loss": 4.0687, | |
| "step": 683520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.880048520568174e-05, | |
| "loss": 4.0794, | |
| "step": 684032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.879209925817122e-05, | |
| "loss": 4.0675, | |
| "step": 684544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.87837133106607e-05, | |
| "loss": 4.0656, | |
| "step": 685056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8775343741953915e-05, | |
| "loss": 4.0752, | |
| "step": 685568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.8766957794443395e-05, | |
| "loss": 4.0682, | |
| "step": 686080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.875858822573661e-05, | |
| "loss": 4.0757, | |
| "step": 686592 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.1041579246521, | |
| "eval_runtime": 301.1615, | |
| "eval_samples_per_second": 1267.064, | |
| "eval_steps_per_second": 39.597, | |
| "step": 686880 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.8750202278226084e-05, | |
| "loss": 4.0618, | |
| "step": 687104 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.8741816330715564e-05, | |
| "loss": 4.0752, | |
| "step": 687616 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.8733430383205044e-05, | |
| "loss": 4.0678, | |
| "step": 688128 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.8725044435694524e-05, | |
| "loss": 4.0712, | |
| "step": 688640 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.8716658488184004e-05, | |
| "loss": 4.0821, | |
| "step": 689152 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.8708272540673483e-05, | |
| "loss": 4.0701, | |
| "step": 689664 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.8699886593162963e-05, | |
| "loss": 4.0644, | |
| "step": 690176 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.869153340325991e-05, | |
| "loss": 4.0595, | |
| "step": 690688 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.868314745574939e-05, | |
| "loss": 4.069, | |
| "step": 691200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.867476150823887e-05, | |
| "loss": 4.0807, | |
| "step": 691712 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.866637556072835e-05, | |
| "loss": 4.0679, | |
| "step": 692224 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.865798961321783e-05, | |
| "loss": 4.0708, | |
| "step": 692736 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.864962004451104e-05, | |
| "loss": 4.073, | |
| "step": 693248 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.864123409700052e-05, | |
| "loss": 4.0634, | |
| "step": 693760 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.863284814949e-05, | |
| "loss": 4.0618, | |
| "step": 694272 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.862446220197948e-05, | |
| "loss": 4.0569, | |
| "step": 694784 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.861607625446896e-05, | |
| "loss": 4.0686, | |
| "step": 695296 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.860769030695844e-05, | |
| "loss": 4.0634, | |
| "step": 695808 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.859930435944792e-05, | |
| "loss": 4.062, | |
| "step": 696320 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.8590934790741126e-05, | |
| "loss": 4.0748, | |
| "step": 696832 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.8582548843230606e-05, | |
| "loss": 4.0759, | |
| "step": 697344 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.857416289572009e-05, | |
| "loss": 4.0682, | |
| "step": 697856 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.856577694820957e-05, | |
| "loss": 4.0668, | |
| "step": 698368 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.855739100069905e-05, | |
| "loss": 4.0655, | |
| "step": 698880 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.854900505318853e-05, | |
| "loss": 4.0651, | |
| "step": 699392 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.854061910567801e-05, | |
| "loss": 4.0641, | |
| "step": 699904 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.853224953697122e-05, | |
| "loss": 4.0492, | |
| "step": 700416 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.85238635894607e-05, | |
| "loss": 4.0699, | |
| "step": 700928 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.851549402075391e-05, | |
| "loss": 4.0692, | |
| "step": 701440 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.850710807324339e-05, | |
| "loss": 4.0527, | |
| "step": 701952 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.849872212573287e-05, | |
| "loss": 4.0599, | |
| "step": 702464 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.849033617822235e-05, | |
| "loss": 4.0664, | |
| "step": 702976 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.848195023071183e-05, | |
| "loss": 4.0714, | |
| "step": 703488 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.847356428320131e-05, | |
| "loss": 4.0718, | |
| "step": 704000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.846517833569079e-05, | |
| "loss": 4.0667, | |
| "step": 704512 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.845679238818027e-05, | |
| "loss": 4.0583, | |
| "step": 705024 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.844840644066975e-05, | |
| "loss": 4.0755, | |
| "step": 705536 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.844002049315923e-05, | |
| "loss": 4.0499, | |
| "step": 706048 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.843163454564871e-05, | |
| "loss": 4.0512, | |
| "step": 706560 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.842324859813819e-05, | |
| "loss": 4.0548, | |
| "step": 707072 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.841486265062767e-05, | |
| "loss": 4.0581, | |
| "step": 707584 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.840649308192088e-05, | |
| "loss": 4.063, | |
| "step": 708096 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.839810713441036e-05, | |
| "loss": 4.0558, | |
| "step": 708608 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.838972118689984e-05, | |
| "loss": 4.0645, | |
| "step": 709120 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.838133523938932e-05, | |
| "loss": 4.0614, | |
| "step": 709632 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.83729492918788e-05, | |
| "loss": 4.0564, | |
| "step": 710144 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.836457972317201e-05, | |
| "loss": 4.0689, | |
| "step": 710656 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8356193775661495e-05, | |
| "loss": 4.065, | |
| "step": 711168 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8347807828150975e-05, | |
| "loss": 4.0378, | |
| "step": 711680 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8339438259444184e-05, | |
| "loss": 4.0635, | |
| "step": 712192 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8331052311933664e-05, | |
| "loss": 4.0644, | |
| "step": 712704 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8322666364423144e-05, | |
| "loss": 4.0722, | |
| "step": 713216 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8314280416912624e-05, | |
| "loss": 4.0584, | |
| "step": 713728 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.830591084820583e-05, | |
| "loss": 4.0562, | |
| "step": 714240 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.829752490069531e-05, | |
| "loss": 4.0515, | |
| "step": 714752 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.828915533198852e-05, | |
| "loss": 4.0571, | |
| "step": 715264 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8280769384478e-05, | |
| "loss": 4.0599, | |
| "step": 715776 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.827238343696748e-05, | |
| "loss": 4.0465, | |
| "step": 716288 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.826399748945696e-05, | |
| "loss": 4.0637, | |
| "step": 716800 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.825561154194645e-05, | |
| "loss": 4.0605, | |
| "step": 717312 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.824722559443593e-05, | |
| "loss": 4.0504, | |
| "step": 717824 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.823883964692541e-05, | |
| "loss": 4.061, | |
| "step": 718336 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.823045369941489e-05, | |
| "loss": 4.0471, | |
| "step": 718848 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.822206775190437e-05, | |
| "loss": 4.0485, | |
| "step": 719360 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.821368180439385e-05, | |
| "loss": 4.0549, | |
| "step": 719872 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.820529585688333e-05, | |
| "loss": 4.0658, | |
| "step": 720384 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.819690990937281e-05, | |
| "loss": 4.0469, | |
| "step": 720896 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.818854034066602e-05, | |
| "loss": 4.0591, | |
| "step": 721408 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.81801543931555e-05, | |
| "loss": 4.0339, | |
| "step": 721920 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8171784824448707e-05, | |
| "loss": 4.06, | |
| "step": 722432 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8163398876938187e-05, | |
| "loss": 4.0591, | |
| "step": 722944 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8155012929427666e-05, | |
| "loss": 4.0596, | |
| "step": 723456 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8146626981917146e-05, | |
| "loss": 4.0547, | |
| "step": 723968 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.813824103440663e-05, | |
| "loss": 4.0606, | |
| "step": 724480 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8129855086896106e-05, | |
| "loss": 4.0756, | |
| "step": 724992 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.812148551818932e-05, | |
| "loss": 4.0485, | |
| "step": 725504 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.81130995706788e-05, | |
| "loss": 4.0606, | |
| "step": 726016 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.810471362316828e-05, | |
| "loss": 4.0521, | |
| "step": 726528 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8096327675657755e-05, | |
| "loss": 4.0545, | |
| "step": 727040 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8087941728147235e-05, | |
| "loss": 4.0559, | |
| "step": 727552 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8079555780636715e-05, | |
| "loss": 4.053, | |
| "step": 728064 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8071186211929924e-05, | |
| "loss": 4.0626, | |
| "step": 728576 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.8062800264419404e-05, | |
| "loss": 4.0695, | |
| "step": 729088 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.805443069571262e-05, | |
| "loss": 4.0435, | |
| "step": 729600 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.80460447482021e-05, | |
| "loss": 4.052, | |
| "step": 730112 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.803765880069158e-05, | |
| "loss": 4.0514, | |
| "step": 730624 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.802927285318106e-05, | |
| "loss": 4.065, | |
| "step": 731136 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.802088690567054e-05, | |
| "loss": 4.048, | |
| "step": 731648 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.801251733696375e-05, | |
| "loss": 4.0623, | |
| "step": 732160 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.800413138945323e-05, | |
| "loss": 4.0538, | |
| "step": 732672 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.799574544194271e-05, | |
| "loss": 4.0519, | |
| "step": 733184 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.798735949443219e-05, | |
| "loss": 4.051, | |
| "step": 733696 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.797897354692167e-05, | |
| "loss": 4.0368, | |
| "step": 734208 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.797058759941115e-05, | |
| "loss": 4.0484, | |
| "step": 734720 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.796220165190063e-05, | |
| "loss": 4.057, | |
| "step": 735232 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.795381570439011e-05, | |
| "loss": 4.064, | |
| "step": 735744 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.794542975687959e-05, | |
| "loss": 4.0421, | |
| "step": 736256 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.793704380936907e-05, | |
| "loss": 4.0461, | |
| "step": 736768 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.792865786185855e-05, | |
| "loss": 4.0492, | |
| "step": 737280 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7920271914348035e-05, | |
| "loss": 4.0541, | |
| "step": 737792 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7911902345641244e-05, | |
| "loss": 4.0593, | |
| "step": 738304 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7903516398130724e-05, | |
| "loss": 4.0496, | |
| "step": 738816 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7895130450620204e-05, | |
| "loss": 4.0545, | |
| "step": 739328 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7886744503109684e-05, | |
| "loss": 4.0514, | |
| "step": 739840 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7878358555599164e-05, | |
| "loss": 4.0499, | |
| "step": 740352 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.786997260808864e-05, | |
| "loss": 4.0393, | |
| "step": 740864 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.786160303938185e-05, | |
| "loss": 4.0443, | |
| "step": 741376 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.785323347067506e-05, | |
| "loss": 4.0554, | |
| "step": 741888 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.784484752316454e-05, | |
| "loss": 4.0452, | |
| "step": 742400 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.783646157565402e-05, | |
| "loss": 4.0505, | |
| "step": 742912 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.78280756281435e-05, | |
| "loss": 4.0399, | |
| "step": 743424 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.781968968063299e-05, | |
| "loss": 4.0501, | |
| "step": 743936 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.78113201119262e-05, | |
| "loss": 4.048, | |
| "step": 744448 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.780293416441568e-05, | |
| "loss": 4.0503, | |
| "step": 744960 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.779454821690516e-05, | |
| "loss": 4.0526, | |
| "step": 745472 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.778616226939464e-05, | |
| "loss": 4.0639, | |
| "step": 745984 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.777777632188411e-05, | |
| "loss": 4.0505, | |
| "step": 746496 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.776940675317733e-05, | |
| "loss": 4.0543, | |
| "step": 747008 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.776102080566681e-05, | |
| "loss": 4.0525, | |
| "step": 747520 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.775263485815629e-05, | |
| "loss": 4.044, | |
| "step": 748032 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.774424891064576e-05, | |
| "loss": 4.055, | |
| "step": 748544 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.773586296313524e-05, | |
| "loss": 4.0454, | |
| "step": 749056 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.772747701562473e-05, | |
| "loss": 4.0437, | |
| "step": 749568 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.771909106811421e-05, | |
| "loss": 4.0516, | |
| "step": 750080 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.771070512060369e-05, | |
| "loss": 4.0474, | |
| "step": 750592 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.770235193070063e-05, | |
| "loss": 4.0523, | |
| "step": 751104 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.769396598319011e-05, | |
| "loss": 4.0454, | |
| "step": 751616 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.768559641448332e-05, | |
| "loss": 4.0567, | |
| "step": 752128 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.76772104669728e-05, | |
| "loss": 4.0445, | |
| "step": 752640 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.766882451946228e-05, | |
| "loss": 4.0583, | |
| "step": 753152 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.766043857195176e-05, | |
| "loss": 4.0433, | |
| "step": 753664 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7652052624441234e-05, | |
| "loss": 4.0461, | |
| "step": 754176 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7643666676930714e-05, | |
| "loss": 4.023, | |
| "step": 754688 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7635280729420194e-05, | |
| "loss": 4.0579, | |
| "step": 755200 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.762689478190968e-05, | |
| "loss": 4.0509, | |
| "step": 755712 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.761850883439916e-05, | |
| "loss": 4.06, | |
| "step": 756224 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.761012288688864e-05, | |
| "loss": 4.0448, | |
| "step": 756736 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.760173693937812e-05, | |
| "loss": 4.057, | |
| "step": 757248 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.75933509918676e-05, | |
| "loss": 4.0471, | |
| "step": 757760 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.758498142316081e-05, | |
| "loss": 4.0642, | |
| "step": 758272 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.757659547565029e-05, | |
| "loss": 4.0389, | |
| "step": 758784 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.756820952813977e-05, | |
| "loss": 4.046, | |
| "step": 759296 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.755982358062925e-05, | |
| "loss": 4.0442, | |
| "step": 759808 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.755143763311873e-05, | |
| "loss": 4.0547, | |
| "step": 760320 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.754306806441194e-05, | |
| "loss": 4.0442, | |
| "step": 760832 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.753468211690142e-05, | |
| "loss": 4.0437, | |
| "step": 761344 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7526312548194634e-05, | |
| "loss": 4.0481, | |
| "step": 761856 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7517926600684114e-05, | |
| "loss": 4.0502, | |
| "step": 762368 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7509540653173594e-05, | |
| "loss": 4.0511, | |
| "step": 762880 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "eval_loss": 4.087827205657959, | |
| "eval_runtime": 300.3169, | |
| "eval_samples_per_second": 1270.628, | |
| "eval_steps_per_second": 39.708, | |
| "step": 763200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7501154705663074e-05, | |
| "loss": 4.0383, | |
| "step": 763392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7492768758152554e-05, | |
| "loss": 4.0522, | |
| "step": 763904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7484382810642034e-05, | |
| "loss": 4.0478, | |
| "step": 764416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7475996863131514e-05, | |
| "loss": 4.0462, | |
| "step": 764928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7467610915620994e-05, | |
| "loss": 4.0573, | |
| "step": 765440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7459224968110474e-05, | |
| "loss": 4.0474, | |
| "step": 765952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.745083902059995e-05, | |
| "loss": 4.0459, | |
| "step": 766464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.744245307308943e-05, | |
| "loss": 4.0352, | |
| "step": 766976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.743406712557891e-05, | |
| "loss": 4.0428, | |
| "step": 767488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.742568117806839e-05, | |
| "loss": 4.0521, | |
| "step": 768000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.741729523055787e-05, | |
| "loss": 4.0503, | |
| "step": 768512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.740890928304735e-05, | |
| "loss": 4.0481, | |
| "step": 769024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.740053971434056e-05, | |
| "loss": 4.0514, | |
| "step": 769536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.739215376683004e-05, | |
| "loss": 4.0392, | |
| "step": 770048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.738376781931952e-05, | |
| "loss": 4.0397, | |
| "step": 770560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7375381871809e-05, | |
| "loss": 4.0316, | |
| "step": 771072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.736699592429848e-05, | |
| "loss": 4.0501, | |
| "step": 771584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.735860997678796e-05, | |
| "loss": 4.0418, | |
| "step": 772096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.735022402927744e-05, | |
| "loss": 4.0382, | |
| "step": 772608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.734185446057065e-05, | |
| "loss": 4.05, | |
| "step": 773120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.733346851306013e-05, | |
| "loss": 4.0584, | |
| "step": 773632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.732508256554961e-05, | |
| "loss": 4.0451, | |
| "step": 774144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.731669661803909e-05, | |
| "loss": 4.0425, | |
| "step": 774656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.730831067052857e-05, | |
| "loss": 4.0455, | |
| "step": 775168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.729992472301805e-05, | |
| "loss": 4.0425, | |
| "step": 775680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.729153877550753e-05, | |
| "loss": 4.0422, | |
| "step": 776192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.728315282799701e-05, | |
| "loss": 4.0254, | |
| "step": 776704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.727476688048649e-05, | |
| "loss": 4.045, | |
| "step": 777216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.726638093297597e-05, | |
| "loss": 4.0475, | |
| "step": 777728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.725799498546545e-05, | |
| "loss": 4.0302, | |
| "step": 778240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.724960903795493e-05, | |
| "loss": 4.0385, | |
| "step": 778752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.724123946924814e-05, | |
| "loss": 4.0462, | |
| "step": 779264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.723285352173762e-05, | |
| "loss": 4.0437, | |
| "step": 779776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.72244675742271e-05, | |
| "loss": 4.0523, | |
| "step": 780288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.721608162671658e-05, | |
| "loss": 4.0433, | |
| "step": 780800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.720771205800979e-05, | |
| "loss": 4.0365, | |
| "step": 781312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.719932611049927e-05, | |
| "loss": 4.0567, | |
| "step": 781824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.719094016298875e-05, | |
| "loss": 4.0236, | |
| "step": 782336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7182554215478236e-05, | |
| "loss": 4.0313, | |
| "step": 782848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7174168267967716e-05, | |
| "loss": 4.0366, | |
| "step": 783360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7165782320457196e-05, | |
| "loss": 4.0339, | |
| "step": 783872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7157396372946676e-05, | |
| "loss": 4.0386, | |
| "step": 784384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7149010425436155e-05, | |
| "loss": 4.0363, | |
| "step": 784896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7140640856729365e-05, | |
| "loss": 4.0416, | |
| "step": 785408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7132254909218845e-05, | |
| "loss": 4.0377, | |
| "step": 785920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7123868961708324e-05, | |
| "loss": 4.0349, | |
| "step": 786432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7115483014197804e-05, | |
| "loss": 4.0463, | |
| "step": 786944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7107097066687284e-05, | |
| "loss": 4.042, | |
| "step": 787456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7098727497980494e-05, | |
| "loss": 4.0185, | |
| "step": 787968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7090341550469973e-05, | |
| "loss": 4.0408, | |
| "step": 788480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7081955602959453e-05, | |
| "loss": 4.0436, | |
| "step": 788992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.707356965544893e-05, | |
| "loss": 4.0514, | |
| "step": 789504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.706518370793842e-05, | |
| "loss": 4.0357, | |
| "step": 790016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.705681413923163e-05, | |
| "loss": 4.0367, | |
| "step": 790528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.704842819172111e-05, | |
| "loss": 4.0285, | |
| "step": 791040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.704004224421059e-05, | |
| "loss": 4.0322, | |
| "step": 791552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.703165629670007e-05, | |
| "loss": 4.0414, | |
| "step": 792064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.702327034918955e-05, | |
| "loss": 4.0242, | |
| "step": 792576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.701488440167902e-05, | |
| "loss": 4.0391, | |
| "step": 793088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.700651483297224e-05, | |
| "loss": 4.0391, | |
| "step": 793600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.699812888546172e-05, | |
| "loss": 4.032, | |
| "step": 794112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.69897429379512e-05, | |
| "loss": 4.0376, | |
| "step": 794624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.698135699044067e-05, | |
| "loss": 4.0254, | |
| "step": 795136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.697297104293016e-05, | |
| "loss": 4.0298, | |
| "step": 795648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.696458509541964e-05, | |
| "loss": 4.0331, | |
| "step": 796160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.695619914790912e-05, | |
| "loss": 4.0434, | |
| "step": 796672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.694782957920233e-05, | |
| "loss": 4.0259, | |
| "step": 797184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.693944363169181e-05, | |
| "loss": 4.0386, | |
| "step": 797696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.693105768418129e-05, | |
| "loss": 4.0148, | |
| "step": 798208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.692267173667077e-05, | |
| "loss": 4.0356, | |
| "step": 798720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.691428578916025e-05, | |
| "loss": 4.0358, | |
| "step": 799232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.690589984164973e-05, | |
| "loss": 4.0415, | |
| "step": 799744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.689751389413921e-05, | |
| "loss": 4.0333, | |
| "step": 800256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6889127946628687e-05, | |
| "loss": 4.0379, | |
| "step": 800768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6880741999118167e-05, | |
| "loss": 4.0546, | |
| "step": 801280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6872372430411376e-05, | |
| "loss": 4.0282, | |
| "step": 801792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6863986482900856e-05, | |
| "loss": 4.044, | |
| "step": 802304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6855600535390336e-05, | |
| "loss": 4.0285, | |
| "step": 802816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.684723096668355e-05, | |
| "loss": 4.0347, | |
| "step": 803328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.683884501917303e-05, | |
| "loss": 4.0354, | |
| "step": 803840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.683045907166251e-05, | |
| "loss": 4.0289, | |
| "step": 804352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.682207312415199e-05, | |
| "loss": 4.042, | |
| "step": 804864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.681368717664147e-05, | |
| "loss": 4.0481, | |
| "step": 805376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.680530122913095e-05, | |
| "loss": 4.0304, | |
| "step": 805888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.679691528162043e-05, | |
| "loss": 4.0263, | |
| "step": 806400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.678852933410991e-05, | |
| "loss": 4.0296, | |
| "step": 806912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.678015976540312e-05, | |
| "loss": 4.0403, | |
| "step": 807424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.67717738178926e-05, | |
| "loss": 4.0288, | |
| "step": 807936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6763420627989545e-05, | |
| "loss": 4.0421, | |
| "step": 808448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6755034680479025e-05, | |
| "loss": 4.0338, | |
| "step": 808960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6746648732968505e-05, | |
| "loss": 4.0336, | |
| "step": 809472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6738262785457985e-05, | |
| "loss": 4.0286, | |
| "step": 809984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6729876837947465e-05, | |
| "loss": 4.0183, | |
| "step": 810496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6721490890436945e-05, | |
| "loss": 4.0255, | |
| "step": 811008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6713104942926425e-05, | |
| "loss": 4.0362, | |
| "step": 811520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6704718995415905e-05, | |
| "loss": 4.0389, | |
| "step": 812032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6696333047905385e-05, | |
| "loss": 4.024, | |
| "step": 812544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.668794710039486e-05, | |
| "loss": 4.0238, | |
| "step": 813056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.667956115288434e-05, | |
| "loss": 4.0296, | |
| "step": 813568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.667117520537382e-05, | |
| "loss": 4.0356, | |
| "step": 814080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6662805636667034e-05, | |
| "loss": 4.042, | |
| "step": 814592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6654419689156514e-05, | |
| "loss": 4.0263, | |
| "step": 815104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6646033741645994e-05, | |
| "loss": 4.0348, | |
| "step": 815616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6637647794135474e-05, | |
| "loss": 4.0282, | |
| "step": 816128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6629261846624954e-05, | |
| "loss": 4.0286, | |
| "step": 816640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.662089227791816e-05, | |
| "loss": 4.0227, | |
| "step": 817152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.661250633040764e-05, | |
| "loss": 4.0268, | |
| "step": 817664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.660412038289712e-05, | |
| "loss": 4.0319, | |
| "step": 818176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.65957344353866e-05, | |
| "loss": 4.0238, | |
| "step": 818688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.658734848787608e-05, | |
| "loss": 4.0285, | |
| "step": 819200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.657896254036556e-05, | |
| "loss": 4.022, | |
| "step": 819712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.657059297165877e-05, | |
| "loss": 4.0279, | |
| "step": 820224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.656220702414825e-05, | |
| "loss": 4.0301, | |
| "step": 820736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.655382107663773e-05, | |
| "loss": 4.0274, | |
| "step": 821248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.654545150793095e-05, | |
| "loss": 4.0306, | |
| "step": 821760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.653706556042043e-05, | |
| "loss": 4.0442, | |
| "step": 822272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.652867961290991e-05, | |
| "loss": 4.0335, | |
| "step": 822784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.652029366539939e-05, | |
| "loss": 4.0337, | |
| "step": 823296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.651190771788887e-05, | |
| "loss": 4.0307, | |
| "step": 823808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.650352177037835e-05, | |
| "loss": 4.0247, | |
| "step": 824320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.649513582286783e-05, | |
| "loss": 4.0388, | |
| "step": 824832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.648674987535731e-05, | |
| "loss": 4.0227, | |
| "step": 825344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.647836392784679e-05, | |
| "loss": 4.0233, | |
| "step": 825856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6469994359139996e-05, | |
| "loss": 4.0298, | |
| "step": 826368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6461608411629476e-05, | |
| "loss": 4.0297, | |
| "step": 826880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6453222464118956e-05, | |
| "loss": 4.0298, | |
| "step": 827392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6444836516608436e-05, | |
| "loss": 4.0293, | |
| "step": 827904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.643646694790165e-05, | |
| "loss": 4.034, | |
| "step": 828416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.642809737919486e-05, | |
| "loss": 4.0251, | |
| "step": 828928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.641971143168434e-05, | |
| "loss": 4.0361, | |
| "step": 829440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.641132548417382e-05, | |
| "loss": 4.0275, | |
| "step": 829952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.64029395366633e-05, | |
| "loss": 4.0227, | |
| "step": 830464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.639455358915278e-05, | |
| "loss": 4.0041, | |
| "step": 830976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.638616764164226e-05, | |
| "loss": 4.0392, | |
| "step": 831488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.637778169413174e-05, | |
| "loss": 4.0306, | |
| "step": 832000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.636941212542495e-05, | |
| "loss": 4.0419, | |
| "step": 832512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.636102617791443e-05, | |
| "loss": 4.0265, | |
| "step": 833024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.635264023040391e-05, | |
| "loss": 4.0324, | |
| "step": 833536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.634425428289339e-05, | |
| "loss": 4.028, | |
| "step": 834048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.633586833538287e-05, | |
| "loss": 4.0455, | |
| "step": 834560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.632748238787235e-05, | |
| "loss": 4.0158, | |
| "step": 835072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.631909644036183e-05, | |
| "loss": 4.0281, | |
| "step": 835584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.631071049285131e-05, | |
| "loss": 4.0283, | |
| "step": 836096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.630232454534079e-05, | |
| "loss": 4.0319, | |
| "step": 836608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6293954976634e-05, | |
| "loss": 4.0257, | |
| "step": 837120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.628556902912348e-05, | |
| "loss": 4.026, | |
| "step": 837632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.627718308161296e-05, | |
| "loss": 4.0266, | |
| "step": 838144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.626881351290617e-05, | |
| "loss": 4.0346, | |
| "step": 838656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.626042756539565e-05, | |
| "loss": 4.0272, | |
| "step": 839168 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.075066089630127, | |
| "eval_runtime": 292.4706, | |
| "eval_samples_per_second": 1304.716, | |
| "eval_steps_per_second": 40.773, | |
| "step": 839520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.625204161788513e-05, | |
| "loss": 4.021, | |
| "step": 839680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6243672049178337e-05, | |
| "loss": 4.0297, | |
| "step": 840192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.623528610166782e-05, | |
| "loss": 4.0255, | |
| "step": 840704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.62269001541573e-05, | |
| "loss": 4.0276, | |
| "step": 841216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.621851420664678e-05, | |
| "loss": 4.0336, | |
| "step": 841728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.621012825913626e-05, | |
| "loss": 4.0261, | |
| "step": 842240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.620174231162574e-05, | |
| "loss": 4.0283, | |
| "step": 842752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.619335636411522e-05, | |
| "loss": 4.0174, | |
| "step": 843264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.61849704166047e-05, | |
| "loss": 4.0219, | |
| "step": 843776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.617658446909418e-05, | |
| "loss": 4.0316, | |
| "step": 844288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.616819852158366e-05, | |
| "loss": 4.034, | |
| "step": 844800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.615982895287687e-05, | |
| "loss": 4.026, | |
| "step": 845312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.615145938417008e-05, | |
| "loss": 4.039, | |
| "step": 845824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.614307343665956e-05, | |
| "loss": 4.0149, | |
| "step": 846336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.613468748914904e-05, | |
| "loss": 4.0209, | |
| "step": 846848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.612630154163852e-05, | |
| "loss": 4.014, | |
| "step": 847360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.611791559412801e-05, | |
| "loss": 4.0311, | |
| "step": 847872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.610952964661749e-05, | |
| "loss": 4.0205, | |
| "step": 848384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.610114369910697e-05, | |
| "loss": 4.0218, | |
| "step": 848896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.609275775159645e-05, | |
| "loss": 4.0281, | |
| "step": 849408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.608437180408593e-05, | |
| "loss": 4.0398, | |
| "step": 849920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.607598585657541e-05, | |
| "loss": 4.028, | |
| "step": 850432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.606759990906488e-05, | |
| "loss": 4.0219, | |
| "step": 850944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.605921396155436e-05, | |
| "loss": 4.0264, | |
| "step": 851456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.605082801404384e-05, | |
| "loss": 4.0261, | |
| "step": 851968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6042458445337056e-05, | |
| "loss": 4.0231, | |
| "step": 852480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.603407249782653e-05, | |
| "loss": 4.0091, | |
| "step": 852992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.602568655031601e-05, | |
| "loss": 4.0236, | |
| "step": 853504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.601730060280549e-05, | |
| "loss": 4.028, | |
| "step": 854016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6008914655294976e-05, | |
| "loss": 4.0094, | |
| "step": 854528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.6000528707784456e-05, | |
| "loss": 4.0211, | |
| "step": 855040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5992142760273936e-05, | |
| "loss": 4.028, | |
| "step": 855552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5983773191567145e-05, | |
| "loss": 4.0239, | |
| "step": 856064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5975387244056625e-05, | |
| "loss": 4.0333, | |
| "step": 856576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5967001296546105e-05, | |
| "loss": 4.0237, | |
| "step": 857088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5958615349035585e-05, | |
| "loss": 4.0198, | |
| "step": 857600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5950229401525065e-05, | |
| "loss": 4.0415, | |
| "step": 858112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5941843454014545e-05, | |
| "loss": 4.0017, | |
| "step": 858624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5933457506504025e-05, | |
| "loss": 4.0134, | |
| "step": 859136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5925071558993505e-05, | |
| "loss": 4.0216, | |
| "step": 859648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5916685611482985e-05, | |
| "loss": 4.0123, | |
| "step": 860160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5908299663972465e-05, | |
| "loss": 4.0203, | |
| "step": 860672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5899930095265674e-05, | |
| "loss": 4.022, | |
| "step": 861184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5891544147755154e-05, | |
| "loss": 4.019, | |
| "step": 861696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.588315820024464e-05, | |
| "loss": 4.0217, | |
| "step": 862208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.587478863153785e-05, | |
| "loss": 4.0143, | |
| "step": 862720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.586640268402733e-05, | |
| "loss": 4.023, | |
| "step": 863232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.585801673651681e-05, | |
| "loss": 4.0271, | |
| "step": 863744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.584963078900629e-05, | |
| "loss": 4.0026, | |
| "step": 864256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.584124484149577e-05, | |
| "loss": 4.0209, | |
| "step": 864768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.583285889398524e-05, | |
| "loss": 4.0257, | |
| "step": 865280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.582447294647472e-05, | |
| "loss": 4.031, | |
| "step": 865792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.58160869989642e-05, | |
| "loss": 4.0164, | |
| "step": 866304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.580771743025742e-05, | |
| "loss": 4.0149, | |
| "step": 866816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.579934786155063e-05, | |
| "loss": 4.016, | |
| "step": 867328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.579096191404011e-05, | |
| "loss": 4.0115, | |
| "step": 867840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5782575966529594e-05, | |
| "loss": 4.023, | |
| "step": 868352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.577419001901907e-05, | |
| "loss": 4.0047, | |
| "step": 868864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.576580407150855e-05, | |
| "loss": 4.0231, | |
| "step": 869376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.575743450280176e-05, | |
| "loss": 4.0172, | |
| "step": 869888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.574904855529124e-05, | |
| "loss": 4.0151, | |
| "step": 870400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5740662607780716e-05, | |
| "loss": 4.0146, | |
| "step": 870912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5732276660270196e-05, | |
| "loss": 4.0145, | |
| "step": 871424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5723890712759676e-05, | |
| "loss": 4.0121, | |
| "step": 871936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5715504765249156e-05, | |
| "loss": 4.0068, | |
| "step": 872448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5707118817738636e-05, | |
| "loss": 4.0291, | |
| "step": 872960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5698749249031845e-05, | |
| "loss": 4.0115, | |
| "step": 873472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.569036330152133e-05, | |
| "loss": 4.0102, | |
| "step": 873984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.568199373281454e-05, | |
| "loss": 4.0006, | |
| "step": 874496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.567360778530402e-05, | |
| "loss": 4.0164, | |
| "step": 875008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.56652218377935e-05, | |
| "loss": 4.018, | |
| "step": 875520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.565685226908672e-05, | |
| "loss": 4.0239, | |
| "step": 876032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.564846632157619e-05, | |
| "loss": 4.0177, | |
| "step": 876544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.564008037406567e-05, | |
| "loss": 4.0207, | |
| "step": 877056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.563169442655515e-05, | |
| "loss": 4.0323, | |
| "step": 877568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.562330847904463e-05, | |
| "loss": 4.0125, | |
| "step": 878080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.561493891033784e-05, | |
| "loss": 4.0277, | |
| "step": 878592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.560655296282732e-05, | |
| "loss": 4.0126, | |
| "step": 879104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.55981670153168e-05, | |
| "loss": 4.0146, | |
| "step": 879616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5589781067806286e-05, | |
| "loss": 4.0169, | |
| "step": 880128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5581395120295766e-05, | |
| "loss": 4.0115, | |
| "step": 880640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5573009172785246e-05, | |
| "loss": 4.0241, | |
| "step": 881152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5564623225274726e-05, | |
| "loss": 4.0311, | |
| "step": 881664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5556237277764206e-05, | |
| "loss": 4.0177, | |
| "step": 882176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5547851330253686e-05, | |
| "loss": 4.0025, | |
| "step": 882688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5539465382743166e-05, | |
| "loss": 4.0133, | |
| "step": 883200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5531079435232645e-05, | |
| "loss": 4.0224, | |
| "step": 883712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5522709866525855e-05, | |
| "loss": 4.0101, | |
| "step": 884224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5514340297819064e-05, | |
| "loss": 4.0213, | |
| "step": 884736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5505954350308544e-05, | |
| "loss": 4.0166, | |
| "step": 885248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5497568402798024e-05, | |
| "loss": 4.0175, | |
| "step": 885760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5489182455287504e-05, | |
| "loss": 4.0117, | |
| "step": 886272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5480796507776983e-05, | |
| "loss": 3.9997, | |
| "step": 886784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.547241056026647e-05, | |
| "loss": 4.009, | |
| "step": 887296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.546402461275595e-05, | |
| "loss": 4.0131, | |
| "step": 887808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.545563866524543e-05, | |
| "loss": 4.0288, | |
| "step": 888320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.54472527177349e-05, | |
| "loss": 4.0034, | |
| "step": 888832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.543886677022438e-05, | |
| "loss": 4.0093, | |
| "step": 889344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.543048082271386e-05, | |
| "loss": 4.0133, | |
| "step": 889856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.542211125400708e-05, | |
| "loss": 4.0137, | |
| "step": 890368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.541372530649655e-05, | |
| "loss": 4.023, | |
| "step": 890880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.540533935898603e-05, | |
| "loss": 4.0107, | |
| "step": 891392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.539695341147551e-05, | |
| "loss": 4.0197, | |
| "step": 891904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.538856746396499e-05, | |
| "loss": 4.0105, | |
| "step": 892416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.538018151645447e-05, | |
| "loss": 4.0102, | |
| "step": 892928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.537181194774769e-05, | |
| "loss": 4.0063, | |
| "step": 893440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.536342600023717e-05, | |
| "loss": 4.0092, | |
| "step": 893952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.535504005272665e-05, | |
| "loss": 4.0159, | |
| "step": 894464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.534665410521613e-05, | |
| "loss": 4.0076, | |
| "step": 894976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.533826815770561e-05, | |
| "loss": 4.0093, | |
| "step": 895488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.532989858899882e-05, | |
| "loss": 4.0028, | |
| "step": 896000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.53215126414883e-05, | |
| "loss": 4.008, | |
| "step": 896512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.531312669397778e-05, | |
| "loss": 4.0141, | |
| "step": 897024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.530474074646726e-05, | |
| "loss": 4.0134, | |
| "step": 897536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5296371177760466e-05, | |
| "loss": 4.0125, | |
| "step": 898048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5287985230249946e-05, | |
| "loss": 4.0177, | |
| "step": 898560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5279599282739426e-05, | |
| "loss": 4.0183, | |
| "step": 899072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5271213335228906e-05, | |
| "loss": 4.0191, | |
| "step": 899584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5262827387718386e-05, | |
| "loss": 4.0164, | |
| "step": 900096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.525444144020787e-05, | |
| "loss": 4.0088, | |
| "step": 900608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.524605549269735e-05, | |
| "loss": 4.0162, | |
| "step": 901120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.523766954518683e-05, | |
| "loss": 4.0055, | |
| "step": 901632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.522928359767631e-05, | |
| "loss": 4.0074, | |
| "step": 902144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.522089765016579e-05, | |
| "loss": 4.0133, | |
| "step": 902656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5212511702655265e-05, | |
| "loss": 4.0127, | |
| "step": 903168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5204125755144745e-05, | |
| "loss": 4.0109, | |
| "step": 903680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.519575618643796e-05, | |
| "loss": 4.01, | |
| "step": 904192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.518738661773117e-05, | |
| "loss": 4.0175, | |
| "step": 904704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.517900067022065e-05, | |
| "loss": 4.0086, | |
| "step": 905216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.517061472271013e-05, | |
| "loss": 4.019, | |
| "step": 905728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.516222877519961e-05, | |
| "loss": 4.01, | |
| "step": 906240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.515384282768909e-05, | |
| "loss": 4.0087, | |
| "step": 906752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5145473258982306e-05, | |
| "loss": 3.9845, | |
| "step": 907264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5137087311471786e-05, | |
| "loss": 4.0201, | |
| "step": 907776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5128701363961266e-05, | |
| "loss": 4.0122, | |
| "step": 908288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.512031541645074e-05, | |
| "loss": 4.0251, | |
| "step": 908800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.511192946894022e-05, | |
| "loss": 4.0127, | |
| "step": 909312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.51035435214297e-05, | |
| "loss": 4.013, | |
| "step": 909824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.509515757391918e-05, | |
| "loss": 4.0152, | |
| "step": 910336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.508677162640866e-05, | |
| "loss": 4.0257, | |
| "step": 910848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.507838567889814e-05, | |
| "loss": 3.9989, | |
| "step": 911360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.507001611019135e-05, | |
| "loss": 4.0118, | |
| "step": 911872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.506163016268083e-05, | |
| "loss": 4.0078, | |
| "step": 912384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.505324421517031e-05, | |
| "loss": 4.015, | |
| "step": 912896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5044874646463524e-05, | |
| "loss": 4.0098, | |
| "step": 913408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5036488698953004e-05, | |
| "loss": 4.01, | |
| "step": 913920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5028102751442484e-05, | |
| "loss": 4.0078, | |
| "step": 914432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5019716803931964e-05, | |
| "loss": 4.0161, | |
| "step": 914944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5011330856421444e-05, | |
| "loss": 4.009, | |
| "step": 915456 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.0643205642700195, | |
| "eval_runtime": 294.1127, | |
| "eval_samples_per_second": 1297.431, | |
| "eval_steps_per_second": 40.546, | |
| "step": 915840 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 3.631465644861328e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |