| { | |
| "best_metric": 4.126788139343262, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-c-command/lstm/0/checkpoints/checkpoint-534240", | |
| "epoch": 0.025000606015738065, | |
| "eval_steps": 10, | |
| "global_step": 534240, | |
| "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": 10.8209, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 7.5535, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 7.0546, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 6.9931, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 6.9443, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 6.894, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 6.727, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994129836742636e-05, | |
| "loss": 6.6298, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993291241991584e-05, | |
| "loss": 6.5262, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992452647240532e-05, | |
| "loss": 6.4594, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99161405248948e-05, | |
| "loss": 6.3927, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990775457738428e-05, | |
| "loss": 6.3337, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989936862987376e-05, | |
| "loss": 6.2739, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989098268236324e-05, | |
| "loss": 6.1987, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988261311365645e-05, | |
| "loss": 6.1549, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987422716614593e-05, | |
| "loss": 6.0934, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986584121863541e-05, | |
| "loss": 6.0554, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985745527112489e-05, | |
| "loss": 6.0083, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984906932361437e-05, | |
| "loss": 5.9618, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984068337610385e-05, | |
| "loss": 5.9254, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.8891, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.982391148108281e-05, | |
| "loss": 5.8527, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.981552553357229e-05, | |
| "loss": 5.8223, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.980713958606178e-05, | |
| "loss": 5.7851, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.979875363855125e-05, | |
| "loss": 5.7676, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.979036769104073e-05, | |
| "loss": 5.7383, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9781998122333946e-05, | |
| "loss": 5.7084, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773612174823426e-05, | |
| "loss": 5.6882, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.97652262273129e-05, | |
| "loss": 5.6597, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.975684027980238e-05, | |
| "loss": 5.6424, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748470711095595e-05, | |
| "loss": 5.6238, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9740101142388804e-05, | |
| "loss": 5.6022, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9731715194878284e-05, | |
| "loss": 5.58, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9723329247367764e-05, | |
| "loss": 5.5579, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714943299857244e-05, | |
| "loss": 5.5529, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9706557352346724e-05, | |
| "loss": 5.5431, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.969818778363994e-05, | |
| "loss": 5.5157, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968980183612942e-05, | |
| "loss": 5.4749, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96814158886189e-05, | |
| "loss": 5.4795, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.967302994110837e-05, | |
| "loss": 5.4488, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966466037240159e-05, | |
| "loss": 5.4497, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965627442489107e-05, | |
| "loss": 5.4329, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964788847738054e-05, | |
| "loss": 5.4199, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963950252987002e-05, | |
| "loss": 5.3947, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963113296116324e-05, | |
| "loss": 5.3985, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962274701365272e-05, | |
| "loss": 5.3857, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96143610661422e-05, | |
| "loss": 5.3831, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.960597511863168e-05, | |
| "loss": 5.366, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.959758917112116e-05, | |
| "loss": 5.3456, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9589219602414374e-05, | |
| "loss": 5.338, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958083365490385e-05, | |
| "loss": 5.3396, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.957244770739333e-05, | |
| "loss": 5.3168, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956406175988281e-05, | |
| "loss": 5.3083, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955567581237229e-05, | |
| "loss": 5.2929, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.954728986486177e-05, | |
| "loss": 5.2902, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953890391735125e-05, | |
| "loss": 5.2727, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9530517969840727e-05, | |
| "loss": 5.2862, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9522148401133936e-05, | |
| "loss": 5.2498, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951376245362342e-05, | |
| "loss": 5.2553, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95053765061129e-05, | |
| "loss": 5.247, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949699055860238e-05, | |
| "loss": 5.2258, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948862098989559e-05, | |
| "loss": 5.2278, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9480267799992536e-05, | |
| "loss": 5.2148, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9471881852482016e-05, | |
| "loss": 5.1968, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946349590497149e-05, | |
| "loss": 5.19, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945510995746097e-05, | |
| "loss": 5.2081, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944672400995045e-05, | |
| "loss": 5.1821, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9438354441243665e-05, | |
| "loss": 5.1719, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9429968493733145e-05, | |
| "loss": 5.1546, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9421582546222625e-05, | |
| "loss": 5.1635, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9413196598712105e-05, | |
| "loss": 5.1683, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9404810651201585e-05, | |
| "loss": 5.1514, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9396424703691065e-05, | |
| "loss": 5.1452, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388038756180545e-05, | |
| "loss": 5.149, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379652808670025e-05, | |
| "loss": 5.1401, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371266861159505e-05, | |
| "loss": 5.1275, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362897292452714e-05, | |
| "loss": 5.1061, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9354511344942194e-05, | |
| "loss": 5.1133, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9346125397431674e-05, | |
| "loss": 5.0958, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9337739449921154e-05, | |
| "loss": 5.0941, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329353502410634e-05, | |
| "loss": 5.0872, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932098393370384e-05, | |
| "loss": 5.0844, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.931259798619332e-05, | |
| "loss": 5.093, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.930421203868281e-05, | |
| "loss": 5.0819, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.929582609117229e-05, | |
| "loss": 5.0549, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.928744014366177e-05, | |
| "loss": 5.0561, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927907057495498e-05, | |
| "loss": 5.0596, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927068462744446e-05, | |
| "loss": 5.0558, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.926229867993394e-05, | |
| "loss": 5.0461, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.925391273242342e-05, | |
| "loss": 5.0399, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.92455267849129e-05, | |
| "loss": 5.0327, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923715721620611e-05, | |
| "loss": 5.0186, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922877126869559e-05, | |
| "loss": 5.0237, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922038532118507e-05, | |
| "loss": 5.0162, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921199937367455e-05, | |
| "loss": 5.0074, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920361342616403e-05, | |
| "loss": 5.0146, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919522747865351e-05, | |
| "loss": 4.9952, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918684153114299e-05, | |
| "loss": 4.9911, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917845558363247e-05, | |
| "loss": 4.982, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917010239372941e-05, | |
| "loss": 4.9771, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.916171644621889e-05, | |
| "loss": 4.9846, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.915333049870837e-05, | |
| "loss": 4.9692, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.914494455119785e-05, | |
| "loss": 4.9635, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9136558603687325e-05, | |
| "loss": 4.9596, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9128172656176805e-05, | |
| "loss": 4.9585, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9119786708666285e-05, | |
| "loss": 4.9404, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.91114171399595e-05, | |
| "loss": 4.9307, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.910303119244898e-05, | |
| "loss": 4.9368, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.909464524493846e-05, | |
| "loss": 4.9431, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.908625929742794e-05, | |
| "loss": 4.9329, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907787334991742e-05, | |
| "loss": 4.9311, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906950378121063e-05, | |
| "loss": 4.9164, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906111783370011e-05, | |
| "loss": 4.9208, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905273188618959e-05, | |
| "loss": 4.9078, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904434593867907e-05, | |
| "loss": 4.913, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903595999116855e-05, | |
| "loss": 4.9062, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.902759042246176e-05, | |
| "loss": 4.9046, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901920447495124e-05, | |
| "loss": 4.8971, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901081852744072e-05, | |
| "loss": 4.8926, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90024325799302e-05, | |
| "loss": 4.8958, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8994046632419686e-05, | |
| "loss": 4.8948, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8985660684909166e-05, | |
| "loss": 4.8828, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8977274737398646e-05, | |
| "loss": 4.8833, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968888789888125e-05, | |
| "loss": 4.8743, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8960519221181335e-05, | |
| "loss": 4.8772, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8952133273670815e-05, | |
| "loss": 4.8568, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8943747326160294e-05, | |
| "loss": 4.8531, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8935361378649774e-05, | |
| "loss": 4.8574, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8926991809942984e-05, | |
| "loss": 4.8574, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918605862432463e-05, | |
| "loss": 4.8475, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8910219914921943e-05, | |
| "loss": 4.8597, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8901833967411423e-05, | |
| "loss": 4.8512, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889346439870464e-05, | |
| "loss": 4.8355, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888507845119412e-05, | |
| "loss": 4.8408, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88766925036836e-05, | |
| "loss": 4.844, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.886830655617308e-05, | |
| "loss": 4.835, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.885993698746629e-05, | |
| "loss": 4.8312, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88515674187595e-05, | |
| "loss": 4.823, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884318147124898e-05, | |
| "loss": 4.826, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883479552373846e-05, | |
| "loss": 4.8298, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882640957622794e-05, | |
| "loss": 4.8112, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881802362871742e-05, | |
| "loss": 4.8051, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88096376812069e-05, | |
| "loss": 4.8115, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8801268112500106e-05, | |
| "loss": 4.8033, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879289854379332e-05, | |
| "loss": 4.8001, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.87845125962828e-05, | |
| "loss": 4.8024, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877612664877228e-05, | |
| "loss": 4.7947, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.876774070126176e-05, | |
| "loss": 4.7939, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875935475375124e-05, | |
| "loss": 4.7902, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875096880624072e-05, | |
| "loss": 4.7849, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.752007484436035, | |
| "eval_runtime": 290.964, | |
| "eval_samples_per_second": 1311.471, | |
| "eval_steps_per_second": 40.984, | |
| "step": 76320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.87425828587302e-05, | |
| "loss": 4.7788, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.873419691121968e-05, | |
| "loss": 4.7817, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.872581096370916e-05, | |
| "loss": 4.7889, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8717425016198635e-05, | |
| "loss": 4.7769, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8709039068688115e-05, | |
| "loss": 4.7747, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.870066949998133e-05, | |
| "loss": 4.77, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.869228355247081e-05, | |
| "loss": 4.7722, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.868389760496029e-05, | |
| "loss": 4.7517, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8675528036253507e-05, | |
| "loss": 4.7741, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8667142088742986e-05, | |
| "loss": 4.7594, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865875614123246e-05, | |
| "loss": 4.767, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865037019372194e-05, | |
| "loss": 4.7652, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8642000625015155e-05, | |
| "loss": 4.7458, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8633614677504635e-05, | |
| "loss": 4.7478, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.862522872999411e-05, | |
| "loss": 4.7412, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.861684278248359e-05, | |
| "loss": 4.7365, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.860845683497307e-05, | |
| "loss": 4.7441, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.860007088746255e-05, | |
| "loss": 4.7316, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.859168493995203e-05, | |
| "loss": 4.7296, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.858329899244151e-05, | |
| "loss": 4.7462, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8574913044930995e-05, | |
| "loss": 4.7306, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8566527097420475e-05, | |
| "loss": 4.7251, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8558141149909955e-05, | |
| "loss": 4.7174, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8549771581203164e-05, | |
| "loss": 4.7329, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.854140201249637e-05, | |
| "loss": 4.7128, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.853301606498585e-05, | |
| "loss": 4.7066, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.852463011747533e-05, | |
| "loss": 4.7081, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.851624416996481e-05, | |
| "loss": 4.7102, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.850785822245429e-05, | |
| "loss": 4.7027, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.84994886537475e-05, | |
| "loss": 4.7035, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.849110270623698e-05, | |
| "loss": 4.704, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.848271675872646e-05, | |
| "loss": 4.7031, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.847433081121595e-05, | |
| "loss": 4.6921, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.846594486370543e-05, | |
| "loss": 4.7028, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.845755891619491e-05, | |
| "loss": 4.7052, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.844918934748812e-05, | |
| "loss": 4.6961, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.84408033999776e-05, | |
| "loss": 4.6691, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.843241745246708e-05, | |
| "loss": 4.6793, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.842403150495656e-05, | |
| "loss": 4.6753, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.841564555744604e-05, | |
| "loss": 4.6816, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.840725960993552e-05, | |
| "loss": 4.6726, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8398873662425e-05, | |
| "loss": 4.6763, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839048771491447e-05, | |
| "loss": 4.6633, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8382118146207687e-05, | |
| "loss": 4.6776, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8373732198697167e-05, | |
| "loss": 4.6726, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8365346251186646e-05, | |
| "loss": 4.6724, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8356960303676126e-05, | |
| "loss": 4.6726, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834859073496934e-05, | |
| "loss": 4.6532, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834020478745882e-05, | |
| "loss": 4.6607, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8331818839948295e-05, | |
| "loss": 4.666, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.832344927124151e-05, | |
| "loss": 4.6579, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.831506332373099e-05, | |
| "loss": 4.6552, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.830667737622047e-05, | |
| "loss": 4.6445, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8298291428709944e-05, | |
| "loss": 4.6515, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8289905481199424e-05, | |
| "loss": 4.6349, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8281519533688904e-05, | |
| "loss": 4.6645, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8273133586178384e-05, | |
| "loss": 4.6361, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.826474763866787e-05, | |
| "loss": 4.6477, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.825637806996108e-05, | |
| "loss": 4.6449, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.824799212245056e-05, | |
| "loss": 4.6229, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823960617494004e-05, | |
| "loss": 4.637, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823122022742952e-05, | |
| "loss": 4.6308, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8222834279919e-05, | |
| "loss": 4.6157, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.821446471121221e-05, | |
| "loss": 4.6161, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.820607876370169e-05, | |
| "loss": 4.6352, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.819769281619117e-05, | |
| "loss": 4.6229, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.818930686868065e-05, | |
| "loss": 4.6132, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.818092092117013e-05, | |
| "loss": 4.6081, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.817253497365961e-05, | |
| "loss": 4.6222, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.816414902614909e-05, | |
| "loss": 4.629, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8155779457442305e-05, | |
| "loss": 4.6211, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8147393509931785e-05, | |
| "loss": 4.6145, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8139007562421265e-05, | |
| "loss": 4.6289, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8130621614910745e-05, | |
| "loss": 4.6209, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8122235667400224e-05, | |
| "loss": 4.6188, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8113866098693434e-05, | |
| "loss": 4.5985, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8105480151182914e-05, | |
| "loss": 4.6106, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8097094203672393e-05, | |
| "loss": 4.604, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8088708256161873e-05, | |
| "loss": 4.599, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808033868745508e-05, | |
| "loss": 4.6007, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.807195273994456e-05, | |
| "loss": 4.6002, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.806356679243404e-05, | |
| "loss": 4.6134, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.805518084492352e-05, | |
| "loss": 4.6092, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8046794897413e-05, | |
| "loss": 4.5839, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803842532870622e-05, | |
| "loss": 4.5904, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.80300393811957e-05, | |
| "loss": 4.5927, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.802165343368518e-05, | |
| "loss": 4.5941, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.801326748617466e-05, | |
| "loss": 4.5926, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.800488153866413e-05, | |
| "loss": 4.5898, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.799649559115361e-05, | |
| "loss": 4.5838, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.798810964364309e-05, | |
| "loss": 4.5761, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.79797400749363e-05, | |
| "loss": 4.5862, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797135412742578e-05, | |
| "loss": 4.5799, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.796296817991526e-05, | |
| "loss": 4.5793, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.795458223240474e-05, | |
| "loss": 4.5857, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.794619628489423e-05, | |
| "loss": 4.5763, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.793781033738371e-05, | |
| "loss": 4.5693, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.792942438987319e-05, | |
| "loss": 4.5674, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7921054821166396e-05, | |
| "loss": 4.5705, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7912668873655876e-05, | |
| "loss": 4.5754, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7904282926145356e-05, | |
| "loss": 4.5621, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7895913357438565e-05, | |
| "loss": 4.5606, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7887527409928045e-05, | |
| "loss": 4.5617, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7879141462417525e-05, | |
| "loss": 4.5627, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7870755514907005e-05, | |
| "loss": 4.5517, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7862369567396485e-05, | |
| "loss": 4.5395, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7853983619885965e-05, | |
| "loss": 4.5551, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7845597672375445e-05, | |
| "loss": 4.557, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7837211724864925e-05, | |
| "loss": 4.5516, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782884215615814e-05, | |
| "loss": 4.5529, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782045620864762e-05, | |
| "loss": 4.545, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.78120702611371e-05, | |
| "loss": 4.5516, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.780368431362658e-05, | |
| "loss": 4.5394, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.779531474491979e-05, | |
| "loss": 4.5431, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.778692879740927e-05, | |
| "loss": 4.5485, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777854284989875e-05, | |
| "loss": 4.5483, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777015690238823e-05, | |
| "loss": 4.5397, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.776177095487771e-05, | |
| "loss": 4.5397, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.775338500736719e-05, | |
| "loss": 4.5462, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.774499905985666e-05, | |
| "loss": 4.5458, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.773661311234615e-05, | |
| "loss": 4.53, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.772822716483563e-05, | |
| "loss": 4.5428, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771984121732511e-05, | |
| "loss": 4.53, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771147164861832e-05, | |
| "loss": 4.5439, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.77030857011078e-05, | |
| "loss": 4.5209, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.769469975359728e-05, | |
| "loss": 4.5201, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7686330184890494e-05, | |
| "loss": 4.5238, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.767794423737997e-05, | |
| "loss": 4.5312, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766955828986945e-05, | |
| "loss": 4.5214, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766117234235893e-05, | |
| "loss": 4.5378, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.765278639484841e-05, | |
| "loss": 4.5275, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.764440044733789e-05, | |
| "loss": 4.5198, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.763601449982737e-05, | |
| "loss": 4.5229, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.762762855231685e-05, | |
| "loss": 4.5313, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761924260480633e-05, | |
| "loss": 4.5215, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7610856657295813e-05, | |
| "loss": 4.5207, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7602470709785293e-05, | |
| "loss": 4.5145, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75941011410785e-05, | |
| "loss": 4.5191, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.758571519356798e-05, | |
| "loss": 4.5227, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.757732924605746e-05, | |
| "loss": 4.511, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756894329854694e-05, | |
| "loss": 4.5087, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756055735103642e-05, | |
| "loss": 4.514, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75521714035259e-05, | |
| "loss": 4.5045, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7543785456015375e-05, | |
| "loss": 4.5049, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7535399508504855e-05, | |
| "loss": 4.5124, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.752702993979807e-05, | |
| "loss": 4.504, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.751864399228755e-05, | |
| "loss": 4.503, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.751025804477703e-05, | |
| "loss": 4.5034, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.750187209726651e-05, | |
| "loss": 4.4971, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.470280647277832, | |
| "eval_runtime": 291.1643, | |
| "eval_samples_per_second": 1310.569, | |
| "eval_steps_per_second": 40.956, | |
| "step": 152640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.749348614975599e-05, | |
| "loss": 4.491, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.748510020224547e-05, | |
| "loss": 4.4977, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.747671425473495e-05, | |
| "loss": 4.5105, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.746832830722443e-05, | |
| "loss": 4.4992, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745994235971391e-05, | |
| "loss": 4.4995, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745155641220339e-05, | |
| "loss": 4.4931, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.744317046469287e-05, | |
| "loss": 4.4997, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.743478451718235e-05, | |
| "loss": 4.4793, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.742639856967183e-05, | |
| "loss": 4.5019, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.741801262216131e-05, | |
| "loss": 4.4983, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.740962667465079e-05, | |
| "loss": 4.4946, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7401240727140264e-05, | |
| "loss": 4.5055, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.739287115843348e-05, | |
| "loss": 4.4785, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7384485210922966e-05, | |
| "loss": 4.4836, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7376099263412446e-05, | |
| "loss": 4.4894, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.736771331590192e-05, | |
| "loss": 4.4716, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.73593273683914e-05, | |
| "loss": 4.4851, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.735094142088088e-05, | |
| "loss": 4.4748, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.734255547337036e-05, | |
| "loss": 4.4754, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.733416952585984e-05, | |
| "loss": 4.4976, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.732578357834932e-05, | |
| "loss": 4.4773, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.73173976308388e-05, | |
| "loss": 4.4797, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730901168332828e-05, | |
| "loss": 4.4679, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730064211462149e-05, | |
| "loss": 4.4857, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7292272545914704e-05, | |
| "loss": 4.4616, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7283886598404184e-05, | |
| "loss": 4.4642, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7275500650893664e-05, | |
| "loss": 4.4627, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7267114703383144e-05, | |
| "loss": 4.4729, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7258728755872624e-05, | |
| "loss": 4.4609, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7250342808362104e-05, | |
| "loss": 4.4615, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7241956860851584e-05, | |
| "loss": 4.4688, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.723358729214479e-05, | |
| "loss": 4.4705, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.722520134463427e-05, | |
| "loss": 4.4535, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.721681539712375e-05, | |
| "loss": 4.4689, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720842944961323e-05, | |
| "loss": 4.4709, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720005988090644e-05, | |
| "loss": 4.4634, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.719169031219965e-05, | |
| "loss": 4.4411, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.718330436468914e-05, | |
| "loss": 4.4541, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.717491841717862e-05, | |
| "loss": 4.4425, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.71665324696681e-05, | |
| "loss": 4.4587, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.715816290096131e-05, | |
| "loss": 4.4488, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714977695345079e-05, | |
| "loss": 4.454, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714139100594027e-05, | |
| "loss": 4.4429, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.713300505842975e-05, | |
| "loss": 4.4539, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.712461911091923e-05, | |
| "loss": 4.4526, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7116249542212436e-05, | |
| "loss": 4.4532, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7107863594701916e-05, | |
| "loss": 4.455, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7099477647191396e-05, | |
| "loss": 4.4335, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7091091699680876e-05, | |
| "loss": 4.442, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7082705752170356e-05, | |
| "loss": 4.457, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7074319804659836e-05, | |
| "loss": 4.4402, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.706593385714932e-05, | |
| "loss": 4.4406, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.70575479096388e-05, | |
| "loss": 4.4359, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704917834093201e-05, | |
| "loss": 4.4403, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704080877222522e-05, | |
| "loss": 4.4288, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.703243920351843e-05, | |
| "loss": 4.4481, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.702405325600791e-05, | |
| "loss": 4.4335, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.701566730849739e-05, | |
| "loss": 4.4374, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.700728136098687e-05, | |
| "loss": 4.4444, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699889541347635e-05, | |
| "loss": 4.416, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699050946596583e-05, | |
| "loss": 4.4332, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.698212351845531e-05, | |
| "loss": 4.4286, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.697373757094479e-05, | |
| "loss": 4.4184, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6965368002238005e-05, | |
| "loss": 4.4183, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6956982054727485e-05, | |
| "loss": 4.4298, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6948596107216965e-05, | |
| "loss": 4.4278, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6940210159706445e-05, | |
| "loss": 4.4157, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6931824212195925e-05, | |
| "loss": 4.409, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6923454643489134e-05, | |
| "loss": 4.4285, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6915068695978614e-05, | |
| "loss": 4.4371, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6906682748468094e-05, | |
| "loss": 4.4251, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6898296800957574e-05, | |
| "loss": 4.4232, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688991085344705e-05, | |
| "loss": 4.433, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688152490593653e-05, | |
| "loss": 4.4297, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6873138958426014e-05, | |
| "loss": 4.4296, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6864753010915494e-05, | |
| "loss": 4.4057, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.685639982101244e-05, | |
| "loss": 4.4219, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.684801387350192e-05, | |
| "loss": 4.4213, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68396279259914e-05, | |
| "loss": 4.4094, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683125835728461e-05, | |
| "loss": 4.4189, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.682287240977409e-05, | |
| "loss": 4.4155, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.681448646226357e-05, | |
| "loss": 4.429, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.680610051475305e-05, | |
| "loss": 4.4213, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.679773094604626e-05, | |
| "loss": 4.4018, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678934499853574e-05, | |
| "loss": 4.4087, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678095905102522e-05, | |
| "loss": 4.4101, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6772573103514697e-05, | |
| "loss": 4.4145, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6764187156004177e-05, | |
| "loss": 4.4137, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6755801208493657e-05, | |
| "loss": 4.408, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6747415260983136e-05, | |
| "loss": 4.4055, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6739029313472616e-05, | |
| "loss": 4.3976, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6730659744765826e-05, | |
| "loss": 4.4107, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6722273797255305e-05, | |
| "loss": 4.3994, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6713887849744785e-05, | |
| "loss": 4.407, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6705501902234265e-05, | |
| "loss": 4.41, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6697115954723745e-05, | |
| "loss": 4.4033, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6688730007213225e-05, | |
| "loss": 4.3971, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6680344059702705e-05, | |
| "loss": 4.3985, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.667197449099592e-05, | |
| "loss": 4.3991, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.66635885434854e-05, | |
| "loss": 4.4031, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.665520259597488e-05, | |
| "loss": 4.3946, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.664681664846436e-05, | |
| "loss": 4.3914, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663843070095384e-05, | |
| "loss": 4.3905, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663004475344332e-05, | |
| "loss": 4.3948, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.66216588059328e-05, | |
| "loss": 4.3856, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.661327285842228e-05, | |
| "loss": 4.3752, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.660490328971549e-05, | |
| "loss": 4.3863, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.659651734220497e-05, | |
| "loss": 4.3932, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.658813139469445e-05, | |
| "loss": 4.3841, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657976182598766e-05, | |
| "loss": 4.3936, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657137587847714e-05, | |
| "loss": 4.3801, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.656298993096662e-05, | |
| "loss": 4.3899, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.65546039834561e-05, | |
| "loss": 4.3805, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6546218035945586e-05, | |
| "loss": 4.3775, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6537848467238795e-05, | |
| "loss": 4.392, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6529462519728275e-05, | |
| "loss": 4.3889, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6521076572217755e-05, | |
| "loss": 4.3819, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6512690624707234e-05, | |
| "loss": 4.3834, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.650430467719671e-05, | |
| "loss": 4.3833, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.649591872968619e-05, | |
| "loss": 4.3879, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.648753278217567e-05, | |
| "loss": 4.3754, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647914683466515e-05, | |
| "loss": 4.3877, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647077726595836e-05, | |
| "loss": 4.3749, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6462391318447837e-05, | |
| "loss": 4.3873, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.645400537093732e-05, | |
| "loss": 4.3633, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.644563580223053e-05, | |
| "loss": 4.3725, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.643724985472001e-05, | |
| "loss": 4.3672, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642886390720949e-05, | |
| "loss": 4.3772, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642047795969897e-05, | |
| "loss": 4.3692, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.641209201218845e-05, | |
| "loss": 4.3884, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.640370606467793e-05, | |
| "loss": 4.3768, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.639532011716741e-05, | |
| "loss": 4.3721, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.638695054846062e-05, | |
| "loss": 4.3654, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637858097975383e-05, | |
| "loss": 4.3805, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637019503224331e-05, | |
| "loss": 4.3738, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.636180908473279e-05, | |
| "loss": 4.3753, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.635342313722228e-05, | |
| "loss": 4.3681, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.634503718971176e-05, | |
| "loss": 4.3686, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.633665124220124e-05, | |
| "loss": 4.3768, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6328281673494446e-05, | |
| "loss": 4.3653, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6319895725983926e-05, | |
| "loss": 4.3619, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6311509778473406e-05, | |
| "loss": 4.3729, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6303123830962886e-05, | |
| "loss": 4.3624, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6294737883452366e-05, | |
| "loss": 4.3586, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6286351935941846e-05, | |
| "loss": 4.3633, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6277965988431326e-05, | |
| "loss": 4.3673, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6269580040920806e-05, | |
| "loss": 4.358, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6261226851017744e-05, | |
| "loss": 4.3579, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.625284090350723e-05, | |
| "loss": 4.3536, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.335472106933594, | |
| "eval_runtime": 293.6337, | |
| "eval_samples_per_second": 1299.548, | |
| "eval_steps_per_second": 40.612, | |
| "step": 228960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.624445495599671e-05, | |
| "loss": 4.3602, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.623606900848619e-05, | |
| "loss": 4.353, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.622768306097567e-05, | |
| "loss": 4.3716, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.621929711346515e-05, | |
| "loss": 4.3565, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.621091116595463e-05, | |
| "loss": 4.3631, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.620252521844411e-05, | |
| "loss": 4.3533, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.619415564973732e-05, | |
| "loss": 4.3598, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.618578608103053e-05, | |
| "loss": 4.3451, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.617740013352001e-05, | |
| "loss": 4.3678, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616901418600949e-05, | |
| "loss": 4.3621, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616062823849897e-05, | |
| "loss": 4.359, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.615224229098845e-05, | |
| "loss": 4.3695, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6143872722281664e-05, | |
| "loss": 4.3428, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6135486774771144e-05, | |
| "loss": 4.3504, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6127100827260624e-05, | |
| "loss": 4.3553, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6118714879750104e-05, | |
| "loss": 4.3428, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6110328932239584e-05, | |
| "loss": 4.3462, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.610195936353279e-05, | |
| "loss": 4.3418, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.609357341602227e-05, | |
| "loss": 4.3438, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.608518746851175e-05, | |
| "loss": 4.3671, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.607680152100123e-05, | |
| "loss": 4.3446, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606841557349071e-05, | |
| "loss": 4.3452, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606002962598019e-05, | |
| "loss": 4.3385, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6051643678469666e-05, | |
| "loss": 4.351, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.604325773095915e-05, | |
| "loss": 4.3354, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.603488816225236e-05, | |
| "loss": 4.3333, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.602650221474184e-05, | |
| "loss": 4.3364, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.601813264603506e-05, | |
| "loss": 4.3444, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.600974669852454e-05, | |
| "loss": 4.3308, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.600136075101402e-05, | |
| "loss": 4.3371, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.599297480350349e-05, | |
| "loss": 4.3369, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.598458885599297e-05, | |
| "loss": 4.3445, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.597620290848245e-05, | |
| "loss": 4.3333, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.596781696097193e-05, | |
| "loss": 4.3391, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595943101346141e-05, | |
| "loss": 4.343, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595106144475462e-05, | |
| "loss": 4.3385, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.59426754972441e-05, | |
| "loss": 4.3181, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5934305928537316e-05, | |
| "loss": 4.3325, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5925919981026796e-05, | |
| "loss": 4.3188, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5917534033516276e-05, | |
| "loss": 4.3351, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5909148086005756e-05, | |
| "loss": 4.3229, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5900762138495235e-05, | |
| "loss": 4.3339, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5892392569788445e-05, | |
| "loss": 4.3248, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5884006622277925e-05, | |
| "loss": 4.3285, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5875620674767404e-05, | |
| "loss": 4.3344, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5867251106060614e-05, | |
| "loss": 4.3298, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5858865158550094e-05, | |
| "loss": 4.333, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5850479211039573e-05, | |
| "loss": 4.3193, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5842093263529053e-05, | |
| "loss": 4.3194, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.583370731601854e-05, | |
| "loss": 4.3371, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.582532136850802e-05, | |
| "loss": 4.3227, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.58169354209975e-05, | |
| "loss": 4.3179, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.580854947348698e-05, | |
| "loss": 4.3195, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.580017990478019e-05, | |
| "loss": 4.3235, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.579179395726967e-05, | |
| "loss": 4.3101, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.578340800975915e-05, | |
| "loss": 4.3288, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.577502206224863e-05, | |
| "loss": 4.3152, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.576663611473811e-05, | |
| "loss": 4.3223, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.575826654603132e-05, | |
| "loss": 4.3252, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.57498805985208e-05, | |
| "loss": 4.3057, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.574149465101028e-05, | |
| "loss": 4.3123, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.573310870349976e-05, | |
| "loss": 4.3163, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5724739134792974e-05, | |
| "loss": 4.3053, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5716353187282454e-05, | |
| "loss": 4.3041, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5707967239771934e-05, | |
| "loss": 4.3149, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5699581292261414e-05, | |
| "loss": 4.316, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5691195344750894e-05, | |
| "loss": 4.3028, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.56828257760441e-05, | |
| "loss": 4.2971, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.567443982853358e-05, | |
| "loss": 4.3132, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.566605388102306e-05, | |
| "loss": 4.324, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.565766793351254e-05, | |
| "loss": 4.3113, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.564928198600202e-05, | |
| "loss": 4.3159, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5640896038491496e-05, | |
| "loss": 4.3197, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5632510090980976e-05, | |
| "loss": 4.3216, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.562412414347046e-05, | |
| "loss": 4.3213, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5615787332371137e-05, | |
| "loss": 4.2949, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5607401384860617e-05, | |
| "loss": 4.3111, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5599015437350096e-05, | |
| "loss": 4.3148, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5590629489839576e-05, | |
| "loss": 4.2972, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5582243542329056e-05, | |
| "loss": 4.3135, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5573857594818536e-05, | |
| "loss": 4.304, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5565471647308016e-05, | |
| "loss": 4.3179, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5557085699797496e-05, | |
| "loss": 4.3161, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554869975228697e-05, | |
| "loss": 4.2923, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5540330183580185e-05, | |
| "loss": 4.3036, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5531944236069665e-05, | |
| "loss": 4.3001, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5523558288559145e-05, | |
| "loss": 4.3059, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5515172341048625e-05, | |
| "loss": 4.3084, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5506786393538105e-05, | |
| "loss": 4.3016, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5498400446027585e-05, | |
| "loss": 4.2996, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5490014498517065e-05, | |
| "loss": 4.2945, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5481628551006545e-05, | |
| "loss": 4.3025, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5473258982299754e-05, | |
| "loss": 4.2939, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5464873034789234e-05, | |
| "loss": 4.304, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5456487087278714e-05, | |
| "loss": 4.3076, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5448101139768194e-05, | |
| "loss": 4.2974, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.54397315710614e-05, | |
| "loss": 4.2935, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.543134562355088e-05, | |
| "loss": 4.2964, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.542295967604037e-05, | |
| "loss": 4.294, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.541457372852985e-05, | |
| "loss": 4.3009, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.540620415982306e-05, | |
| "loss": 4.2949, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.539781821231254e-05, | |
| "loss": 4.289, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.538943226480202e-05, | |
| "loss": 4.2848, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.538106269609523e-05, | |
| "loss": 4.2979, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.537267674858471e-05, | |
| "loss": 4.2783, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.536429080107419e-05, | |
| "loss": 4.2774, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.535590485356367e-05, | |
| "loss": 4.2833, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.534751890605315e-05, | |
| "loss": 4.2896, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533913295854263e-05, | |
| "loss": 4.2862, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533074701103211e-05, | |
| "loss": 4.2978, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.532237744232532e-05, | |
| "loss": 4.2827, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.53139914948148e-05, | |
| "loss": 4.2812, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.530560554730428e-05, | |
| "loss": 4.2844, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.529723597859749e-05, | |
| "loss": 4.2753, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528885003108697e-05, | |
| "loss": 4.2964, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528046408357645e-05, | |
| "loss": 4.2898, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.527207813606593e-05, | |
| "loss": 4.287, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.526369218855541e-05, | |
| "loss": 4.2877, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.525530624104489e-05, | |
| "loss": 4.2787, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.524692029353437e-05, | |
| "loss": 4.2907, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.523853434602385e-05, | |
| "loss": 4.2788, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.523014839851333e-05, | |
| "loss": 4.295, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5221762451002805e-05, | |
| "loss": 4.2765, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5213376503492285e-05, | |
| "loss": 4.286, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.520499055598177e-05, | |
| "loss": 4.2741, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.519662098727499e-05, | |
| "loss": 4.2764, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.518823503976446e-05, | |
| "loss": 4.2703, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517984909225394e-05, | |
| "loss": 4.2831, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517147952354716e-05, | |
| "loss": 4.2731, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.516309357603663e-05, | |
| "loss": 4.2906, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.515470762852611e-05, | |
| "loss": 4.2842, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.514632168101559e-05, | |
| "loss": 4.2796, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.513793573350507e-05, | |
| "loss": 4.2695, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512956616479828e-05, | |
| "loss": 4.2858, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512118021728776e-05, | |
| "loss": 4.2791, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.511279426977724e-05, | |
| "loss": 4.2867, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5104408322266726e-05, | |
| "loss": 4.2747, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5096022374756206e-05, | |
| "loss": 4.2741, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5087652806049415e-05, | |
| "loss": 4.2825, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5079266858538895e-05, | |
| "loss": 4.2769, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5070880911028375e-05, | |
| "loss": 4.2669, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5062494963517855e-05, | |
| "loss": 4.2787, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5054109016007334e-05, | |
| "loss": 4.2693, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5045723068496814e-05, | |
| "loss": 4.2686, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5037353499790024e-05, | |
| "loss": 4.2724, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.502898393108323e-05, | |
| "loss": 4.279, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.502059798357271e-05, | |
| "loss": 4.265, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.501221203606219e-05, | |
| "loss": 4.2639, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.500382608855168e-05, | |
| "loss": 4.2709, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.252410411834717, | |
| "eval_runtime": 293.7422, | |
| "eval_samples_per_second": 1299.068, | |
| "eval_steps_per_second": 40.597, | |
| "step": 305280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.499544014104116e-05, | |
| "loss": 4.2651, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.498705419353064e-05, | |
| "loss": 4.264, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.497866824602012e-05, | |
| "loss": 4.2782, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.49702822985096e-05, | |
| "loss": 4.2651, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.496189635099908e-05, | |
| "loss": 4.2766, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.495351040348856e-05, | |
| "loss": 4.2635, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.494512445597804e-05, | |
| "loss": 4.2684, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.493673850846752e-05, | |
| "loss": 4.2561, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.492835256095699e-05, | |
| "loss": 4.2797, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491996661344647e-05, | |
| "loss": 4.2705, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491158066593595e-05, | |
| "loss": 4.2685, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.490319471842543e-05, | |
| "loss": 4.2813, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.489482514971865e-05, | |
| "loss": 4.2536, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.488643920220813e-05, | |
| "loss": 4.2671, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.487805325469761e-05, | |
| "loss": 4.267, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486966730718709e-05, | |
| "loss": 4.2557, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486128135967657e-05, | |
| "loss": 4.2588, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.485289541216605e-05, | |
| "loss": 4.2566, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.484450946465553e-05, | |
| "loss": 4.2586, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.483612351714501e-05, | |
| "loss": 4.2765, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4827753948438217e-05, | |
| "loss": 4.2592, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4819368000927697e-05, | |
| "loss": 4.2597, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4810982053417177e-05, | |
| "loss": 4.2536, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4802596105906656e-05, | |
| "loss": 4.2666, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4794210158396136e-05, | |
| "loss": 4.2521, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4785824210885616e-05, | |
| "loss": 4.2512, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4777438263375096e-05, | |
| "loss": 4.2496, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4769052315864576e-05, | |
| "loss": 4.2593, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.476068274715779e-05, | |
| "loss": 4.2454, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.475229679964727e-05, | |
| "loss": 4.2551, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.474391085213675e-05, | |
| "loss": 4.2538, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.473552490462623e-05, | |
| "loss": 4.2608, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.472715533591944e-05, | |
| "loss": 4.2489, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.471876938840892e-05, | |
| "loss": 4.2544, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.471039981970213e-05, | |
| "loss": 4.2596, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.470201387219161e-05, | |
| "loss": 4.2548, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.469362792468109e-05, | |
| "loss": 4.2393, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.468524197717057e-05, | |
| "loss": 4.2478, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.467685602966005e-05, | |
| "loss": 4.2374, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466847008214953e-05, | |
| "loss": 4.2498, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466008413463901e-05, | |
| "loss": 4.2431, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.465169818712849e-05, | |
| "loss": 4.2519, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4643328618421706e-05, | |
| "loss": 4.2434, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.463494267091118e-05, | |
| "loss": 4.25, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.462655672340066e-05, | |
| "loss": 4.2503, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4618187154693875e-05, | |
| "loss": 4.2485, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4609817585987084e-05, | |
| "loss": 4.2536, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4601431638476564e-05, | |
| "loss": 4.2425, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4593045690966044e-05, | |
| "loss": 4.232, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4584659743455524e-05, | |
| "loss": 4.2586, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4576273795945004e-05, | |
| "loss": 4.2447, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4567887848434484e-05, | |
| "loss": 4.2388, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4559501900923964e-05, | |
| "loss": 4.2401, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4551115953413444e-05, | |
| "loss": 4.2455, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4542730005902923e-05, | |
| "loss": 4.2297, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4534344058392403e-05, | |
| "loss": 4.2491, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4525958110881883e-05, | |
| "loss": 4.2366, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.451757216337136e-05, | |
| "loss": 4.2423, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.45092189734683e-05, | |
| "loss": 4.2507, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.450083302595778e-05, | |
| "loss": 4.2327, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.449244707844726e-05, | |
| "loss": 4.2311, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.448406113093674e-05, | |
| "loss": 4.2453, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.447567518342622e-05, | |
| "loss": 4.2224, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.44672892359157e-05, | |
| "loss": 4.2272, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445891966720892e-05, | |
| "loss": 4.2367, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.44505337196984e-05, | |
| "loss": 4.2395, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.444214777218788e-05, | |
| "loss": 4.2249, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.443376182467736e-05, | |
| "loss": 4.223, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4425392255970566e-05, | |
| "loss": 4.2341, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4417006308460046e-05, | |
| "loss": 4.245, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4408620360949526e-05, | |
| "loss": 4.238, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4400234413439006e-05, | |
| "loss": 4.237, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4391848465928486e-05, | |
| "loss": 4.2448, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4383462518417966e-05, | |
| "loss": 4.2427, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4375092949711175e-05, | |
| "loss": 4.2502, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4366707002200655e-05, | |
| "loss": 4.2219, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.435833743349387e-05, | |
| "loss": 4.2368, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.434995148598335e-05, | |
| "loss": 4.2387, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.434156553847283e-05, | |
| "loss": 4.2215, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.433317959096231e-05, | |
| "loss": 4.2409, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.432479364345179e-05, | |
| "loss": 4.2309, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.431640769594127e-05, | |
| "loss": 4.2425, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.430802174843075e-05, | |
| "loss": 4.2449, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.429963580092023e-05, | |
| "loss": 4.2193, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.429124985340971e-05, | |
| "loss": 4.2324, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.428286390589919e-05, | |
| "loss": 4.2236, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.42744943371924e-05, | |
| "loss": 4.2358, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.426610838968188e-05, | |
| "loss": 4.2342, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.425772244217136e-05, | |
| "loss": 4.2279, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424933649466084e-05, | |
| "loss": 4.2282, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424095054715032e-05, | |
| "loss": 4.2241, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.42325645996398e-05, | |
| "loss": 4.2272, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.422417865212928e-05, | |
| "loss": 4.2226, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.421580908342249e-05, | |
| "loss": 4.2348, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.420742313591197e-05, | |
| "loss": 4.2359, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419903718840145e-05, | |
| "loss": 4.2267, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4190667619694664e-05, | |
| "loss": 4.2209, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.418228167218414e-05, | |
| "loss": 4.2248, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.417389572467362e-05, | |
| "loss": 4.2253, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.41655097771631e-05, | |
| "loss": 4.2304, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.415714020845631e-05, | |
| "loss": 4.2263, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.414875426094579e-05, | |
| "loss": 4.219, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.414036831343527e-05, | |
| "loss": 4.2132, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.413199874472849e-05, | |
| "loss": 4.2267, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.412361279721796e-05, | |
| "loss": 4.2102, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.411522684970744e-05, | |
| "loss": 4.2139, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.410684090219692e-05, | |
| "loss": 4.2092, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.40984549546864e-05, | |
| "loss": 4.2235, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.409006900717588e-05, | |
| "loss": 4.2116, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.408168305966536e-05, | |
| "loss": 4.2321, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.407329711215484e-05, | |
| "loss": 4.2104, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.406491116464432e-05, | |
| "loss": 4.2157, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.40565252171338e-05, | |
| "loss": 4.2179, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.404813926962328e-05, | |
| "loss": 4.205, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403975332211276e-05, | |
| "loss": 4.2286, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403138375340598e-05, | |
| "loss": 4.2233, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.402301418469919e-05, | |
| "loss": 4.2186, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.401462823718867e-05, | |
| "loss": 4.2187, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4006242289678147e-05, | |
| "loss": 4.2111, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3997856342167627e-05, | |
| "loss": 4.2234, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3989470394657106e-05, | |
| "loss": 4.2026, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3981084447146586e-05, | |
| "loss": 4.2287, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3972698499636066e-05, | |
| "loss": 4.2127, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3964312552125546e-05, | |
| "loss": 4.2154, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3955942983418755e-05, | |
| "loss": 4.2096, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3947557035908235e-05, | |
| "loss": 4.2092, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3939171088397715e-05, | |
| "loss": 4.2018, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3930785140887195e-05, | |
| "loss": 4.2179, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3922399193376675e-05, | |
| "loss": 4.2069, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3914013245866155e-05, | |
| "loss": 4.223, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3905627298355635e-05, | |
| "loss": 4.2167, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3897241350845115e-05, | |
| "loss": 4.2153, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3888871782138324e-05, | |
| "loss": 4.2026, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3880485834627804e-05, | |
| "loss": 4.2176, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3872099887117284e-05, | |
| "loss": 4.2153, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3863713939606764e-05, | |
| "loss": 4.2186, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.385534437089997e-05, | |
| "loss": 4.2088, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.384695842338945e-05, | |
| "loss": 4.2078, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.383857247587893e-05, | |
| "loss": 4.2148, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.383020290717215e-05, | |
| "loss": 4.2171, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.382181695966163e-05, | |
| "loss": 4.1986, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.381343101215111e-05, | |
| "loss": 4.2147, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.380504506464059e-05, | |
| "loss": 4.2021, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.379665911713007e-05, | |
| "loss": 4.2035, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.378828954842328e-05, | |
| "loss": 4.2144, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.377990360091276e-05, | |
| "loss": 4.2064, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.377151765340224e-05, | |
| "loss": 4.2033, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.376313170589172e-05, | |
| "loss": 4.202, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.37547457583812e-05, | |
| "loss": 4.206, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.197500705718994, | |
| "eval_runtime": 351.9654, | |
| "eval_samples_per_second": 1084.172, | |
| "eval_steps_per_second": 33.881, | |
| "step": 381600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.374637618967441e-05, | |
| "loss": 4.2067, | |
| "step": 381952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.373799024216389e-05, | |
| "loss": 4.2, | |
| "step": 382464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3729604294653374e-05, | |
| "loss": 4.2165, | |
| "step": 382976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3721218347142853e-05, | |
| "loss": 4.2049, | |
| "step": 383488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3712832399632333e-05, | |
| "loss": 4.2084, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3704446452121813e-05, | |
| "loss": 4.2055, | |
| "step": 384512 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.369606050461129e-05, | |
| "loss": 4.2012, | |
| "step": 385024 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.36876909359045e-05, | |
| "loss": 4.1959, | |
| "step": 385536 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367930498839398e-05, | |
| "loss": 4.2108, | |
| "step": 386048 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367091904088346e-05, | |
| "loss": 4.2134, | |
| "step": 386560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.366254947217667e-05, | |
| "loss": 4.2047, | |
| "step": 387072 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.365416352466615e-05, | |
| "loss": 4.217, | |
| "step": 387584 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.364579395595936e-05, | |
| "loss": 4.1919, | |
| "step": 388096 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.363740800844884e-05, | |
| "loss": 4.2019, | |
| "step": 388608 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.362902206093832e-05, | |
| "loss": 4.2088, | |
| "step": 389120 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.362063611342781e-05, | |
| "loss": 4.1912, | |
| "step": 389632 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.361225016591729e-05, | |
| "loss": 4.197, | |
| "step": 390144 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.360386421840677e-05, | |
| "loss": 4.1959, | |
| "step": 390656 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.359547827089625e-05, | |
| "loss": 4.1965, | |
| "step": 391168 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.358709232338573e-05, | |
| "loss": 4.2129, | |
| "step": 391680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3578722754678936e-05, | |
| "loss": 4.2029, | |
| "step": 392192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3570336807168416e-05, | |
| "loss": 4.2014, | |
| "step": 392704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3561950859657896e-05, | |
| "loss": 4.1907, | |
| "step": 393216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3553564912147376e-05, | |
| "loss": 4.2073, | |
| "step": 393728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3545178964636856e-05, | |
| "loss": 4.1901, | |
| "step": 394240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3536809395930065e-05, | |
| "loss": 4.1897, | |
| "step": 394752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3528423448419545e-05, | |
| "loss": 4.1912, | |
| "step": 395264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3520037500909025e-05, | |
| "loss": 4.1977, | |
| "step": 395776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3511651553398505e-05, | |
| "loss": 4.1853, | |
| "step": 396288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3503265605887985e-05, | |
| "loss": 4.1948, | |
| "step": 396800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3494879658377465e-05, | |
| "loss": 4.198, | |
| "step": 397312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3486493710866945e-05, | |
| "loss": 4.1981, | |
| "step": 397824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3478107763356425e-05, | |
| "loss": 4.191, | |
| "step": 398336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.346975457345337e-05, | |
| "loss": 4.1945, | |
| "step": 398848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.346136862594285e-05, | |
| "loss": 4.2021, | |
| "step": 399360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.345298267843233e-05, | |
| "loss": 4.193, | |
| "step": 399872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.34445967309218e-05, | |
| "loss": 4.1838, | |
| "step": 400384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.343621078341128e-05, | |
| "loss": 4.1849, | |
| "step": 400896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.342782483590076e-05, | |
| "loss": 4.1803, | |
| "step": 401408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.341943888839024e-05, | |
| "loss": 4.1883, | |
| "step": 401920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.341106931968346e-05, | |
| "loss": 4.1857, | |
| "step": 402432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.340268337217294e-05, | |
| "loss": 4.1934, | |
| "step": 402944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.339429742466242e-05, | |
| "loss": 4.1851, | |
| "step": 403456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.33859114771519e-05, | |
| "loss": 4.1887, | |
| "step": 403968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.337754190844511e-05, | |
| "loss": 4.1964, | |
| "step": 404480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.336915596093459e-05, | |
| "loss": 4.187, | |
| "step": 404992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.336077001342407e-05, | |
| "loss": 4.1962, | |
| "step": 405504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.335238406591355e-05, | |
| "loss": 4.1817, | |
| "step": 406016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.334399811840303e-05, | |
| "loss": 4.1752, | |
| "step": 406528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.333561217089251e-05, | |
| "loss": 4.2035, | |
| "step": 407040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.332722622338199e-05, | |
| "loss": 4.185, | |
| "step": 407552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331884027587147e-05, | |
| "loss": 4.1831, | |
| "step": 408064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331045432836095e-05, | |
| "loss": 4.1793, | |
| "step": 408576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.330206838085043e-05, | |
| "loss": 4.189, | |
| "step": 409088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3293682433339914e-05, | |
| "loss": 4.1715, | |
| "step": 409600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3285296485829394e-05, | |
| "loss": 4.1895, | |
| "step": 410112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.32769269171226e-05, | |
| "loss": 4.1815, | |
| "step": 410624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326855734841581e-05, | |
| "loss": 4.1812, | |
| "step": 411136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326017140090529e-05, | |
| "loss": 4.1983, | |
| "step": 411648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.325178545339477e-05, | |
| "loss": 4.1733, | |
| "step": 412160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.324339950588425e-05, | |
| "loss": 4.1748, | |
| "step": 412672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.323502993717746e-05, | |
| "loss": 4.1896, | |
| "step": 413184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.322664398966694e-05, | |
| "loss": 4.1702, | |
| "step": 413696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.321825804215642e-05, | |
| "loss": 4.1736, | |
| "step": 414208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.32098720946459e-05, | |
| "loss": 4.1765, | |
| "step": 414720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320148614713538e-05, | |
| "loss": 4.1843, | |
| "step": 415232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.319310019962487e-05, | |
| "loss": 4.1722, | |
| "step": 415744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.318471425211435e-05, | |
| "loss": 4.1695, | |
| "step": 416256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.317632830460382e-05, | |
| "loss": 4.1742, | |
| "step": 416768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3167975114700766e-05, | |
| "loss": 4.186, | |
| "step": 417280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3159589167190246e-05, | |
| "loss": 4.1838, | |
| "step": 417792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3151203219679726e-05, | |
| "loss": 4.1843, | |
| "step": 418304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3142817272169205e-05, | |
| "loss": 4.1858, | |
| "step": 418816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3134431324658685e-05, | |
| "loss": 4.1877, | |
| "step": 419328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3126045377148165e-05, | |
| "loss": 4.1957, | |
| "step": 419840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.311765942963764e-05, | |
| "loss": 4.1651, | |
| "step": 420352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310927348212712e-05, | |
| "loss": 4.1853, | |
| "step": 420864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3100903913420334e-05, | |
| "loss": 4.1867, | |
| "step": 421376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3092517965909814e-05, | |
| "loss": 4.1605, | |
| "step": 421888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3084132018399294e-05, | |
| "loss": 4.1893, | |
| "step": 422400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3075746070888774e-05, | |
| "loss": 4.1753, | |
| "step": 422912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3067360123378254e-05, | |
| "loss": 4.1867, | |
| "step": 423424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3058974175867734e-05, | |
| "loss": 4.1917, | |
| "step": 423936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3050588228357214e-05, | |
| "loss": 4.1668, | |
| "step": 424448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.304221865965042e-05, | |
| "loss": 4.1718, | |
| "step": 424960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.30338327121399e-05, | |
| "loss": 4.1705, | |
| "step": 425472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.302544676462938e-05, | |
| "loss": 4.1806, | |
| "step": 425984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.301706081711886e-05, | |
| "loss": 4.1773, | |
| "step": 426496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.300869124841207e-05, | |
| "loss": 4.1768, | |
| "step": 427008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.300030530090155e-05, | |
| "loss": 4.1713, | |
| "step": 427520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.299191935339104e-05, | |
| "loss": 4.172, | |
| "step": 428032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.298353340588052e-05, | |
| "loss": 4.1738, | |
| "step": 428544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.297514745837e-05, | |
| "loss": 4.1677, | |
| "step": 429056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.296677788966321e-05, | |
| "loss": 4.1844, | |
| "step": 429568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.295839194215269e-05, | |
| "loss": 4.1808, | |
| "step": 430080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.295000599464217e-05, | |
| "loss": 4.1782, | |
| "step": 430592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294162004713165e-05, | |
| "loss": 4.1646, | |
| "step": 431104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.293323409962113e-05, | |
| "loss": 4.1718, | |
| "step": 431616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.292484815211061e-05, | |
| "loss": 4.1719, | |
| "step": 432128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.291646220460009e-05, | |
| "loss": 4.1735, | |
| "step": 432640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.29080926358933e-05, | |
| "loss": 4.1755, | |
| "step": 433152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2899723067186506e-05, | |
| "loss": 4.1693, | |
| "step": 433664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289133711967599e-05, | |
| "loss": 4.1607, | |
| "step": 434176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.288295117216547e-05, | |
| "loss": 4.1733, | |
| "step": 434688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.287456522465495e-05, | |
| "loss": 4.1598, | |
| "step": 435200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.286617927714443e-05, | |
| "loss": 4.1594, | |
| "step": 435712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.285779332963391e-05, | |
| "loss": 4.1523, | |
| "step": 436224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284940738212339e-05, | |
| "loss": 4.169, | |
| "step": 436736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284102143461287e-05, | |
| "loss": 4.1644, | |
| "step": 437248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.283265186590608e-05, | |
| "loss": 4.1774, | |
| "step": 437760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.282426591839556e-05, | |
| "loss": 4.1605, | |
| "step": 438272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.281587997088504e-05, | |
| "loss": 4.165, | |
| "step": 438784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.280749402337452e-05, | |
| "loss": 4.1668, | |
| "step": 439296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.279912445466773e-05, | |
| "loss": 4.1527, | |
| "step": 439808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.279073850715721e-05, | |
| "loss": 4.1769, | |
| "step": 440320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.278235255964669e-05, | |
| "loss": 4.1703, | |
| "step": 440832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2773982990939906e-05, | |
| "loss": 4.1662, | |
| "step": 441344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2765597043429386e-05, | |
| "loss": 4.1666, | |
| "step": 441856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2757211095918866e-05, | |
| "loss": 4.1619, | |
| "step": 442368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2748825148408346e-05, | |
| "loss": 4.1726, | |
| "step": 442880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2740439200897826e-05, | |
| "loss": 4.1492, | |
| "step": 443392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.27320532533873e-05, | |
| "loss": 4.1791, | |
| "step": 443904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.272366730587678e-05, | |
| "loss": 4.16, | |
| "step": 444416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.271528135836626e-05, | |
| "loss": 4.1623, | |
| "step": 444928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2706911789659475e-05, | |
| "loss": 4.1615, | |
| "step": 445440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.269852584214895e-05, | |
| "loss": 4.1568, | |
| "step": 445952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.269013989463843e-05, | |
| "loss": 4.1574, | |
| "step": 446464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2681753947127915e-05, | |
| "loss": 4.1644, | |
| "step": 446976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.267338437842113e-05, | |
| "loss": 4.1611, | |
| "step": 447488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2664998430910604e-05, | |
| "loss": 4.1703, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2656612483400084e-05, | |
| "loss": 4.1627, | |
| "step": 448512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2648226535889564e-05, | |
| "loss": 4.1688, | |
| "step": 449024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2639840588379044e-05, | |
| "loss": 4.149, | |
| "step": 449536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2631454640868524e-05, | |
| "loss": 4.1739, | |
| "step": 450048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.262308507216173e-05, | |
| "loss": 4.1616, | |
| "step": 450560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.261469912465121e-05, | |
| "loss": 4.1692, | |
| "step": 451072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.260631317714069e-05, | |
| "loss": 4.1606, | |
| "step": 451584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.259792722963017e-05, | |
| "loss": 4.1573, | |
| "step": 452096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.258954128211965e-05, | |
| "loss": 4.1698, | |
| "step": 452608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.258115533460913e-05, | |
| "loss": 4.1672, | |
| "step": 453120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.257278576590235e-05, | |
| "loss": 4.1513, | |
| "step": 453632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.256439981839183e-05, | |
| "loss": 4.1606, | |
| "step": 454144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.255601387088131e-05, | |
| "loss": 4.1572, | |
| "step": 454656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.254762792337079e-05, | |
| "loss": 4.1528, | |
| "step": 455168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253924197586027e-05, | |
| "loss": 4.1658, | |
| "step": 455680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.253085602834975e-05, | |
| "loss": 4.1555, | |
| "step": 456192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.252247008083923e-05, | |
| "loss": 4.1608, | |
| "step": 456704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.251410051213244e-05, | |
| "loss": 4.147, | |
| "step": 457216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.250571456462192e-05, | |
| "loss": 4.1616, | |
| "step": 457728 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.157258033752441, | |
| "eval_runtime": 324.8956, | |
| "eval_samples_per_second": 1174.503, | |
| "eval_steps_per_second": 36.704, | |
| "step": 457920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.24973286171114e-05, | |
| "loss": 4.1593, | |
| "step": 458240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.248894266960088e-05, | |
| "loss": 4.149, | |
| "step": 458752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.248055672209036e-05, | |
| "loss": 4.1675, | |
| "step": 459264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.247217077457984e-05, | |
| "loss": 4.1591, | |
| "step": 459776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.246378482706932e-05, | |
| "loss": 4.1597, | |
| "step": 460288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.24553988795588e-05, | |
| "loss": 4.1589, | |
| "step": 460800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.244701293204828e-05, | |
| "loss": 4.1483, | |
| "step": 461312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243862698453776e-05, | |
| "loss": 4.1523, | |
| "step": 461824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243024103702724e-05, | |
| "loss": 4.1594, | |
| "step": 462336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.242185508951672e-05, | |
| "loss": 4.1684, | |
| "step": 462848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.24134691420062e-05, | |
| "loss": 4.1582, | |
| "step": 463360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.240508319449568e-05, | |
| "loss": 4.1695, | |
| "step": 463872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2396713625788886e-05, | |
| "loss": 4.1488, | |
| "step": 464384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2388327678278366e-05, | |
| "loss": 4.151, | |
| "step": 464896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2379941730767846e-05, | |
| "loss": 4.163, | |
| "step": 465408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2371555783257326e-05, | |
| "loss": 4.1437, | |
| "step": 465920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2363169835746806e-05, | |
| "loss": 4.1509, | |
| "step": 466432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2354783888236286e-05, | |
| "loss": 4.1477, | |
| "step": 466944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2346397940725765e-05, | |
| "loss": 4.1529, | |
| "step": 467456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2338011993215245e-05, | |
| "loss": 4.1618, | |
| "step": 467968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.232962604570473e-05, | |
| "loss": 4.1586, | |
| "step": 468480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.232125647699794e-05, | |
| "loss": 4.1558, | |
| "step": 468992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.231287052948742e-05, | |
| "loss": 4.141, | |
| "step": 469504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.23044845819769e-05, | |
| "loss": 4.1598, | |
| "step": 470016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2296098634466374e-05, | |
| "loss": 4.1484, | |
| "step": 470528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2287712686955854e-05, | |
| "loss": 4.1457, | |
| "step": 471040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2279326739445334e-05, | |
| "loss": 4.1426, | |
| "step": 471552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2270940791934814e-05, | |
| "loss": 4.147, | |
| "step": 472064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2262554844424294e-05, | |
| "loss": 4.1486, | |
| "step": 472576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2254168896913774e-05, | |
| "loss": 4.1425, | |
| "step": 473088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.224579932820698e-05, | |
| "loss": 4.1522, | |
| "step": 473600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.223741338069647e-05, | |
| "loss": 4.1537, | |
| "step": 474112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.222902743318595e-05, | |
| "loss": 4.146, | |
| "step": 474624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.222064148567543e-05, | |
| "loss": 4.1527, | |
| "step": 475136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.221227191696864e-05, | |
| "loss": 4.1536, | |
| "step": 475648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.220388596945812e-05, | |
| "loss": 4.1489, | |
| "step": 476160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.219551640075133e-05, | |
| "loss": 4.1444, | |
| "step": 476672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.218713045324081e-05, | |
| "loss": 4.1334, | |
| "step": 477184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.217874450573029e-05, | |
| "loss": 4.1331, | |
| "step": 477696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.217035855821977e-05, | |
| "loss": 4.1432, | |
| "step": 478208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.216197261070925e-05, | |
| "loss": 4.1426, | |
| "step": 478720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.215361942080619e-05, | |
| "loss": 4.1463, | |
| "step": 479232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.214523347329567e-05, | |
| "loss": 4.1409, | |
| "step": 479744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.213684752578515e-05, | |
| "loss": 4.1435, | |
| "step": 480256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.212846157827463e-05, | |
| "loss": 4.1533, | |
| "step": 480768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.212007563076411e-05, | |
| "loss": 4.1425, | |
| "step": 481280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.211168968325359e-05, | |
| "loss": 4.1534, | |
| "step": 481792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.210330373574307e-05, | |
| "loss": 4.1382, | |
| "step": 482304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.209491778823255e-05, | |
| "loss": 4.1285, | |
| "step": 482816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.208653184072203e-05, | |
| "loss": 4.1589, | |
| "step": 483328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.207814589321151e-05, | |
| "loss": 4.1435, | |
| "step": 483840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.206977632450472e-05, | |
| "loss": 4.1347, | |
| "step": 484352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.20613903769942e-05, | |
| "loss": 4.1391, | |
| "step": 484864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.205300442948368e-05, | |
| "loss": 4.1409, | |
| "step": 485376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.204461848197316e-05, | |
| "loss": 4.1318, | |
| "step": 485888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.203623253446264e-05, | |
| "loss": 4.1412, | |
| "step": 486400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.202784658695212e-05, | |
| "loss": 4.1389, | |
| "step": 486912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.20194606394416e-05, | |
| "loss": 4.1407, | |
| "step": 487424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201109107073482e-05, | |
| "loss": 4.151, | |
| "step": 487936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.20027051232243e-05, | |
| "loss": 4.128, | |
| "step": 488448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.199431917571378e-05, | |
| "loss": 4.1321, | |
| "step": 488960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.198593322820326e-05, | |
| "loss": 4.1484, | |
| "step": 489472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.197754728069274e-05, | |
| "loss": 4.1276, | |
| "step": 489984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.196916133318221e-05, | |
| "loss": 4.1258, | |
| "step": 490496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.196077538567169e-05, | |
| "loss": 4.1337, | |
| "step": 491008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1952405816964906e-05, | |
| "loss": 4.1398, | |
| "step": 491520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1944019869454386e-05, | |
| "loss": 4.1275, | |
| "step": 492032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.193563392194386e-05, | |
| "loss": 4.1306, | |
| "step": 492544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.192724797443334e-05, | |
| "loss": 4.1252, | |
| "step": 493056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1918862026922826e-05, | |
| "loss": 4.1476, | |
| "step": 493568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1910476079412306e-05, | |
| "loss": 4.1388, | |
| "step": 494080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1902090131901786e-05, | |
| "loss": 4.1397, | |
| "step": 494592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1893720563194995e-05, | |
| "loss": 4.1416, | |
| "step": 495104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1885334615684475e-05, | |
| "loss": 4.144, | |
| "step": 495616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1876948668173955e-05, | |
| "loss": 4.153, | |
| "step": 496128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1868579099467164e-05, | |
| "loss": 4.124, | |
| "step": 496640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1860193151956644e-05, | |
| "loss": 4.1412, | |
| "step": 497152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.185182358324986e-05, | |
| "loss": 4.1442, | |
| "step": 497664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.184343763573933e-05, | |
| "loss": 4.1222, | |
| "step": 498176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.183505168822881e-05, | |
| "loss": 4.1409, | |
| "step": 498688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.182666574071829e-05, | |
| "loss": 4.1341, | |
| "step": 499200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.181827979320778e-05, | |
| "loss": 4.1458, | |
| "step": 499712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.180989384569726e-05, | |
| "loss": 4.1474, | |
| "step": 500224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.180150789818674e-05, | |
| "loss": 4.1294, | |
| "step": 500736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.179312195067622e-05, | |
| "loss": 4.1256, | |
| "step": 501248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.17847360031657e-05, | |
| "loss": 4.1288, | |
| "step": 501760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.177635005565518e-05, | |
| "loss": 4.1401, | |
| "step": 502272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.176796410814466e-05, | |
| "loss": 4.1318, | |
| "step": 502784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.175959453943787e-05, | |
| "loss": 4.1338, | |
| "step": 503296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.175120859192735e-05, | |
| "loss": 4.1327, | |
| "step": 503808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.174282264441683e-05, | |
| "loss": 4.1278, | |
| "step": 504320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.173443669690631e-05, | |
| "loss": 4.1343, | |
| "step": 504832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.172605074939579e-05, | |
| "loss": 4.1273, | |
| "step": 505344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1717681180689e-05, | |
| "loss": 4.1422, | |
| "step": 505856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.170929523317848e-05, | |
| "loss": 4.1375, | |
| "step": 506368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1700909285667964e-05, | |
| "loss": 4.1367, | |
| "step": 506880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1692523338157444e-05, | |
| "loss": 4.1233, | |
| "step": 507392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.168415376945065e-05, | |
| "loss": 4.1294, | |
| "step": 507904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.167576782194013e-05, | |
| "loss": 4.1331, | |
| "step": 508416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.166738187442961e-05, | |
| "loss": 4.1321, | |
| "step": 508928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165899592691909e-05, | |
| "loss": 4.1351, | |
| "step": 509440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165060997940857e-05, | |
| "loss": 4.1231, | |
| "step": 509952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1642224031898046e-05, | |
| "loss": 4.1248, | |
| "step": 510464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1633838084387526e-05, | |
| "loss": 4.1294, | |
| "step": 510976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1625452136877006e-05, | |
| "loss": 4.1166, | |
| "step": 511488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1617066189366486e-05, | |
| "loss": 4.1266, | |
| "step": 512000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.16086966206597e-05, | |
| "loss": 4.1047, | |
| "step": 512512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160031067314918e-05, | |
| "loss": 4.1308, | |
| "step": 513024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.159192472563866e-05, | |
| "loss": 4.125, | |
| "step": 513536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.158353877812814e-05, | |
| "loss": 4.1329, | |
| "step": 514048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.157515283061762e-05, | |
| "loss": 4.1183, | |
| "step": 514560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.15667668831071e-05, | |
| "loss": 4.125, | |
| "step": 515072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.155838093559658e-05, | |
| "loss": 4.1274, | |
| "step": 515584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.154999498808606e-05, | |
| "loss": 4.1097, | |
| "step": 516096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.154162541937927e-05, | |
| "loss": 4.1399, | |
| "step": 516608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.153325585067248e-05, | |
| "loss": 4.1284, | |
| "step": 517120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.152486990316196e-05, | |
| "loss": 4.1275, | |
| "step": 517632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.151648395565144e-05, | |
| "loss": 4.1304, | |
| "step": 518144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1508114386944655e-05, | |
| "loss": 4.1141, | |
| "step": 518656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1499728439434135e-05, | |
| "loss": 4.1355, | |
| "step": 519168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1491342491923615e-05, | |
| "loss": 4.1141, | |
| "step": 519680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1482956544413095e-05, | |
| "loss": 4.1367, | |
| "step": 520192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1474586975706304e-05, | |
| "loss": 4.1168, | |
| "step": 520704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1466201028195784e-05, | |
| "loss": 4.1257, | |
| "step": 521216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1457815080685264e-05, | |
| "loss": 4.119, | |
| "step": 521728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1449429133174744e-05, | |
| "loss": 4.1202, | |
| "step": 522240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1441043185664224e-05, | |
| "loss": 4.1193, | |
| "step": 522752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1432657238153704e-05, | |
| "loss": 4.1212, | |
| "step": 523264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1424271290643184e-05, | |
| "loss": 4.1236, | |
| "step": 523776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1415885343132664e-05, | |
| "loss": 4.1292, | |
| "step": 524288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1407499395622144e-05, | |
| "loss": 4.1228, | |
| "step": 524800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1399113448111624e-05, | |
| "loss": 4.1322, | |
| "step": 525312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.139074387940483e-05, | |
| "loss": 4.1102, | |
| "step": 525824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.138235793189432e-05, | |
| "loss": 4.1325, | |
| "step": 526336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.13739719843838e-05, | |
| "loss": 4.1271, | |
| "step": 526848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.136558603687328e-05, | |
| "loss": 4.1262, | |
| "step": 527360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.135720008936276e-05, | |
| "loss": 4.1209, | |
| "step": 527872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.134881414185223e-05, | |
| "loss": 4.1219, | |
| "step": 528384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.134044457314545e-05, | |
| "loss": 4.1253, | |
| "step": 528896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.133205862563493e-05, | |
| "loss": 4.1314, | |
| "step": 529408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.132367267812441e-05, | |
| "loss": 4.1108, | |
| "step": 529920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.131528673061388e-05, | |
| "loss": 4.1239, | |
| "step": 530432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.130690078310336e-05, | |
| "loss": 4.1181, | |
| "step": 530944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.129851483559284e-05, | |
| "loss": 4.1166, | |
| "step": 531456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.129016164568979e-05, | |
| "loss": 4.1221, | |
| "step": 531968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1281775698179273e-05, | |
| "loss": 4.1175, | |
| "step": 532480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1273389750668753e-05, | |
| "loss": 4.1216, | |
| "step": 532992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.126500380315823e-05, | |
| "loss": 4.1088, | |
| "step": 533504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1256617855647707e-05, | |
| "loss": 4.1284, | |
| "step": 534016 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.126788139343262, | |
| "eval_runtime": 311.1873, | |
| "eval_samples_per_second": 1226.242, | |
| "eval_steps_per_second": 38.321, | |
| "step": 534240 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 2.2164435046184458e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |