| { | |
| "best_metric": 4.200494289398193, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/passive/lstm/2/checkpoints/checkpoint-381600", | |
| "epoch": 1.0250006060157382, | |
| "eval_steps": 10, | |
| "global_step": 381600, | |
| "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": 10.8206, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 7.5517, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 7.0492, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 6.9882, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 6.943, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 6.8668, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 6.7139, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994131474623009e-05, | |
| "loss": 6.615, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993292879871958e-05, | |
| "loss": 6.5194, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992455923001279e-05, | |
| "loss": 6.4393, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.991617328250227e-05, | |
| "loss": 6.3772, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990778733499175e-05, | |
| "loss": 6.3056, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989940138748123e-05, | |
| "loss": 6.2402, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989101543997071e-05, | |
| "loss": 6.1777, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988262949246018e-05, | |
| "loss": 6.1283, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987424354494966e-05, | |
| "loss": 6.0656, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986585759743914e-05, | |
| "loss": 6.0332, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985747164992862e-05, | |
| "loss": 5.9864, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.98490857024181e-05, | |
| "loss": 5.9467, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984069975490758e-05, | |
| "loss": 5.9056, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983231380739706e-05, | |
| "loss": 5.8737, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9823927859886547e-05, | |
| "loss": 5.8322, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9815558291179756e-05, | |
| "loss": 5.8041, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9807172343669236e-05, | |
| "loss": 5.7614, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9798786396158716e-05, | |
| "loss": 5.7496, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9790400448648195e-05, | |
| "loss": 5.7106, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9782014501137675e-05, | |
| "loss": 5.6924, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773628553627155e-05, | |
| "loss": 5.6523, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9765242606116635e-05, | |
| "loss": 5.6354, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9756856658606115e-05, | |
| "loss": 5.6148, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748470711095595e-05, | |
| "loss": 5.5896, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.974008476358507e-05, | |
| "loss": 5.58, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.973169881607455e-05, | |
| "loss": 5.5538, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9723329247367764e-05, | |
| "loss": 5.5377, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714943299857244e-05, | |
| "loss": 5.5294, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9706557352346724e-05, | |
| "loss": 5.506, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9698171404836204e-05, | |
| "loss": 5.4977, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9689785457325684e-05, | |
| "loss": 5.4519, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9681399509815164e-05, | |
| "loss": 5.4578, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9673013562304644e-05, | |
| "loss": 5.4188, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9664627614794124e-05, | |
| "loss": 5.4256, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965625804608733e-05, | |
| "loss": 5.4043, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964787209857681e-05, | |
| "loss": 5.4025, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963948615106629e-05, | |
| "loss": 5.3707, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963110020355577e-05, | |
| "loss": 5.3789, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962271425604525e-05, | |
| "loss": 5.3669, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.961434468733847e-05, | |
| "loss": 5.3595, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.960595873982795e-05, | |
| "loss": 5.3521, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.959757279231743e-05, | |
| "loss": 5.3231, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958918684480691e-05, | |
| "loss": 5.3145, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958080089729639e-05, | |
| "loss": 5.3136, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.957241494978587e-05, | |
| "loss": 5.3003, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956402900227535e-05, | |
| "loss": 5.2938, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955564305476483e-05, | |
| "loss": 5.271, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.954728986486177e-05, | |
| "loss": 5.2684, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953890391735125e-05, | |
| "loss": 5.2476, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9530517969840727e-05, | |
| "loss": 5.2694, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9522132022330207e-05, | |
| "loss": 5.2219, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9513746074819686e-05, | |
| "loss": 5.2387, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95053765061129e-05, | |
| "loss": 5.228, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949699055860238e-05, | |
| "loss": 5.2077, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948860461109186e-05, | |
| "loss": 5.2068, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948021866358134e-05, | |
| "loss": 5.1939, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947184909487455e-05, | |
| "loss": 5.1821, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946346314736403e-05, | |
| "loss": 5.1751, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945507719985351e-05, | |
| "loss": 5.189, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944669125234299e-05, | |
| "loss": 5.1628, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.943830530483247e-05, | |
| "loss": 5.1535, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942993573612568e-05, | |
| "loss": 5.144, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942154978861516e-05, | |
| "loss": 5.1415, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.941316384110464e-05, | |
| "loss": 5.1472, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.940477789359412e-05, | |
| "loss": 5.1426, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.93963919460836e-05, | |
| "loss": 5.1174, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388022377376816e-05, | |
| "loss": 5.1341, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379636429866296e-05, | |
| "loss": 5.1269, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371250482355776e-05, | |
| "loss": 5.0984, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362864534845256e-05, | |
| "loss": 5.0971, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.935447858733473e-05, | |
| "loss": 5.0869, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9346109018627945e-05, | |
| "loss": 5.0785, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9337723071117425e-05, | |
| "loss": 5.0734, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329337123606905e-05, | |
| "loss": 5.0779, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932095117609638e-05, | |
| "loss": 5.0732, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9312581607389594e-05, | |
| "loss": 5.0751, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9304195659879074e-05, | |
| "loss": 5.0513, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.929582609117229e-05, | |
| "loss": 5.0322, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.928744014366177e-05, | |
| "loss": 5.0385, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927905419615125e-05, | |
| "loss": 5.0402, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927066824864073e-05, | |
| "loss": 5.039, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.92622823011302e-05, | |
| "loss": 5.0211, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.925389635361968e-05, | |
| "loss": 5.0188, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.924551040610916e-05, | |
| "loss": 5.0187, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923712445859864e-05, | |
| "loss": 5.0121, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922875488989185e-05, | |
| "loss": 5.0, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922036894238133e-05, | |
| "loss": 4.9985, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921198299487081e-05, | |
| "loss": 4.9846, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920361342616403e-05, | |
| "loss": 4.9976, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919524385745724e-05, | |
| "loss": 4.9831, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918685790994672e-05, | |
| "loss": 4.9624, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.91784719624362e-05, | |
| "loss": 4.9672, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9170086014925676e-05, | |
| "loss": 4.9667, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9161700067415156e-05, | |
| "loss": 4.9722, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9153314119904636e-05, | |
| "loss": 4.9544, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9144928172394116e-05, | |
| "loss": 4.9512, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9136542224883596e-05, | |
| "loss": 4.937, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9128172656176805e-05, | |
| "loss": 4.9429, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9119786708666285e-05, | |
| "loss": 4.9218, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9111400761155765e-05, | |
| "loss": 4.9317, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9103014813645245e-05, | |
| "loss": 4.9198, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.909462886613473e-05, | |
| "loss": 4.9228, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.908624291862421e-05, | |
| "loss": 4.911, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907785697111369e-05, | |
| "loss": 4.9157, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906947102360317e-05, | |
| "loss": 4.9017, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906110145489638e-05, | |
| "loss": 4.9064, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905271550738586e-05, | |
| "loss": 4.8975, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904432955987534e-05, | |
| "loss": 4.8909, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903594361236482e-05, | |
| "loss": 4.9032, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90275576648543e-05, | |
| "loss": 4.8873, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901917171734378e-05, | |
| "loss": 4.8853, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901078576983326e-05, | |
| "loss": 4.874, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.900239982232274e-05, | |
| "loss": 4.8793, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899403025361595e-05, | |
| "loss": 4.8742, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.898564430610543e-05, | |
| "loss": 4.8674, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.897725835859491e-05, | |
| "loss": 4.8735, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968872411084396e-05, | |
| "loss": 4.8595, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.896048646357387e-05, | |
| "loss": 4.8598, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.895210051606335e-05, | |
| "loss": 4.8395, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.894371456855283e-05, | |
| "loss": 4.8452, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.893534499984604e-05, | |
| "loss": 4.8457, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.892695905233552e-05, | |
| "loss": 4.838, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918573104825e-05, | |
| "loss": 4.8288, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.891018715731448e-05, | |
| "loss": 4.8508, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890180120980396e-05, | |
| "loss": 4.8343, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889341526229344e-05, | |
| "loss": 4.8247, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888502931478292e-05, | |
| "loss": 4.8251, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8876659746076134e-05, | |
| "loss": 4.8201, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8868273798565614e-05, | |
| "loss": 4.822, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8859887851055094e-05, | |
| "loss": 4.8172, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8851501903544574e-05, | |
| "loss": 4.8133, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8843115956034054e-05, | |
| "loss": 4.8159, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8834730008523534e-05, | |
| "loss": 4.8135, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8826344061013014e-05, | |
| "loss": 4.8021, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8817958113502494e-05, | |
| "loss": 4.7927, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880960492359943e-05, | |
| "loss": 4.8037, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880123535489264e-05, | |
| "loss": 4.7904, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879286578618586e-05, | |
| "loss": 4.793, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.878447983867534e-05, | |
| "loss": 4.7851, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877609389116482e-05, | |
| "loss": 4.7825, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.87677079436543e-05, | |
| "loss": 4.787, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875932199614378e-05, | |
| "loss": 4.7761, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875093604863326e-05, | |
| "loss": 4.7721, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.7402849197387695, | |
| "eval_runtime": 293.2378, | |
| "eval_samples_per_second": 1301.302, | |
| "eval_steps_per_second": 40.667, | |
| "step": 76320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.874255010112274e-05, | |
| "loss": 4.7586, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.873416415361222e-05, | |
| "loss": 4.759, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.87257782061017e-05, | |
| "loss": 4.7823, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.871739225859118e-05, | |
| "loss": 4.7667, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8709006311080657e-05, | |
| "loss": 4.7598, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8700620363570137e-05, | |
| "loss": 4.756, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8692234416059616e-05, | |
| "loss": 4.7551, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8683848468549096e-05, | |
| "loss": 4.7379, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8675462521038576e-05, | |
| "loss": 4.7479, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8667076573528056e-05, | |
| "loss": 4.7477, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865870700482127e-05, | |
| "loss": 4.7495, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865032105731075e-05, | |
| "loss": 4.7478, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.864195148860396e-05, | |
| "loss": 4.7303, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.863356554109344e-05, | |
| "loss": 4.7299, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.862517959358292e-05, | |
| "loss": 4.7187, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.86167936460724e-05, | |
| "loss": 4.7209, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8608407698561874e-05, | |
| "loss": 4.7267, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8600021751051354e-05, | |
| "loss": 4.7171, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8591635803540834e-05, | |
| "loss": 4.7229, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.858326623483405e-05, | |
| "loss": 4.7366, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.857488028732352e-05, | |
| "loss": 4.7178, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.856649433981301e-05, | |
| "loss": 4.7206, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.855810839230249e-05, | |
| "loss": 4.704, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.85497388235957e-05, | |
| "loss": 4.724, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8541369254888915e-05, | |
| "loss": 4.7004, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8532983307378395e-05, | |
| "loss": 4.7044, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8524597359867875e-05, | |
| "loss": 4.6906, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.851621141235735e-05, | |
| "loss": 4.6998, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.850782546484683e-05, | |
| "loss": 4.6908, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8499455896140044e-05, | |
| "loss": 4.6874, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8491069948629524e-05, | |
| "loss": 4.6928, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8482684001119e-05, | |
| "loss": 4.6957, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.847429805360848e-05, | |
| "loss": 4.6893, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8465912106097964e-05, | |
| "loss": 4.6908, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.845754253739117e-05, | |
| "loss": 4.68, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.844915658988065e-05, | |
| "loss": 4.6897, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.844077064237013e-05, | |
| "loss": 4.6626, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.843238469485961e-05, | |
| "loss": 4.6769, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.842399874734909e-05, | |
| "loss": 4.6532, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.841561279983857e-05, | |
| "loss": 4.6755, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.840724323113178e-05, | |
| "loss": 4.6572, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839885728362126e-05, | |
| "loss": 4.6693, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839047133611074e-05, | |
| "loss": 4.6497, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.838210176740395e-05, | |
| "loss": 4.6681, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.837371581989343e-05, | |
| "loss": 4.6645, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.836532987238292e-05, | |
| "loss": 4.6615, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.83569439248724e-05, | |
| "loss": 4.6675, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834855797736188e-05, | |
| "loss": 4.6374, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834017202985136e-05, | |
| "loss": 4.6415, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.833178608234084e-05, | |
| "loss": 4.655, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.832340013483032e-05, | |
| "loss": 4.6475, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.83150141873198e-05, | |
| "loss": 4.645, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.830662823980928e-05, | |
| "loss": 4.6372, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8298258671102486e-05, | |
| "loss": 4.6328, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8289872723591966e-05, | |
| "loss": 4.6271, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8281486776081446e-05, | |
| "loss": 4.6424, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8273100828570926e-05, | |
| "loss": 4.6192, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8264714881060406e-05, | |
| "loss": 4.6311, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8256328933549886e-05, | |
| "loss": 4.635, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8247942986039366e-05, | |
| "loss": 4.6181, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8239557038528846e-05, | |
| "loss": 4.6194, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823120384862579e-05, | |
| "loss": 4.6182, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.822281790111527e-05, | |
| "loss": 4.6053, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.821443195360475e-05, | |
| "loss": 4.6096, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.820604600609423e-05, | |
| "loss": 4.6276, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.819766005858371e-05, | |
| "loss": 4.609, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8189274111073184e-05, | |
| "loss": 4.6016, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8180888163562664e-05, | |
| "loss": 4.6028, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.817251859485588e-05, | |
| "loss": 4.6033, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.816413264734535e-05, | |
| "loss": 4.6169, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.815574669983483e-05, | |
| "loss": 4.6107, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.814736075232432e-05, | |
| "loss": 4.5989, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.81389748048138e-05, | |
| "loss": 4.6144, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.813060523610701e-05, | |
| "loss": 4.6132, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.812221928859649e-05, | |
| "loss": 4.5955, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.811383334108597e-05, | |
| "loss": 4.5952, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.810544739357545e-05, | |
| "loss": 4.5939, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.809706144606493e-05, | |
| "loss": 4.5882, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808869187735814e-05, | |
| "loss": 4.5865, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808030592984762e-05, | |
| "loss": 4.5895, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.80719199823371e-05, | |
| "loss": 4.596, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.806353403482658e-05, | |
| "loss": 4.5971, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.805514808731606e-05, | |
| "loss": 4.5855, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.804676213980554e-05, | |
| "loss": 4.5624, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803837619229502e-05, | |
| "loss": 4.5784, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8029990244784504e-05, | |
| "loss": 4.583, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.802162067607771e-05, | |
| "loss": 4.5864, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.801323472856719e-05, | |
| "loss": 4.5765, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.80048651598604e-05, | |
| "loss": 4.5728, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.799647921234988e-05, | |
| "loss": 4.5787, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.798809326483936e-05, | |
| "loss": 4.5711, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797970731732884e-05, | |
| "loss": 4.5665, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797132136981832e-05, | |
| "loss": 4.5731, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.79629354223078e-05, | |
| "loss": 4.5579, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.795454947479728e-05, | |
| "loss": 4.576, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.794617990609049e-05, | |
| "loss": 4.5685, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.793779395857997e-05, | |
| "loss": 4.5468, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.792940801106946e-05, | |
| "loss": 4.5608, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.792102206355894e-05, | |
| "loss": 4.5584, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.791263611604842e-05, | |
| "loss": 4.5696, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.790426654734163e-05, | |
| "loss": 4.5567, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7895880599831107e-05, | |
| "loss": 4.5517, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7887494652320587e-05, | |
| "loss": 4.5448, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7879108704810067e-05, | |
| "loss": 4.5547, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7870722757299546e-05, | |
| "loss": 4.5367, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7862353188592756e-05, | |
| "loss": 4.5446, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7853967241082236e-05, | |
| "loss": 4.5442, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7845581293571715e-05, | |
| "loss": 4.5443, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7837195346061195e-05, | |
| "loss": 4.5342, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7828809398550675e-05, | |
| "loss": 4.5442, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7820423451040155e-05, | |
| "loss": 4.5361, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7812037503529635e-05, | |
| "loss": 4.5378, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7803651556019115e-05, | |
| "loss": 4.5388, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7795281987312324e-05, | |
| "loss": 4.5295, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7786896039801804e-05, | |
| "loss": 4.5529, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7778510092291284e-05, | |
| "loss": 4.5383, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777014052358449e-05, | |
| "loss": 4.5356, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.776175457607397e-05, | |
| "loss": 4.5252, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.775336862856345e-05, | |
| "loss": 4.5332, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.774498268105293e-05, | |
| "loss": 4.5307, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.773659673354241e-05, | |
| "loss": 4.5222, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.772821078603189e-05, | |
| "loss": 4.5387, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771984121732511e-05, | |
| "loss": 4.5258, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771145526981459e-05, | |
| "loss": 4.5203, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.770306932230407e-05, | |
| "loss": 4.5165, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.769468337479355e-05, | |
| "loss": 4.5162, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.768629742728303e-05, | |
| "loss": 4.5163, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.767791147977251e-05, | |
| "loss": 4.5189, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766952553226199e-05, | |
| "loss": 4.5055, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766113958475147e-05, | |
| "loss": 4.5329, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.765277001604468e-05, | |
| "loss": 4.5228, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.764440044733789e-05, | |
| "loss": 4.5119, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.763601449982737e-05, | |
| "loss": 4.511, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.762762855231685e-05, | |
| "loss": 4.5062, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761924260480633e-05, | |
| "loss": 4.5181, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7610856657295813e-05, | |
| "loss": 4.5148, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7602470709785293e-05, | |
| "loss": 4.5101, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75941011410785e-05, | |
| "loss": 4.5149, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.758571519356798e-05, | |
| "loss": 4.5104, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.757734562486119e-05, | |
| "loss": 4.5089, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756895967735067e-05, | |
| "loss": 4.4995, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756057372984015e-05, | |
| "loss": 4.5121, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.755218778232963e-05, | |
| "loss": 4.501, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.754380183481911e-05, | |
| "loss": 4.504, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.753541588730859e-05, | |
| "loss": 4.499, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.752702993979807e-05, | |
| "loss": 4.5013, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.751864399228755e-05, | |
| "loss": 4.5006, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.751025804477703e-05, | |
| "loss": 4.4889, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.750188847607025e-05, | |
| "loss": 4.4938, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.468594074249268, | |
| "eval_runtime": 293.1863, | |
| "eval_samples_per_second": 1301.531, | |
| "eval_steps_per_second": 40.674, | |
| "step": 152640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.749350252855973e-05, | |
| "loss": 4.4813, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.748511658104921e-05, | |
| "loss": 4.4803, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.747673063353868e-05, | |
| "loss": 4.5069, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.746834468602816e-05, | |
| "loss": 4.4979, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745995873851764e-05, | |
| "loss": 4.4866, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745157279100712e-05, | |
| "loss": 4.486, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.74431868434966e-05, | |
| "loss": 4.488, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.743480089598608e-05, | |
| "loss": 4.4733, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.742641494847556e-05, | |
| "loss": 4.4907, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.741802900096504e-05, | |
| "loss": 4.4829, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.740964305345452e-05, | |
| "loss": 4.4869, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7401257105944e-05, | |
| "loss": 4.4882, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7392887537237216e-05, | |
| "loss": 4.4709, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7384501589726696e-05, | |
| "loss": 4.4747, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7376115642216176e-05, | |
| "loss": 4.4685, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7367729694705656e-05, | |
| "loss": 4.4692, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7359343747195135e-05, | |
| "loss": 4.473, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7350957799684615e-05, | |
| "loss": 4.4655, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7342571852174095e-05, | |
| "loss": 4.4732, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.733418590466357e-05, | |
| "loss": 4.4934, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.732579995715305e-05, | |
| "loss": 4.4702, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7317430388446264e-05, | |
| "loss": 4.4742, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730904444093574e-05, | |
| "loss": 4.468, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730065849342522e-05, | |
| "loss": 4.4812, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7292272545914704e-05, | |
| "loss": 4.4606, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7283886598404184e-05, | |
| "loss": 4.4703, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7275500650893664e-05, | |
| "loss": 4.4543, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7267114703383144e-05, | |
| "loss": 4.4588, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7258728755872624e-05, | |
| "loss": 4.4567, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.725035918716583e-05, | |
| "loss": 4.4534, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.724197323965531e-05, | |
| "loss": 4.4594, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.723358729214479e-05, | |
| "loss": 4.4657, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.722520134463427e-05, | |
| "loss": 4.4593, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.721683177592748e-05, | |
| "loss": 4.4649, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720846220722069e-05, | |
| "loss": 4.4488, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720007625971017e-05, | |
| "loss": 4.4673, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.719169031219965e-05, | |
| "loss": 4.4346, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.718330436468914e-05, | |
| "loss": 4.4572, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.717491841717862e-05, | |
| "loss": 4.4322, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.71665324696681e-05, | |
| "loss": 4.4539, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.715814652215758e-05, | |
| "loss": 4.4405, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714976057464706e-05, | |
| "loss": 4.4541, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714139100594027e-05, | |
| "loss": 4.4352, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7133021437233476e-05, | |
| "loss": 4.4473, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7124635489722956e-05, | |
| "loss": 4.4495, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7116249542212436e-05, | |
| "loss": 4.4501, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7107863594701916e-05, | |
| "loss": 4.4512, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7099477647191396e-05, | |
| "loss": 4.4295, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7091091699680876e-05, | |
| "loss": 4.4266, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7082705752170356e-05, | |
| "loss": 4.4488, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7074319804659836e-05, | |
| "loss": 4.4405, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.706595023595305e-05, | |
| "loss": 4.4344, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.705756428844253e-05, | |
| "loss": 4.4331, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704917834093201e-05, | |
| "loss": 4.429, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704079239342149e-05, | |
| "loss": 4.4185, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.70324228247147e-05, | |
| "loss": 4.4373, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.702405325600791e-05, | |
| "loss": 4.4161, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.701566730849739e-05, | |
| "loss": 4.4264, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.700728136098687e-05, | |
| "loss": 4.4377, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699889541347635e-05, | |
| "loss": 4.4191, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699050946596583e-05, | |
| "loss": 4.4202, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.698212351845531e-05, | |
| "loss": 4.4228, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.697373757094479e-05, | |
| "loss": 4.4064, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6965351623434276e-05, | |
| "loss": 4.41, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6956982054727485e-05, | |
| "loss": 4.4309, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6948596107216965e-05, | |
| "loss": 4.4157, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6940210159706445e-05, | |
| "loss": 4.4091, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6931824212195925e-05, | |
| "loss": 4.4105, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.69234382646854e-05, | |
| "loss": 4.4147, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.691505231717488e-05, | |
| "loss": 4.4253, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.690666636966436e-05, | |
| "loss": 4.4222, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.689828042215384e-05, | |
| "loss": 4.4104, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688989447464332e-05, | |
| "loss": 4.4255, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68815085271328e-05, | |
| "loss": 4.4213, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.687312257962228e-05, | |
| "loss": 4.4096, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.686473663211176e-05, | |
| "loss": 4.4071, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6856367063404974e-05, | |
| "loss": 4.4138, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6847981115894454e-05, | |
| "loss": 4.406, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6839595168383934e-05, | |
| "loss": 4.4043, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6831209220873414e-05, | |
| "loss": 4.406, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.682283965216662e-05, | |
| "loss": 4.4132, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68144537046561e-05, | |
| "loss": 4.4174, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.680606775714558e-05, | |
| "loss": 4.4032, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.679768180963506e-05, | |
| "loss": 4.389, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678929586212454e-05, | |
| "loss": 4.3971, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678090991461402e-05, | |
| "loss": 4.4031, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.677254034590723e-05, | |
| "loss": 4.4097, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.676415439839671e-05, | |
| "loss": 4.3995, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.675576845088619e-05, | |
| "loss": 4.3979, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.674738250337568e-05, | |
| "loss": 4.4082, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.673899655586516e-05, | |
| "loss": 4.3933, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.673062698715837e-05, | |
| "loss": 4.3946, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.672224103964785e-05, | |
| "loss": 4.3985, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.671385509213733e-05, | |
| "loss": 4.3899, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.670546914462681e-05, | |
| "loss": 4.4069, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.669708319711629e-05, | |
| "loss": 4.3936, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.668869724960576e-05, | |
| "loss": 4.3809, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.668031130209524e-05, | |
| "loss": 4.3924, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6671941733388456e-05, | |
| "loss": 4.3895, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6663555785877936e-05, | |
| "loss": 4.4059, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6655186217171145e-05, | |
| "loss": 4.3932, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.664680026966063e-05, | |
| "loss": 4.3851, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663841432215011e-05, | |
| "loss": 4.3777, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663002837463959e-05, | |
| "loss": 4.3908, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6621642427129065e-05, | |
| "loss": 4.3738, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6613256479618545e-05, | |
| "loss": 4.3838, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6604870532108025e-05, | |
| "loss": 4.3826, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6596484584597505e-05, | |
| "loss": 4.3815, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6588098637086985e-05, | |
| "loss": 4.3703, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6579712689576465e-05, | |
| "loss": 4.3896, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6571326742065945e-05, | |
| "loss": 4.3745, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6562940794555425e-05, | |
| "loss": 4.3746, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6554554847044905e-05, | |
| "loss": 4.3826, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6546185278338114e-05, | |
| "loss": 4.3708, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.653781570963133e-05, | |
| "loss": 4.3925, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.652942976212081e-05, | |
| "loss": 4.3814, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.652104381461029e-05, | |
| "loss": 4.3816, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.651265786709977e-05, | |
| "loss": 4.3709, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.650427191958925e-05, | |
| "loss": 4.3764, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.649590235088246e-05, | |
| "loss": 4.3765, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.648751640337194e-05, | |
| "loss": 4.3663, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647913045586142e-05, | |
| "loss": 4.3814, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.64707445083509e-05, | |
| "loss": 4.3756, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.646235856084038e-05, | |
| "loss": 4.3677, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.645397261332986e-05, | |
| "loss": 4.3699, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.644558666581934e-05, | |
| "loss": 4.3683, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.643720071830882e-05, | |
| "loss": 4.3599, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6428831149602034e-05, | |
| "loss": 4.3733, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642046158089524e-05, | |
| "loss": 4.3581, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.641207563338472e-05, | |
| "loss": 4.3795, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.64036896858742e-05, | |
| "loss": 4.3767, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.639530373836368e-05, | |
| "loss": 4.3633, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.638693416965689e-05, | |
| "loss": 4.3616, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637854822214637e-05, | |
| "loss": 4.362, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637016227463585e-05, | |
| "loss": 4.3713, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.636177632712533e-05, | |
| "loss": 4.3675, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.635339037961481e-05, | |
| "loss": 4.3657, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.634502081090802e-05, | |
| "loss": 4.3682, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.633663486339751e-05, | |
| "loss": 4.3681, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.632824891588699e-05, | |
| "loss": 4.3675, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.631986296837647e-05, | |
| "loss": 4.3531, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.631147702086595e-05, | |
| "loss": 4.3731, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.630310745215916e-05, | |
| "loss": 4.3616, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6294737883452366e-05, | |
| "loss": 4.3583, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6286351935941846e-05, | |
| "loss": 4.3602, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6277965988431326e-05, | |
| "loss": 4.3614, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6269580040920806e-05, | |
| "loss": 4.3575, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6261194093410286e-05, | |
| "loss": 4.3473, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6252808145899766e-05, | |
| "loss": 4.3497, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.337387561798096, | |
| "eval_runtime": 294.5321, | |
| "eval_samples_per_second": 1295.584, | |
| "eval_steps_per_second": 40.488, | |
| "step": 228960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6244422198389246e-05, | |
| "loss": 4.3456, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6236036250878726e-05, | |
| "loss": 4.3415, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6227650303368205e-05, | |
| "loss": 4.3677, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6219264355857685e-05, | |
| "loss": 4.3592, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6210878408347165e-05, | |
| "loss": 4.3566, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6202492460836645e-05, | |
| "loss": 4.3438, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6194106513326125e-05, | |
| "loss": 4.3543, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6185720565815605e-05, | |
| "loss": 4.3369, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6177334618305085e-05, | |
| "loss": 4.3582, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6168948670794565e-05, | |
| "loss": 4.3503, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6160562723284045e-05, | |
| "loss": 4.3497, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6152176775773525e-05, | |
| "loss": 4.3589, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6143807207066734e-05, | |
| "loss": 4.337, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6135421259556214e-05, | |
| "loss": 4.3454, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6127035312045694e-05, | |
| "loss": 4.334, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6118649364535174e-05, | |
| "loss": 4.3405, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6110263417024654e-05, | |
| "loss": 4.3385, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.610187746951414e-05, | |
| "loss": 4.3373, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6093491522003614e-05, | |
| "loss": 4.3406, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.608512195329683e-05, | |
| "loss": 4.3606, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.607673600578631e-05, | |
| "loss": 4.3408, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.606835005827578e-05, | |
| "loss": 4.3454, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.605996411076526e-05, | |
| "loss": 4.3405, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.605159454205848e-05, | |
| "loss": 4.3532, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.604320859454796e-05, | |
| "loss": 4.3331, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.603482264703743e-05, | |
| "loss": 4.3407, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.602643669952691e-05, | |
| "loss": 4.3299, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.601805075201639e-05, | |
| "loss": 4.33, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.600966480450588e-05, | |
| "loss": 4.3291, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.600127885699536e-05, | |
| "loss": 4.3346, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.599289290948484e-05, | |
| "loss": 4.3326, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.598450696197432e-05, | |
| "loss": 4.3404, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.597613739326753e-05, | |
| "loss": 4.3351, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.596775144575701e-05, | |
| "loss": 4.337, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.595936549824649e-05, | |
| "loss": 4.331, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.595097955073597e-05, | |
| "loss": 4.3413, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5942609982029176e-05, | |
| "loss": 4.3191, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5934224034518656e-05, | |
| "loss": 4.332, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5925838087008136e-05, | |
| "loss": 4.3097, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5917452139497616e-05, | |
| "loss": 4.329, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5909066191987096e-05, | |
| "loss": 4.3225, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5900680244476576e-05, | |
| "loss": 4.3361, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.589229429696606e-05, | |
| "loss": 4.3157, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.588390834945554e-05, | |
| "loss": 4.3273, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.587555515955248e-05, | |
| "loss": 4.3298, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.586716921204196e-05, | |
| "loss": 4.3327, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.585879964333517e-05, | |
| "loss": 4.3274, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.585041369582465e-05, | |
| "loss": 4.3155, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.584202774831413e-05, | |
| "loss": 4.3055, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.583364180080361e-05, | |
| "loss": 4.3294, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.582525585329309e-05, | |
| "loss": 4.3249, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.581686990578257e-05, | |
| "loss": 4.3175, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.580848395827205e-05, | |
| "loss": 4.314, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.580009801076153e-05, | |
| "loss": 4.3152, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5791712063251017e-05, | |
| "loss": 4.2995, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5783342494544226e-05, | |
| "loss": 4.3246, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5774956547033706e-05, | |
| "loss": 4.2991, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5766570599523186e-05, | |
| "loss": 4.3125, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5758184652012666e-05, | |
| "loss": 4.3201, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5749798704502146e-05, | |
| "loss": 4.3085, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5741429135795355e-05, | |
| "loss": 4.3036, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5733043188284835e-05, | |
| "loss": 4.3106, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5724657240774315e-05, | |
| "loss": 4.2951, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5716271293263794e-05, | |
| "loss": 4.3025, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.570788534575327e-05, | |
| "loss": 4.314, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5699515777046484e-05, | |
| "loss": 4.3045, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5691129829535963e-05, | |
| "loss": 4.2974, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5682743882025443e-05, | |
| "loss": 4.2966, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5674357934514923e-05, | |
| "loss": 4.3046, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.56659719870044e-05, | |
| "loss": 4.3094, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.565758603949388e-05, | |
| "loss": 4.3111, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.564921647078709e-05, | |
| "loss": 4.3035, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.564083052327657e-05, | |
| "loss": 4.3132, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.563244457576605e-05, | |
| "loss": 4.3137, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.562405862825553e-05, | |
| "loss": 4.2984, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.561568905954874e-05, | |
| "loss": 4.2969, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.560730311203822e-05, | |
| "loss": 4.3066, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.559891716452771e-05, | |
| "loss": 4.3015, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.559053121701719e-05, | |
| "loss": 4.2992, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.55821616483104e-05, | |
| "loss": 4.2977, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.557377570079988e-05, | |
| "loss": 4.3051, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.556538975328936e-05, | |
| "loss": 4.3084, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.555700380577884e-05, | |
| "loss": 4.2993, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.554861785826832e-05, | |
| "loss": 4.2852, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5540248289561526e-05, | |
| "loss": 4.287, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5531862342051006e-05, | |
| "loss": 4.2956, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5523476394540486e-05, | |
| "loss": 4.3041, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5515090447029966e-05, | |
| "loss": 4.2966, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5506720878323175e-05, | |
| "loss": 4.2875, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5498334930812655e-05, | |
| "loss": 4.3015, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.548994898330214e-05, | |
| "loss": 4.2926, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.548156303579162e-05, | |
| "loss": 4.2889, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.54731770882811e-05, | |
| "loss": 4.2903, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.546480751957431e-05, | |
| "loss": 4.295, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.545642157206379e-05, | |
| "loss": 4.3046, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.544803562455327e-05, | |
| "loss": 4.2842, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.543964967704275e-05, | |
| "loss": 4.2767, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.543126372953223e-05, | |
| "loss": 4.2932, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.542287778202171e-05, | |
| "loss": 4.2885, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.541449183451119e-05, | |
| "loss": 4.3016, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.54061222658044e-05, | |
| "loss": 4.2949, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.539773631829388e-05, | |
| "loss": 4.2814, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.538935037078336e-05, | |
| "loss": 4.2765, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.538096442327284e-05, | |
| "loss": 4.2919, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5372578475762326e-05, | |
| "loss": 4.2723, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5364192528251806e-05, | |
| "loss": 4.2831, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.535580658074128e-05, | |
| "loss": 4.2785, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5347437012034495e-05, | |
| "loss": 4.2865, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5339051064523975e-05, | |
| "loss": 4.2703, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5330665117013455e-05, | |
| "loss": 4.2896, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5322295548306664e-05, | |
| "loss": 4.2745, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5313909600796144e-05, | |
| "loss": 4.2729, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5305523653285624e-05, | |
| "loss": 4.2861, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5297137705775104e-05, | |
| "loss": 4.2715, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.528875175826458e-05, | |
| "loss": 4.2964, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5280365810754064e-05, | |
| "loss": 4.2857, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5271979863243544e-05, | |
| "loss": 4.282, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5263593915733024e-05, | |
| "loss": 4.2743, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5255207968222504e-05, | |
| "loss": 4.2772, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5246822020711984e-05, | |
| "loss": 4.2796, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5238436073201464e-05, | |
| "loss": 4.2711, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5230050125690944e-05, | |
| "loss": 4.285, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.522168055698415e-05, | |
| "loss": 4.2775, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.521329460947363e-05, | |
| "loss": 4.272, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.520492504076684e-05, | |
| "loss": 4.2733, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.519653909325632e-05, | |
| "loss": 4.2738, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.51881531457458e-05, | |
| "loss": 4.267, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.517976719823528e-05, | |
| "loss": 4.2818, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.51713976295285e-05, | |
| "loss": 4.2616, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.516301168201798e-05, | |
| "loss": 4.284, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.515462573450746e-05, | |
| "loss": 4.2771, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.514623978699694e-05, | |
| "loss": 4.2774, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5137870218290147e-05, | |
| "loss": 4.2674, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5129484270779626e-05, | |
| "loss": 4.2707, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5121098323269106e-05, | |
| "loss": 4.2796, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5112712375758586e-05, | |
| "loss": 4.2771, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5104326428248066e-05, | |
| "loss": 4.2686, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5095940480737546e-05, | |
| "loss": 4.2748, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5087554533227026e-05, | |
| "loss": 4.2728, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5079168585716506e-05, | |
| "loss": 4.2783, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5070782638205986e-05, | |
| "loss": 4.2637, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5062396690695466e-05, | |
| "loss": 4.282, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5054010743184946e-05, | |
| "loss": 4.2726, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.504564117447816e-05, | |
| "loss": 4.2688, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.503725522696764e-05, | |
| "loss": 4.2677, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5028869279457115e-05, | |
| "loss": 4.2746, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.502049971075033e-05, | |
| "loss": 4.2644, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.501211376323981e-05, | |
| "loss": 4.2543, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.500372781572929e-05, | |
| "loss": 4.2699, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.256018161773682, | |
| "eval_runtime": 291.5284, | |
| "eval_samples_per_second": 1308.933, | |
| "eval_steps_per_second": 40.905, | |
| "step": 305280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4995341868218764e-05, | |
| "loss": 4.2624, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4986955920708244e-05, | |
| "loss": 4.2517, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4978569973197724e-05, | |
| "loss": 4.2775, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.497020040449093e-05, | |
| "loss": 4.2703, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.496181445698042e-05, | |
| "loss": 4.269, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.49534285094699e-05, | |
| "loss": 4.2614, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.494504256195938e-05, | |
| "loss": 4.2643, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.493667299325259e-05, | |
| "loss": 4.245, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.492828704574207e-05, | |
| "loss": 4.2699, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.491990109823155e-05, | |
| "loss": 4.265, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.491151515072103e-05, | |
| "loss": 4.2611, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.490312920321051e-05, | |
| "loss": 4.2689, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.489475963450372e-05, | |
| "loss": 4.2513, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.48863736869932e-05, | |
| "loss": 4.258, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.487798773948268e-05, | |
| "loss": 4.2515, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.486960179197216e-05, | |
| "loss": 4.2492, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.486121584446164e-05, | |
| "loss": 4.2536, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.485282989695112e-05, | |
| "loss": 4.252, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4844443949440604e-05, | |
| "loss": 4.2547, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4836058001930084e-05, | |
| "loss": 4.2733, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.482768843322329e-05, | |
| "loss": 4.2594, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.481930248571277e-05, | |
| "loss": 4.2636, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.481091653820225e-05, | |
| "loss": 4.2544, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.480254696949546e-05, | |
| "loss": 4.2641, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.479416102198494e-05, | |
| "loss": 4.2509, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.478577507447442e-05, | |
| "loss": 4.2562, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.47773891269639e-05, | |
| "loss": 4.2499, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.476900317945338e-05, | |
| "loss": 4.2432, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.476063361074659e-05, | |
| "loss": 4.2428, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.475224766323607e-05, | |
| "loss": 4.2518, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.474386171572556e-05, | |
| "loss": 4.2502, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.473547576821504e-05, | |
| "loss": 4.2576, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.472708982070452e-05, | |
| "loss": 4.2546, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4718703873194e-05, | |
| "loss": 4.2558, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.471033430448721e-05, | |
| "loss": 4.2459, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.470194835697669e-05, | |
| "loss": 4.2598, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4693578788269896e-05, | |
| "loss": 4.2405, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4685192840759376e-05, | |
| "loss": 4.2466, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4676806893248856e-05, | |
| "loss": 4.2316, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4668420945738336e-05, | |
| "loss": 4.2477, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4660034998227816e-05, | |
| "loss": 4.2439, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4651649050717296e-05, | |
| "loss": 4.2523, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4643263103206776e-05, | |
| "loss": 4.2367, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4634877155696256e-05, | |
| "loss": 4.2473, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4626491208185736e-05, | |
| "loss": 4.2493, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4618105260675215e-05, | |
| "loss": 4.2517, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4609719313164695e-05, | |
| "loss": 4.2482, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4601349744457905e-05, | |
| "loss": 4.2353, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4592963796947384e-05, | |
| "loss": 4.2249, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4584577849436864e-05, | |
| "loss": 4.2531, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4576191901926344e-05, | |
| "loss": 4.2457, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4567805954415824e-05, | |
| "loss": 4.2382, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4559420006905304e-05, | |
| "loss": 4.2378, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4551034059394784e-05, | |
| "loss": 4.2371, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4542648111884264e-05, | |
| "loss": 4.2224, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.453427854317748e-05, | |
| "loss": 4.2402, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.452589259566696e-05, | |
| "loss": 4.2252, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.451750664815644e-05, | |
| "loss": 4.2323, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.450912070064592e-05, | |
| "loss": 4.2451, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.450075113193913e-05, | |
| "loss": 4.2303, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.449236518442861e-05, | |
| "loss": 4.2259, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.448397923691809e-05, | |
| "loss": 4.2393, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.447559328940757e-05, | |
| "loss": 4.2169, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.446720734189705e-05, | |
| "loss": 4.2264, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.445882139438653e-05, | |
| "loss": 4.2354, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.445043544687601e-05, | |
| "loss": 4.2309, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.444204949936548e-05, | |
| "loss": 4.2222, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.443366355185496e-05, | |
| "loss": 4.2231, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.442529398314818e-05, | |
| "loss": 4.2249, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.441690803563766e-05, | |
| "loss": 4.2335, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.440852208812714e-05, | |
| "loss": 4.2343, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.440013614061662e-05, | |
| "loss": 4.2284, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4391766571909834e-05, | |
| "loss": 4.238, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4383380624399314e-05, | |
| "loss": 4.2404, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.437499467688879e-05, | |
| "loss": 4.225, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.436664148698573e-05, | |
| "loss": 4.2183, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.435825553947521e-05, | |
| "loss": 4.237, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.434986959196469e-05, | |
| "loss": 4.2296, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.434148364445417e-05, | |
| "loss": 4.2204, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.433309769694365e-05, | |
| "loss": 4.2254, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.432471174943313e-05, | |
| "loss": 4.2303, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.431634218072635e-05, | |
| "loss": 4.2391, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.430795623321583e-05, | |
| "loss": 4.2207, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.429957028570531e-05, | |
| "loss": 4.2117, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.429118433819479e-05, | |
| "loss": 4.2216, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.428279839068426e-05, | |
| "loss": 4.2176, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.427441244317374e-05, | |
| "loss": 4.2306, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.426602649566322e-05, | |
| "loss": 4.2267, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.42576405481527e-05, | |
| "loss": 4.2138, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.424925460064218e-05, | |
| "loss": 4.2309, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.424086865313166e-05, | |
| "loss": 4.2201, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.423248270562114e-05, | |
| "loss": 4.2223, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.422409675811062e-05, | |
| "loss": 4.2152, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4215727189403836e-05, | |
| "loss": 4.2225, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4207341241893316e-05, | |
| "loss": 4.2342, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4198955294382796e-05, | |
| "loss": 4.2192, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4190569346872276e-05, | |
| "loss": 4.2028, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4182199778165485e-05, | |
| "loss": 4.2181, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4173813830654965e-05, | |
| "loss": 4.2198, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4165427883144445e-05, | |
| "loss": 4.2254, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4157041935633925e-05, | |
| "loss": 4.2287, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4148655988123405e-05, | |
| "loss": 4.2125, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4140270040612885e-05, | |
| "loss": 4.2086, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4131884093102365e-05, | |
| "loss": 4.2207, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4123514524395574e-05, | |
| "loss": 4.1986, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4115128576885054e-05, | |
| "loss": 4.215, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4106742629374534e-05, | |
| "loss": 4.2089, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.409835668186402e-05, | |
| "loss": 4.2173, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.40899707343535e-05, | |
| "loss": 4.2046, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4081584786842974e-05, | |
| "loss": 4.2201, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4073198839332453e-05, | |
| "loss": 4.2048, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.406482927062567e-05, | |
| "loss": 4.2035, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.405644332311514e-05, | |
| "loss": 4.2185, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.404805737560462e-05, | |
| "loss": 4.2076, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.40396714280941e-05, | |
| "loss": 4.2269, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.403128548058358e-05, | |
| "loss": 4.2169, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.402293229068053e-05, | |
| "loss": 4.2101, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.401454634317001e-05, | |
| "loss": 4.2159, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.400616039565949e-05, | |
| "loss": 4.2031, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.399777444814897e-05, | |
| "loss": 4.2133, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.398938850063845e-05, | |
| "loss": 4.2043, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.398100255312793e-05, | |
| "loss": 4.2203, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.397263298442114e-05, | |
| "loss": 4.2063, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3964247036910616e-05, | |
| "loss": 4.2041, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3955861089400096e-05, | |
| "loss": 4.205, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3947475141889576e-05, | |
| "loss": 4.2093, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3939089194379056e-05, | |
| "loss": 4.1999, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3930703246868536e-05, | |
| "loss": 4.2149, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3922317299358016e-05, | |
| "loss": 4.1964, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3913931351847496e-05, | |
| "loss": 4.2167, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3905545404336976e-05, | |
| "loss": 4.21, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.389717583563019e-05, | |
| "loss": 4.2113, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.38888062669234e-05, | |
| "loss": 4.2027, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.388042031941288e-05, | |
| "loss": 4.2042, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.387203437190236e-05, | |
| "loss": 4.2159, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.386364842439184e-05, | |
| "loss": 4.2077, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.385526247688132e-05, | |
| "loss": 4.2003, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.38468765293708e-05, | |
| "loss": 4.2116, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.383849058186028e-05, | |
| "loss": 4.2067, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.383010463434976e-05, | |
| "loss": 4.2158, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.382171868683924e-05, | |
| "loss": 4.2013, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.381333273932872e-05, | |
| "loss": 4.2138, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.38049467918182e-05, | |
| "loss": 4.206, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.379657722311141e-05, | |
| "loss": 4.2031, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3788207654404625e-05, | |
| "loss": 4.2048, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3779821706894105e-05, | |
| "loss": 4.2052, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3771435759383585e-05, | |
| "loss": 4.2049, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3763066190676794e-05, | |
| "loss": 4.1903, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3754680243166274e-05, | |
| "loss": 4.2042, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.200494289398193, | |
| "eval_runtime": 292.0312, | |
| "eval_samples_per_second": 1306.679, | |
| "eval_steps_per_second": 40.835, | |
| "step": 381600 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 1.576778732850636e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |