| { | |
| "best_metric": 4.138632297515869, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/pp-mod-subj/lstm/0/checkpoints/checkpoint-686871", | |
| "epoch": 0.025000278439663435, | |
| "eval_steps": 10, | |
| "global_step": 686871, | |
| "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": 10.8212, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 7.5833, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 7.0572, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 6.9837, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 6.9532, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 6.8624, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 6.7187, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994129836742636e-05, | |
| "loss": 6.6081, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993291241991584e-05, | |
| "loss": 6.5231, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992452647240532e-05, | |
| "loss": 6.4374, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99161405248948e-05, | |
| "loss": 6.3822, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990775457738428e-05, | |
| "loss": 6.3146, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989936862987376e-05, | |
| "loss": 6.2475, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989098268236324e-05, | |
| "loss": 6.1734, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988259673485272e-05, | |
| "loss": 6.1072, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.98742107873422e-05, | |
| "loss": 6.0606, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986582483983168e-05, | |
| "loss": 6.0118, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985743889232116e-05, | |
| "loss": 5.9605, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984905294481064e-05, | |
| "loss": 5.9258, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984066699730012e-05, | |
| "loss": 5.8818, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.8523, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.982391148108281e-05, | |
| "loss": 5.8137, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.981552553357229e-05, | |
| "loss": 5.7773, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.980713958606178e-05, | |
| "loss": 5.7553, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9798770017354986e-05, | |
| "loss": 5.7238, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9790384069844466e-05, | |
| "loss": 5.6962, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9781998122333946e-05, | |
| "loss": 5.6734, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773612174823426e-05, | |
| "loss": 5.6524, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9765242606116635e-05, | |
| "loss": 5.619, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9756856658606115e-05, | |
| "loss": 5.605, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748470711095595e-05, | |
| "loss": 5.579, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.974008476358507e-05, | |
| "loss": 5.5615, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9731715194878284e-05, | |
| "loss": 5.5501, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9723329247367764e-05, | |
| "loss": 5.5333, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714943299857244e-05, | |
| "loss": 5.5129, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9706557352346724e-05, | |
| "loss": 5.5041, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.969818778363994e-05, | |
| "loss": 5.4898, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968980183612942e-05, | |
| "loss": 5.4715, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96814158886189e-05, | |
| "loss": 5.4539, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.967302994110837e-05, | |
| "loss": 5.4452, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966464399359785e-05, | |
| "loss": 5.4191, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965625804608733e-05, | |
| "loss": 5.4201, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964787209857681e-05, | |
| "loss": 5.391, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963950252987002e-05, | |
| "loss": 5.3906, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96311165823595e-05, | |
| "loss": 5.3845, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962273063484898e-05, | |
| "loss": 5.3753, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.961434468733847e-05, | |
| "loss": 5.3496, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9605991497435414e-05, | |
| "loss": 5.3482, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9597605549924894e-05, | |
| "loss": 5.3439, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9589219602414374e-05, | |
| "loss": 5.3209, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958083365490385e-05, | |
| "loss": 5.3069, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.957244770739333e-05, | |
| "loss": 5.3047, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956406175988281e-05, | |
| "loss": 5.3114, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955567581237229e-05, | |
| "loss": 5.2782, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.954728986486177e-05, | |
| "loss": 5.2988, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953890391735125e-05, | |
| "loss": 5.2592, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9530534348644456e-05, | |
| "loss": 5.2698, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9522148401133936e-05, | |
| "loss": 5.2482, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951376245362342e-05, | |
| "loss": 5.2445, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95053765061129e-05, | |
| "loss": 5.2362, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949699055860238e-05, | |
| "loss": 5.2305, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948860461109186e-05, | |
| "loss": 5.2104, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948023504238507e-05, | |
| "loss": 5.2179, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947184909487455e-05, | |
| "loss": 5.2004, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946346314736403e-05, | |
| "loss": 5.2091, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945507719985351e-05, | |
| "loss": 5.19, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944670763114672e-05, | |
| "loss": 5.1804, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94383216836362e-05, | |
| "loss": 5.1635, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942995211492941e-05, | |
| "loss": 5.1712, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942156616741889e-05, | |
| "loss": 5.165, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.941318021990837e-05, | |
| "loss": 5.1462, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9404794272397856e-05, | |
| "loss": 5.1487, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9396408324887336e-05, | |
| "loss": 5.14, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388022377376816e-05, | |
| "loss": 5.1364, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379636429866296e-05, | |
| "loss": 5.1337, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371250482355776e-05, | |
| "loss": 5.1279, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362880913648985e-05, | |
| "loss": 5.114, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9354494966138465e-05, | |
| "loss": 5.0973, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9346109018627945e-05, | |
| "loss": 5.1051, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9337723071117425e-05, | |
| "loss": 5.0962, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329337123606905e-05, | |
| "loss": 5.0985, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932095117609638e-05, | |
| "loss": 5.0895, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.931256522858586e-05, | |
| "loss": 5.0805, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9304195659879074e-05, | |
| "loss": 5.0587, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9295809712368554e-05, | |
| "loss": 5.0656, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9287423764858034e-05, | |
| "loss": 5.067, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9279037817347514e-05, | |
| "loss": 5.0617, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9270651869836994e-05, | |
| "loss": 5.0511, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9262265922326474e-05, | |
| "loss": 5.0447, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.925391273242342e-05, | |
| "loss": 5.0329, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.92455267849129e-05, | |
| "loss": 5.0276, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923714083740238e-05, | |
| "loss": 5.0249, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922875488989185e-05, | |
| "loss": 5.0319, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922038532118507e-05, | |
| "loss": 5.0224, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921199937367455e-05, | |
| "loss": 4.9982, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920361342616403e-05, | |
| "loss": 5.0003, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919522747865351e-05, | |
| "loss": 5.0063, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918684153114299e-05, | |
| "loss": 5.0023, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917845558363247e-05, | |
| "loss": 4.9909, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917006963612195e-05, | |
| "loss": 4.9903, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.916168368861143e-05, | |
| "loss": 4.978, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.915329774110091e-05, | |
| "loss": 4.9765, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.914491179359039e-05, | |
| "loss": 4.9828, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.913652584607987e-05, | |
| "loss": 4.978, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9128156277373076e-05, | |
| "loss": 4.952, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9119770329862556e-05, | |
| "loss": 4.9629, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9111384382352036e-05, | |
| "loss": 4.9433, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9102998434841516e-05, | |
| "loss": 4.947, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9094612487330996e-05, | |
| "loss": 4.9444, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9086226539820476e-05, | |
| "loss": 4.9349, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907785697111369e-05, | |
| "loss": 4.9321, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906947102360317e-05, | |
| "loss": 4.9225, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906108507609265e-05, | |
| "loss": 4.922, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905269912858213e-05, | |
| "loss": 4.9196, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904432955987534e-05, | |
| "loss": 4.9309, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903594361236482e-05, | |
| "loss": 4.907, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90275576648543e-05, | |
| "loss": 4.8897, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901917171734378e-05, | |
| "loss": 4.9012, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901078576983326e-05, | |
| "loss": 4.9114, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.900239982232274e-05, | |
| "loss": 4.8947, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899403025361595e-05, | |
| "loss": 4.9021, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8985660684909166e-05, | |
| "loss": 4.8903, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8977274737398646e-05, | |
| "loss": 4.8903, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968888789888125e-05, | |
| "loss": 4.8759, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8960502842377605e-05, | |
| "loss": 4.8793, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8952116894867085e-05, | |
| "loss": 4.874, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8943730947356565e-05, | |
| "loss": 4.8775, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.893534499984604e-05, | |
| "loss": 4.8567, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.892695905233552e-05, | |
| "loss": 4.8656, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918573104825e-05, | |
| "loss": 4.8632, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8910203536118214e-05, | |
| "loss": 4.8488, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890181758860769e-05, | |
| "loss": 4.8573, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889343164109717e-05, | |
| "loss": 4.8501, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888504569358665e-05, | |
| "loss": 4.8548, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8876659746076134e-05, | |
| "loss": 4.8508, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8868273798565614e-05, | |
| "loss": 4.8334, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8859887851055094e-05, | |
| "loss": 4.828, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88515182823483e-05, | |
| "loss": 4.828, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884314871364151e-05, | |
| "loss": 4.833, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883476276613099e-05, | |
| "loss": 4.8406, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882637681862047e-05, | |
| "loss": 4.8283, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881799087110995e-05, | |
| "loss": 4.8189, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880960492359943e-05, | |
| "loss": 4.8147, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880121897608891e-05, | |
| "loss": 4.8139, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879283302857839e-05, | |
| "loss": 4.81, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.878444708106787e-05, | |
| "loss": 4.8075, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877607751236109e-05, | |
| "loss": 4.8196, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.87677079436543e-05, | |
| "loss": 4.8047, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875932199614378e-05, | |
| "loss": 4.7968, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875093604863326e-05, | |
| "loss": 4.7923, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.810235500335693, | |
| "eval_runtime": 552.8952, | |
| "eval_samples_per_second": 690.169, | |
| "eval_steps_per_second": 21.568, | |
| "step": 76319 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8742566479926466e-05, | |
| "loss": 4.8064, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8734180532415946e-05, | |
| "loss": 4.7989, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8725794584905426e-05, | |
| "loss": 4.7829, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8717408637394906e-05, | |
| "loss": 4.7851, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8709022689884386e-05, | |
| "loss": 4.7746, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8700636742373866e-05, | |
| "loss": 4.7713, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8692250794863346e-05, | |
| "loss": 4.768, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8683864847352826e-05, | |
| "loss": 4.7704, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.867549527864604e-05, | |
| "loss": 4.766, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.866710933113552e-05, | |
| "loss": 4.7674, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8658723383625e-05, | |
| "loss": 4.7656, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865033743611448e-05, | |
| "loss": 4.7718, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.864195148860396e-05, | |
| "loss": 4.7582, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.863358191989717e-05, | |
| "loss": 4.7493, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.862519597238665e-05, | |
| "loss": 4.7476, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.861681002487613e-05, | |
| "loss": 4.7293, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.860842407736561e-05, | |
| "loss": 4.742, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.860005450865882e-05, | |
| "loss": 4.7399, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.85916685611483e-05, | |
| "loss": 4.7377, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.858328261363778e-05, | |
| "loss": 4.7475, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.857489666612726e-05, | |
| "loss": 4.735, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.856651071861674e-05, | |
| "loss": 4.7337, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8558124771106226e-05, | |
| "loss": 4.736, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.85497388235957e-05, | |
| "loss": 4.7317, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8541369254888915e-05, | |
| "loss": 4.7214, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8532983307378395e-05, | |
| "loss": 4.7209, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8524597359867875e-05, | |
| "loss": 4.7255, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.851621141235735e-05, | |
| "loss": 4.7128, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.850782546484683e-05, | |
| "loss": 4.7133, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.849943951733631e-05, | |
| "loss": 4.6998, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.849105356982579e-05, | |
| "loss": 4.703, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.848266762231527e-05, | |
| "loss": 4.7132, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.847429805360848e-05, | |
| "loss": 4.7053, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.846594486370543e-05, | |
| "loss": 4.7041, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.845755891619491e-05, | |
| "loss": 4.7007, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.844917296868439e-05, | |
| "loss": 4.704, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.844078702117387e-05, | |
| "loss": 4.6991, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.843240107366335e-05, | |
| "loss": 4.6963, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.842401512615282e-05, | |
| "loss": 4.697, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.84156291786423e-05, | |
| "loss": 4.6842, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.840724323113178e-05, | |
| "loss": 4.6882, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839885728362126e-05, | |
| "loss": 4.6749, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839047133611074e-05, | |
| "loss": 4.6808, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.838210176740395e-05, | |
| "loss": 4.6897, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.837371581989343e-05, | |
| "loss": 4.6873, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.836532987238292e-05, | |
| "loss": 4.673, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.83569439248724e-05, | |
| "loss": 4.6754, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834855797736188e-05, | |
| "loss": 4.6796, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834017202985136e-05, | |
| "loss": 4.6585, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.833178608234084e-05, | |
| "loss": 4.6583, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.832340013483032e-05, | |
| "loss": 4.6611, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8315030566123526e-05, | |
| "loss": 4.6742, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8306644618613006e-05, | |
| "loss": 4.6483, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8298258671102486e-05, | |
| "loss": 4.6754, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8289872723591966e-05, | |
| "loss": 4.6436, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8281503154885175e-05, | |
| "loss": 4.6568, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8273133586178384e-05, | |
| "loss": 4.6493, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.826474763866787e-05, | |
| "loss": 4.6525, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.825636169115735e-05, | |
| "loss": 4.6438, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.824797574364683e-05, | |
| "loss": 4.6482, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823958979613631e-05, | |
| "loss": 4.6368, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823120384862579e-05, | |
| "loss": 4.6451, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8222834279919e-05, | |
| "loss": 4.6366, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.821444833240848e-05, | |
| "loss": 4.6434, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.820606238489796e-05, | |
| "loss": 4.635, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.819767643738744e-05, | |
| "loss": 4.6322, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.818929048987692e-05, | |
| "loss": 4.6151, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.81809045423664e-05, | |
| "loss": 4.6404, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.817253497365961e-05, | |
| "loss": 4.6284, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.816414902614909e-05, | |
| "loss": 4.6173, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.815576307863857e-05, | |
| "loss": 4.6271, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.814737713112805e-05, | |
| "loss": 4.6216, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8139007562421265e-05, | |
| "loss": 4.623, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8130621614910745e-05, | |
| "loss": 4.6289, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8122235667400224e-05, | |
| "loss": 4.6274, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8113849719889704e-05, | |
| "loss": 4.6135, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8105463772379184e-05, | |
| "loss": 4.6047, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.809707782486866e-05, | |
| "loss": 4.6119, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8088708256161873e-05, | |
| "loss": 4.6131, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808032230865135e-05, | |
| "loss": 4.6206, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8071936361140827e-05, | |
| "loss": 4.6123, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8063550413630307e-05, | |
| "loss": 4.6078, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8055164466119786e-05, | |
| "loss": 4.5875, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8046794897413e-05, | |
| "loss": 4.599, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803840894990248e-05, | |
| "loss": 4.6039, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803002300239196e-05, | |
| "loss": 4.6011, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.802163705488144e-05, | |
| "loss": 4.6059, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.801326748617466e-05, | |
| "loss": 4.5966, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.800488153866413e-05, | |
| "loss": 4.5867, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.799649559115361e-05, | |
| "loss": 4.5921, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.798812602244683e-05, | |
| "loss": 4.5863, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.79797400749363e-05, | |
| "loss": 4.5929, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797135412742578e-05, | |
| "loss": 4.5947, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.796296817991526e-05, | |
| "loss": 4.5717, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.795458223240474e-05, | |
| "loss": 4.5739, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.794619628489423e-05, | |
| "loss": 4.5843, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7937826716187436e-05, | |
| "loss": 4.5896, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7929440768676916e-05, | |
| "loss": 4.5759, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7921054821166396e-05, | |
| "loss": 4.5821, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7912668873655876e-05, | |
| "loss": 4.5747, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7904282926145356e-05, | |
| "loss": 4.5727, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7895896978634836e-05, | |
| "loss": 4.5819, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7887511031124316e-05, | |
| "loss": 4.5783, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7879125083613796e-05, | |
| "loss": 4.56, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7870739136103276e-05, | |
| "loss": 4.5745, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7862353188592756e-05, | |
| "loss": 4.5556, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7853967241082236e-05, | |
| "loss": 4.5593, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7845581293571715e-05, | |
| "loss": 4.5654, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7837211724864925e-05, | |
| "loss": 4.5552, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782884215615814e-05, | |
| "loss": 4.5559, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782045620864762e-05, | |
| "loss": 4.5508, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.78120702611371e-05, | |
| "loss": 4.5517, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.780370069243031e-05, | |
| "loss": 4.5506, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.779531474491979e-05, | |
| "loss": 4.5659, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.778692879740927e-05, | |
| "loss": 4.547, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777854284989875e-05, | |
| "loss": 4.5376, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777015690238823e-05, | |
| "loss": 4.5441, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.776177095487771e-05, | |
| "loss": 4.552, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.775340138617092e-05, | |
| "loss": 4.5468, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.77450154386604e-05, | |
| "loss": 4.5502, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.773662949114988e-05, | |
| "loss": 4.5462, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.772824354363936e-05, | |
| "loss": 4.5486, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7719857596128845e-05, | |
| "loss": 4.539, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771147164861832e-05, | |
| "loss": 4.5351, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.77030857011078e-05, | |
| "loss": 4.5392, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.769469975359728e-05, | |
| "loss": 4.5443, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.768631380608676e-05, | |
| "loss": 4.5291, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.767792785857624e-05, | |
| "loss": 4.5335, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766954191106572e-05, | |
| "loss": 4.5462, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766117234235893e-05, | |
| "loss": 4.5204, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.765278639484841e-05, | |
| "loss": 4.5349, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.764440044733789e-05, | |
| "loss": 4.5302, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7636030878631096e-05, | |
| "loss": 4.5409, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.762764493112058e-05, | |
| "loss": 4.5333, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761925898361006e-05, | |
| "loss": 4.5209, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761087303609954e-05, | |
| "loss": 4.5195, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.760248708858902e-05, | |
| "loss": 4.5187, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75941011410785e-05, | |
| "loss": 4.5281, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.758571519356798e-05, | |
| "loss": 4.535, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.757732924605746e-05, | |
| "loss": 4.5289, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756895967735067e-05, | |
| "loss": 4.5147, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756057372984015e-05, | |
| "loss": 4.5162, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.755218778232963e-05, | |
| "loss": 4.518, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.754381821362284e-05, | |
| "loss": 4.5166, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.753543226611232e-05, | |
| "loss": 4.5138, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75270463186018e-05, | |
| "loss": 4.5307, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.751866037109128e-05, | |
| "loss": 4.5096, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.751027442358077e-05, | |
| "loss": 4.5127, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.750188847607025e-05, | |
| "loss": 4.5063, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.530740261077881, | |
| "eval_runtime": 587.9687, | |
| "eval_samples_per_second": 648.999, | |
| "eval_steps_per_second": 20.282, | |
| "step": 152638 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.749350252855973e-05, | |
| "loss": 4.522, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.748511658104921e-05, | |
| "loss": 4.5171, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.747673063353868e-05, | |
| "loss": 4.5035, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.746834468602816e-05, | |
| "loss": 4.5042, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745995873851764e-05, | |
| "loss": 4.4969, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745157279100712e-05, | |
| "loss": 4.4991, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.74431868434966e-05, | |
| "loss": 4.4972, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.743480089598608e-05, | |
| "loss": 4.4973, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.742641494847556e-05, | |
| "loss": 4.4959, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.741804537976877e-05, | |
| "loss": 4.5006, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.740965943225825e-05, | |
| "loss": 4.5014, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7401273484747736e-05, | |
| "loss": 4.5065, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7392887537237216e-05, | |
| "loss": 4.4941, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7384501589726696e-05, | |
| "loss": 4.4843, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7376115642216176e-05, | |
| "loss": 4.4911, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7367729694705656e-05, | |
| "loss": 4.47, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7359343747195135e-05, | |
| "loss": 4.4869, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7350957799684615e-05, | |
| "loss": 4.4835, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7342571852174095e-05, | |
| "loss": 4.4828, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.733418590466357e-05, | |
| "loss": 4.4979, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.732579995715305e-05, | |
| "loss": 4.4854, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.731741400964253e-05, | |
| "loss": 4.4853, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730902806213201e-05, | |
| "loss": 4.4879, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730064211462149e-05, | |
| "loss": 4.4815, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.729225616711097e-05, | |
| "loss": 4.4794, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7283886598404184e-05, | |
| "loss": 4.4756, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7275500650893664e-05, | |
| "loss": 4.4805, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.726713108218687e-05, | |
| "loss": 4.4708, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.725874513467635e-05, | |
| "loss": 4.4751, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.725035918716583e-05, | |
| "loss": 4.4601, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.724197323965531e-05, | |
| "loss": 4.468, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.723358729214479e-05, | |
| "loss": 4.4786, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.722520134463427e-05, | |
| "loss": 4.4658, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.721681539712375e-05, | |
| "loss": 4.4717, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720842944961323e-05, | |
| "loss": 4.4676, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720007625971017e-05, | |
| "loss": 4.4696, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.719169031219965e-05, | |
| "loss": 4.4722, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.718332074349287e-05, | |
| "loss": 4.4701, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.717493479598235e-05, | |
| "loss": 4.4688, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.716654884847183e-05, | |
| "loss": 4.4568, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.715816290096131e-05, | |
| "loss": 4.4651, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714977695345079e-05, | |
| "loss": 4.4513, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714139100594027e-05, | |
| "loss": 4.466, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.713300505842975e-05, | |
| "loss": 4.4636, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.712461911091923e-05, | |
| "loss": 4.4667, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7116249542212436e-05, | |
| "loss": 4.4515, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7107863594701916e-05, | |
| "loss": 4.4538, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7099477647191396e-05, | |
| "loss": 4.4628, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7091091699680876e-05, | |
| "loss": 4.4426, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7082705752170356e-05, | |
| "loss": 4.4466, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.707433618346357e-05, | |
| "loss": 4.4435, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.706595023595305e-05, | |
| "loss": 4.4636, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.705756428844253e-05, | |
| "loss": 4.4391, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704917834093201e-05, | |
| "loss": 4.4653, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704080877222522e-05, | |
| "loss": 4.435, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.70324228247147e-05, | |
| "loss": 4.4435, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.702403687720418e-05, | |
| "loss": 4.446, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.701565092969366e-05, | |
| "loss": 4.4457, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.700726498218314e-05, | |
| "loss": 4.4407, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699887903467262e-05, | |
| "loss": 4.4506, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.69904930871621e-05, | |
| "loss": 4.4256, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.698212351845531e-05, | |
| "loss": 4.4424, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.697373757094479e-05, | |
| "loss": 4.4342, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6965351623434276e-05, | |
| "loss": 4.4449, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6956965675923756e-05, | |
| "loss": 4.4329, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.694857972841323e-05, | |
| "loss": 4.4342, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6940210159706445e-05, | |
| "loss": 4.417, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6931824212195925e-05, | |
| "loss": 4.4441, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.69234382646854e-05, | |
| "loss": 4.434, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.691505231717488e-05, | |
| "loss": 4.4253, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.690666636966436e-05, | |
| "loss": 4.4313, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.689828042215384e-05, | |
| "loss": 4.4304, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688989447464332e-05, | |
| "loss": 4.4287, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68815085271328e-05, | |
| "loss": 4.4382, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6873138958426014e-05, | |
| "loss": 4.4371, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.686476938971923e-05, | |
| "loss": 4.4265, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68563834422087e-05, | |
| "loss": 4.4175, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.684799749469818e-05, | |
| "loss": 4.4224, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683961154718766e-05, | |
| "loss": 4.4291, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683122559967714e-05, | |
| "loss": 4.4324, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.682283965216662e-05, | |
| "loss": 4.4253, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68144537046561e-05, | |
| "loss": 4.4272, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.680606775714558e-05, | |
| "loss": 4.4052, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.679769818843879e-05, | |
| "loss": 4.4169, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6789328619732e-05, | |
| "loss": 4.415, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678094267222148e-05, | |
| "loss": 4.4199, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.677255672471097e-05, | |
| "loss": 4.427, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.676417077720045e-05, | |
| "loss": 4.4174, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.675578482968993e-05, | |
| "loss": 4.4057, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.674739888217941e-05, | |
| "loss": 4.416, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6739029313472616e-05, | |
| "loss": 4.4084, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6730643365962096e-05, | |
| "loss": 4.4158, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6722257418451576e-05, | |
| "loss": 4.4149, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6713871470941056e-05, | |
| "loss": 4.3973, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6705485523430536e-05, | |
| "loss": 4.4005, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6697115954723745e-05, | |
| "loss": 4.4073, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6688730007213225e-05, | |
| "loss": 4.4142, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6680344059702705e-05, | |
| "loss": 4.4053, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6671958112192185e-05, | |
| "loss": 4.4133, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6663572164681665e-05, | |
| "loss": 4.4, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6655186217171145e-05, | |
| "loss": 4.4021, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.664681664846436e-05, | |
| "loss": 4.4096, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663843070095384e-05, | |
| "loss": 4.4077, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663004475344332e-05, | |
| "loss": 4.3933, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.66216588059328e-05, | |
| "loss": 4.41, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.661327285842228e-05, | |
| "loss": 4.3849, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.660488691091176e-05, | |
| "loss": 4.3895, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6596500963401234e-05, | |
| "loss": 4.403, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6588115015890714e-05, | |
| "loss": 4.3871, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657974544718393e-05, | |
| "loss": 4.3981, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657135949967341e-05, | |
| "loss": 4.3872, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.656297355216288e-05, | |
| "loss": 4.3864, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.655458760465237e-05, | |
| "loss": 4.3857, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.654620165714185e-05, | |
| "loss": 4.4062, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.653781570963133e-05, | |
| "loss": 4.3855, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.652944614092454e-05, | |
| "loss": 4.3807, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.652106019341402e-05, | |
| "loss": 4.3785, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.65126742459035e-05, | |
| "loss": 4.388, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.650428829839298e-05, | |
| "loss": 4.3934, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.649591872968619e-05, | |
| "loss": 4.3904, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.648753278217567e-05, | |
| "loss": 4.3865, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647914683466515e-05, | |
| "loss": 4.3892, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647076088715463e-05, | |
| "loss": 4.3829, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.646237493964411e-05, | |
| "loss": 4.3845, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.645398899213359e-05, | |
| "loss": 4.3823, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.644560304462307e-05, | |
| "loss": 4.3867, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6437217097112554e-05, | |
| "loss": 4.3766, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642884752840576e-05, | |
| "loss": 4.3788, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642046158089524e-05, | |
| "loss": 4.3918, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.641209201218845e-05, | |
| "loss": 4.368, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.640370606467793e-05, | |
| "loss": 4.3769, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.639532011716741e-05, | |
| "loss": 4.3787, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.638693416965689e-05, | |
| "loss": 4.3915, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637854822214637e-05, | |
| "loss": 4.3809, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637016227463585e-05, | |
| "loss": 4.3763, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.636177632712533e-05, | |
| "loss": 4.3662, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.635339037961481e-05, | |
| "loss": 4.3688, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.634500443210429e-05, | |
| "loss": 4.3784, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.633661848459377e-05, | |
| "loss": 4.3859, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.632823253708325e-05, | |
| "loss": 4.3803, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.631984658957273e-05, | |
| "loss": 4.37, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.631147702086595e-05, | |
| "loss": 4.3667, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.630309107335542e-05, | |
| "loss": 4.3788, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.62947051258449e-05, | |
| "loss": 4.3685, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.628631917833438e-05, | |
| "loss": 4.3711, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6277949609627597e-05, | |
| "loss": 4.3831, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.626956366211707e-05, | |
| "loss": 4.3659, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.626117771460655e-05, | |
| "loss": 4.3702, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.625279176709603e-05, | |
| "loss": 4.365, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.396077632904053, | |
| "eval_runtime": 542.5284, | |
| "eval_samples_per_second": 703.357, | |
| "eval_steps_per_second": 21.98, | |
| "step": 228957 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6244422198389246e-05, | |
| "loss": 4.3755, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.623605262968246e-05, | |
| "loss": 4.3765, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.622766668217194e-05, | |
| "loss": 4.3623, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.621928073466142e-05, | |
| "loss": 4.3622, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6210894787150895e-05, | |
| "loss": 4.3602, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6202508839640374e-05, | |
| "loss": 4.3561, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.619413927093359e-05, | |
| "loss": 4.3598, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.618575332342307e-05, | |
| "loss": 4.3564, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6177367375912543e-05, | |
| "loss": 4.3584, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616899780720576e-05, | |
| "loss": 4.3672, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616061185969524e-05, | |
| "loss": 4.3623, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.615222591218472e-05, | |
| "loss": 4.365, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61438399646742e-05, | |
| "loss": 4.3591, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.613545401716368e-05, | |
| "loss": 4.352, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.612706806965316e-05, | |
| "loss": 4.3534, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611868212214264e-05, | |
| "loss": 4.3371, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611029617463212e-05, | |
| "loss": 4.3493, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61019102271216e-05, | |
| "loss": 4.3495, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.609352427961108e-05, | |
| "loss": 4.3494, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.608513833210056e-05, | |
| "loss": 4.3631, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.607675238459004e-05, | |
| "loss": 4.3533, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606838281588325e-05, | |
| "loss": 4.3557, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606001324717646e-05, | |
| "loss": 4.3519, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.605162729966594e-05, | |
| "loss": 4.3478, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.604324135215542e-05, | |
| "loss": 4.347, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.60348554046449e-05, | |
| "loss": 4.3505, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.602646945713438e-05, | |
| "loss": 4.3468, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.601809988842759e-05, | |
| "loss": 4.3421, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.60097303197208e-05, | |
| "loss": 4.3435, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.600134437221028e-05, | |
| "loss": 4.333, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.599295842469976e-05, | |
| "loss": 4.3419, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.598457247718924e-05, | |
| "loss": 4.3485, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.597618652967872e-05, | |
| "loss": 4.3429, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.59678005821682e-05, | |
| "loss": 4.3427, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595941463465768e-05, | |
| "loss": 4.3401, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595102868714716e-05, | |
| "loss": 4.3455, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.594264273963664e-05, | |
| "loss": 4.3423, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.593425679212612e-05, | |
| "loss": 4.3506, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.592588722341933e-05, | |
| "loss": 4.3405, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.591750127590882e-05, | |
| "loss": 4.3335, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.59091153283983e-05, | |
| "loss": 4.3399, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.590072938088778e-05, | |
| "loss": 4.3233, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.589234343337726e-05, | |
| "loss": 4.3439, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.588395748586673e-05, | |
| "loss": 4.3425, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5875587917159946e-05, | |
| "loss": 4.3412, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5867201969649426e-05, | |
| "loss": 4.3304, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5858816022138906e-05, | |
| "loss": 4.3311, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.585043007462838e-05, | |
| "loss": 4.3427, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.584204412711786e-05, | |
| "loss": 4.32, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.583365817960734e-05, | |
| "loss": 4.3265, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.582527223209682e-05, | |
| "loss": 4.3256, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.58168862845863e-05, | |
| "loss": 4.3417, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5808516715879515e-05, | |
| "loss": 4.3155, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5800130768368995e-05, | |
| "loss": 4.3465, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5791761199662204e-05, | |
| "loss": 4.318, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5783375252151684e-05, | |
| "loss": 4.3191, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5774989304641164e-05, | |
| "loss": 4.3321, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5766603357130644e-05, | |
| "loss": 4.3194, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5758217409620124e-05, | |
| "loss": 4.3317, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5749831462109604e-05, | |
| "loss": 4.3333, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5741445514599084e-05, | |
| "loss": 4.3132, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5733059567088564e-05, | |
| "loss": 4.3205, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5724673619578044e-05, | |
| "loss": 4.3232, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.571630405087125e-05, | |
| "loss": 4.3278, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.570791810336074e-05, | |
| "loss": 4.316, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.569953215585022e-05, | |
| "loss": 4.3201, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.56911462083397e-05, | |
| "loss": 4.3017, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.568276026082918e-05, | |
| "loss": 4.3306, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.567439069212239e-05, | |
| "loss": 4.3176, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.566600474461187e-05, | |
| "loss": 4.3104, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.565761879710135e-05, | |
| "loss": 4.3186, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.564923284959083e-05, | |
| "loss": 4.3178, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.564086328088404e-05, | |
| "loss": 4.3171, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.563247733337352e-05, | |
| "loss": 4.3219, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5624091385863e-05, | |
| "loss": 4.328, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5615721817156206e-05, | |
| "loss": 4.3149, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.560733586964569e-05, | |
| "loss": 4.3097, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.559894992213517e-05, | |
| "loss": 4.3124, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.559056397462465e-05, | |
| "loss": 4.3135, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.558217802711413e-05, | |
| "loss": 4.325, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.557380845840734e-05, | |
| "loss": 4.3174, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.556542251089682e-05, | |
| "loss": 4.3116, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.55570365633863e-05, | |
| "loss": 4.3003, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554865061587578e-05, | |
| "loss": 4.307, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554026466836526e-05, | |
| "loss": 4.3075, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.553189509965847e-05, | |
| "loss": 4.3119, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.552350915214795e-05, | |
| "loss": 4.312, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.551512320463743e-05, | |
| "loss": 4.3103, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.550673725712691e-05, | |
| "loss": 4.3019, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.549835130961639e-05, | |
| "loss": 4.3121, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.548996536210587e-05, | |
| "loss": 4.3028, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.548157941459535e-05, | |
| "loss": 4.3048, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.547319346708483e-05, | |
| "loss": 4.3085, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.546482389837804e-05, | |
| "loss": 4.2943, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.545643795086752e-05, | |
| "loss": 4.293, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5448052003357e-05, | |
| "loss": 4.3089, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.543966605584648e-05, | |
| "loss": 4.3018, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.543128010833596e-05, | |
| "loss": 4.3075, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.542289416082544e-05, | |
| "loss": 4.3057, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.541450821331492e-05, | |
| "loss": 4.2974, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.54061222658044e-05, | |
| "loss": 4.2974, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.539773631829388e-05, | |
| "loss": 4.3064, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5389366749587095e-05, | |
| "loss": 4.3058, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5380997180880304e-05, | |
| "loss": 4.2891, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5372611233369784e-05, | |
| "loss": 4.3072, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5364225285859264e-05, | |
| "loss": 4.2809, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5355839338348744e-05, | |
| "loss": 4.2901, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5347453390838224e-05, | |
| "loss": 4.3013, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533908382213143e-05, | |
| "loss": 4.2847, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533069787462091e-05, | |
| "loss": 4.2987, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.532231192711039e-05, | |
| "loss": 4.2833, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.531392597959987e-05, | |
| "loss": 4.2822, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.530554003208935e-05, | |
| "loss": 4.2845, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.529715408457883e-05, | |
| "loss": 4.3069, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528876813706831e-05, | |
| "loss": 4.2905, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528038218955779e-05, | |
| "loss": 4.2798, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.527201262085101e-05, | |
| "loss": 4.2804, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.526362667334049e-05, | |
| "loss": 4.2896, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.525524072582997e-05, | |
| "loss": 4.2945, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.524685477831945e-05, | |
| "loss": 4.294, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.523848520961266e-05, | |
| "loss": 4.2845, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.523009926210214e-05, | |
| "loss": 4.2954, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.522171331459162e-05, | |
| "loss": 4.2849, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.52133273670811e-05, | |
| "loss": 4.2873, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.520494141957057e-05, | |
| "loss": 4.2846, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.519657185086379e-05, | |
| "loss": 4.289, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.518818590335327e-05, | |
| "loss": 4.2802, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517979995584275e-05, | |
| "loss": 4.2825, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517141400833223e-05, | |
| "loss": 4.2937, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.516304443962544e-05, | |
| "loss": 4.274, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.515465849211492e-05, | |
| "loss": 4.2833, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.51462725446044e-05, | |
| "loss": 4.2867, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5137886597093876e-05, | |
| "loss": 4.2923, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5129500649583356e-05, | |
| "loss": 4.285, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512113108087657e-05, | |
| "loss": 4.2832, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5112745133366045e-05, | |
| "loss": 4.2724, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5104359185855525e-05, | |
| "loss": 4.275, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5095973238345005e-05, | |
| "loss": 4.2859, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5087587290834485e-05, | |
| "loss": 4.2906, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.507920134332397e-05, | |
| "loss": 4.2893, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.507081539581345e-05, | |
| "loss": 4.2774, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.506242944830293e-05, | |
| "loss": 4.2719, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.505405987959614e-05, | |
| "loss": 4.284, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.504567393208562e-05, | |
| "loss": 4.2761, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.503730436337883e-05, | |
| "loss": 4.2799, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.502891841586831e-05, | |
| "loss": 4.2903, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.502053246835779e-05, | |
| "loss": 4.2789, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.501214652084727e-05, | |
| "loss": 4.2772, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.500376057333675e-05, | |
| "loss": 4.2747, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.312550067901611, | |
| "eval_runtime": 539.9506, | |
| "eval_samples_per_second": 706.715, | |
| "eval_steps_per_second": 22.085, | |
| "step": 305276 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.499537462582623e-05, | |
| "loss": 4.2907, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.498698867831571e-05, | |
| "loss": 4.285, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.497860273080519e-05, | |
| "loss": 4.271, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.497021678329467e-05, | |
| "loss": 4.2698, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.496183083578415e-05, | |
| "loss": 4.2729, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4953444888273636e-05, | |
| "loss": 4.2673, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4945058940763116e-05, | |
| "loss": 4.2706, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.493667299325259e-05, | |
| "loss": 4.2644, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.492828704574207e-05, | |
| "loss": 4.271, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4919917477035285e-05, | |
| "loss": 4.2824, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4911531529524765e-05, | |
| "loss": 4.2702, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.490314558201424e-05, | |
| "loss": 4.2704, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.489475963450372e-05, | |
| "loss": 4.2742, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.48863736869932e-05, | |
| "loss": 4.2665, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.487798773948268e-05, | |
| "loss": 4.2663, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486960179197216e-05, | |
| "loss": 4.2482, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486121584446164e-05, | |
| "loss": 4.2613, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.485282989695112e-05, | |
| "loss": 4.2615, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4844443949440604e-05, | |
| "loss": 4.2624, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4836058001930084e-05, | |
| "loss": 4.2818, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4827672054419564e-05, | |
| "loss": 4.2641, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.481930248571277e-05, | |
| "loss": 4.2712, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.481091653820225e-05, | |
| "loss": 4.2617, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.480253059069173e-05, | |
| "loss": 4.2671, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.479414464318121e-05, | |
| "loss": 4.2569, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.478575869567069e-05, | |
| "loss": 4.2656, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.47773891269639e-05, | |
| "loss": 4.2629, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.476900317945338e-05, | |
| "loss": 4.2602, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.476061723194286e-05, | |
| "loss": 4.2574, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.475223128443234e-05, | |
| "loss": 4.2495, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.474384533692182e-05, | |
| "loss": 4.2522, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.47354593894113e-05, | |
| "loss": 4.2654, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.472708982070452e-05, | |
| "loss": 4.2596, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4718703873194e-05, | |
| "loss": 4.2611, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.471031792568348e-05, | |
| "loss": 4.2583, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.470194835697669e-05, | |
| "loss": 4.2621, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.469356240946617e-05, | |
| "loss": 4.2599, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.468517646195565e-05, | |
| "loss": 4.267, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.467679051444512e-05, | |
| "loss": 4.2593, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.46684045669346e-05, | |
| "loss": 4.2502, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466001861942408e-05, | |
| "loss": 4.2618, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.465163267191356e-05, | |
| "loss": 4.2421, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.464324672440304e-05, | |
| "loss": 4.2606, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4634860776892526e-05, | |
| "loss": 4.2583, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4626474829382006e-05, | |
| "loss": 4.2646, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4618088881871486e-05, | |
| "loss": 4.2511, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4609702934360966e-05, | |
| "loss": 4.2457, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4601333365654175e-05, | |
| "loss": 4.2612, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4592947418143655e-05, | |
| "loss": 4.2439, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4584561470633135e-05, | |
| "loss": 4.2431, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4576175523122615e-05, | |
| "loss": 4.2493, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4567805954415824e-05, | |
| "loss": 4.2578, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4559420006905304e-05, | |
| "loss": 4.2404, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4551034059394784e-05, | |
| "loss": 4.2598, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4542648111884264e-05, | |
| "loss": 4.2395, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.453427854317748e-05, | |
| "loss": 4.2429, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.452589259566696e-05, | |
| "loss": 4.252, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.451750664815644e-05, | |
| "loss": 4.2372, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.450912070064592e-05, | |
| "loss": 4.2559, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.450075113193913e-05, | |
| "loss": 4.2578, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.449236518442861e-05, | |
| "loss": 4.2335, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.448397923691809e-05, | |
| "loss": 4.24, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.447559328940757e-05, | |
| "loss": 4.2465, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.446720734189705e-05, | |
| "loss": 4.2534, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445882139438653e-05, | |
| "loss": 4.2363, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445043544687601e-05, | |
| "loss": 4.2405, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.444204949936548e-05, | |
| "loss": 4.2268, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.443366355185496e-05, | |
| "loss": 4.2492, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.442527760434444e-05, | |
| "loss": 4.2411, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.441689165683393e-05, | |
| "loss": 4.2324, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440850570932341e-05, | |
| "loss": 4.2433, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440013614061662e-05, | |
| "loss": 4.2409, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.43917501931061e-05, | |
| "loss": 4.2384, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4383380624399314e-05, | |
| "loss": 4.2468, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.437499467688879e-05, | |
| "loss": 4.2498, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.436660872937827e-05, | |
| "loss": 4.2401, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.435823916067148e-05, | |
| "loss": 4.2374, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4349853213160956e-05, | |
| "loss": 4.2377, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4341467265650436e-05, | |
| "loss": 4.236, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4333081318139916e-05, | |
| "loss": 4.2541, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4324695370629396e-05, | |
| "loss": 4.2409, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.431630942311888e-05, | |
| "loss": 4.2386, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.430792347560836e-05, | |
| "loss": 4.223, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.429955390690157e-05, | |
| "loss": 4.2365, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.429116795939105e-05, | |
| "loss": 4.229, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.428278201188053e-05, | |
| "loss": 4.2427, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.427439606437001e-05, | |
| "loss": 4.2354, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.426601011685949e-05, | |
| "loss": 4.2362, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.425762416934897e-05, | |
| "loss": 4.2289, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424923822183845e-05, | |
| "loss": 4.2417, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424085227432793e-05, | |
| "loss": 4.2278, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.423246632681741e-05, | |
| "loss": 4.2312, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.422408037930689e-05, | |
| "loss": 4.2396, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.42157108106001e-05, | |
| "loss": 4.2173, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.420732486308958e-05, | |
| "loss": 4.221, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419893891557907e-05, | |
| "loss": 4.2373, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419055296806855e-05, | |
| "loss": 4.2267, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.418216702055803e-05, | |
| "loss": 4.2427, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.41737810730475e-05, | |
| "loss": 4.2324, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4165411504340716e-05, | |
| "loss": 4.2216, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4157025556830196e-05, | |
| "loss": 4.2268, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4148639609319676e-05, | |
| "loss": 4.2344, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.414025366180915e-05, | |
| "loss": 4.2324, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4131884093102365e-05, | |
| "loss": 4.2201, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4123498145591845e-05, | |
| "loss": 4.2356, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.411511219808132e-05, | |
| "loss": 4.2097, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4106726250570805e-05, | |
| "loss": 4.2194, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4098340303060284e-05, | |
| "loss": 4.2314, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4089954355549764e-05, | |
| "loss": 4.2133, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4081568408039244e-05, | |
| "loss": 4.2286, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4073182460528724e-05, | |
| "loss": 4.2225, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4064812891821933e-05, | |
| "loss": 4.206, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.405644332311514e-05, | |
| "loss": 4.2164, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.404805737560462e-05, | |
| "loss": 4.2373, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.40396714280941e-05, | |
| "loss": 4.2204, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403128548058358e-05, | |
| "loss": 4.2154, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.402289953307306e-05, | |
| "loss": 4.2117, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.401451358556254e-05, | |
| "loss": 4.2194, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.400614401685576e-05, | |
| "loss": 4.2276, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.399775806934524e-05, | |
| "loss": 4.2257, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.398937212183472e-05, | |
| "loss": 4.2183, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.39809861743242e-05, | |
| "loss": 4.2233, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.397260022681368e-05, | |
| "loss": 4.2194, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.396421427930316e-05, | |
| "loss": 4.2181, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.395582833179264e-05, | |
| "loss": 4.2191, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.394744238428212e-05, | |
| "loss": 4.2203, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.393907281557533e-05, | |
| "loss": 4.2095, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3930703246868536e-05, | |
| "loss": 4.2212, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3922317299358016e-05, | |
| "loss": 4.2219, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3913931351847496e-05, | |
| "loss": 4.2097, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3905545404336976e-05, | |
| "loss": 4.2134, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3897159456826456e-05, | |
| "loss": 4.2218, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3888773509315936e-05, | |
| "loss": 4.2232, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.388038756180542e-05, | |
| "loss": 4.223, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.387201799309863e-05, | |
| "loss": 4.2165, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.386363204558811e-05, | |
| "loss": 4.2051, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.385524609807759e-05, | |
| "loss": 4.2092, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.384686015056707e-05, | |
| "loss": 4.2177, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.383847420305655e-05, | |
| "loss": 4.224, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.383008825554603e-05, | |
| "loss": 4.2262, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3821702308035505e-05, | |
| "loss": 4.2145, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3813316360524985e-05, | |
| "loss": 4.2052, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.38049467918182e-05, | |
| "loss": 4.2189, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.379656084430768e-05, | |
| "loss": 4.2082, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.378817489679716e-05, | |
| "loss": 4.2121, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.377978894928664e-05, | |
| "loss": 4.2253, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.377140300177612e-05, | |
| "loss": 4.2169, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3763049811873065e-05, | |
| "loss": 4.207, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3754663864362545e-05, | |
| "loss": 4.212, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.255102634429932, | |
| "eval_runtime": 527.9768, | |
| "eval_samples_per_second": 722.742, | |
| "eval_steps_per_second": 22.586, | |
| "step": 381595 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3746277916852025e-05, | |
| "loss": 4.2194, | |
| "step": 381952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3737908348145234e-05, | |
| "loss": 4.2235, | |
| "step": 382464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3729522400634714e-05, | |
| "loss": 4.2062, | |
| "step": 382976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3721136453124194e-05, | |
| "loss": 4.2042, | |
| "step": 383488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3712750505613674e-05, | |
| "loss": 4.2124, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3704364558103154e-05, | |
| "loss": 4.2036, | |
| "step": 384512 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.369597861059263e-05, | |
| "loss": 4.2058, | |
| "step": 385024 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3687592663082114e-05, | |
| "loss": 4.1959, | |
| "step": 385536 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3679206715571594e-05, | |
| "loss": 4.2098, | |
| "step": 386048 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367083714686481e-05, | |
| "loss": 4.2244, | |
| "step": 386560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.366245119935428e-05, | |
| "loss": 4.2032, | |
| "step": 387072 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.365406525184376e-05, | |
| "loss": 4.208, | |
| "step": 387584 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.364567930433324e-05, | |
| "loss": 4.2142, | |
| "step": 388096 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.363729335682272e-05, | |
| "loss": 4.2021, | |
| "step": 388608 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.36289074093122e-05, | |
| "loss": 4.2059, | |
| "step": 389120 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.362052146180168e-05, | |
| "loss": 4.1862, | |
| "step": 389632 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.361213551429116e-05, | |
| "loss": 4.1967, | |
| "step": 390144 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.360374956678064e-05, | |
| "loss": 4.198, | |
| "step": 390656 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.359539637687758e-05, | |
| "loss": 4.1986, | |
| "step": 391168 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.358701042936707e-05, | |
| "loss": 4.2218, | |
| "step": 391680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357862448185655e-05, | |
| "loss": 4.2006, | |
| "step": 392192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357023853434603e-05, | |
| "loss": 4.2103, | |
| "step": 392704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.356186896563924e-05, | |
| "loss": 4.2008, | |
| "step": 393216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.355348301812872e-05, | |
| "loss": 4.2055, | |
| "step": 393728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.35450970706182e-05, | |
| "loss": 4.194, | |
| "step": 394240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3536711123107677e-05, | |
| "loss": 4.2015, | |
| "step": 394752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3528325175597157e-05, | |
| "loss": 4.2036, | |
| "step": 395264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3519939228086636e-05, | |
| "loss": 4.1989, | |
| "step": 395776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3511553280576116e-05, | |
| "loss": 4.1959, | |
| "step": 396288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3503167333065596e-05, | |
| "loss": 4.1919, | |
| "step": 396800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3494797764358805e-05, | |
| "loss": 4.1874, | |
| "step": 397312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3486411816848285e-05, | |
| "loss": 4.2068, | |
| "step": 397824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3478025869337765e-05, | |
| "loss": 4.2016, | |
| "step": 398336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.346963992182725e-05, | |
| "loss": 4.1978, | |
| "step": 398848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.346125397431673e-05, | |
| "loss": 4.1971, | |
| "step": 399360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.345286802680621e-05, | |
| "loss": 4.201, | |
| "step": 399872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.344451483690315e-05, | |
| "loss": 4.2036, | |
| "step": 400384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.343612888939263e-05, | |
| "loss": 4.2028, | |
| "step": 400896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.342774294188211e-05, | |
| "loss": 4.1994, | |
| "step": 401408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.341935699437159e-05, | |
| "loss": 4.1902, | |
| "step": 401920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.341097104686107e-05, | |
| "loss": 4.1989, | |
| "step": 402432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.340258509935055e-05, | |
| "loss": 4.1845, | |
| "step": 402944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.339419915184003e-05, | |
| "loss": 4.1999, | |
| "step": 403456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.338581320432951e-05, | |
| "loss": 4.1965, | |
| "step": 403968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.337744363562272e-05, | |
| "loss": 4.2055, | |
| "step": 404480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.33690576881122e-05, | |
| "loss": 4.1929, | |
| "step": 404992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3360671740601686e-05, | |
| "loss": 4.1865, | |
| "step": 405504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3352285793091166e-05, | |
| "loss": 4.2043, | |
| "step": 406016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.334389984558064e-05, | |
| "loss": 4.1816, | |
| "step": 406528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.333551389807012e-05, | |
| "loss": 4.1885, | |
| "step": 407040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.33271279505596e-05, | |
| "loss": 4.1906, | |
| "step": 407552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331874200304908e-05, | |
| "loss": 4.197, | |
| "step": 408064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331035605553856e-05, | |
| "loss": 4.182, | |
| "step": 408576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.330197010802804e-05, | |
| "loss": 4.1979, | |
| "step": 409088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.329358416051752e-05, | |
| "loss": 4.1855, | |
| "step": 409600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.328521459181073e-05, | |
| "loss": 4.1833, | |
| "step": 410112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.327682864430021e-05, | |
| "loss": 4.1956, | |
| "step": 410624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326844269678969e-05, | |
| "loss": 4.1812, | |
| "step": 411136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3260073128082904e-05, | |
| "loss": 4.1968, | |
| "step": 411648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3251687180572383e-05, | |
| "loss": 4.1999, | |
| "step": 412160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3243301233061863e-05, | |
| "loss": 4.1813, | |
| "step": 412672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.323491528555134e-05, | |
| "loss": 4.1824, | |
| "step": 413184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.322652933804082e-05, | |
| "loss": 4.185, | |
| "step": 413696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.32181433905303e-05, | |
| "loss": 4.1999, | |
| "step": 414208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320975744301978e-05, | |
| "loss": 4.1813, | |
| "step": 414720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320137149550926e-05, | |
| "loss": 4.1862, | |
| "step": 415232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.319300192680247e-05, | |
| "loss": 4.1687, | |
| "step": 415744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.318461597929195e-05, | |
| "loss": 4.1937, | |
| "step": 416256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.317623003178143e-05, | |
| "loss": 4.1834, | |
| "step": 416768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.316784408427091e-05, | |
| "loss": 4.1772, | |
| "step": 417280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315947451556412e-05, | |
| "loss": 4.1846, | |
| "step": 417792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315108856805361e-05, | |
| "loss": 4.1867, | |
| "step": 418304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.314271899934682e-05, | |
| "loss": 4.182, | |
| "step": 418816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.31343330518363e-05, | |
| "loss": 4.1875, | |
| "step": 419328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.312594710432578e-05, | |
| "loss": 4.1924, | |
| "step": 419840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.311756115681526e-05, | |
| "loss": 4.1835, | |
| "step": 420352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310917520930474e-05, | |
| "loss": 4.1805, | |
| "step": 420864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310078926179422e-05, | |
| "loss": 4.1854, | |
| "step": 421376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.30924033142837e-05, | |
| "loss": 4.1791, | |
| "step": 421888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.308401736677318e-05, | |
| "loss": 4.1948, | |
| "step": 422400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3075647798066386e-05, | |
| "loss": 4.1837, | |
| "step": 422912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3067261850555866e-05, | |
| "loss": 4.1866, | |
| "step": 423424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3058875903045346e-05, | |
| "loss": 4.1696, | |
| "step": 423936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.305050633433856e-05, | |
| "loss": 4.1799, | |
| "step": 424448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.304212038682804e-05, | |
| "loss": 4.1705, | |
| "step": 424960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.303373443931752e-05, | |
| "loss": 4.1886, | |
| "step": 425472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3025348491807e-05, | |
| "loss": 4.181, | |
| "step": 425984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3016962544296475e-05, | |
| "loss": 4.1812, | |
| "step": 426496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.300859297558969e-05, | |
| "loss": 4.1772, | |
| "step": 427008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.300020702807917e-05, | |
| "loss": 4.1881, | |
| "step": 427520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.299182108056865e-05, | |
| "loss": 4.1665, | |
| "step": 428032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2983435133058124e-05, | |
| "loss": 4.176, | |
| "step": 428544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2975049185547604e-05, | |
| "loss": 4.187, | |
| "step": 429056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.296667961684082e-05, | |
| "loss": 4.1636, | |
| "step": 429568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.29582936693303e-05, | |
| "loss": 4.1669, | |
| "step": 430080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294990772181978e-05, | |
| "loss": 4.1824, | |
| "step": 430592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294152177430926e-05, | |
| "loss": 4.1708, | |
| "step": 431104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.293313582679874e-05, | |
| "loss": 4.1915, | |
| "step": 431616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.292474987928822e-05, | |
| "loss": 4.1768, | |
| "step": 432128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.291638031058143e-05, | |
| "loss": 4.1684, | |
| "step": 432640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.290799436307091e-05, | |
| "loss": 4.1704, | |
| "step": 433152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289960841556039e-05, | |
| "loss": 4.1839, | |
| "step": 433664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289122246804987e-05, | |
| "loss": 4.1794, | |
| "step": 434176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.288285289934308e-05, | |
| "loss": 4.1663, | |
| "step": 434688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.287446695183256e-05, | |
| "loss": 4.1795, | |
| "step": 435200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.286608100432204e-05, | |
| "loss": 4.1565, | |
| "step": 435712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.285769505681152e-05, | |
| "loss": 4.166, | |
| "step": 436224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2849309109301e-05, | |
| "loss": 4.1795, | |
| "step": 436736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284092316179048e-05, | |
| "loss": 4.1616, | |
| "step": 437248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2832537214279964e-05, | |
| "loss": 4.1742, | |
| "step": 437760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2824151266769444e-05, | |
| "loss": 4.1754, | |
| "step": 438272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.281578169806265e-05, | |
| "loss": 4.1512, | |
| "step": 438784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.280739575055213e-05, | |
| "loss": 4.1662, | |
| "step": 439296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.279900980304161e-05, | |
| "loss": 4.182, | |
| "step": 439808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.279062385553109e-05, | |
| "loss": 4.1729, | |
| "step": 440320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.27822542868243e-05, | |
| "loss": 4.1604, | |
| "step": 440832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.277386833931378e-05, | |
| "loss": 4.1612, | |
| "step": 441344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.276548239180326e-05, | |
| "loss": 4.1679, | |
| "step": 441856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.275711282309647e-05, | |
| "loss": 4.1738, | |
| "step": 442368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.274872687558595e-05, | |
| "loss": 4.1756, | |
| "step": 442880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.274034092807543e-05, | |
| "loss": 4.1649, | |
| "step": 443392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.273195498056492e-05, | |
| "loss": 4.1706, | |
| "step": 443904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.27235690330544e-05, | |
| "loss": 4.1655, | |
| "step": 444416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.271518308554388e-05, | |
| "loss": 4.1695, | |
| "step": 444928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.270679713803336e-05, | |
| "loss": 4.165, | |
| "step": 445440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.269841119052284e-05, | |
| "loss": 4.1694, | |
| "step": 445952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2690041621816046e-05, | |
| "loss": 4.1617, | |
| "step": 446464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2681655674305526e-05, | |
| "loss": 4.1696, | |
| "step": 446976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2673269726795006e-05, | |
| "loss": 4.1703, | |
| "step": 447488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2664883779284486e-05, | |
| "loss": 4.1566, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2656514210577695e-05, | |
| "loss": 4.1632, | |
| "step": 448512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2648128263067175e-05, | |
| "loss": 4.1735, | |
| "step": 449024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2639742315556655e-05, | |
| "loss": 4.1716, | |
| "step": 449536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2631356368046135e-05, | |
| "loss": 4.1707, | |
| "step": 450048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.262300317814308e-05, | |
| "loss": 4.1651, | |
| "step": 450560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.261461723063256e-05, | |
| "loss": 4.1523, | |
| "step": 451072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.260623128312204e-05, | |
| "loss": 4.1579, | |
| "step": 451584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.259784533561152e-05, | |
| "loss": 4.1672, | |
| "step": 452096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2589459388101e-05, | |
| "loss": 4.1788, | |
| "step": 452608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.258107344059048e-05, | |
| "loss": 4.1734, | |
| "step": 453120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.257268749307996e-05, | |
| "loss": 4.1592, | |
| "step": 453632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.256430154556943e-05, | |
| "loss": 4.1624, | |
| "step": 454144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.255591559805891e-05, | |
| "loss": 4.1659, | |
| "step": 454656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.254754602935213e-05, | |
| "loss": 4.1599, | |
| "step": 455168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253916008184161e-05, | |
| "loss": 4.1644, | |
| "step": 455680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253077413433109e-05, | |
| "loss": 4.1777, | |
| "step": 456192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.252238818682057e-05, | |
| "loss": 4.1668, | |
| "step": 456704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2514018618113785e-05, | |
| "loss": 4.1544, | |
| "step": 457216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2505649049406994e-05, | |
| "loss": 4.1642, | |
| "step": 457728 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.213168144226074, | |
| "eval_runtime": 549.6612, | |
| "eval_samples_per_second": 694.229, | |
| "eval_steps_per_second": 21.695, | |
| "step": 457914 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2497263101896474e-05, | |
| "loss": 4.1757, | |
| "step": 458240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2488877154385954e-05, | |
| "loss": 4.174, | |
| "step": 458752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2480491206875434e-05, | |
| "loss": 4.1595, | |
| "step": 459264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.247210525936491e-05, | |
| "loss": 4.1568, | |
| "step": 459776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.246371931185439e-05, | |
| "loss": 4.1609, | |
| "step": 460288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.245533336434387e-05, | |
| "loss": 4.1544, | |
| "step": 460800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.244694741683335e-05, | |
| "loss": 4.1574, | |
| "step": 461312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243856146932283e-05, | |
| "loss": 4.1478, | |
| "step": 461824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243017552181231e-05, | |
| "loss": 4.1577, | |
| "step": 462336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2421789574301793e-05, | |
| "loss": 4.1756, | |
| "step": 462848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2413420005595e-05, | |
| "loss": 4.1564, | |
| "step": 463360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.240503405808448e-05, | |
| "loss": 4.1565, | |
| "step": 463872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.239664811057396e-05, | |
| "loss": 4.1675, | |
| "step": 464384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.238826216306344e-05, | |
| "loss": 4.1528, | |
| "step": 464896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.237987621555292e-05, | |
| "loss": 4.1602, | |
| "step": 465408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.23714902680424e-05, | |
| "loss": 4.1374, | |
| "step": 465920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.236310432053188e-05, | |
| "loss": 4.1452, | |
| "step": 466432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.235471837302136e-05, | |
| "loss": 4.1529, | |
| "step": 466944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.234633242551084e-05, | |
| "loss": 4.1506, | |
| "step": 467456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2337946478000315e-05, | |
| "loss": 4.1756, | |
| "step": 467968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2329560530489795e-05, | |
| "loss": 4.1555, | |
| "step": 468480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2321174582979275e-05, | |
| "loss": 4.1657, | |
| "step": 468992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.231278863546876e-05, | |
| "loss": 4.1501, | |
| "step": 469504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.230440268795824e-05, | |
| "loss": 4.1557, | |
| "step": 470016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.229601674044772e-05, | |
| "loss": 4.1447, | |
| "step": 470528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.22876307929372e-05, | |
| "loss": 4.1559, | |
| "step": 471040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.227926122423041e-05, | |
| "loss": 4.1558, | |
| "step": 471552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.227087527671989e-05, | |
| "loss": 4.1473, | |
| "step": 472064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.22625057080131e-05, | |
| "loss": 4.1556, | |
| "step": 472576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.225411976050258e-05, | |
| "loss": 4.1396, | |
| "step": 473088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.224573381299206e-05, | |
| "loss": 4.138, | |
| "step": 473600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.223734786548154e-05, | |
| "loss": 4.1594, | |
| "step": 474112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.222896191797102e-05, | |
| "loss": 4.154, | |
| "step": 474624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.222059234926423e-05, | |
| "loss": 4.1544, | |
| "step": 475136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.221220640175371e-05, | |
| "loss": 4.1466, | |
| "step": 475648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2203820454243196e-05, | |
| "loss": 4.1573, | |
| "step": 476160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2195434506732676e-05, | |
| "loss": 4.1523, | |
| "step": 476672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2187048559222156e-05, | |
| "loss": 4.1587, | |
| "step": 477184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2178662611711635e-05, | |
| "loss": 4.1513, | |
| "step": 477696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2170276664201115e-05, | |
| "loss": 4.1456, | |
| "step": 478208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2161890716690595e-05, | |
| "loss": 4.1531, | |
| "step": 478720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2153521147983804e-05, | |
| "loss": 4.1362, | |
| "step": 479232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2145135200473284e-05, | |
| "loss": 4.153, | |
| "step": 479744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2136749252962764e-05, | |
| "loss": 4.1528, | |
| "step": 480256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2128363305452244e-05, | |
| "loss": 4.1589, | |
| "step": 480768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2119993736745453e-05, | |
| "loss": 4.1498, | |
| "step": 481280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.211160778923493e-05, | |
| "loss": 4.1398, | |
| "step": 481792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.210322184172441e-05, | |
| "loss": 4.1572, | |
| "step": 482304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.209483589421389e-05, | |
| "loss": 4.1396, | |
| "step": 482816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.208644994670338e-05, | |
| "loss": 4.1402, | |
| "step": 483328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.207808037799659e-05, | |
| "loss": 4.1444, | |
| "step": 483840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.206969443048607e-05, | |
| "loss": 4.1496, | |
| "step": 484352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.206130848297555e-05, | |
| "loss": 4.1406, | |
| "step": 484864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.205292253546503e-05, | |
| "loss": 4.1476, | |
| "step": 485376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.204453658795451e-05, | |
| "loss": 4.1407, | |
| "step": 485888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.203615064044398e-05, | |
| "loss": 4.143, | |
| "step": 486400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.20277810717372e-05, | |
| "loss": 4.1455, | |
| "step": 486912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201939512422668e-05, | |
| "loss": 4.1391, | |
| "step": 487424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201100917671615e-05, | |
| "loss": 4.1493, | |
| "step": 487936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.200262322920563e-05, | |
| "loss": 4.1542, | |
| "step": 488448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.199423728169512e-05, | |
| "loss": 4.136, | |
| "step": 488960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.19858513341846e-05, | |
| "loss": 4.1367, | |
| "step": 489472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.197746538667408e-05, | |
| "loss": 4.1364, | |
| "step": 489984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.196907943916356e-05, | |
| "loss": 4.1578, | |
| "step": 490496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.196070987045677e-05, | |
| "loss": 4.1387, | |
| "step": 491008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.195232392294625e-05, | |
| "loss": 4.1392, | |
| "step": 491520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.194393797543573e-05, | |
| "loss": 4.1261, | |
| "step": 492032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.193555202792521e-05, | |
| "loss": 4.1445, | |
| "step": 492544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1927182459218416e-05, | |
| "loss": 4.141, | |
| "step": 493056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1918796511707896e-05, | |
| "loss": 4.1339, | |
| "step": 493568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1910426943001105e-05, | |
| "loss": 4.1425, | |
| "step": 494080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1902040995490585e-05, | |
| "loss": 4.1408, | |
| "step": 494592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.189365504798007e-05, | |
| "loss": 4.1401, | |
| "step": 495104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.188526910046955e-05, | |
| "loss": 4.1423, | |
| "step": 495616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.187688315295903e-05, | |
| "loss": 4.1522, | |
| "step": 496128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.186849720544851e-05, | |
| "loss": 4.1405, | |
| "step": 496640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.186012763674172e-05, | |
| "loss": 4.137, | |
| "step": 497152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.18517416892312e-05, | |
| "loss": 4.1375, | |
| "step": 497664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.184335574172068e-05, | |
| "loss": 4.1365, | |
| "step": 498176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.183496979421016e-05, | |
| "loss": 4.1513, | |
| "step": 498688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.182658384669964e-05, | |
| "loss": 4.1397, | |
| "step": 499200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.181819789918912e-05, | |
| "loss": 4.1457, | |
| "step": 499712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.18098119516786e-05, | |
| "loss": 4.1253, | |
| "step": 500224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.180142600416808e-05, | |
| "loss": 4.1404, | |
| "step": 500736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.179305643546129e-05, | |
| "loss": 4.1247, | |
| "step": 501248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1784686866754505e-05, | |
| "loss": 4.1465, | |
| "step": 501760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1776300919243985e-05, | |
| "loss": 4.1382, | |
| "step": 502272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1767914971733465e-05, | |
| "loss": 4.139, | |
| "step": 502784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1759529024222945e-05, | |
| "loss": 4.1338, | |
| "step": 503296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1751143076712425e-05, | |
| "loss": 4.1431, | |
| "step": 503808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1742773508005634e-05, | |
| "loss": 4.1276, | |
| "step": 504320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1734387560495114e-05, | |
| "loss": 4.1331, | |
| "step": 504832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1726001612984594e-05, | |
| "loss": 4.1432, | |
| "step": 505344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1717615665474074e-05, | |
| "loss": 4.1229, | |
| "step": 505856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1709229717963554e-05, | |
| "loss": 4.1252, | |
| "step": 506368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.170086014925676e-05, | |
| "loss": 4.1364, | |
| "step": 506880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.169247420174624e-05, | |
| "loss": 4.1273, | |
| "step": 507392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.168408825423572e-05, | |
| "loss": 4.1503, | |
| "step": 507904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.16757023067252e-05, | |
| "loss": 4.1316, | |
| "step": 508416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.166731635921469e-05, | |
| "loss": 4.13, | |
| "step": 508928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165893041170417e-05, | |
| "loss": 4.1269, | |
| "step": 509440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165054446419364e-05, | |
| "loss": 4.1415, | |
| "step": 509952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.164215851668312e-05, | |
| "loss": 4.1384, | |
| "step": 510464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.163380532678007e-05, | |
| "loss": 4.127, | |
| "step": 510976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.162541937926955e-05, | |
| "loss": 4.1352, | |
| "step": 511488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.161703343175903e-05, | |
| "loss": 4.12, | |
| "step": 512000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160864748424851e-05, | |
| "loss": 4.1216, | |
| "step": 512512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160026153673799e-05, | |
| "loss": 4.1359, | |
| "step": 513024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.159187558922746e-05, | |
| "loss": 4.1241, | |
| "step": 513536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1583506020520677e-05, | |
| "loss": 4.1322, | |
| "step": 514048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1575120073010156e-05, | |
| "loss": 4.1322, | |
| "step": 514560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.156673412549964e-05, | |
| "loss": 4.1105, | |
| "step": 515072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1558348177989116e-05, | |
| "loss": 4.1228, | |
| "step": 515584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1549962230478596e-05, | |
| "loss": 4.1437, | |
| "step": 516096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.154159266177181e-05, | |
| "loss": 4.135, | |
| "step": 516608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1533206714261285e-05, | |
| "loss": 4.1157, | |
| "step": 517120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1524820766750765e-05, | |
| "loss": 4.1184, | |
| "step": 517632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1516434819240245e-05, | |
| "loss": 4.1258, | |
| "step": 518144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1508048871729725e-05, | |
| "loss": 4.1317, | |
| "step": 518656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1499662924219205e-05, | |
| "loss": 4.1345, | |
| "step": 519168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1491276976708685e-05, | |
| "loss": 4.126, | |
| "step": 519680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1482891029198165e-05, | |
| "loss": 4.1301, | |
| "step": 520192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.147452146049138e-05, | |
| "loss": 4.1264, | |
| "step": 520704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.146613551298086e-05, | |
| "loss": 4.1279, | |
| "step": 521216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.145774956547034e-05, | |
| "loss": 4.1254, | |
| "step": 521728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.144936361795982e-05, | |
| "loss": 4.1281, | |
| "step": 522240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.144099404925303e-05, | |
| "loss": 4.1232, | |
| "step": 522752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.143262448054624e-05, | |
| "loss": 4.1285, | |
| "step": 523264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.142423853303572e-05, | |
| "loss": 4.128, | |
| "step": 523776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.14158525855252e-05, | |
| "loss": 4.1141, | |
| "step": 524288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.140746663801468e-05, | |
| "loss": 4.1242, | |
| "step": 524800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.139908069050416e-05, | |
| "loss": 4.1316, | |
| "step": 525312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.139069474299364e-05, | |
| "loss": 4.1324, | |
| "step": 525824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.138230879548312e-05, | |
| "loss": 4.134, | |
| "step": 526336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1373939226776335e-05, | |
| "loss": 4.1293, | |
| "step": 526848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1365553279265815e-05, | |
| "loss": 4.1138, | |
| "step": 527360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1357167331755295e-05, | |
| "loss": 4.1149, | |
| "step": 527872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1348781384244775e-05, | |
| "loss": 4.1246, | |
| "step": 528384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1340395436734254e-05, | |
| "loss": 4.1394, | |
| "step": 528896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1332009489223734e-05, | |
| "loss": 4.1351, | |
| "step": 529408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1323639920516944e-05, | |
| "loss": 4.12, | |
| "step": 529920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1315253973006424e-05, | |
| "loss": 4.117, | |
| "step": 530432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1306868025495903e-05, | |
| "loss": 4.1277, | |
| "step": 530944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1298482077985383e-05, | |
| "loss": 4.1211, | |
| "step": 531456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.129009613047486e-05, | |
| "loss": 4.1216, | |
| "step": 531968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.128171018296434e-05, | |
| "loss": 4.1391, | |
| "step": 532480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.127332423545382e-05, | |
| "loss": 4.128, | |
| "step": 532992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.12649382879433e-05, | |
| "loss": 4.115, | |
| "step": 533504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.125655234043278e-05, | |
| "loss": 4.1266, | |
| "step": 534016 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.180727005004883, | |
| "eval_runtime": 547.4766, | |
| "eval_samples_per_second": 697.0, | |
| "eval_steps_per_second": 21.782, | |
| "step": 534233 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1248182771726e-05, | |
| "loss": 4.1259, | |
| "step": 534528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.123979682421548e-05, | |
| "loss": 4.1349, | |
| "step": 535040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.123141087670495e-05, | |
| "loss": 4.1219, | |
| "step": 535552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.122302492919443e-05, | |
| "loss": 4.1183, | |
| "step": 536064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.121465536048765e-05, | |
| "loss": 4.1184, | |
| "step": 536576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.120626941297712e-05, | |
| "loss": 4.1224, | |
| "step": 537088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.119789984427034e-05, | |
| "loss": 4.1159, | |
| "step": 537600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.118951389675982e-05, | |
| "loss": 4.1104, | |
| "step": 538112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.11811279492493e-05, | |
| "loss": 4.1176, | |
| "step": 538624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.117274200173877e-05, | |
| "loss": 4.133, | |
| "step": 539136 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1164372433031986e-05, | |
| "loss": 4.1183, | |
| "step": 539648 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.115598648552147e-05, | |
| "loss": 4.1167, | |
| "step": 540160 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.114760053801095e-05, | |
| "loss": 4.1311, | |
| "step": 540672 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1139214590500426e-05, | |
| "loss": 4.1116, | |
| "step": 541184 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1130828642989906e-05, | |
| "loss": 4.125, | |
| "step": 541696 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1122442695479386e-05, | |
| "loss": 4.1013, | |
| "step": 542208 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1114056747968866e-05, | |
| "loss": 4.1042, | |
| "step": 542720 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1105687179262075e-05, | |
| "loss": 4.1168, | |
| "step": 543232 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1097301231751555e-05, | |
| "loss": 4.1101, | |
| "step": 543744 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1088915284241035e-05, | |
| "loss": 4.1358, | |
| "step": 544256 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1080529336730515e-05, | |
| "loss": 4.1201, | |
| "step": 544768 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1072143389219995e-05, | |
| "loss": 4.1224, | |
| "step": 545280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.106377382051321e-05, | |
| "loss": 4.1139, | |
| "step": 545792 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.105538787300269e-05, | |
| "loss": 4.123, | |
| "step": 546304 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.104700192549217e-05, | |
| "loss": 4.106, | |
| "step": 546816 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.103861597798165e-05, | |
| "loss": 4.1168, | |
| "step": 547328 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.103023003047113e-05, | |
| "loss": 4.12, | |
| "step": 547840 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.102184408296061e-05, | |
| "loss": 4.115, | |
| "step": 548352 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.101347451425382e-05, | |
| "loss": 4.1119, | |
| "step": 548864 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.10050885667433e-05, | |
| "loss": 4.1065, | |
| "step": 549376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.099670261923278e-05, | |
| "loss": 4.1004, | |
| "step": 549888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.098831667172226e-05, | |
| "loss": 4.1211, | |
| "step": 550400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.097993072421174e-05, | |
| "loss": 4.1157, | |
| "step": 550912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.097154477670122e-05, | |
| "loss": 4.1157, | |
| "step": 551424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.09631588291907e-05, | |
| "loss": 4.1113, | |
| "step": 551936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.095477288168018e-05, | |
| "loss": 4.1207, | |
| "step": 552448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.094640331297339e-05, | |
| "loss": 4.1151, | |
| "step": 552960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0938017365462875e-05, | |
| "loss": 4.1162, | |
| "step": 553472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0929631417952355e-05, | |
| "loss": 4.1129, | |
| "step": 553984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0921245470441835e-05, | |
| "loss": 4.1137, | |
| "step": 554496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.091285952293131e-05, | |
| "loss": 4.1102, | |
| "step": 555008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.090447357542079e-05, | |
| "loss": 4.1014, | |
| "step": 555520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.089608762791027e-05, | |
| "loss": 4.1182, | |
| "step": 556032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.088770168039975e-05, | |
| "loss": 4.1134, | |
| "step": 556544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.087933211169296e-05, | |
| "loss": 4.1223, | |
| "step": 557056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.087096254298617e-05, | |
| "loss": 4.115, | |
| "step": 557568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.086257659547565e-05, | |
| "loss": 4.1004, | |
| "step": 558080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.085419064796513e-05, | |
| "loss": 4.1226, | |
| "step": 558592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.084580470045461e-05, | |
| "loss": 4.1017, | |
| "step": 559104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.083743513174783e-05, | |
| "loss": 4.1089, | |
| "step": 559616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.082904918423731e-05, | |
| "loss": 4.1038, | |
| "step": 560128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.082066323672678e-05, | |
| "loss": 4.1153, | |
| "step": 560640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.081227728921626e-05, | |
| "loss": 4.104, | |
| "step": 561152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.080389134170574e-05, | |
| "loss": 4.1098, | |
| "step": 561664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.079550539419522e-05, | |
| "loss": 4.1063, | |
| "step": 562176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.07871194466847e-05, | |
| "loss": 4.103, | |
| "step": 562688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.077874987797791e-05, | |
| "loss": 4.11, | |
| "step": 563200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.077036393046739e-05, | |
| "loss": 4.1047, | |
| "step": 563712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.076197798295687e-05, | |
| "loss": 4.1166, | |
| "step": 564224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.075359203544635e-05, | |
| "loss": 4.1134, | |
| "step": 564736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.074520608793583e-05, | |
| "loss": 4.1049, | |
| "step": 565248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.073682014042531e-05, | |
| "loss": 4.0974, | |
| "step": 565760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.07284341929148e-05, | |
| "loss": 4.1039, | |
| "step": 566272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.072004824540428e-05, | |
| "loss": 4.1201, | |
| "step": 566784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0711678676697486e-05, | |
| "loss": 4.1046, | |
| "step": 567296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0703292729186966e-05, | |
| "loss": 4.1021, | |
| "step": 567808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0694906781676446e-05, | |
| "loss": 4.0957, | |
| "step": 568320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0686520834165926e-05, | |
| "loss": 4.1103, | |
| "step": 568832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0678151265459135e-05, | |
| "loss": 4.1004, | |
| "step": 569344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0669765317948615e-05, | |
| "loss": 4.1028, | |
| "step": 569856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0661379370438095e-05, | |
| "loss": 4.1068, | |
| "step": 570368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0652993422927575e-05, | |
| "loss": 4.1012, | |
| "step": 570880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0644607475417055e-05, | |
| "loss": 4.1077, | |
| "step": 571392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0636221527906535e-05, | |
| "loss": 4.1095, | |
| "step": 571904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.062785195919975e-05, | |
| "loss": 4.1141, | |
| "step": 572416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.061946601168923e-05, | |
| "loss": 4.105, | |
| "step": 572928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.061108006417871e-05, | |
| "loss": 4.1021, | |
| "step": 573440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.060269411666819e-05, | |
| "loss": 4.1029, | |
| "step": 573952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.059434092676513e-05, | |
| "loss": 4.1018, | |
| "step": 574464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.058595497925461e-05, | |
| "loss": 4.1137, | |
| "step": 574976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.057756903174409e-05, | |
| "loss": 4.1094, | |
| "step": 575488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.056918308423357e-05, | |
| "loss": 4.1121, | |
| "step": 576000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.056079713672305e-05, | |
| "loss": 4.0903, | |
| "step": 576512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.055241118921253e-05, | |
| "loss": 4.1023, | |
| "step": 577024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.054402524170201e-05, | |
| "loss": 4.0916, | |
| "step": 577536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.053563929419149e-05, | |
| "loss": 4.1101, | |
| "step": 578048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.052725334668097e-05, | |
| "loss": 4.1049, | |
| "step": 578560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.051886739917045e-05, | |
| "loss": 4.1044, | |
| "step": 579072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.051048145165993e-05, | |
| "loss": 4.1009, | |
| "step": 579584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.050209550414941e-05, | |
| "loss": 4.1055, | |
| "step": 580096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.049370955663889e-05, | |
| "loss": 4.0951, | |
| "step": 580608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.04853399879321e-05, | |
| "loss": 4.0998, | |
| "step": 581120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.047695404042158e-05, | |
| "loss": 4.1069, | |
| "step": 581632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.046856809291106e-05, | |
| "loss": 4.0914, | |
| "step": 582144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.046018214540054e-05, | |
| "loss": 4.0933, | |
| "step": 582656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0451812576693746e-05, | |
| "loss": 4.0966, | |
| "step": 583168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0443426629183226e-05, | |
| "loss": 4.0987, | |
| "step": 583680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0435040681672706e-05, | |
| "loss": 4.1155, | |
| "step": 584192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0426654734162186e-05, | |
| "loss": 4.0979, | |
| "step": 584704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.04182851654554e-05, | |
| "loss": 4.0993, | |
| "step": 585216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.040991559674862e-05, | |
| "loss": 4.0919, | |
| "step": 585728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.040152964923809e-05, | |
| "loss": 4.1105, | |
| "step": 586240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.039314370172757e-05, | |
| "loss": 4.1007, | |
| "step": 586752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.038475775421705e-05, | |
| "loss": 4.0941, | |
| "step": 587264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.037637180670653e-05, | |
| "loss": 4.1036, | |
| "step": 587776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.036798585919601e-05, | |
| "loss": 4.0842, | |
| "step": 588288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.035959991168549e-05, | |
| "loss": 4.086, | |
| "step": 588800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.035121396417497e-05, | |
| "loss": 4.1075, | |
| "step": 589312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.034284439546818e-05, | |
| "loss": 4.0909, | |
| "step": 589824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.033445844795766e-05, | |
| "loss": 4.1012, | |
| "step": 590336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.032607250044714e-05, | |
| "loss": 4.0981, | |
| "step": 590848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.031768655293662e-05, | |
| "loss": 4.0742, | |
| "step": 591360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0309316984229836e-05, | |
| "loss": 4.0915, | |
| "step": 591872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0300931036719316e-05, | |
| "loss": 4.1083, | |
| "step": 592384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0292545089208796e-05, | |
| "loss": 4.1005, | |
| "step": 592896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0284159141698276e-05, | |
| "loss": 4.0866, | |
| "step": 593408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0275773194187756e-05, | |
| "loss": 4.0859, | |
| "step": 593920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0267403625480965e-05, | |
| "loss": 4.0888, | |
| "step": 594432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0259017677970445e-05, | |
| "loss": 4.1, | |
| "step": 594944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0250631730459925e-05, | |
| "loss": 4.1024, | |
| "step": 595456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0242262161753134e-05, | |
| "loss": 4.0936, | |
| "step": 595968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0233876214242614e-05, | |
| "loss": 4.0933, | |
| "step": 596480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0225490266732094e-05, | |
| "loss": 4.099, | |
| "step": 596992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0217104319221574e-05, | |
| "loss": 4.0934, | |
| "step": 597504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.020871837171106e-05, | |
| "loss": 4.0895, | |
| "step": 598016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.020033242420054e-05, | |
| "loss": 4.0984, | |
| "step": 598528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.019194647669002e-05, | |
| "loss": 4.0919, | |
| "step": 599040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.018357690798323e-05, | |
| "loss": 4.0925, | |
| "step": 599552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.017519096047271e-05, | |
| "loss": 4.1048, | |
| "step": 600064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.016680501296219e-05, | |
| "loss": 4.0755, | |
| "step": 600576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.015841906545167e-05, | |
| "loss": 4.096, | |
| "step": 601088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.015003311794115e-05, | |
| "loss": 4.1012, | |
| "step": 601600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.014164717043063e-05, | |
| "loss": 4.0979, | |
| "step": 602112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.013327760172384e-05, | |
| "loss": 4.0978, | |
| "step": 602624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.012489165421332e-05, | |
| "loss": 4.0985, | |
| "step": 603136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.01165057067028e-05, | |
| "loss": 4.0853, | |
| "step": 603648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.010811975919228e-05, | |
| "loss": 4.0829, | |
| "step": 604160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.009973381168176e-05, | |
| "loss": 4.0951, | |
| "step": 604672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.009134786417124e-05, | |
| "loss": 4.1027, | |
| "step": 605184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.008296191666072e-05, | |
| "loss": 4.106, | |
| "step": 605696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.00745759691502e-05, | |
| "loss": 4.0873, | |
| "step": 606208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.006620640044341e-05, | |
| "loss": 4.0862, | |
| "step": 606720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.005782045293289e-05, | |
| "loss": 4.0972, | |
| "step": 607232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.004943450542237e-05, | |
| "loss": 4.089, | |
| "step": 607744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.004104855791185e-05, | |
| "loss": 4.0935, | |
| "step": 608256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0032678989205056e-05, | |
| "loss": 4.1014, | |
| "step": 608768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0024293041694536e-05, | |
| "loss": 4.1002, | |
| "step": 609280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.001592347298775e-05, | |
| "loss": 4.0836, | |
| "step": 609792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.000753752547723e-05, | |
| "loss": 4.0965, | |
| "step": 610304 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.157375812530518, | |
| "eval_runtime": 547.0603, | |
| "eval_samples_per_second": 697.53, | |
| "eval_steps_per_second": 21.798, | |
| "step": 610552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.999915157796671e-05, | |
| "loss": 4.0996, | |
| "step": 610816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.999076563045619e-05, | |
| "loss": 4.1056, | |
| "step": 611328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.998237968294567e-05, | |
| "loss": 4.0923, | |
| "step": 611840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.997399373543515e-05, | |
| "loss": 4.0904, | |
| "step": 612352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.996560778792463e-05, | |
| "loss": 4.0847, | |
| "step": 612864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.995722184041411e-05, | |
| "loss": 4.0913, | |
| "step": 613376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.994883589290359e-05, | |
| "loss": 4.0893, | |
| "step": 613888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.994044994539307e-05, | |
| "loss": 4.0733, | |
| "step": 614400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.993206399788255e-05, | |
| "loss": 4.0846, | |
| "step": 614912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.992367805037203e-05, | |
| "loss": 4.1052, | |
| "step": 615424 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.991530848166524e-05, | |
| "loss": 4.088, | |
| "step": 615936 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.990692253415472e-05, | |
| "loss": 4.0851, | |
| "step": 616448 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.98985365866442e-05, | |
| "loss": 4.1003, | |
| "step": 616960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.989015063913368e-05, | |
| "loss": 4.0815, | |
| "step": 617472 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.988176469162316e-05, | |
| "loss": 4.0924, | |
| "step": 617984 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.987337874411264e-05, | |
| "loss": 4.0708, | |
| "step": 618496 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.986499279660212e-05, | |
| "loss": 4.0712, | |
| "step": 619008 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.98566068490916e-05, | |
| "loss": 4.0893, | |
| "step": 619520 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.984822090158108e-05, | |
| "loss": 4.0778, | |
| "step": 620032 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.983983495407056e-05, | |
| "loss": 4.1042, | |
| "step": 620544 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.983144900656004e-05, | |
| "loss": 4.0906, | |
| "step": 621056 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.982306305904952e-05, | |
| "loss": 4.0936, | |
| "step": 621568 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9814677111539e-05, | |
| "loss": 4.0828, | |
| "step": 622080 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.980629116402848e-05, | |
| "loss": 4.0952, | |
| "step": 622592 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.979790521651796e-05, | |
| "loss": 4.0767, | |
| "step": 623104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.978951926900744e-05, | |
| "loss": 4.0868, | |
| "step": 623616 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.978113332149692e-05, | |
| "loss": 4.0824, | |
| "step": 624128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.977276375279013e-05, | |
| "loss": 4.0864, | |
| "step": 624640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9764377805279616e-05, | |
| "loss": 4.0865, | |
| "step": 625152 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9755991857769096e-05, | |
| "loss": 4.0739, | |
| "step": 625664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9747605910258575e-05, | |
| "loss": 4.0749, | |
| "step": 626176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9739252720355514e-05, | |
| "loss": 4.0879, | |
| "step": 626688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9730866772844994e-05, | |
| "loss": 4.0891, | |
| "step": 627200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9722480825334474e-05, | |
| "loss": 4.0844, | |
| "step": 627712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9714094877823954e-05, | |
| "loss": 4.0838, | |
| "step": 628224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9705708930313434e-05, | |
| "loss": 4.088, | |
| "step": 628736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.969733936160664e-05, | |
| "loss": 4.0858, | |
| "step": 629248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.968895341409612e-05, | |
| "loss": 4.0881, | |
| "step": 629760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.96805674665856e-05, | |
| "loss": 4.0841, | |
| "step": 630272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.967218151907508e-05, | |
| "loss": 4.0817, | |
| "step": 630784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.966379557156457e-05, | |
| "loss": 4.0829, | |
| "step": 631296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.965540962405405e-05, | |
| "loss": 4.0715, | |
| "step": 631808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.964702367654353e-05, | |
| "loss": 4.0852, | |
| "step": 632320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9638637729033e-05, | |
| "loss": 4.0888, | |
| "step": 632832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.963028453912995e-05, | |
| "loss": 4.0904, | |
| "step": 633344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.962189859161943e-05, | |
| "loss": 4.0876, | |
| "step": 633856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9613529022912636e-05, | |
| "loss": 4.0663, | |
| "step": 634368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9605143075402116e-05, | |
| "loss": 4.0944, | |
| "step": 634880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9596757127891596e-05, | |
| "loss": 4.0727, | |
| "step": 635392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9588371180381076e-05, | |
| "loss": 4.0807, | |
| "step": 635904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9579985232870556e-05, | |
| "loss": 4.0763, | |
| "step": 636416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9571599285360036e-05, | |
| "loss": 4.0796, | |
| "step": 636928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.956321333784952e-05, | |
| "loss": 4.0829, | |
| "step": 637440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9554827390339e-05, | |
| "loss": 4.0733, | |
| "step": 637952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9546441442828476e-05, | |
| "loss": 4.0818, | |
| "step": 638464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9538055495317956e-05, | |
| "loss": 4.0722, | |
| "step": 638976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9529669547807436e-05, | |
| "loss": 4.0847, | |
| "step": 639488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.952129997910065e-05, | |
| "loss": 4.0728, | |
| "step": 640000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9512914031590125e-05, | |
| "loss": 4.0874, | |
| "step": 640512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9504528084079605e-05, | |
| "loss": 4.0811, | |
| "step": 641024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9496142136569085e-05, | |
| "loss": 4.0778, | |
| "step": 641536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9487756189058565e-05, | |
| "loss": 4.0722, | |
| "step": 642048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9479370241548045e-05, | |
| "loss": 4.0757, | |
| "step": 642560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9470984294037525e-05, | |
| "loss": 4.0843, | |
| "step": 643072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.946261472533074e-05, | |
| "loss": 4.0767, | |
| "step": 643584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.945422877782022e-05, | |
| "loss": 4.0766, | |
| "step": 644096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.94458428303097e-05, | |
| "loss": 4.0693, | |
| "step": 644608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.943745688279918e-05, | |
| "loss": 4.0766, | |
| "step": 645120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.942908731409239e-05, | |
| "loss": 4.0733, | |
| "step": 645632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.942070136658187e-05, | |
| "loss": 4.0746, | |
| "step": 646144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.941231541907135e-05, | |
| "loss": 4.0747, | |
| "step": 646656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.940392947156083e-05, | |
| "loss": 4.0754, | |
| "step": 647168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.939554352405031e-05, | |
| "loss": 4.0794, | |
| "step": 647680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.938715757653979e-05, | |
| "loss": 4.0866, | |
| "step": 648192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9378788007833e-05, | |
| "loss": 4.0806, | |
| "step": 648704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.937040206032248e-05, | |
| "loss": 4.0753, | |
| "step": 649216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.936201611281196e-05, | |
| "loss": 4.0767, | |
| "step": 649728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.935363016530144e-05, | |
| "loss": 4.0759, | |
| "step": 650240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9345260596594654e-05, | |
| "loss": 4.0717, | |
| "step": 650752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9336874649084134e-05, | |
| "loss": 4.0847, | |
| "step": 651264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9328488701573614e-05, | |
| "loss": 4.0834, | |
| "step": 651776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9320102754063094e-05, | |
| "loss": 4.0833, | |
| "step": 652288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9311716806552574e-05, | |
| "loss": 4.0642, | |
| "step": 652800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.930336361664951e-05, | |
| "loss": 4.0714, | |
| "step": 653312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.929497766913899e-05, | |
| "loss": 4.0616, | |
| "step": 653824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.928659172162847e-05, | |
| "loss": 4.0834, | |
| "step": 654336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.927820577411795e-05, | |
| "loss": 4.079, | |
| "step": 654848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.926981982660743e-05, | |
| "loss": 4.0721, | |
| "step": 655360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.926143387909691e-05, | |
| "loss": 4.0737, | |
| "step": 655872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.925304793158639e-05, | |
| "loss": 4.0775, | |
| "step": 656384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.924466198407588e-05, | |
| "loss": 4.068, | |
| "step": 656896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.923627603656536e-05, | |
| "loss": 4.0705, | |
| "step": 657408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.922790646785857e-05, | |
| "loss": 4.0783, | |
| "step": 657920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.921952052034805e-05, | |
| "loss": 4.064, | |
| "step": 658432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.921113457283753e-05, | |
| "loss": 4.0667, | |
| "step": 658944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.920274862532701e-05, | |
| "loss": 4.0644, | |
| "step": 659456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.919437905662022e-05, | |
| "loss": 4.0725, | |
| "step": 659968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.91859931091097e-05, | |
| "loss": 4.0859, | |
| "step": 660480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.917760716159918e-05, | |
| "loss": 4.0705, | |
| "step": 660992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.916922121408866e-05, | |
| "loss": 4.0724, | |
| "step": 661504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.916083526657813e-05, | |
| "loss": 4.0671, | |
| "step": 662016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9152449319067617e-05, | |
| "loss": 4.0798, | |
| "step": 662528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9144063371557096e-05, | |
| "loss": 4.0729, | |
| "step": 663040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.913569380285031e-05, | |
| "loss": 4.068, | |
| "step": 663552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.912732423414352e-05, | |
| "loss": 4.0738, | |
| "step": 664064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9118938286633e-05, | |
| "loss": 4.0562, | |
| "step": 664576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.911056871792621e-05, | |
| "loss": 4.0592, | |
| "step": 665088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.910218277041569e-05, | |
| "loss": 4.0783, | |
| "step": 665600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.90938132017089e-05, | |
| "loss": 4.0695, | |
| "step": 666112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.908542725419838e-05, | |
| "loss": 4.0639, | |
| "step": 666624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.907704130668786e-05, | |
| "loss": 4.0726, | |
| "step": 667136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.906865535917734e-05, | |
| "loss": 4.0494, | |
| "step": 667648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.906026941166682e-05, | |
| "loss": 4.0678, | |
| "step": 668160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.90518834641563e-05, | |
| "loss": 4.0766, | |
| "step": 668672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9043497516645786e-05, | |
| "loss": 4.0762, | |
| "step": 669184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.903511156913526e-05, | |
| "loss": 4.0579, | |
| "step": 669696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.902672562162474e-05, | |
| "loss": 4.061, | |
| "step": 670208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.901833967411422e-05, | |
| "loss": 4.0598, | |
| "step": 670720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.90099537266037e-05, | |
| "loss": 4.0732, | |
| "step": 671232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.900156777909318e-05, | |
| "loss": 4.0742, | |
| "step": 671744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.899319821038639e-05, | |
| "loss": 4.068, | |
| "step": 672256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.898481226287587e-05, | |
| "loss": 4.0651, | |
| "step": 672768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.897642631536535e-05, | |
| "loss": 4.0718, | |
| "step": 673280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.896804036785483e-05, | |
| "loss": 4.0682, | |
| "step": 673792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.895965442034431e-05, | |
| "loss": 4.0637, | |
| "step": 674304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.895126847283379e-05, | |
| "loss": 4.0665, | |
| "step": 674816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.894288252532327e-05, | |
| "loss": 4.065, | |
| "step": 675328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8934512956616484e-05, | |
| "loss": 4.0672, | |
| "step": 675840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8926127009105964e-05, | |
| "loss": 4.0726, | |
| "step": 676352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8917741061595444e-05, | |
| "loss": 4.055, | |
| "step": 676864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8909355114084924e-05, | |
| "loss": 4.0663, | |
| "step": 677376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8900969166574404e-05, | |
| "loss": 4.0754, | |
| "step": 677888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8892583219063884e-05, | |
| "loss": 4.0672, | |
| "step": 678400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8884197271553364e-05, | |
| "loss": 4.0722, | |
| "step": 678912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8875811324042843e-05, | |
| "loss": 4.0702, | |
| "step": 679424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.886745813413978e-05, | |
| "loss": 4.0604, | |
| "step": 679936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.885907218662926e-05, | |
| "loss": 4.0572, | |
| "step": 680448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.885068623911874e-05, | |
| "loss": 4.0649, | |
| "step": 680960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.884230029160822e-05, | |
| "loss": 4.0785, | |
| "step": 681472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.88339143440977e-05, | |
| "loss": 4.0769, | |
| "step": 681984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.882552839658719e-05, | |
| "loss": 4.0601, | |
| "step": 682496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.881714244907667e-05, | |
| "loss": 4.0639, | |
| "step": 683008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.880875650156615e-05, | |
| "loss": 4.0657, | |
| "step": 683520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.880037055405562e-05, | |
| "loss": 4.0647, | |
| "step": 684032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.879200098534884e-05, | |
| "loss": 4.0675, | |
| "step": 684544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8783631416642046e-05, | |
| "loss": 4.074, | |
| "step": 685056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8775245469131526e-05, | |
| "loss": 4.0755, | |
| "step": 685568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8766875900424735e-05, | |
| "loss": 4.0574, | |
| "step": 686080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8758489952914215e-05, | |
| "loss": 4.0676, | |
| "step": 686592 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.138632297515869, | |
| "eval_runtime": 548.0588, | |
| "eval_samples_per_second": 696.259, | |
| "eval_steps_per_second": 21.759, | |
| "step": 686871 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 2.6432065212073062e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |