| { | |
| "best_metric": 4.048644065856934, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl2/lstm/1/checkpoints/checkpoint-992160", | |
| "epoch": 0.025000606015738065, | |
| "eval_steps": 10, | |
| "global_step": 992160, | |
| "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": 10.8195, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 7.5561, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 7.0599, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 6.9972, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 6.9475, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 6.9188, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 6.7574, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994131474623009e-05, | |
| "loss": 6.6474, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993292879871958e-05, | |
| "loss": 6.5534, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992454285120906e-05, | |
| "loss": 6.4733, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.991615690369854e-05, | |
| "loss": 6.4162, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990777095618801e-05, | |
| "loss": 6.3506, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989938500867749e-05, | |
| "loss": 6.2788, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989099906116697e-05, | |
| "loss": 6.2098, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988261311365645e-05, | |
| "loss": 6.154, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987422716614593e-05, | |
| "loss": 6.0923, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986584121863541e-05, | |
| "loss": 6.0503, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985745527112489e-05, | |
| "loss": 6.0096, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984906932361437e-05, | |
| "loss": 5.963, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984068337610385e-05, | |
| "loss": 5.9206, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.8923, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.982391148108281e-05, | |
| "loss": 5.8495, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.981552553357229e-05, | |
| "loss": 5.8265, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.980713958606178e-05, | |
| "loss": 5.788, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.979875363855125e-05, | |
| "loss": 5.7724, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.979036769104073e-05, | |
| "loss": 5.7327, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.978198174353021e-05, | |
| "loss": 5.7102, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773612174823426e-05, | |
| "loss": 5.6808, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.97652262273129e-05, | |
| "loss": 5.6655, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.975684027980238e-05, | |
| "loss": 5.6388, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.974845433229186e-05, | |
| "loss": 5.6265, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9740101142388804e-05, | |
| "loss": 5.6087, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9731715194878284e-05, | |
| "loss": 5.5848, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9723329247367764e-05, | |
| "loss": 5.5705, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714943299857244e-05, | |
| "loss": 5.5555, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.970657373115046e-05, | |
| "loss": 5.5334, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.969818778363994e-05, | |
| "loss": 5.5301, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968980183612942e-05, | |
| "loss": 5.4959, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96814158886189e-05, | |
| "loss": 5.4843, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.967302994110837e-05, | |
| "loss": 5.4547, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966467675120532e-05, | |
| "loss": 5.4546, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96562908036948e-05, | |
| "loss": 5.4389, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964790485618428e-05, | |
| "loss": 5.4366, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963951890867376e-05, | |
| "loss": 5.409, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963113296116324e-05, | |
| "loss": 5.3987, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962274701365272e-05, | |
| "loss": 5.3976, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96143610661422e-05, | |
| "loss": 5.3891, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9605991497435414e-05, | |
| "loss": 5.3773, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9597605549924894e-05, | |
| "loss": 5.3578, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9589219602414374e-05, | |
| "loss": 5.3439, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958083365490385e-05, | |
| "loss": 5.3611, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.957244770739333e-05, | |
| "loss": 5.3472, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956406175988281e-05, | |
| "loss": 5.3271, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955567581237229e-05, | |
| "loss": 5.3072, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.954728986486177e-05, | |
| "loss": 5.3151, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9538920296154976e-05, | |
| "loss": 5.294, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9530534348644456e-05, | |
| "loss": 5.3014, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9522148401133936e-05, | |
| "loss": 5.27, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951376245362342e-05, | |
| "loss": 5.2815, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.950539288491663e-05, | |
| "loss": 5.266, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949702331620985e-05, | |
| "loss": 5.2482, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948863736869932e-05, | |
| "loss": 5.235, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94802514211888e-05, | |
| "loss": 5.2471, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947186547367828e-05, | |
| "loss": 5.2046, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946347952616776e-05, | |
| "loss": 5.2193, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945509357865724e-05, | |
| "loss": 5.2267, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944670763114672e-05, | |
| "loss": 5.2094, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94383216836362e-05, | |
| "loss": 5.2021, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942995211492941e-05, | |
| "loss": 5.1768, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942156616741889e-05, | |
| "loss": 5.1731, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.941318021990837e-05, | |
| "loss": 5.171, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9404794272397856e-05, | |
| "loss": 5.1818, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9396408324887336e-05, | |
| "loss": 5.1619, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388038756180545e-05, | |
| "loss": 5.1702, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379652808670025e-05, | |
| "loss": 5.1592, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371283239963234e-05, | |
| "loss": 5.1568, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362897292452714e-05, | |
| "loss": 5.1342, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9354511344942194e-05, | |
| "loss": 5.1321, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9346125397431674e-05, | |
| "loss": 5.1243, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9337739449921154e-05, | |
| "loss": 5.1194, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329353502410634e-05, | |
| "loss": 5.1185, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9320967554900114e-05, | |
| "loss": 5.1013, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9312581607389594e-05, | |
| "loss": 5.1026, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.930421203868281e-05, | |
| "loss": 5.1017, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.929582609117229e-05, | |
| "loss": 5.0774, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.928744014366177e-05, | |
| "loss": 5.0845, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927905419615125e-05, | |
| "loss": 5.0835, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927066824864073e-05, | |
| "loss": 5.0844, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.92622823011302e-05, | |
| "loss": 5.0655, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.925389635361968e-05, | |
| "loss": 5.062, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.924551040610916e-05, | |
| "loss": 5.0485, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923712445859864e-05, | |
| "loss": 5.0456, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922875488989185e-05, | |
| "loss": 5.0451, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922036894238133e-05, | |
| "loss": 5.0429, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921198299487081e-05, | |
| "loss": 5.0349, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920359704736029e-05, | |
| "loss": 5.037, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919522747865351e-05, | |
| "loss": 5.0276, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918684153114299e-05, | |
| "loss": 5.0119, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917845558363247e-05, | |
| "loss": 5.0076, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917006963612195e-05, | |
| "loss": 5.0109, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.916168368861143e-05, | |
| "loss": 5.0064, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9153314119904636e-05, | |
| "loss": 5.0103, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9144928172394116e-05, | |
| "loss": 4.9923, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9136558603687325e-05, | |
| "loss": 4.9852, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9128172656176805e-05, | |
| "loss": 4.9892, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9119786708666285e-05, | |
| "loss": 4.9695, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9111400761155765e-05, | |
| "loss": 4.975, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9103014813645245e-05, | |
| "loss": 4.9649, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.909462886613473e-05, | |
| "loss": 4.9654, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.908624291862421e-05, | |
| "loss": 4.9521, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907785697111369e-05, | |
| "loss": 4.9555, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906947102360317e-05, | |
| "loss": 4.9455, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906110145489638e-05, | |
| "loss": 4.9481, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905271550738586e-05, | |
| "loss": 4.9428, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904432955987534e-05, | |
| "loss": 4.9335, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903594361236482e-05, | |
| "loss": 4.9442, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.902757404365803e-05, | |
| "loss": 4.9418, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901918809614751e-05, | |
| "loss": 4.9294, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901080214863699e-05, | |
| "loss": 4.9164, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.900241620112647e-05, | |
| "loss": 4.9174, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899403025361595e-05, | |
| "loss": 4.9151, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8985660684909166e-05, | |
| "loss": 4.9168, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8977274737398646e-05, | |
| "loss": 4.9108, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968888789888125e-05, | |
| "loss": 4.9012, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8960502842377605e-05, | |
| "loss": 4.9047, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8952116894867085e-05, | |
| "loss": 4.8905, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8943730947356565e-05, | |
| "loss": 4.8952, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.893534499984604e-05, | |
| "loss": 4.8811, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.892695905233552e-05, | |
| "loss": 4.8873, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918573104825e-05, | |
| "loss": 4.876, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8910203536118214e-05, | |
| "loss": 4.8856, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890181758860769e-05, | |
| "loss": 4.8846, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889343164109717e-05, | |
| "loss": 4.882, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888504569358665e-05, | |
| "loss": 4.8584, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8876659746076134e-05, | |
| "loss": 4.8639, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8868273798565614e-05, | |
| "loss": 4.862, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8859887851055094e-05, | |
| "loss": 4.8639, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88515182823483e-05, | |
| "loss": 4.852, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884313233483778e-05, | |
| "loss": 4.8572, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883474638732726e-05, | |
| "loss": 4.8458, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882636043981674e-05, | |
| "loss": 4.861, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881797449230622e-05, | |
| "loss": 4.8289, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880960492359943e-05, | |
| "loss": 4.8484, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880121897608891e-05, | |
| "loss": 4.8358, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879283302857839e-05, | |
| "loss": 4.8287, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.878444708106787e-05, | |
| "loss": 4.8282, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877607751236109e-05, | |
| "loss": 4.813, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.876769156485057e-05, | |
| "loss": 4.8183, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875932199614378e-05, | |
| "loss": 4.8269, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875093604863326e-05, | |
| "loss": 4.8123, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.779831886291504, | |
| "eval_runtime": 288.1178, | |
| "eval_samples_per_second": 1324.427, | |
| "eval_steps_per_second": 41.389, | |
| "step": 76320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.874255010112274e-05, | |
| "loss": 4.8111, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.873416415361222e-05, | |
| "loss": 4.8048, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.87257782061017e-05, | |
| "loss": 4.8175, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.871739225859118e-05, | |
| "loss": 4.8068, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8709006311080657e-05, | |
| "loss": 4.8127, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8700636742373866e-05, | |
| "loss": 4.7888, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8692250794863346e-05, | |
| "loss": 4.796, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8683864847352826e-05, | |
| "loss": 4.7801, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8675478899842306e-05, | |
| "loss": 4.7839, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8667092952331785e-05, | |
| "loss": 4.788, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865870700482127e-05, | |
| "loss": 4.7842, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865033743611448e-05, | |
| "loss": 4.7876, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.864196786740769e-05, | |
| "loss": 4.7664, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.863358191989717e-05, | |
| "loss": 4.7803, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.862519597238665e-05, | |
| "loss": 4.7618, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.861682640367986e-05, | |
| "loss": 4.7667, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.860844045616934e-05, | |
| "loss": 4.7686, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.860005450865882e-05, | |
| "loss": 4.7593, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.85916685611483e-05, | |
| "loss": 4.7627, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.858328261363778e-05, | |
| "loss": 4.775, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.857489666612726e-05, | |
| "loss": 4.7505, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.856651071861674e-05, | |
| "loss": 4.7575, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8558124771106226e-05, | |
| "loss": 4.7503, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.85497388235957e-05, | |
| "loss": 4.7591, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.854135287608518e-05, | |
| "loss": 4.748, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.853296692857466e-05, | |
| "loss": 4.7478, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.852458098106414e-05, | |
| "loss": 4.7309, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.851619503355362e-05, | |
| "loss": 4.7403, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.85078090860431e-05, | |
| "loss": 4.7271, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.849942313853258e-05, | |
| "loss": 4.738, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.849105356982579e-05, | |
| "loss": 4.7333, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.848266762231527e-05, | |
| "loss": 4.7318, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.847428167480475e-05, | |
| "loss": 4.7342, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.846589572729423e-05, | |
| "loss": 4.725, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8457526158587444e-05, | |
| "loss": 4.7303, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8449140211076924e-05, | |
| "loss": 4.7232, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8440754263566404e-05, | |
| "loss": 4.7129, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8432368316055884e-05, | |
| "loss": 4.7066, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8423982368545363e-05, | |
| "loss": 4.6953, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8415596421034843e-05, | |
| "loss": 4.7051, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.840721047352432e-05, | |
| "loss": 4.7039, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839884090481753e-05, | |
| "loss": 4.7105, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839045495730701e-05, | |
| "loss": 4.6933, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.838206900979649e-05, | |
| "loss": 4.6952, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.837368306228597e-05, | |
| "loss": 4.6984, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.836529711477545e-05, | |
| "loss": 4.6911, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.835691116726493e-05, | |
| "loss": 4.6973, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834852521975441e-05, | |
| "loss": 4.6833, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834013927224389e-05, | |
| "loss": 4.675, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.833176970353711e-05, | |
| "loss": 4.7029, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.832338375602659e-05, | |
| "loss": 4.6967, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.831499780851606e-05, | |
| "loss": 4.6819, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.830661186100554e-05, | |
| "loss": 4.678, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.829822591349502e-05, | |
| "loss": 4.6853, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.828985634478824e-05, | |
| "loss": 4.6695, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.828147039727771e-05, | |
| "loss": 4.687, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.827308444976719e-05, | |
| "loss": 4.6578, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.826469850225667e-05, | |
| "loss": 4.682, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.825631255474615e-05, | |
| "loss": 4.6739, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.824792660723563e-05, | |
| "loss": 4.6568, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823954065972511e-05, | |
| "loss": 4.6549, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.82311547122146e-05, | |
| "loss": 4.6691, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8222801522311535e-05, | |
| "loss": 4.6315, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8214415574801015e-05, | |
| "loss": 4.654, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8206029627290495e-05, | |
| "loss": 4.6645, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8197643679779975e-05, | |
| "loss": 4.656, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8189274111073184e-05, | |
| "loss": 4.6474, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8180888163562664e-05, | |
| "loss": 4.6334, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8172502216052144e-05, | |
| "loss": 4.639, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8164116268541624e-05, | |
| "loss": 4.6418, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.815574669983483e-05, | |
| "loss": 4.6485, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.814736075232432e-05, | |
| "loss": 4.6363, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.81389748048138e-05, | |
| "loss": 4.6456, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.813058885730328e-05, | |
| "loss": 4.649, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.812220290979276e-05, | |
| "loss": 4.6462, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.811381696228224e-05, | |
| "loss": 4.6283, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.810543101477172e-05, | |
| "loss": 4.6338, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.80970450672612e-05, | |
| "loss": 4.6326, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808867549855441e-05, | |
| "loss": 4.625, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808028955104389e-05, | |
| "loss": 4.6308, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.807190360353337e-05, | |
| "loss": 4.6205, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.806351765602285e-05, | |
| "loss": 4.6291, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.805514808731606e-05, | |
| "loss": 4.6264, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.804676213980554e-05, | |
| "loss": 4.6066, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803837619229502e-05, | |
| "loss": 4.6191, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8029990244784504e-05, | |
| "loss": 4.6188, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8021604297273984e-05, | |
| "loss": 4.6249, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.801323472856719e-05, | |
| "loss": 4.6154, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.800484878105667e-05, | |
| "loss": 4.6108, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.799646283354615e-05, | |
| "loss": 4.6036, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.798807688603563e-05, | |
| "loss": 4.5998, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797970731732884e-05, | |
| "loss": 4.604, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797132136981832e-05, | |
| "loss": 4.6104, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.79629354223078e-05, | |
| "loss": 4.6054, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.795454947479728e-05, | |
| "loss": 4.6106, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.794616352728676e-05, | |
| "loss": 4.6005, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.793779395857997e-05, | |
| "loss": 4.595, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.792940801106946e-05, | |
| "loss": 4.5925, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.792102206355894e-05, | |
| "loss": 4.5972, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.791263611604842e-05, | |
| "loss": 4.6017, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.79042501685379e-05, | |
| "loss": 4.6028, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7895880599831107e-05, | |
| "loss": 4.5901, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7887494652320587e-05, | |
| "loss": 4.5824, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7879108704810067e-05, | |
| "loss": 4.5922, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7870722757299546e-05, | |
| "loss": 4.577, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.786233680978902e-05, | |
| "loss": 4.584, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7853967241082236e-05, | |
| "loss": 4.5802, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7845581293571715e-05, | |
| "loss": 4.5804, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7837195346061195e-05, | |
| "loss": 4.5706, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7828809398550675e-05, | |
| "loss": 4.5814, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7820423451040155e-05, | |
| "loss": 4.5675, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.781205388233337e-05, | |
| "loss": 4.5722, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7803667934822844e-05, | |
| "loss": 4.5762, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7795281987312324e-05, | |
| "loss": 4.5642, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7786896039801804e-05, | |
| "loss": 4.5851, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777852647109502e-05, | |
| "loss": 4.5813, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777014052358449e-05, | |
| "loss": 4.5737, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.776175457607397e-05, | |
| "loss": 4.5632, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.775336862856345e-05, | |
| "loss": 4.5608, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.774499905985666e-05, | |
| "loss": 4.5673, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.773661311234615e-05, | |
| "loss": 4.5617, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.772822716483563e-05, | |
| "loss": 4.5727, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771984121732511e-05, | |
| "loss": 4.5525, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771145526981459e-05, | |
| "loss": 4.5608, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.770306932230407e-05, | |
| "loss": 4.5584, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.769468337479355e-05, | |
| "loss": 4.5635, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.768629742728303e-05, | |
| "loss": 4.546, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.767792785857624e-05, | |
| "loss": 4.5561, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766954191106572e-05, | |
| "loss": 4.5529, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.76611559635552e-05, | |
| "loss": 4.555, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.765277001604468e-05, | |
| "loss": 4.5589, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.764438406853416e-05, | |
| "loss": 4.5623, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.763599812102364e-05, | |
| "loss": 4.5398, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.762761217351312e-05, | |
| "loss": 4.5484, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.76192262260026e-05, | |
| "loss": 4.5526, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7610856657295813e-05, | |
| "loss": 4.5491, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7602470709785293e-05, | |
| "loss": 4.5443, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75941011410785e-05, | |
| "loss": 4.5479, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.758571519356798e-05, | |
| "loss": 4.5424, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.757732924605746e-05, | |
| "loss": 4.5595, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756894329854694e-05, | |
| "loss": 4.5306, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756055735103642e-05, | |
| "loss": 4.5488, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.75521714035259e-05, | |
| "loss": 4.5361, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7543785456015375e-05, | |
| "loss": 4.5327, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.753541588730859e-05, | |
| "loss": 4.5343, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.752702993979807e-05, | |
| "loss": 4.5265, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.751864399228755e-05, | |
| "loss": 4.5283, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.751025804477703e-05, | |
| "loss": 4.534, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.750187209726651e-05, | |
| "loss": 4.5274, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.494524002075195, | |
| "eval_runtime": 288.449, | |
| "eval_samples_per_second": 1322.906, | |
| "eval_steps_per_second": 41.342, | |
| "step": 152640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.749348614975599e-05, | |
| "loss": 4.5262, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.748510020224547e-05, | |
| "loss": 4.5186, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.747671425473495e-05, | |
| "loss": 4.5382, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.746832830722443e-05, | |
| "loss": 4.5264, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745994235971391e-05, | |
| "loss": 4.5361, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.745155641220339e-05, | |
| "loss": 4.5141, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.744317046469287e-05, | |
| "loss": 4.5224, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.743478451718235e-05, | |
| "loss": 4.5078, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.742639856967183e-05, | |
| "loss": 4.5174, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.741801262216131e-05, | |
| "loss": 4.5193, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.740962667465079e-05, | |
| "loss": 4.5152, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7401240727140264e-05, | |
| "loss": 4.5262, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.739287115843348e-05, | |
| "loss": 4.5013, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7384485210922966e-05, | |
| "loss": 4.5182, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7376099263412446e-05, | |
| "loss": 4.5033, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.736771331590192e-05, | |
| "loss": 4.5029, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.73593273683914e-05, | |
| "loss": 4.5036, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.735094142088088e-05, | |
| "loss": 4.5115, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.734255547337036e-05, | |
| "loss": 4.5026, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.733416952585984e-05, | |
| "loss": 4.5215, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.732578357834932e-05, | |
| "loss": 4.5029, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.73173976308388e-05, | |
| "loss": 4.5042, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730901168332828e-05, | |
| "loss": 4.503, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730062573581776e-05, | |
| "loss": 4.5126, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.729225616711097e-05, | |
| "loss": 4.4995, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.728387021960045e-05, | |
| "loss": 4.503, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.727548427208993e-05, | |
| "loss": 4.4894, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7267098324579415e-05, | |
| "loss": 4.4941, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.725874513467635e-05, | |
| "loss": 4.4833, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.725035918716583e-05, | |
| "loss": 4.4986, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.724197323965531e-05, | |
| "loss": 4.4974, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.723358729214479e-05, | |
| "loss": 4.4935, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.722520134463427e-05, | |
| "loss": 4.5015, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.721681539712375e-05, | |
| "loss": 4.4897, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720844582841696e-05, | |
| "loss": 4.4901, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720005988090644e-05, | |
| "loss": 4.4915, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.719169031219965e-05, | |
| "loss": 4.4862, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.718330436468914e-05, | |
| "loss": 4.4764, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.717491841717862e-05, | |
| "loss": 4.4676, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.71665324696681e-05, | |
| "loss": 4.4776, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.715814652215758e-05, | |
| "loss": 4.4798, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714976057464706e-05, | |
| "loss": 4.4838, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714137462713654e-05, | |
| "loss": 4.4753, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.713298867962602e-05, | |
| "loss": 4.4687, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.71246027321155e-05, | |
| "loss": 4.4825, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.711623316340871e-05, | |
| "loss": 4.473, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.710784721589819e-05, | |
| "loss": 4.4749, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7099461268387667e-05, | |
| "loss": 4.4669, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7091075320877147e-05, | |
| "loss": 4.4554, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7082689373366627e-05, | |
| "loss": 4.4871, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7074319804659836e-05, | |
| "loss": 4.4797, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.706593385714932e-05, | |
| "loss": 4.4676, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.70575479096388e-05, | |
| "loss": 4.4653, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.704916196212828e-05, | |
| "loss": 4.4741, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7040776014617755e-05, | |
| "loss": 4.4556, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.703240644591097e-05, | |
| "loss": 4.4764, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.702402049840045e-05, | |
| "loss": 4.4508, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7015634550889924e-05, | |
| "loss": 4.469, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7007248603379404e-05, | |
| "loss": 4.4738, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699887903467262e-05, | |
| "loss": 4.4546, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.69904930871621e-05, | |
| "loss": 4.4467, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6982107139651573e-05, | |
| "loss": 4.465, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.697372119214106e-05, | |
| "loss": 4.4352, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.696533524463054e-05, | |
| "loss": 4.4534, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.695694929712002e-05, | |
| "loss": 4.4594, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.694857972841323e-05, | |
| "loss": 4.4581, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.694019378090271e-05, | |
| "loss": 4.4516, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.693180783339219e-05, | |
| "loss": 4.4364, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.692342188588167e-05, | |
| "loss": 4.4406, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.691503593837115e-05, | |
| "loss": 4.4428, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.690666636966436e-05, | |
| "loss": 4.4552, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.689828042215384e-05, | |
| "loss": 4.4428, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688989447464332e-05, | |
| "loss": 4.4518, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68815085271328e-05, | |
| "loss": 4.458, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.687312257962228e-05, | |
| "loss": 4.455, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.686473663211176e-05, | |
| "loss": 4.4371, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6856367063404974e-05, | |
| "loss": 4.4433, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6847981115894454e-05, | |
| "loss": 4.4492, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6839595168383934e-05, | |
| "loss": 4.4361, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683122559967714e-05, | |
| "loss": 4.4411, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.682283965216662e-05, | |
| "loss": 4.4373, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.68144537046561e-05, | |
| "loss": 4.4382, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.680606775714558e-05, | |
| "loss": 4.4438, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.679769818843879e-05, | |
| "loss": 4.4227, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678931224092827e-05, | |
| "loss": 4.4412, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678092629341775e-05, | |
| "loss": 4.4312, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.677254034590723e-05, | |
| "loss": 4.4441, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.676415439839671e-05, | |
| "loss": 4.4348, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.675576845088619e-05, | |
| "loss": 4.4311, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.674738250337568e-05, | |
| "loss": 4.4247, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.673899655586516e-05, | |
| "loss": 4.4232, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.673061060835464e-05, | |
| "loss": 4.4229, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.672222466084412e-05, | |
| "loss": 4.4311, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.671383871333359e-05, | |
| "loss": 4.4378, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.670545276582307e-05, | |
| "loss": 4.4318, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.669708319711629e-05, | |
| "loss": 4.4237, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6688713628409496e-05, | |
| "loss": 4.4198, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6680327680898976e-05, | |
| "loss": 4.4215, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6671941733388456e-05, | |
| "loss": 4.4234, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6663555785877936e-05, | |
| "loss": 4.4318, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6655169838367416e-05, | |
| "loss": 4.4312, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6646783890856896e-05, | |
| "loss": 4.4199, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663841432215011e-05, | |
| "loss": 4.4116, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663002837463959e-05, | |
| "loss": 4.4286, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6621642427129065e-05, | |
| "loss": 4.4094, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6613256479618545e-05, | |
| "loss": 4.4189, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6604870532108025e-05, | |
| "loss": 4.411, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6596500963401234e-05, | |
| "loss": 4.4159, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6588115015890714e-05, | |
| "loss": 4.4065, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6579729068380194e-05, | |
| "loss": 4.417, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6571343120869674e-05, | |
| "loss": 4.4063, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6562957173359154e-05, | |
| "loss": 4.4024, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6554571225848634e-05, | |
| "loss": 4.4189, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6546185278338114e-05, | |
| "loss": 4.3986, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.65377993308276e-05, | |
| "loss": 4.4243, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.652941338331708e-05, | |
| "loss": 4.4206, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.652104381461029e-05, | |
| "loss": 4.4142, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.651265786709977e-05, | |
| "loss": 4.4049, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.650427191958925e-05, | |
| "loss": 4.403, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.649588597207873e-05, | |
| "loss": 4.4038, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.648751640337194e-05, | |
| "loss": 4.4043, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647914683466515e-05, | |
| "loss": 4.4138, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647076088715463e-05, | |
| "loss": 4.3966, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.646237493964411e-05, | |
| "loss": 4.4073, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.645398899213359e-05, | |
| "loss": 4.4052, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.644560304462307e-05, | |
| "loss": 4.4055, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6437217097112554e-05, | |
| "loss": 4.3951, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6428831149602034e-05, | |
| "loss": 4.4013, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6420445202091514e-05, | |
| "loss": 4.401, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6412059254580994e-05, | |
| "loss": 4.403, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6403673307070474e-05, | |
| "loss": 4.4031, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.639528735955995e-05, | |
| "loss": 4.4114, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.638690141204943e-05, | |
| "loss": 4.3866, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637853184334264e-05, | |
| "loss": 4.3996, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.637014589583212e-05, | |
| "loss": 4.4052, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6361759948321596e-05, | |
| "loss": 4.3959, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6353374000811076e-05, | |
| "loss": 4.3934, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.634500443210429e-05, | |
| "loss": 4.4041, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.633663486339751e-05, | |
| "loss": 4.3935, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.632824891588699e-05, | |
| "loss": 4.4104, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.63198793471802e-05, | |
| "loss": 4.3835, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.631149339966968e-05, | |
| "loss": 4.397, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.630310745215916e-05, | |
| "loss": 4.3932, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.629472150464864e-05, | |
| "loss": 4.3882, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.628633555713812e-05, | |
| "loss": 4.3865, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6277949609627597e-05, | |
| "loss": 4.3833, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.626956366211707e-05, | |
| "loss": 4.3806, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.626117771460655e-05, | |
| "loss": 4.3906, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.625279176709603e-05, | |
| "loss": 4.3857, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.3582868576049805, | |
| "eval_runtime": 293.5432, | |
| "eval_samples_per_second": 1299.948, | |
| "eval_steps_per_second": 40.624, | |
| "step": 228960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6244422198389246e-05, | |
| "loss": 4.3887, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6236036250878726e-05, | |
| "loss": 4.3754, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6227650303368205e-05, | |
| "loss": 4.3947, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6219264355857685e-05, | |
| "loss": 4.3826, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6210878408347165e-05, | |
| "loss": 4.3988, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6202492460836645e-05, | |
| "loss": 4.3769, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6194106513326125e-05, | |
| "loss": 4.3828, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6185720565815605e-05, | |
| "loss": 4.3684, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6177350997108814e-05, | |
| "loss": 4.3816, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6168965049598294e-05, | |
| "loss": 4.382, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6160579102087774e-05, | |
| "loss": 4.3795, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6152193154577254e-05, | |
| "loss": 4.3837, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.614382358587046e-05, | |
| "loss": 4.3677, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.613545401716368e-05, | |
| "loss": 4.3831, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.612706806965316e-05, | |
| "loss": 4.3696, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611868212214264e-05, | |
| "loss": 4.3639, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611029617463212e-05, | |
| "loss": 4.3692, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61019102271216e-05, | |
| "loss": 4.3744, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.609354065841481e-05, | |
| "loss": 4.3704, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.608515471090429e-05, | |
| "loss": 4.3866, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.607676876339377e-05, | |
| "loss": 4.3714, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606838281588325e-05, | |
| "loss": 4.3769, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.605999686837273e-05, | |
| "loss": 4.3679, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.605161092086221e-05, | |
| "loss": 4.3785, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.604322497335169e-05, | |
| "loss": 4.368, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.603483902584117e-05, | |
| "loss": 4.3728, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.602645307833065e-05, | |
| "loss": 4.3629, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.601806713082013e-05, | |
| "loss": 4.3584, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6009697562113344e-05, | |
| "loss": 4.3592, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6001311614602824e-05, | |
| "loss": 4.3679, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5992925667092303e-05, | |
| "loss": 4.37, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5984539719581783e-05, | |
| "loss": 4.3664, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5976153772071257e-05, | |
| "loss": 4.3754, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.59678005821682e-05, | |
| "loss": 4.3645, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595941463465768e-05, | |
| "loss": 4.36, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.595102868714716e-05, | |
| "loss": 4.3633, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.594264273963664e-05, | |
| "loss": 4.3668, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.593425679212612e-05, | |
| "loss": 4.3479, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.59258708446156e-05, | |
| "loss": 4.3428, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.591748489710508e-05, | |
| "loss": 4.3506, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.590909894959456e-05, | |
| "loss": 4.3588, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.590071300208404e-05, | |
| "loss": 4.3584, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.589232705457352e-05, | |
| "loss": 4.3526, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5883941107063e-05, | |
| "loss": 4.3427, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.587557153835621e-05, | |
| "loss": 4.3586, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.586718559084569e-05, | |
| "loss": 4.3524, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.585879964333517e-05, | |
| "loss": 4.3523, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.585041369582465e-05, | |
| "loss": 4.3456, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.584202774831413e-05, | |
| "loss": 4.3306, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.583364180080361e-05, | |
| "loss": 4.3749, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.582527223209682e-05, | |
| "loss": 4.3568, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.58168862845863e-05, | |
| "loss": 4.3493, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5808500337075786e-05, | |
| "loss": 4.3493, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5800114389565266e-05, | |
| "loss": 4.3509, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5791728442054746e-05, | |
| "loss": 4.3403, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5783358873347955e-05, | |
| "loss": 4.3533, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5774972925837435e-05, | |
| "loss": 4.3367, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5766586978326915e-05, | |
| "loss": 4.3481, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5758201030816395e-05, | |
| "loss": 4.3593, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5749815083305875e-05, | |
| "loss": 4.338, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5741429135795355e-05, | |
| "loss": 4.3261, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5733043188284835e-05, | |
| "loss": 4.3532, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5724657240774315e-05, | |
| "loss": 4.3204, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5716287672067524e-05, | |
| "loss": 4.3331, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.570791810336074e-05, | |
| "loss": 4.346, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.569953215585022e-05, | |
| "loss": 4.3441, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.56911462083397e-05, | |
| "loss": 4.3354, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.568276026082918e-05, | |
| "loss": 4.3288, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.567437431331866e-05, | |
| "loss": 4.3209, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.566598836580814e-05, | |
| "loss": 4.3281, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.565760241829762e-05, | |
| "loss": 4.3445, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.564921647078709e-05, | |
| "loss": 4.3284, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.564084690208031e-05, | |
| "loss": 4.3444, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.563246095456979e-05, | |
| "loss": 4.3405, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.562407500705927e-05, | |
| "loss": 4.3471, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.561568905954874e-05, | |
| "loss": 4.3282, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.560731949084196e-05, | |
| "loss": 4.3286, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.559893354333144e-05, | |
| "loss": 4.3388, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.559054759582092e-05, | |
| "loss": 4.324, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.558217802711413e-05, | |
| "loss": 4.3378, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.557379207960361e-05, | |
| "loss": 4.3254, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.556540613209309e-05, | |
| "loss": 4.3274, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5557020184582566e-05, | |
| "loss": 4.3353, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5548634237072046e-05, | |
| "loss": 4.3179, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5540248289561526e-05, | |
| "loss": 4.3285, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5531862342051006e-05, | |
| "loss": 4.3276, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5523476394540486e-05, | |
| "loss": 4.3344, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5515090447029966e-05, | |
| "loss": 4.3233, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5506720878323175e-05, | |
| "loss": 4.327, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5498334930812655e-05, | |
| "loss": 4.3156, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.548994898330214e-05, | |
| "loss": 4.3153, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.548156303579162e-05, | |
| "loss": 4.3196, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.54731770882811e-05, | |
| "loss": 4.324, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.546479114077058e-05, | |
| "loss": 4.3319, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.545642157206379e-05, | |
| "loss": 4.3303, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.544803562455327e-05, | |
| "loss": 4.3201, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.543964967704275e-05, | |
| "loss": 4.3138, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.543126372953223e-05, | |
| "loss": 4.3171, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.542289416082544e-05, | |
| "loss": 4.3218, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.541450821331492e-05, | |
| "loss": 4.3253, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.54061222658044e-05, | |
| "loss": 4.3271, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.539773631829388e-05, | |
| "loss": 4.317, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.538935037078336e-05, | |
| "loss": 4.3099, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.538096442327284e-05, | |
| "loss": 4.328, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5372578475762326e-05, | |
| "loss": 4.3029, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5364192528251806e-05, | |
| "loss": 4.3195, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5355822959545015e-05, | |
| "loss": 4.3065, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5347437012034495e-05, | |
| "loss": 4.3141, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5339051064523975e-05, | |
| "loss": 4.3064, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5330665117013455e-05, | |
| "loss": 4.314, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.532227916950293e-05, | |
| "loss": 4.3067, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5313909600796144e-05, | |
| "loss": 4.302, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5305523653285624e-05, | |
| "loss": 4.3139, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5297137705775104e-05, | |
| "loss": 4.3027, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528875175826458e-05, | |
| "loss": 4.3223, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5280365810754064e-05, | |
| "loss": 4.3231, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.527199624204728e-05, | |
| "loss": 4.3175, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.526361029453675e-05, | |
| "loss": 4.3067, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.525522434702623e-05, | |
| "loss": 4.3067, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.524683839951571e-05, | |
| "loss": 4.3046, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.523845245200519e-05, | |
| "loss": 4.3072, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.523006650449467e-05, | |
| "loss": 4.3175, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.522168055698415e-05, | |
| "loss": 4.2948, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.521329460947363e-05, | |
| "loss": 4.3092, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.520492504076684e-05, | |
| "loss": 4.3055, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.519655547206005e-05, | |
| "loss": 4.312, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.518816952454953e-05, | |
| "loss": 4.2991, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517978357703902e-05, | |
| "loss": 4.3067, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517141400833223e-05, | |
| "loss": 4.2991, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.516302806082171e-05, | |
| "loss": 4.3151, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5154642113311187e-05, | |
| "loss": 4.3065, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5146256165800667e-05, | |
| "loss": 4.3162, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5137870218290147e-05, | |
| "loss": 4.2916, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5129484270779626e-05, | |
| "loss": 4.3078, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5121098323269106e-05, | |
| "loss": 4.3099, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5112712375758586e-05, | |
| "loss": 4.296, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5104326428248066e-05, | |
| "loss": 4.3028, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5095940480737546e-05, | |
| "loss": 4.3106, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5087570912030755e-05, | |
| "loss": 4.3006, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5079184964520235e-05, | |
| "loss": 4.3186, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5070799017009715e-05, | |
| "loss": 4.2898, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5062413069499195e-05, | |
| "loss": 4.3069, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.505404350079241e-05, | |
| "loss": 4.2947, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.504565755328189e-05, | |
| "loss": 4.2977, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.503727160577137e-05, | |
| "loss": 4.2981, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.502888565826085e-05, | |
| "loss": 4.2905, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.502049971075033e-05, | |
| "loss": 4.288, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.501211376323981e-05, | |
| "loss": 4.2969, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.500372781572929e-05, | |
| "loss": 4.2964, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.2750773429870605, | |
| "eval_runtime": 293.1439, | |
| "eval_samples_per_second": 1301.719, | |
| "eval_steps_per_second": 40.68, | |
| "step": 305280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4995341868218764e-05, | |
| "loss": 4.2958, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4986955920708244e-05, | |
| "loss": 4.2852, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4978569973197724e-05, | |
| "loss": 4.3043, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4970184025687204e-05, | |
| "loss": 4.293, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4961798078176684e-05, | |
| "loss": 4.3037, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4953412130666164e-05, | |
| "loss": 4.2926, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.494502618315565e-05, | |
| "loss": 4.2921, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.493664023564513e-05, | |
| "loss": 4.2802, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.492825428813461e-05, | |
| "loss": 4.2896, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491986834062409e-05, | |
| "loss": 4.2923, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.491148239311357e-05, | |
| "loss": 4.2916, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.490309644560305e-05, | |
| "loss": 4.2929, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.489472687689626e-05, | |
| "loss": 4.2806, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.488634092938574e-05, | |
| "loss": 4.2963, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.487795498187522e-05, | |
| "loss": 4.2803, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486958541316843e-05, | |
| "loss": 4.2775, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.486119946565791e-05, | |
| "loss": 4.281, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.485281351814739e-05, | |
| "loss": 4.2898, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.484442757063687e-05, | |
| "loss": 4.2827, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.483604162312635e-05, | |
| "loss": 4.2987, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.482765567561583e-05, | |
| "loss": 4.2862, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.481926972810531e-05, | |
| "loss": 4.2932, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.481088378059479e-05, | |
| "loss": 4.2809, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.480249783308427e-05, | |
| "loss": 4.2902, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.479411188557375e-05, | |
| "loss": 4.2834, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.478572593806323e-05, | |
| "loss": 4.288, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.477733999055271e-05, | |
| "loss": 4.2782, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.476897042184592e-05, | |
| "loss": 4.2767, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.47605844743354e-05, | |
| "loss": 4.2726, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.475219852682488e-05, | |
| "loss": 4.2809, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.474381257931436e-05, | |
| "loss": 4.2861, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.473542663180384e-05, | |
| "loss": 4.2828, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.472704068429332e-05, | |
| "loss": 4.2886, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.471867111558653e-05, | |
| "loss": 4.2777, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.471028516807601e-05, | |
| "loss": 4.2777, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.470189922056549e-05, | |
| "loss": 4.2856, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.469351327305497e-05, | |
| "loss": 4.2844, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.468512732554445e-05, | |
| "loss": 4.2607, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.467674137803393e-05, | |
| "loss": 4.2671, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.466837180932714e-05, | |
| "loss": 4.265, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.465998586181662e-05, | |
| "loss": 4.2714, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.46515999143061e-05, | |
| "loss": 4.2813, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.464321396679558e-05, | |
| "loss": 4.2696, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.463484439808879e-05, | |
| "loss": 4.2651, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.462645845057827e-05, | |
| "loss": 4.2755, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.461807250306775e-05, | |
| "loss": 4.2689, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.460968655555724e-05, | |
| "loss": 4.2737, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.460130060804672e-05, | |
| "loss": 4.268, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4592931039339926e-05, | |
| "loss": 4.2446, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4584545091829406e-05, | |
| "loss": 4.2936, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4576159144318886e-05, | |
| "loss": 4.2772, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4567773196808366e-05, | |
| "loss": 4.2751, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.455938724929784e-05, | |
| "loss": 4.2675, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.455100130178732e-05, | |
| "loss": 4.2674, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4542631733080535e-05, | |
| "loss": 4.2608, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4534262164373744e-05, | |
| "loss": 4.2699, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4525876216863224e-05, | |
| "loss": 4.2617, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4517490269352704e-05, | |
| "loss": 4.2662, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.450910432184219e-05, | |
| "loss": 4.2816, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4500718374331664e-05, | |
| "loss": 4.2594, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4492332426821144e-05, | |
| "loss": 4.2508, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4483946479310624e-05, | |
| "loss": 4.2747, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4475560531800104e-05, | |
| "loss": 4.2426, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.446719096309331e-05, | |
| "loss": 4.2581, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445880501558279e-05, | |
| "loss": 4.2659, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.445041906807227e-05, | |
| "loss": 4.2655, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.444203312056175e-05, | |
| "loss": 4.2577, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.443366355185496e-05, | |
| "loss": 4.2551, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.442527760434444e-05, | |
| "loss": 4.2392, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.441689165683393e-05, | |
| "loss": 4.2539, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440850570932341e-05, | |
| "loss": 4.2646, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.440011976181289e-05, | |
| "loss": 4.255, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.439173381430237e-05, | |
| "loss": 4.2615, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.438334786679185e-05, | |
| "loss": 4.2676, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.437497829808506e-05, | |
| "loss": 4.2746, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.436660872937827e-05, | |
| "loss": 4.2516, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4358222781867747e-05, | |
| "loss": 4.2493, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4349836834357227e-05, | |
| "loss": 4.2652, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4341450886846707e-05, | |
| "loss": 4.2448, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4333064939336186e-05, | |
| "loss": 4.2649, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4324678991825666e-05, | |
| "loss": 4.25, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4316293044315146e-05, | |
| "loss": 4.2513, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4307907096804626e-05, | |
| "loss": 4.2634, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.429952114929411e-05, | |
| "loss": 4.2422, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.429113520178359e-05, | |
| "loss": 4.2523, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.428274925427307e-05, | |
| "loss": 4.2503, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.427436330676255e-05, | |
| "loss": 4.2621, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.426599373805576e-05, | |
| "loss": 4.2458, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.425760779054524e-05, | |
| "loss": 4.2544, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.424922184303472e-05, | |
| "loss": 4.2432, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.42408358955242e-05, | |
| "loss": 4.2444, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.423248270562114e-05, | |
| "loss": 4.2422, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.422409675811062e-05, | |
| "loss": 4.2521, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.42157108106001e-05, | |
| "loss": 4.2584, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.420732486308958e-05, | |
| "loss": 4.2536, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419893891557907e-05, | |
| "loss": 4.2526, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.419055296806855e-05, | |
| "loss": 4.2402, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4182183399361756e-05, | |
| "loss": 4.2437, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4173797451851236e-05, | |
| "loss": 4.2472, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4165411504340716e-05, | |
| "loss": 4.2536, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4157025556830196e-05, | |
| "loss": 4.254, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4148672366927134e-05, | |
| "loss": 4.2441, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4140286419416614e-05, | |
| "loss": 4.2397, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4131900471906094e-05, | |
| "loss": 4.2568, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4123514524395574e-05, | |
| "loss": 4.2326, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4115128576885054e-05, | |
| "loss": 4.247, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4106742629374534e-05, | |
| "loss": 4.238, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.409835668186402e-05, | |
| "loss": 4.2452, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.40899707343535e-05, | |
| "loss": 4.2346, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4081584786842974e-05, | |
| "loss": 4.2447, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4073198839332453e-05, | |
| "loss": 4.2333, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4064812891821933e-05, | |
| "loss": 4.2345, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4056426944311413e-05, | |
| "loss": 4.243, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.404804099680089e-05, | |
| "loss": 4.2353, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403965504929037e-05, | |
| "loss": 4.2501, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.403126910177985e-05, | |
| "loss": 4.254, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.402289953307306e-05, | |
| "loss": 4.2439, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.401451358556254e-05, | |
| "loss": 4.2417, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.400612763805202e-05, | |
| "loss": 4.2321, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.39977416905415e-05, | |
| "loss": 4.2376, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.398935574303098e-05, | |
| "loss": 4.2368, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.39809861743242e-05, | |
| "loss": 4.2531, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.397260022681368e-05, | |
| "loss": 4.2227, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.396421427930316e-05, | |
| "loss": 4.2415, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.395582833179264e-05, | |
| "loss": 4.2369, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.394745876308585e-05, | |
| "loss": 4.2417, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.393907281557533e-05, | |
| "loss": 4.2327, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3930703246868536e-05, | |
| "loss": 4.2389, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3922317299358016e-05, | |
| "loss": 4.2284, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3913931351847496e-05, | |
| "loss": 4.2459, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3905545404336976e-05, | |
| "loss": 4.2412, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3897159456826456e-05, | |
| "loss": 4.2438, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3888773509315936e-05, | |
| "loss": 4.2273, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.388038756180542e-05, | |
| "loss": 4.2393, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.38720016142949e-05, | |
| "loss": 4.2423, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.386361566678438e-05, | |
| "loss": 4.2302, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.385522971927386e-05, | |
| "loss": 4.2351, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3846843771763336e-05, | |
| "loss": 4.2437, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3838457824252816e-05, | |
| "loss": 4.231, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3830071876742296e-05, | |
| "loss": 4.2556, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3821685929231775e-05, | |
| "loss": 4.2251, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3813299981721255e-05, | |
| "loss": 4.2382, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3804914034210735e-05, | |
| "loss": 4.2279, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3796544465503944e-05, | |
| "loss": 4.2325, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.378817489679716e-05, | |
| "loss": 4.2333, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.377978894928664e-05, | |
| "loss": 4.2245, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3771419380579856e-05, | |
| "loss": 4.2265, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3763033433069336e-05, | |
| "loss": 4.2206, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.375464748555881e-05, | |
| "loss": 4.2348, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.217353343963623, | |
| "eval_runtime": 294.6691, | |
| "eval_samples_per_second": 1294.981, | |
| "eval_steps_per_second": 40.469, | |
| "step": 381600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.374626153804829e-05, | |
| "loss": 4.2351, | |
| "step": 381952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.373787559053777e-05, | |
| "loss": 4.2208, | |
| "step": 382464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372948964302725e-05, | |
| "loss": 4.2363, | |
| "step": 382976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.372110369551673e-05, | |
| "loss": 4.2326, | |
| "step": 383488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.371271774800621e-05, | |
| "loss": 4.2341, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.370433180049569e-05, | |
| "loss": 4.226, | |
| "step": 384512 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.369594585298517e-05, | |
| "loss": 4.227, | |
| "step": 385024 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.368755990547465e-05, | |
| "loss": 4.2156, | |
| "step": 385536 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.367917395796413e-05, | |
| "loss": 4.2251, | |
| "step": 386048 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3670804389257345e-05, | |
| "loss": 4.228, | |
| "step": 386560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3662418441746825e-05, | |
| "loss": 4.2298, | |
| "step": 387072 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3654032494236305e-05, | |
| "loss": 4.2242, | |
| "step": 387584 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3645662925529514e-05, | |
| "loss": 4.2222, | |
| "step": 388096 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3637276978018994e-05, | |
| "loss": 4.2263, | |
| "step": 388608 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3628891030508474e-05, | |
| "loss": 4.2174, | |
| "step": 389120 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3620505082997954e-05, | |
| "loss": 4.2101, | |
| "step": 389632 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3612119135487434e-05, | |
| "loss": 4.2188, | |
| "step": 390144 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.360374956678064e-05, | |
| "loss": 4.2245, | |
| "step": 390656 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.359536361927012e-05, | |
| "loss": 4.2206, | |
| "step": 391168 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.35869776717596e-05, | |
| "loss": 4.2341, | |
| "step": 391680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357859172424908e-05, | |
| "loss": 4.2259, | |
| "step": 392192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.357020577673856e-05, | |
| "loss": 4.2302, | |
| "step": 392704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.356181982922804e-05, | |
| "loss": 4.218, | |
| "step": 393216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.355343388171752e-05, | |
| "loss": 4.2259, | |
| "step": 393728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.3545047934207e-05, | |
| "loss": 4.2199, | |
| "step": 394240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.353667836550022e-05, | |
| "loss": 4.227, | |
| "step": 394752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.352830879679343e-05, | |
| "loss": 4.2175, | |
| "step": 395264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.351992284928291e-05, | |
| "loss": 4.2122, | |
| "step": 395776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.351153690177239e-05, | |
| "loss": 4.211, | |
| "step": 396288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.350315095426187e-05, | |
| "loss": 4.222, | |
| "step": 396800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.349476500675134e-05, | |
| "loss": 4.2209, | |
| "step": 397312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3486395438044556e-05, | |
| "loss": 4.2222, | |
| "step": 397824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3478009490534036e-05, | |
| "loss": 4.225, | |
| "step": 398336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3469623543023516e-05, | |
| "loss": 4.2197, | |
| "step": 398848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3461237595512996e-05, | |
| "loss": 4.2161, | |
| "step": 399360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3452851648002476e-05, | |
| "loss": 4.2242, | |
| "step": 399872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3444465700491956e-05, | |
| "loss": 4.2207, | |
| "step": 400384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3436079752981436e-05, | |
| "loss": 4.2044, | |
| "step": 400896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3427710184274645e-05, | |
| "loss": 4.2058, | |
| "step": 401408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3419324236764125e-05, | |
| "loss": 4.2041, | |
| "step": 401920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3410938289253605e-05, | |
| "loss": 4.2059, | |
| "step": 402432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3402552341743085e-05, | |
| "loss": 4.2213, | |
| "step": 402944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3394182773036294e-05, | |
| "loss": 4.2101, | |
| "step": 403456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3385796825525774e-05, | |
| "loss": 4.2036, | |
| "step": 403968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3377410878015254e-05, | |
| "loss": 4.2159, | |
| "step": 404480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3369024930504734e-05, | |
| "loss": 4.208, | |
| "step": 404992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3360638982994214e-05, | |
| "loss": 4.2149, | |
| "step": 405504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.33522530354837e-05, | |
| "loss": 4.2049, | |
| "step": 406016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.334386708797318e-05, | |
| "loss": 4.1849, | |
| "step": 406528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.333548114046266e-05, | |
| "loss": 4.2359, | |
| "step": 407040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.332711157175587e-05, | |
| "loss": 4.2134, | |
| "step": 407552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331874200304908e-05, | |
| "loss": 4.2177, | |
| "step": 408064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.331035605553856e-05, | |
| "loss": 4.2031, | |
| "step": 408576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.330197010802804e-05, | |
| "loss": 4.2129, | |
| "step": 409088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.329358416051752e-05, | |
| "loss": 4.2021, | |
| "step": 409600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3285198213007e-05, | |
| "loss": 4.2069, | |
| "step": 410112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.327681226549648e-05, | |
| "loss": 4.2063, | |
| "step": 410624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326842631798596e-05, | |
| "loss": 4.2058, | |
| "step": 411136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.326005674927917e-05, | |
| "loss": 4.2249, | |
| "step": 411648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3251670801768654e-05, | |
| "loss": 4.1977, | |
| "step": 412160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3243301233061863e-05, | |
| "loss": 4.1894, | |
| "step": 412672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.323491528555134e-05, | |
| "loss": 4.2168, | |
| "step": 413184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.322652933804082e-05, | |
| "loss": 4.1854, | |
| "step": 413696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.32181433905303e-05, | |
| "loss": 4.1956, | |
| "step": 414208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320975744301978e-05, | |
| "loss": 4.2117, | |
| "step": 414720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.320137149550926e-05, | |
| "loss": 4.2112, | |
| "step": 415232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.319298554799874e-05, | |
| "loss": 4.195, | |
| "step": 415744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.318459960048822e-05, | |
| "loss": 4.2029, | |
| "step": 416256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.317623003178143e-05, | |
| "loss": 4.1802, | |
| "step": 416768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.316784408427091e-05, | |
| "loss": 4.1993, | |
| "step": 417280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315945813676039e-05, | |
| "loss": 4.2054, | |
| "step": 417792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.315107218924987e-05, | |
| "loss": 4.1953, | |
| "step": 418304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.314270262054309e-05, | |
| "loss": 4.2062, | |
| "step": 418816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.313431667303257e-05, | |
| "loss": 4.2077, | |
| "step": 419328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.312593072552205e-05, | |
| "loss": 4.2196, | |
| "step": 419840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.311754477801153e-05, | |
| "loss": 4.1921, | |
| "step": 420352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3109158830501e-05, | |
| "loss": 4.197, | |
| "step": 420864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.310077288299048e-05, | |
| "loss": 4.2104, | |
| "step": 421376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.309238693547996e-05, | |
| "loss": 4.1893, | |
| "step": 421888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.308400098796944e-05, | |
| "loss": 4.2046, | |
| "step": 422400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.307561504045892e-05, | |
| "loss": 4.1967, | |
| "step": 422912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.30672290929484e-05, | |
| "loss": 4.1952, | |
| "step": 423424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.305885952424161e-05, | |
| "loss": 4.2095, | |
| "step": 423936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.305047357673109e-05, | |
| "loss": 4.1889, | |
| "step": 424448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3042087629220577e-05, | |
| "loss": 4.1922, | |
| "step": 424960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3033701681710057e-05, | |
| "loss": 4.2001, | |
| "step": 425472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3025315734199536e-05, | |
| "loss": 4.2043, | |
| "step": 425984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3016929786689016e-05, | |
| "loss": 4.191, | |
| "step": 426496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.3008543839178496e-05, | |
| "loss": 4.2, | |
| "step": 427008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3000157891667976e-05, | |
| "loss": 4.1827, | |
| "step": 427520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2991788322961185e-05, | |
| "loss": 4.1934, | |
| "step": 428032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2983418754254395e-05, | |
| "loss": 4.1892, | |
| "step": 428544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2975032806743874e-05, | |
| "loss": 4.1968, | |
| "step": 429056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2966646859233354e-05, | |
| "loss": 4.209, | |
| "step": 429568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2958260911722834e-05, | |
| "loss": 4.1961, | |
| "step": 430080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2949874964212314e-05, | |
| "loss": 4.1971, | |
| "step": 430592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.294150539550553e-05, | |
| "loss": 4.187, | |
| "step": 431104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.293311944799501e-05, | |
| "loss": 4.19, | |
| "step": 431616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.292473350048449e-05, | |
| "loss": 4.1928, | |
| "step": 432128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.291634755297397e-05, | |
| "loss": 4.2012, | |
| "step": 432640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.290796160546345e-05, | |
| "loss": 4.1966, | |
| "step": 433152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289957565795293e-05, | |
| "loss": 4.1976, | |
| "step": 433664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.289118971044241e-05, | |
| "loss": 4.1873, | |
| "step": 434176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.288280376293189e-05, | |
| "loss": 4.2008, | |
| "step": 434688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.28744341942251e-05, | |
| "loss": 4.1791, | |
| "step": 435200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.286604824671458e-05, | |
| "loss": 4.1943, | |
| "step": 435712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.285767867800779e-05, | |
| "loss": 4.1794, | |
| "step": 436224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284929273049727e-05, | |
| "loss": 4.1966, | |
| "step": 436736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.284090678298675e-05, | |
| "loss": 4.1835, | |
| "step": 437248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.283252083547623e-05, | |
| "loss": 4.1888, | |
| "step": 437760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2824151266769444e-05, | |
| "loss": 4.1834, | |
| "step": 438272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2815765319258924e-05, | |
| "loss": 4.1824, | |
| "step": 438784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2807379371748404e-05, | |
| "loss": 4.1866, | |
| "step": 439296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2798993424237884e-05, | |
| "loss": 4.1842, | |
| "step": 439808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2790607476727364e-05, | |
| "loss": 4.1954, | |
| "step": 440320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.278222152921684e-05, | |
| "loss": 4.2001, | |
| "step": 440832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.277383558170632e-05, | |
| "loss": 4.1935, | |
| "step": 441344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.27654496341958e-05, | |
| "loss": 4.1945, | |
| "step": 441856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.275706368668528e-05, | |
| "loss": 4.1763, | |
| "step": 442368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.274867773917476e-05, | |
| "loss": 4.1841, | |
| "step": 442880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2740291791664237e-05, | |
| "loss": 4.1874, | |
| "step": 443392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2731905844153717e-05, | |
| "loss": 4.1977, | |
| "step": 443904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.272353627544693e-05, | |
| "loss": 4.1705, | |
| "step": 444416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.271515032793641e-05, | |
| "loss": 4.1887, | |
| "step": 444928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.270676438042589e-05, | |
| "loss": 4.1893, | |
| "step": 445440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.26983948117191e-05, | |
| "loss": 4.1849, | |
| "step": 445952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.269000886420858e-05, | |
| "loss": 4.1823, | |
| "step": 446464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.268162291669806e-05, | |
| "loss": 4.1897, | |
| "step": 446976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.267323696918754e-05, | |
| "loss": 4.1784, | |
| "step": 447488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.266485102167702e-05, | |
| "loss": 4.1931, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.26564650741665e-05, | |
| "loss": 4.1902, | |
| "step": 448512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.264809550545971e-05, | |
| "loss": 4.1922, | |
| "step": 449024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.263970955794919e-05, | |
| "loss": 4.1765, | |
| "step": 449536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.263132361043867e-05, | |
| "loss": 4.1892, | |
| "step": 450048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.262293766292815e-05, | |
| "loss": 4.1914, | |
| "step": 450560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.261455171541763e-05, | |
| "loss": 4.1797, | |
| "step": 451072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.260616576790712e-05, | |
| "loss": 4.1846, | |
| "step": 451584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2597796199200326e-05, | |
| "loss": 4.1905, | |
| "step": 452096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2589426630493535e-05, | |
| "loss": 4.1858, | |
| "step": 452608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2581040682983015e-05, | |
| "loss": 4.2009, | |
| "step": 453120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2572671114276224e-05, | |
| "loss": 4.178, | |
| "step": 453632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2564285166765704e-05, | |
| "loss": 4.1862, | |
| "step": 454144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2555899219255184e-05, | |
| "loss": 4.1804, | |
| "step": 454656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2547513271744664e-05, | |
| "loss": 4.1835, | |
| "step": 455168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2539127324234144e-05, | |
| "loss": 4.1771, | |
| "step": 455680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2530741376723624e-05, | |
| "loss": 4.1728, | |
| "step": 456192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2522355429213104e-05, | |
| "loss": 4.1814, | |
| "step": 456704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.2513969481702584e-05, | |
| "loss": 4.1694, | |
| "step": 457216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.250558353419207e-05, | |
| "loss": 4.1869, | |
| "step": 457728 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.17576265335083, | |
| "eval_runtime": 291.3672, | |
| "eval_samples_per_second": 1309.657, | |
| "eval_steps_per_second": 40.928, | |
| "step": 457920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.249719758668155e-05, | |
| "loss": 4.1877, | |
| "step": 458240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2488811639171024e-05, | |
| "loss": 4.1711, | |
| "step": 458752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2480425691660504e-05, | |
| "loss": 4.1857, | |
| "step": 459264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2472039744149984e-05, | |
| "loss": 4.18, | |
| "step": 459776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2463653796639464e-05, | |
| "loss": 4.1882, | |
| "step": 460288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2455267849128943e-05, | |
| "loss": 4.1773, | |
| "step": 460800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2446881901618423e-05, | |
| "loss": 4.1795, | |
| "step": 461312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.24384959541079e-05, | |
| "loss": 4.1674, | |
| "step": 461824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.243011000659738e-05, | |
| "loss": 4.1771, | |
| "step": 462336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.242172405908686e-05, | |
| "loss": 4.1802, | |
| "step": 462848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.241333811157634e-05, | |
| "loss": 4.1802, | |
| "step": 463360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.240495216406582e-05, | |
| "loss": 4.1757, | |
| "step": 463872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.239658259535903e-05, | |
| "loss": 4.1744, | |
| "step": 464384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.238819664784852e-05, | |
| "loss": 4.1768, | |
| "step": 464896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2379810700338e-05, | |
| "loss": 4.1746, | |
| "step": 465408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.237142475282748e-05, | |
| "loss": 4.1608, | |
| "step": 465920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.236303880531696e-05, | |
| "loss": 4.1717, | |
| "step": 466432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.235465285780644e-05, | |
| "loss": 4.1765, | |
| "step": 466944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.234626691029591e-05, | |
| "loss": 4.1743, | |
| "step": 467456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.233788096278539e-05, | |
| "loss": 4.1847, | |
| "step": 467968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.232949501527487e-05, | |
| "loss": 4.1816, | |
| "step": 468480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.232110906776435e-05, | |
| "loss": 4.1841, | |
| "step": 468992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.231272312025383e-05, | |
| "loss": 4.1666, | |
| "step": 469504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.230433717274331e-05, | |
| "loss": 4.1787, | |
| "step": 470016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.229598398284026e-05, | |
| "loss": 4.1754, | |
| "step": 470528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.228759803532974e-05, | |
| "loss": 4.1786, | |
| "step": 471040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.227921208781922e-05, | |
| "loss": 4.1684, | |
| "step": 471552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.22708261403087e-05, | |
| "loss": 4.1692, | |
| "step": 472064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.226244019279818e-05, | |
| "loss": 4.1661, | |
| "step": 472576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2254054245287657e-05, | |
| "loss": 4.1704, | |
| "step": 473088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2245684676580866e-05, | |
| "loss": 4.1714, | |
| "step": 473600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2237298729070346e-05, | |
| "loss": 4.1789, | |
| "step": 474112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2228912781559826e-05, | |
| "loss": 4.1795, | |
| "step": 474624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2220526834049306e-05, | |
| "loss": 4.1757, | |
| "step": 475136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2212157265342515e-05, | |
| "loss": 4.1699, | |
| "step": 475648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2203771317831995e-05, | |
| "loss": 4.1722, | |
| "step": 476160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.219540174912521e-05, | |
| "loss": 4.1773, | |
| "step": 476672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.218701580161469e-05, | |
| "loss": 4.1612, | |
| "step": 477184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.217862985410417e-05, | |
| "loss": 4.1565, | |
| "step": 477696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.217024390659365e-05, | |
| "loss": 4.1565, | |
| "step": 478208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.216185795908313e-05, | |
| "loss": 4.1639, | |
| "step": 478720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.215347201157261e-05, | |
| "loss": 4.1701, | |
| "step": 479232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.214508606406209e-05, | |
| "loss": 4.1666, | |
| "step": 479744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.213670011655157e-05, | |
| "loss": 4.1627, | |
| "step": 480256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.212831416904105e-05, | |
| "loss": 4.1669, | |
| "step": 480768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.211992822153053e-05, | |
| "loss": 4.165, | |
| "step": 481280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.211154227402001e-05, | |
| "loss": 4.1681, | |
| "step": 481792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.210315632650949e-05, | |
| "loss": 4.1597, | |
| "step": 482304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.209477037899897e-05, | |
| "loss": 4.1424, | |
| "step": 482816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.208638443148845e-05, | |
| "loss": 4.1858, | |
| "step": 483328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.207799848397792e-05, | |
| "loss": 4.17, | |
| "step": 483840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.206961253646741e-05, | |
| "loss": 4.1745, | |
| "step": 484352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2061242967760626e-05, | |
| "loss": 4.1618, | |
| "step": 484864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.20528570202501e-05, | |
| "loss": 4.1659, | |
| "step": 485376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.204447107273958e-05, | |
| "loss": 4.1618, | |
| "step": 485888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.203608512522906e-05, | |
| "loss": 4.1619, | |
| "step": 486400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.202769917771854e-05, | |
| "loss": 4.1652, | |
| "step": 486912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201932960901175e-05, | |
| "loss": 4.1595, | |
| "step": 487424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.201094366150123e-05, | |
| "loss": 4.1808, | |
| "step": 487936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.200255771399071e-05, | |
| "loss": 4.1569, | |
| "step": 488448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1994188145283924e-05, | |
| "loss": 4.1468, | |
| "step": 488960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.19858021977734e-05, | |
| "loss": 4.1718, | |
| "step": 489472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.197741625026288e-05, | |
| "loss": 4.1426, | |
| "step": 489984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1969030302752363e-05, | |
| "loss": 4.1453, | |
| "step": 490496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1960644355241843e-05, | |
| "loss": 4.1707, | |
| "step": 491008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1952258407731323e-05, | |
| "loss": 4.1662, | |
| "step": 491520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.19438724602208e-05, | |
| "loss": 4.1493, | |
| "step": 492032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.193548651271028e-05, | |
| "loss": 4.1625, | |
| "step": 492544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.192711694400349e-05, | |
| "loss": 4.137, | |
| "step": 493056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.191873099649297e-05, | |
| "loss": 4.1559, | |
| "step": 493568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.191034504898245e-05, | |
| "loss": 4.1601, | |
| "step": 494080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.190195910147193e-05, | |
| "loss": 4.155, | |
| "step": 494592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.189358953276514e-05, | |
| "loss": 4.1625, | |
| "step": 495104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.188520358525462e-05, | |
| "loss": 4.1623, | |
| "step": 495616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.18768176377441e-05, | |
| "loss": 4.1737, | |
| "step": 496128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.186843169023358e-05, | |
| "loss": 4.1514, | |
| "step": 496640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.18600621215268e-05, | |
| "loss": 4.1538, | |
| "step": 497152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.185167617401628e-05, | |
| "loss": 4.1641, | |
| "step": 497664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.184329022650576e-05, | |
| "loss": 4.1487, | |
| "step": 498176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.183490427899524e-05, | |
| "loss": 4.1605, | |
| "step": 498688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1826551089092175e-05, | |
| "loss": 4.1556, | |
| "step": 499200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1818165141581655e-05, | |
| "loss": 4.1518, | |
| "step": 499712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1809779194071135e-05, | |
| "loss": 4.1717, | |
| "step": 500224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1801393246560615e-05, | |
| "loss": 4.1442, | |
| "step": 500736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1793007299050095e-05, | |
| "loss": 4.1477, | |
| "step": 501248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1784621351539575e-05, | |
| "loss": 4.1588, | |
| "step": 501760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1776235404029055e-05, | |
| "loss": 4.1613, | |
| "step": 502272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1767849456518535e-05, | |
| "loss": 4.1514, | |
| "step": 502784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.175947988781175e-05, | |
| "loss": 4.1579, | |
| "step": 503296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.175111031910496e-05, | |
| "loss": 4.1396, | |
| "step": 503808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.174272437159444e-05, | |
| "loss": 4.1488, | |
| "step": 504320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.173433842408392e-05, | |
| "loss": 4.1485, | |
| "step": 504832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.17259524765734e-05, | |
| "loss": 4.154, | |
| "step": 505344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.171756652906288e-05, | |
| "loss": 4.1663, | |
| "step": 505856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.170918058155236e-05, | |
| "loss": 4.1526, | |
| "step": 506368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.170079463404184e-05, | |
| "loss": 4.1557, | |
| "step": 506880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.169240868653132e-05, | |
| "loss": 4.1477, | |
| "step": 507392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.16840227390208e-05, | |
| "loss": 4.1466, | |
| "step": 507904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.167563679151028e-05, | |
| "loss": 4.1489, | |
| "step": 508416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.166725084399975e-05, | |
| "loss": 4.1606, | |
| "step": 508928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165886489648923e-05, | |
| "loss": 4.156, | |
| "step": 509440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.165049532778245e-05, | |
| "loss": 4.1564, | |
| "step": 509952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1642125759075664e-05, | |
| "loss": 4.143, | |
| "step": 510464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1633739811565144e-05, | |
| "loss": 4.1564, | |
| "step": 510976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1625353864054624e-05, | |
| "loss": 4.1427, | |
| "step": 511488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1616967916544104e-05, | |
| "loss": 4.1521, | |
| "step": 512000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160859834783731e-05, | |
| "loss": 4.1375, | |
| "step": 512512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.160021240032679e-05, | |
| "loss": 4.1538, | |
| "step": 513024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.159182645281627e-05, | |
| "loss": 4.1422, | |
| "step": 513536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.158344050530575e-05, | |
| "loss": 4.1489, | |
| "step": 514048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1575054557795226e-05, | |
| "loss": 4.139, | |
| "step": 514560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1566668610284706e-05, | |
| "loss": 4.148, | |
| "step": 515072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.155829904157792e-05, | |
| "loss": 4.1429, | |
| "step": 515584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.15499130940674e-05, | |
| "loss": 4.1403, | |
| "step": 516096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.154152714655688e-05, | |
| "loss": 4.1584, | |
| "step": 516608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.153314119904636e-05, | |
| "loss": 4.1587, | |
| "step": 517120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.152475525153584e-05, | |
| "loss": 4.1541, | |
| "step": 517632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.151636930402532e-05, | |
| "loss": 4.1491, | |
| "step": 518144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.15079833565148e-05, | |
| "loss": 4.1393, | |
| "step": 518656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.149959740900428e-05, | |
| "loss": 4.1439, | |
| "step": 519168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.149122784029749e-05, | |
| "loss": 4.1479, | |
| "step": 519680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.148284189278697e-05, | |
| "loss": 4.1516, | |
| "step": 520192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.147445594527645e-05, | |
| "loss": 4.1333, | |
| "step": 520704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.146606999776593e-05, | |
| "loss": 4.149, | |
| "step": 521216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.145768405025541e-05, | |
| "loss": 4.1486, | |
| "step": 521728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.144931448154863e-05, | |
| "loss": 4.1454, | |
| "step": 522240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.144092853403811e-05, | |
| "loss": 4.1418, | |
| "step": 522752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1432542586527587e-05, | |
| "loss": 4.1462, | |
| "step": 523264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1424156639017067e-05, | |
| "loss": 4.1381, | |
| "step": 523776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1415787070310276e-05, | |
| "loss": 4.1563, | |
| "step": 524288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1407401122799756e-05, | |
| "loss": 4.1516, | |
| "step": 524800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1399015175289236e-05, | |
| "loss": 4.1502, | |
| "step": 525312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1390629227778716e-05, | |
| "loss": 4.1346, | |
| "step": 525824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1382259659071925e-05, | |
| "loss": 4.1536, | |
| "step": 526336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1373873711561405e-05, | |
| "loss": 4.1497, | |
| "step": 526848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1365487764050885e-05, | |
| "loss": 4.1393, | |
| "step": 527360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.1357118195344094e-05, | |
| "loss": 4.1481, | |
| "step": 527872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.134874862663731e-05, | |
| "loss": 4.148, | |
| "step": 528384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.134036267912679e-05, | |
| "loss": 4.1444, | |
| "step": 528896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.133197673161627e-05, | |
| "loss": 4.1619, | |
| "step": 529408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.132359078410575e-05, | |
| "loss": 4.1375, | |
| "step": 529920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.131520483659523e-05, | |
| "loss": 4.1472, | |
| "step": 530432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.130681888908471e-05, | |
| "loss": 4.1446, | |
| "step": 530944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.129843294157419e-05, | |
| "loss": 4.1437, | |
| "step": 531456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.129004699406367e-05, | |
| "loss": 4.136, | |
| "step": 531968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.128166104655315e-05, | |
| "loss": 4.133, | |
| "step": 532480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.127327509904263e-05, | |
| "loss": 4.1396, | |
| "step": 532992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.126488915153211e-05, | |
| "loss": 4.1337, | |
| "step": 533504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.125650320402159e-05, | |
| "loss": 4.1461, | |
| "step": 534016 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.14478063583374, | |
| "eval_runtime": 294.6857, | |
| "eval_samples_per_second": 1294.908, | |
| "eval_steps_per_second": 40.467, | |
| "step": 534240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1248150014118534e-05, | |
| "loss": 4.1446, | |
| "step": 534528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1239764066608014e-05, | |
| "loss": 4.1332, | |
| "step": 535040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1231378119097494e-05, | |
| "loss": 4.1452, | |
| "step": 535552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1222992171586974e-05, | |
| "loss": 4.1421, | |
| "step": 536064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1214606224076454e-05, | |
| "loss": 4.1471, | |
| "step": 536576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1206220276565934e-05, | |
| "loss": 4.1392, | |
| "step": 537088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1197834329055414e-05, | |
| "loss": 4.1432, | |
| "step": 537600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.118944838154489e-05, | |
| "loss": 4.1318, | |
| "step": 538112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.11810788128381e-05, | |
| "loss": 4.1304, | |
| "step": 538624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.117270924413131e-05, | |
| "loss": 4.1464, | |
| "step": 539136 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.116432329662079e-05, | |
| "loss": 4.1415, | |
| "step": 539648 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.115593734911027e-05, | |
| "loss": 4.1344, | |
| "step": 540160 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.114756778040348e-05, | |
| "loss": 4.1455, | |
| "step": 540672 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.113918183289297e-05, | |
| "loss": 4.1318, | |
| "step": 541184 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.113079588538245e-05, | |
| "loss": 4.1317, | |
| "step": 541696 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.112240993787193e-05, | |
| "loss": 4.1251, | |
| "step": 542208 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.111402399036141e-05, | |
| "loss": 4.1336, | |
| "step": 542720 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.110563804285089e-05, | |
| "loss": 4.1379, | |
| "step": 543232 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.109725209534036e-05, | |
| "loss": 4.1321, | |
| "step": 543744 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.108886614782984e-05, | |
| "loss": 4.1482, | |
| "step": 544256 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.108048020031932e-05, | |
| "loss": 4.1428, | |
| "step": 544768 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.10720942528088e-05, | |
| "loss": 4.1445, | |
| "step": 545280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.106370830529828e-05, | |
| "loss": 4.132, | |
| "step": 545792 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.105532235778776e-05, | |
| "loss": 4.1363, | |
| "step": 546304 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.104695278908097e-05, | |
| "loss": 4.1452, | |
| "step": 546816 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.103856684157045e-05, | |
| "loss": 4.1374, | |
| "step": 547328 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1030197272863665e-05, | |
| "loss": 4.1305, | |
| "step": 547840 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1021811325353145e-05, | |
| "loss": 4.1346, | |
| "step": 548352 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1013425377842625e-05, | |
| "loss": 4.1241, | |
| "step": 548864 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1005039430332105e-05, | |
| "loss": 4.1335, | |
| "step": 549376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0996653482821585e-05, | |
| "loss": 4.1343, | |
| "step": 549888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0988267535311065e-05, | |
| "loss": 4.1413, | |
| "step": 550400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0979881587800545e-05, | |
| "loss": 4.1411, | |
| "step": 550912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0971512019093754e-05, | |
| "loss": 4.1401, | |
| "step": 551424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0963126071583234e-05, | |
| "loss": 4.1335, | |
| "step": 551936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0954740124072714e-05, | |
| "loss": 4.1329, | |
| "step": 552448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0946354176562194e-05, | |
| "loss": 4.1388, | |
| "step": 552960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.09379846078554e-05, | |
| "loss": 4.1242, | |
| "step": 553472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.092959866034489e-05, | |
| "loss": 4.1217, | |
| "step": 553984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.092121271283437e-05, | |
| "loss": 4.1188, | |
| "step": 554496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.091282676532385e-05, | |
| "loss": 4.1273, | |
| "step": 555008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.090444081781333e-05, | |
| "loss": 4.1297, | |
| "step": 555520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.089605487030281e-05, | |
| "loss": 4.1322, | |
| "step": 556032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.088766892279229e-05, | |
| "loss": 4.1268, | |
| "step": 556544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.087928297528177e-05, | |
| "loss": 4.1279, | |
| "step": 557056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.087089702777125e-05, | |
| "loss": 4.1302, | |
| "step": 557568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.086251108026072e-05, | |
| "loss": 4.1349, | |
| "step": 558080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.08541251327502e-05, | |
| "loss": 4.1225, | |
| "step": 558592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.084575556404342e-05, | |
| "loss": 4.1062, | |
| "step": 559104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.08373696165329e-05, | |
| "loss": 4.1441, | |
| "step": 559616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.082898366902237e-05, | |
| "loss": 4.1347, | |
| "step": 560128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.082059772151186e-05, | |
| "loss": 4.1403, | |
| "step": 560640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.081221177400134e-05, | |
| "loss": 4.1244, | |
| "step": 561152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.080382582649082e-05, | |
| "loss": 4.1288, | |
| "step": 561664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.079545625778403e-05, | |
| "loss": 4.1246, | |
| "step": 562176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.078707031027351e-05, | |
| "loss": 4.1247, | |
| "step": 562688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.077868436276299e-05, | |
| "loss": 4.1326, | |
| "step": 563200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.077029841525247e-05, | |
| "loss": 4.117, | |
| "step": 563712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.076191246774195e-05, | |
| "loss": 4.1466, | |
| "step": 564224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.075355927783889e-05, | |
| "loss": 4.1258, | |
| "step": 564736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.074517333032837e-05, | |
| "loss": 4.105, | |
| "step": 565248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0736787382817845e-05, | |
| "loss": 4.1362, | |
| "step": 565760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0728401435307325e-05, | |
| "loss": 4.112, | |
| "step": 566272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.072001548779681e-05, | |
| "loss": 4.1093, | |
| "step": 566784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.071162954028629e-05, | |
| "loss": 4.1355, | |
| "step": 567296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.070324359277577e-05, | |
| "loss": 4.128, | |
| "step": 567808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.069485764526525e-05, | |
| "loss": 4.115, | |
| "step": 568320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.068647169775473e-05, | |
| "loss": 4.1249, | |
| "step": 568832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.067808575024421e-05, | |
| "loss": 4.1047, | |
| "step": 569344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.066969980273369e-05, | |
| "loss": 4.118, | |
| "step": 569856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.066131385522317e-05, | |
| "loss": 4.1249, | |
| "step": 570368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.065294428651638e-05, | |
| "loss": 4.1176, | |
| "step": 570880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.064455833900586e-05, | |
| "loss": 4.1265, | |
| "step": 571392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.063617239149534e-05, | |
| "loss": 4.1232, | |
| "step": 571904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.062778644398482e-05, | |
| "loss": 4.138, | |
| "step": 572416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.06194004964743e-05, | |
| "loss": 4.1163, | |
| "step": 572928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.061101454896378e-05, | |
| "loss": 4.1201, | |
| "step": 573440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.060262860145326e-05, | |
| "loss": 4.1279, | |
| "step": 573952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.059424265394274e-05, | |
| "loss": 4.1145, | |
| "step": 574464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0585873085235956e-05, | |
| "loss": 4.1257, | |
| "step": 574976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0577487137725436e-05, | |
| "loss": 4.1205, | |
| "step": 575488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.056910119021491e-05, | |
| "loss": 4.1129, | |
| "step": 576000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.056071524270439e-05, | |
| "loss": 4.1341, | |
| "step": 576512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.055232929519387e-05, | |
| "loss": 4.1084, | |
| "step": 577024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0543959726487085e-05, | |
| "loss": 4.1121, | |
| "step": 577536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.053557377897656e-05, | |
| "loss": 4.124, | |
| "step": 578048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.052718783146604e-05, | |
| "loss": 4.1229, | |
| "step": 578560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.051880188395552e-05, | |
| "loss": 4.1149, | |
| "step": 579072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0510448694052463e-05, | |
| "loss": 4.1241, | |
| "step": 579584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0502062746541943e-05, | |
| "loss": 4.1037, | |
| "step": 580096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.049367679903143e-05, | |
| "loss": 4.1108, | |
| "step": 580608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.048529085152091e-05, | |
| "loss": 4.114, | |
| "step": 581120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.047690490401038e-05, | |
| "loss": 4.1163, | |
| "step": 581632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.046851895649986e-05, | |
| "loss": 4.1319, | |
| "step": 582144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.046013300898934e-05, | |
| "loss": 4.119, | |
| "step": 582656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.045174706147882e-05, | |
| "loss": 4.1228, | |
| "step": 583168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.044337749277203e-05, | |
| "loss": 4.112, | |
| "step": 583680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.043500792406525e-05, | |
| "loss": 4.1136, | |
| "step": 584192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.042662197655473e-05, | |
| "loss": 4.1149, | |
| "step": 584704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.041823602904421e-05, | |
| "loss": 4.1245, | |
| "step": 585216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.040985008153368e-05, | |
| "loss": 4.1223, | |
| "step": 585728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.040146413402317e-05, | |
| "loss": 4.1232, | |
| "step": 586240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.039307818651265e-05, | |
| "loss": 4.113, | |
| "step": 586752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.038470861780586e-05, | |
| "loss": 4.117, | |
| "step": 587264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.037633904909907e-05, | |
| "loss": 4.1115, | |
| "step": 587776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.036795310158855e-05, | |
| "loss": 4.1152, | |
| "step": 588288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.035956715407803e-05, | |
| "loss": 4.1038, | |
| "step": 588800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0351181206567506e-05, | |
| "loss": 4.1212, | |
| "step": 589312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0342795259056986e-05, | |
| "loss": 4.1083, | |
| "step": 589824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0334409311546466e-05, | |
| "loss": 4.1142, | |
| "step": 590336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0326023364035946e-05, | |
| "loss": 4.1057, | |
| "step": 590848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0317637416525426e-05, | |
| "loss": 4.1124, | |
| "step": 591360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0309251469014906e-05, | |
| "loss": 4.1111, | |
| "step": 591872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0300865521504386e-05, | |
| "loss": 4.1085, | |
| "step": 592384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0292479573993866e-05, | |
| "loss": 4.1208, | |
| "step": 592896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.028409362648335e-05, | |
| "loss": 4.1266, | |
| "step": 593408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.027572405777656e-05, | |
| "loss": 4.1251, | |
| "step": 593920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.026733811026604e-05, | |
| "loss": 4.1141, | |
| "step": 594432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.025895216275552e-05, | |
| "loss": 4.105, | |
| "step": 594944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.025058259404873e-05, | |
| "loss": 4.109, | |
| "step": 595456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.024219664653821e-05, | |
| "loss": 4.1143, | |
| "step": 595968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.023381069902769e-05, | |
| "loss": 4.117, | |
| "step": 596480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.022542475151717e-05, | |
| "loss": 4.0999, | |
| "step": 596992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.021703880400665e-05, | |
| "loss": 4.1167, | |
| "step": 597504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.020866923529986e-05, | |
| "loss": 4.1165, | |
| "step": 598016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.020028328778934e-05, | |
| "loss": 4.1088, | |
| "step": 598528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.019189734027882e-05, | |
| "loss": 4.1084, | |
| "step": 599040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0183527771572035e-05, | |
| "loss": 4.1148, | |
| "step": 599552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0175141824061515e-05, | |
| "loss": 4.1057, | |
| "step": 600064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0166755876550995e-05, | |
| "loss": 4.1213, | |
| "step": 600576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0158369929040475e-05, | |
| "loss": 4.1133, | |
| "step": 601088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0149983981529955e-05, | |
| "loss": 4.12, | |
| "step": 601600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0141598034019435e-05, | |
| "loss": 4.102, | |
| "step": 602112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0133212086508915e-05, | |
| "loss": 4.1204, | |
| "step": 602624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0124826138998395e-05, | |
| "loss": 4.1174, | |
| "step": 603136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.011644019148787e-05, | |
| "loss": 4.1076, | |
| "step": 603648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.010805424397735e-05, | |
| "loss": 4.1149, | |
| "step": 604160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0099684675270564e-05, | |
| "loss": 4.1132, | |
| "step": 604672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0091298727760044e-05, | |
| "loss": 4.1143, | |
| "step": 605184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0082912780249524e-05, | |
| "loss": 4.1307, | |
| "step": 605696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0074526832739004e-05, | |
| "loss": 4.1025, | |
| "step": 606208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0066140885228484e-05, | |
| "loss": 4.1165, | |
| "step": 606720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0057754937717964e-05, | |
| "loss": 4.1122, | |
| "step": 607232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0049368990207444e-05, | |
| "loss": 4.1119, | |
| "step": 607744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0040983042696924e-05, | |
| "loss": 4.1041, | |
| "step": 608256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.003261347399013e-05, | |
| "loss": 4.0998, | |
| "step": 608768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.002422752647961e-05, | |
| "loss": 4.1098, | |
| "step": 609280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.001585795777282e-05, | |
| "loss": 4.1006, | |
| "step": 609792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.00074720102623e-05, | |
| "loss": 4.1147, | |
| "step": 610304 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.119741916656494, | |
| "eval_runtime": 295.1931, | |
| "eval_samples_per_second": 1292.683, | |
| "eval_steps_per_second": 40.397, | |
| "step": 610560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.999908606275178e-05, | |
| "loss": 4.1065, | |
| "step": 610816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.999070011524126e-05, | |
| "loss": 4.1047, | |
| "step": 611328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.998231416773074e-05, | |
| "loss": 4.112, | |
| "step": 611840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.997392822022022e-05, | |
| "loss": 4.1107, | |
| "step": 612352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.996554227270971e-05, | |
| "loss": 4.1156, | |
| "step": 612864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.995715632519919e-05, | |
| "loss": 4.1081, | |
| "step": 613376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.994877037768867e-05, | |
| "loss": 4.1144, | |
| "step": 613888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.994038443017815e-05, | |
| "loss": 4.0992, | |
| "step": 614400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.993199848266763e-05, | |
| "loss": 4.099, | |
| "step": 614912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.992361253515711e-05, | |
| "loss": 4.116, | |
| "step": 615424 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.991522658764658e-05, | |
| "loss": 4.109, | |
| "step": 615936 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.990684064013606e-05, | |
| "loss": 4.1025, | |
| "step": 616448 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.989847107142928e-05, | |
| "loss": 4.115, | |
| "step": 616960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.989008512391876e-05, | |
| "loss": 4.0985, | |
| "step": 617472 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.988169917640823e-05, | |
| "loss": 4.103, | |
| "step": 617984 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.987331322889771e-05, | |
| "loss": 4.0881, | |
| "step": 618496 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.986492728138719e-05, | |
| "loss": 4.1038, | |
| "step": 619008 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.985654133387668e-05, | |
| "loss": 4.1096, | |
| "step": 619520 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.984815538636616e-05, | |
| "loss": 4.0968, | |
| "step": 620032 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.983976943885564e-05, | |
| "loss": 4.116, | |
| "step": 620544 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.983138349134512e-05, | |
| "loss": 4.1141, | |
| "step": 621056 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9822997543834597e-05, | |
| "loss": 4.1105, | |
| "step": 621568 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9814611596324077e-05, | |
| "loss": 4.1042, | |
| "step": 622080 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9806225648813557e-05, | |
| "loss": 4.1066, | |
| "step": 622592 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9797856080106766e-05, | |
| "loss": 4.1101, | |
| "step": 623104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9789470132596246e-05, | |
| "loss": 4.1053, | |
| "step": 623616 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9781084185085726e-05, | |
| "loss": 4.096, | |
| "step": 624128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9772714616378935e-05, | |
| "loss": 4.1058, | |
| "step": 624640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9764328668868415e-05, | |
| "loss": 4.0938, | |
| "step": 625152 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.9755942721357895e-05, | |
| "loss": 4.1017, | |
| "step": 625664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9747556773847374e-05, | |
| "loss": 4.1015, | |
| "step": 626176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.973917082633686e-05, | |
| "loss": 4.1066, | |
| "step": 626688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.973080125763007e-05, | |
| "loss": 4.1113, | |
| "step": 627200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.972241531011955e-05, | |
| "loss": 4.1132, | |
| "step": 627712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.971402936260903e-05, | |
| "loss": 4.0965, | |
| "step": 628224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.970565979390224e-05, | |
| "loss": 4.1042, | |
| "step": 628736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.969727384639172e-05, | |
| "loss": 4.1102, | |
| "step": 629248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.96888878988812e-05, | |
| "loss": 4.0916, | |
| "step": 629760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.968050195137068e-05, | |
| "loss": 4.093, | |
| "step": 630272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.967211600386016e-05, | |
| "loss": 4.0922, | |
| "step": 630784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.966373005634964e-05, | |
| "loss": 4.0909, | |
| "step": 631296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.965534410883912e-05, | |
| "loss": 4.1006, | |
| "step": 631808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.96469581613286e-05, | |
| "loss": 4.1019, | |
| "step": 632320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.963857221381808e-05, | |
| "loss": 4.0933, | |
| "step": 632832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.963018626630756e-05, | |
| "loss": 4.0953, | |
| "step": 633344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.962180031879704e-05, | |
| "loss": 4.097, | |
| "step": 633856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.961341437128652e-05, | |
| "loss": 4.1074, | |
| "step": 634368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.960504480257973e-05, | |
| "loss": 4.0875, | |
| "step": 634880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9596675233872944e-05, | |
| "loss": 4.08, | |
| "step": 635392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.958828928636242e-05, | |
| "loss": 4.1094, | |
| "step": 635904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.95799033388519e-05, | |
| "loss": 4.1031, | |
| "step": 636416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.957151739134138e-05, | |
| "loss": 4.108, | |
| "step": 636928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.956313144383086e-05, | |
| "loss": 4.0943, | |
| "step": 637440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9554761875124066e-05, | |
| "loss": 4.0969, | |
| "step": 637952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9546375927613546e-05, | |
| "loss": 4.0959, | |
| "step": 638464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.953800635890676e-05, | |
| "loss": 4.0944, | |
| "step": 638976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.952962041139624e-05, | |
| "loss": 4.1015, | |
| "step": 639488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.952123446388572e-05, | |
| "loss": 4.0845, | |
| "step": 640000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.95128485163752e-05, | |
| "loss": 4.1204, | |
| "step": 640512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.950446256886468e-05, | |
| "loss": 4.091, | |
| "step": 641024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.949607662135416e-05, | |
| "loss": 4.0762, | |
| "step": 641536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.948769067384364e-05, | |
| "loss": 4.1063, | |
| "step": 642048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.947930472633312e-05, | |
| "loss": 4.087, | |
| "step": 642560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.94709187788226e-05, | |
| "loss": 4.0771, | |
| "step": 643072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.946253283131208e-05, | |
| "loss": 4.1038, | |
| "step": 643584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.945414688380156e-05, | |
| "loss": 4.0997, | |
| "step": 644096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.944576093629104e-05, | |
| "loss": 4.0903, | |
| "step": 644608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.943739136758425e-05, | |
| "loss": 4.0919, | |
| "step": 645120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.942900542007373e-05, | |
| "loss": 4.0741, | |
| "step": 645632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.942061947256322e-05, | |
| "loss": 4.0861, | |
| "step": 646144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.94122335250527e-05, | |
| "loss": 4.0976, | |
| "step": 646656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9403863956345906e-05, | |
| "loss": 4.0905, | |
| "step": 647168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9395478008835386e-05, | |
| "loss": 4.0943, | |
| "step": 647680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9387108440128595e-05, | |
| "loss": 4.0916, | |
| "step": 648192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9378722492618075e-05, | |
| "loss": 4.1084, | |
| "step": 648704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9370336545107555e-05, | |
| "loss": 4.0897, | |
| "step": 649216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9361950597597035e-05, | |
| "loss": 4.0918, | |
| "step": 649728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9353564650086515e-05, | |
| "loss": 4.0946, | |
| "step": 650240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9345178702575995e-05, | |
| "loss": 4.0886, | |
| "step": 650752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9336792755065475e-05, | |
| "loss": 4.0975, | |
| "step": 651264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9328406807554955e-05, | |
| "loss": 4.0896, | |
| "step": 651776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.932003723884817e-05, | |
| "loss": 4.0875, | |
| "step": 652288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.931165129133765e-05, | |
| "loss": 4.1027, | |
| "step": 652800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.930326534382713e-05, | |
| "loss": 4.079, | |
| "step": 653312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.929489577512034e-05, | |
| "loss": 4.0844, | |
| "step": 653824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.928650982760982e-05, | |
| "loss": 4.0909, | |
| "step": 654336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.92781238800993e-05, | |
| "loss": 4.0941, | |
| "step": 654848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.926973793258878e-05, | |
| "loss": 4.0831, | |
| "step": 655360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.926135198507825e-05, | |
| "loss": 4.0965, | |
| "step": 655872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.925296603756773e-05, | |
| "loss": 4.0806, | |
| "step": 656384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.924458009005721e-05, | |
| "loss": 4.0834, | |
| "step": 656896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.923619414254669e-05, | |
| "loss": 4.0809, | |
| "step": 657408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.922784095264364e-05, | |
| "loss": 4.0859, | |
| "step": 657920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9219471383936854e-05, | |
| "loss": 4.1019, | |
| "step": 658432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9211085436426334e-05, | |
| "loss": 4.0876, | |
| "step": 658944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9202699488915813e-05, | |
| "loss": 4.0921, | |
| "step": 659456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9194313541405293e-05, | |
| "loss": 4.0838, | |
| "step": 659968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.918592759389477e-05, | |
| "loss": 4.0835, | |
| "step": 660480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.917754164638425e-05, | |
| "loss": 4.0864, | |
| "step": 660992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9169155698873727e-05, | |
| "loss": 4.0957, | |
| "step": 661504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9160769751363206e-05, | |
| "loss": 4.0937, | |
| "step": 662016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9152383803852686e-05, | |
| "loss": 4.0913, | |
| "step": 662528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9143997856342166e-05, | |
| "loss": 4.0912, | |
| "step": 663040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9135611908831646e-05, | |
| "loss": 4.0873, | |
| "step": 663552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9127225961321126e-05, | |
| "loss": 4.0851, | |
| "step": 664064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.911885639261434e-05, | |
| "loss": 4.0859, | |
| "step": 664576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.911047044510382e-05, | |
| "loss": 4.0751, | |
| "step": 665088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.91020844975933e-05, | |
| "loss": 4.0911, | |
| "step": 665600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.909369855008278e-05, | |
| "loss": 4.0809, | |
| "step": 666112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.908531260257226e-05, | |
| "loss": 4.0834, | |
| "step": 666624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.907694303386547e-05, | |
| "loss": 4.0807, | |
| "step": 667136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.906855708635495e-05, | |
| "loss": 4.0781, | |
| "step": 667648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.906017113884443e-05, | |
| "loss": 4.0847, | |
| "step": 668160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.905178519133391e-05, | |
| "loss": 4.0793, | |
| "step": 668672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.904341562262712e-05, | |
| "loss": 4.0959, | |
| "step": 669184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.90350296751166e-05, | |
| "loss": 4.0958, | |
| "step": 669696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9026660106409816e-05, | |
| "loss": 4.0943, | |
| "step": 670208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9018274158899296e-05, | |
| "loss": 4.0855, | |
| "step": 670720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9009904590192505e-05, | |
| "loss": 4.0805, | |
| "step": 671232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9001518642681985e-05, | |
| "loss": 4.079, | |
| "step": 671744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8993132695171465e-05, | |
| "loss": 4.0876, | |
| "step": 672256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8984746747660945e-05, | |
| "loss": 4.0915, | |
| "step": 672768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8976360800150425e-05, | |
| "loss": 4.0741, | |
| "step": 673280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8967974852639905e-05, | |
| "loss": 4.0845, | |
| "step": 673792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8959588905129385e-05, | |
| "loss": 4.0881, | |
| "step": 674304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8951219336422594e-05, | |
| "loss": 4.0844, | |
| "step": 674816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8942833388912074e-05, | |
| "loss": 4.0793, | |
| "step": 675328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8934447441401554e-05, | |
| "loss": 4.0856, | |
| "step": 675840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8926061493891034e-05, | |
| "loss": 4.0762, | |
| "step": 676352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8917675546380514e-05, | |
| "loss": 4.0941, | |
| "step": 676864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8909289598869994e-05, | |
| "loss": 4.0846, | |
| "step": 677376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.890090365135948e-05, | |
| "loss": 4.0962, | |
| "step": 677888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.889251770384896e-05, | |
| "loss": 4.0769, | |
| "step": 678400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.888413175633844e-05, | |
| "loss": 4.0879, | |
| "step": 678912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.887574580882791e-05, | |
| "loss": 4.0925, | |
| "step": 679424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.886737624012113e-05, | |
| "loss": 4.0814, | |
| "step": 679936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.885899029261061e-05, | |
| "loss": 4.0849, | |
| "step": 680448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.885060434510009e-05, | |
| "loss": 4.0867, | |
| "step": 680960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.884221839758956e-05, | |
| "loss": 4.0832, | |
| "step": 681472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.883383245007904e-05, | |
| "loss": 4.1008, | |
| "step": 681984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.882544650256852e-05, | |
| "loss": 4.0796, | |
| "step": 682496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8817060555058e-05, | |
| "loss": 4.0891, | |
| "step": 683008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.880867460754748e-05, | |
| "loss": 4.0861, | |
| "step": 683520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.88003050388407e-05, | |
| "loss": 4.0833, | |
| "step": 684032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.879191909133018e-05, | |
| "loss": 4.0766, | |
| "step": 684544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.878353314381966e-05, | |
| "loss": 4.0749, | |
| "step": 685056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.877516357511287e-05, | |
| "loss": 4.08, | |
| "step": 685568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.876677762760235e-05, | |
| "loss": 4.0777, | |
| "step": 686080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.875839168009183e-05, | |
| "loss": 4.0844, | |
| "step": 686592 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.099898815155029, | |
| "eval_runtime": 286.0052, | |
| "eval_samples_per_second": 1334.21, | |
| "eval_steps_per_second": 41.695, | |
| "step": 686880 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.875000573258131e-05, | |
| "loss": 4.0793, | |
| "step": 687104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.874161978507079e-05, | |
| "loss": 4.0783, | |
| "step": 687616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.873323383756027e-05, | |
| "loss": 4.0808, | |
| "step": 688128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.872484789004975e-05, | |
| "loss": 4.0872, | |
| "step": 688640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.871646194253923e-05, | |
| "loss": 4.092, | |
| "step": 689152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.870807599502871e-05, | |
| "loss": 4.0819, | |
| "step": 689664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8699706426321916e-05, | |
| "loss": 4.083, | |
| "step": 690176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.86913204788114e-05, | |
| "loss": 4.0733, | |
| "step": 690688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.868293453130088e-05, | |
| "loss": 4.0744, | |
| "step": 691200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.867454858379036e-05, | |
| "loss": 4.0843, | |
| "step": 691712 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.866617901508357e-05, | |
| "loss": 4.0827, | |
| "step": 692224 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.865779306757305e-05, | |
| "loss": 4.0779, | |
| "step": 692736 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.864943987766999e-05, | |
| "loss": 4.0875, | |
| "step": 693248 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.864105393015947e-05, | |
| "loss": 4.0729, | |
| "step": 693760 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.863266798264895e-05, | |
| "loss": 4.0749, | |
| "step": 694272 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.862428203513843e-05, | |
| "loss": 4.0593, | |
| "step": 694784 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.861589608762791e-05, | |
| "loss": 4.0812, | |
| "step": 695296 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.860751014011739e-05, | |
| "loss": 4.0814, | |
| "step": 695808 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.859912419260687e-05, | |
| "loss": 4.0708, | |
| "step": 696320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8590738245096356e-05, | |
| "loss": 4.0884, | |
| "step": 696832 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8582368676389565e-05, | |
| "loss": 4.0857, | |
| "step": 697344 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8573982728879045e-05, | |
| "loss": 4.0865, | |
| "step": 697856 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8565596781368525e-05, | |
| "loss": 4.0796, | |
| "step": 698368 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8557210833858005e-05, | |
| "loss": 4.0802, | |
| "step": 698880 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8548824886347485e-05, | |
| "loss": 4.0814, | |
| "step": 699392 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8540438938836965e-05, | |
| "loss": 4.0832, | |
| "step": 699904 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8532052991326445e-05, | |
| "loss": 4.0658, | |
| "step": 700416 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.852366704381592e-05, | |
| "loss": 4.0804, | |
| "step": 700928 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.85152810963054e-05, | |
| "loss": 4.0673, | |
| "step": 701440 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.8506911527598614e-05, | |
| "loss": 4.0764, | |
| "step": 701952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.849854195889182e-05, | |
| "loss": 4.0779, | |
| "step": 702464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.849015601138131e-05, | |
| "loss": 4.0833, | |
| "step": 702976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.848177006387079e-05, | |
| "loss": 4.0767, | |
| "step": 703488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.847338411636027e-05, | |
| "loss": 4.0888, | |
| "step": 704000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.846499816884975e-05, | |
| "loss": 4.0721, | |
| "step": 704512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.845662860014296e-05, | |
| "loss": 4.0776, | |
| "step": 705024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.844824265263244e-05, | |
| "loss": 4.0824, | |
| "step": 705536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.843985670512192e-05, | |
| "loss": 4.0621, | |
| "step": 706048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.843147075761139e-05, | |
| "loss": 4.0662, | |
| "step": 706560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.842308481010087e-05, | |
| "loss": 4.0706, | |
| "step": 707072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.841469886259035e-05, | |
| "loss": 4.0652, | |
| "step": 707584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.840631291507983e-05, | |
| "loss": 4.0737, | |
| "step": 708096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.839794334637305e-05, | |
| "loss": 4.0811, | |
| "step": 708608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.838955739886253e-05, | |
| "loss": 4.0672, | |
| "step": 709120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.838117145135201e-05, | |
| "loss": 4.0732, | |
| "step": 709632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.837278550384149e-05, | |
| "loss": 4.0705, | |
| "step": 710144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.836439955633097e-05, | |
| "loss": 4.0775, | |
| "step": 710656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.835601360882045e-05, | |
| "loss": 4.0632, | |
| "step": 711168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.834762766130993e-05, | |
| "loss": 4.0565, | |
| "step": 711680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.833924171379941e-05, | |
| "loss": 4.0833, | |
| "step": 712192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8330872145092616e-05, | |
| "loss": 4.078, | |
| "step": 712704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8322486197582096e-05, | |
| "loss": 4.0885, | |
| "step": 713216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8314100250071576e-05, | |
| "loss": 4.0657, | |
| "step": 713728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8305714302561056e-05, | |
| "loss": 4.0702, | |
| "step": 714240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8297344733854265e-05, | |
| "loss": 4.0718, | |
| "step": 714752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.828897516514748e-05, | |
| "loss": 4.0668, | |
| "step": 715264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.828058921763696e-05, | |
| "loss": 4.0815, | |
| "step": 715776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.827220327012644e-05, | |
| "loss": 4.0546, | |
| "step": 716288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.826381732261592e-05, | |
| "loss": 4.0959, | |
| "step": 716800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.82554313751054e-05, | |
| "loss": 4.07, | |
| "step": 717312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.824704542759488e-05, | |
| "loss": 4.0487, | |
| "step": 717824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.823865948008436e-05, | |
| "loss": 4.0773, | |
| "step": 718336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.823027353257384e-05, | |
| "loss": 4.0637, | |
| "step": 718848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.822188758506332e-05, | |
| "loss": 4.0512, | |
| "step": 719360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.82135016375528e-05, | |
| "loss": 4.0754, | |
| "step": 719872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.820511569004228e-05, | |
| "loss": 4.0782, | |
| "step": 720384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8196729742531754e-05, | |
| "loss": 4.0615, | |
| "step": 720896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8188343795021234e-05, | |
| "loss": 4.0701, | |
| "step": 721408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.817999060511818e-05, | |
| "loss": 4.048, | |
| "step": 721920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8171604657607666e-05, | |
| "loss": 4.063, | |
| "step": 722432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8163218710097146e-05, | |
| "loss": 4.0685, | |
| "step": 722944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8154832762586626e-05, | |
| "loss": 4.064, | |
| "step": 723456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8146463193879835e-05, | |
| "loss": 4.0759, | |
| "step": 723968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8138077246369315e-05, | |
| "loss": 4.062, | |
| "step": 724480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8129691298858795e-05, | |
| "loss": 4.0861, | |
| "step": 724992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8121305351348275e-05, | |
| "loss": 4.0655, | |
| "step": 725504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8112919403837755e-05, | |
| "loss": 4.0691, | |
| "step": 726016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.810453345632723e-05, | |
| "loss": 4.0687, | |
| "step": 726528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.809614750881671e-05, | |
| "loss": 4.0621, | |
| "step": 727040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.808776156130619e-05, | |
| "loss": 4.0697, | |
| "step": 727552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.807937561379567e-05, | |
| "loss": 4.0626, | |
| "step": 728064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.807098966628515e-05, | |
| "loss": 4.0696, | |
| "step": 728576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8062603718774634e-05, | |
| "loss": 4.0726, | |
| "step": 729088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.8054217771264114e-05, | |
| "loss": 4.0626, | |
| "step": 729600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.804584820255732e-05, | |
| "loss": 4.0558, | |
| "step": 730112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.803747863385053e-05, | |
| "loss": 4.066, | |
| "step": 730624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.802909268634001e-05, | |
| "loss": 4.0727, | |
| "step": 731136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.802070673882949e-05, | |
| "loss": 4.0582, | |
| "step": 731648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.801232079131897e-05, | |
| "loss": 4.0673, | |
| "step": 732160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.800393484380845e-05, | |
| "loss": 4.0543, | |
| "step": 732672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.799554889629793e-05, | |
| "loss": 4.0614, | |
| "step": 733184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.798717932759114e-05, | |
| "loss": 4.0591, | |
| "step": 733696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.797879338008062e-05, | |
| "loss": 4.0631, | |
| "step": 734208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.79704074325701e-05, | |
| "loss": 4.0759, | |
| "step": 734720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.796202148505959e-05, | |
| "loss": 4.0648, | |
| "step": 735232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.795363553754907e-05, | |
| "loss": 4.069, | |
| "step": 735744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.794526596884228e-05, | |
| "loss": 4.0595, | |
| "step": 736256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.793688002133176e-05, | |
| "loss": 4.059, | |
| "step": 736768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.792849407382124e-05, | |
| "loss": 4.0625, | |
| "step": 737280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.792010812631072e-05, | |
| "loss": 4.0707, | |
| "step": 737792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.79117221788002e-05, | |
| "loss": 4.0704, | |
| "step": 738304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.790333623128968e-05, | |
| "loss": 4.067, | |
| "step": 738816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.789495028377916e-05, | |
| "loss": 4.065, | |
| "step": 739328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.788656433626864e-05, | |
| "loss": 4.0606, | |
| "step": 739840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7878194767561846e-05, | |
| "loss": 4.0643, | |
| "step": 740352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7869808820051326e-05, | |
| "loss": 4.0611, | |
| "step": 740864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7861422872540806e-05, | |
| "loss": 4.0562, | |
| "step": 741376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7853036925030286e-05, | |
| "loss": 4.0648, | |
| "step": 741888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.78446673563235e-05, | |
| "loss": 4.0572, | |
| "step": 742400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.783628140881298e-05, | |
| "loss": 4.0591, | |
| "step": 742912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.782789546130246e-05, | |
| "loss": 4.0551, | |
| "step": 743424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.781950951379194e-05, | |
| "loss": 4.0535, | |
| "step": 743936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.781113994508515e-05, | |
| "loss": 4.0611, | |
| "step": 744448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.780275399757463e-05, | |
| "loss": 4.0538, | |
| "step": 744960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.779436805006411e-05, | |
| "loss": 4.0689, | |
| "step": 745472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.778598210255359e-05, | |
| "loss": 4.069, | |
| "step": 745984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.77776125338468e-05, | |
| "loss": 4.0754, | |
| "step": 746496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.776922658633628e-05, | |
| "loss": 4.0627, | |
| "step": 747008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.776084063882576e-05, | |
| "loss": 4.0582, | |
| "step": 747520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7752471070118975e-05, | |
| "loss": 4.0529, | |
| "step": 748032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7744085122608455e-05, | |
| "loss": 4.0687, | |
| "step": 748544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7735699175097935e-05, | |
| "loss": 4.0651, | |
| "step": 749056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7727313227587415e-05, | |
| "loss": 4.048, | |
| "step": 749568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.771892728007689e-05, | |
| "loss": 4.0606, | |
| "step": 750080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.771054133256637e-05, | |
| "loss": 4.0626, | |
| "step": 750592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.770215538505585e-05, | |
| "loss": 4.0616, | |
| "step": 751104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.769376943754533e-05, | |
| "loss": 4.0568, | |
| "step": 751616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.768539986883854e-05, | |
| "loss": 4.0607, | |
| "step": 752128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.767701392132802e-05, | |
| "loss": 4.0522, | |
| "step": 752640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.76686279738175e-05, | |
| "loss": 4.0723, | |
| "step": 753152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.766024202630698e-05, | |
| "loss": 4.0616, | |
| "step": 753664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.765187245760019e-05, | |
| "loss": 4.0699, | |
| "step": 754176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.764348651008967e-05, | |
| "loss": 4.0527, | |
| "step": 754688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.763510056257915e-05, | |
| "loss": 4.0665, | |
| "step": 755200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.762671461506863e-05, | |
| "loss": 4.0693, | |
| "step": 755712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.761834504636184e-05, | |
| "loss": 4.0549, | |
| "step": 756224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.760995909885132e-05, | |
| "loss": 4.0639, | |
| "step": 756736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.76015731513408e-05, | |
| "loss": 4.0618, | |
| "step": 757248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.759320358263401e-05, | |
| "loss": 4.0657, | |
| "step": 757760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.758481763512349e-05, | |
| "loss": 4.0739, | |
| "step": 758272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.757643168761297e-05, | |
| "loss": 4.0595, | |
| "step": 758784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.756804574010245e-05, | |
| "loss": 4.0626, | |
| "step": 759296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.755965979259193e-05, | |
| "loss": 4.063, | |
| "step": 759808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.755127384508141e-05, | |
| "loss": 4.061, | |
| "step": 760320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7542904276374627e-05, | |
| "loss": 4.0539, | |
| "step": 760832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7534518328864107e-05, | |
| "loss": 4.0523, | |
| "step": 761344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7526132381353586e-05, | |
| "loss": 4.0547, | |
| "step": 761856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7517746433843066e-05, | |
| "loss": 4.055, | |
| "step": 762368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7509360486332546e-05, | |
| "loss": 4.0589, | |
| "step": 762880 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.083510875701904, | |
| "eval_runtime": 283.4154, | |
| "eval_samples_per_second": 1346.402, | |
| "eval_steps_per_second": 42.076, | |
| "step": 763200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7500974538822026e-05, | |
| "loss": 4.059, | |
| "step": 763392 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7492588591311506e-05, | |
| "loss": 4.0533, | |
| "step": 763904 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7484202643800986e-05, | |
| "loss": 4.0576, | |
| "step": 764416 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7475816696290466e-05, | |
| "loss": 4.0637, | |
| "step": 764928 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7467430748779946e-05, | |
| "loss": 4.0705, | |
| "step": 765440 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7459044801269426e-05, | |
| "loss": 4.0576, | |
| "step": 765952 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.74506588537589e-05, | |
| "loss": 4.0631, | |
| "step": 766464 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.744227290624838e-05, | |
| "loss": 4.0519, | |
| "step": 766976 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7433886958737866e-05, | |
| "loss": 4.0487, | |
| "step": 767488 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7425501011227346e-05, | |
| "loss": 4.0568, | |
| "step": 768000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7417115063716826e-05, | |
| "loss": 4.0647, | |
| "step": 768512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7408729116206306e-05, | |
| "loss": 4.0564, | |
| "step": 769024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7400359547499515e-05, | |
| "loss": 4.0634, | |
| "step": 769536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7391973599988995e-05, | |
| "loss": 4.0482, | |
| "step": 770048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7383587652478475e-05, | |
| "loss": 4.0552, | |
| "step": 770560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7375201704967955e-05, | |
| "loss": 4.0329, | |
| "step": 771072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7366815757457435e-05, | |
| "loss": 4.0593, | |
| "step": 771584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7358429809946915e-05, | |
| "loss": 4.0581, | |
| "step": 772096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7350043862436395e-05, | |
| "loss": 4.0472, | |
| "step": 772608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7341657914925875e-05, | |
| "loss": 4.0639, | |
| "step": 773120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7333271967415355e-05, | |
| "loss": 4.0652, | |
| "step": 773632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7324886019904835e-05, | |
| "loss": 4.0639, | |
| "step": 774144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7316500072394315e-05, | |
| "loss": 4.053, | |
| "step": 774656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.7308114124883794e-05, | |
| "loss": 4.0578, | |
| "step": 775168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.729976093498074e-05, | |
| "loss": 4.0586, | |
| "step": 775680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.729137498747022e-05, | |
| "loss": 4.0619, | |
| "step": 776192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.72829890399597e-05, | |
| "loss": 4.0421, | |
| "step": 776704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.727460309244918e-05, | |
| "loss": 4.0555, | |
| "step": 777216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.726621714493866e-05, | |
| "loss": 4.0482, | |
| "step": 777728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.725784757623187e-05, | |
| "loss": 4.049, | |
| "step": 778240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.724946162872135e-05, | |
| "loss": 4.0575, | |
| "step": 778752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.724107568121083e-05, | |
| "loss": 4.0592, | |
| "step": 779264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.723268973370031e-05, | |
| "loss": 4.0563, | |
| "step": 779776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.722430378618979e-05, | |
| "loss": 4.0661, | |
| "step": 780288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7215934217483e-05, | |
| "loss": 4.0524, | |
| "step": 780800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.720756464877621e-05, | |
| "loss": 4.0565, | |
| "step": 781312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.719917870126569e-05, | |
| "loss": 4.0632, | |
| "step": 781824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.719079275375517e-05, | |
| "loss": 4.0386, | |
| "step": 782336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.718240680624465e-05, | |
| "loss": 4.0428, | |
| "step": 782848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.717402085873413e-05, | |
| "loss": 4.0472, | |
| "step": 783360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.716565129002734e-05, | |
| "loss": 4.0377, | |
| "step": 783872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.715726534251682e-05, | |
| "loss": 4.0525, | |
| "step": 784384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.71488793950063e-05, | |
| "loss": 4.0597, | |
| "step": 784896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.714049344749578e-05, | |
| "loss": 4.0439, | |
| "step": 785408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.713212387878899e-05, | |
| "loss": 4.0504, | |
| "step": 785920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.712373793127847e-05, | |
| "loss": 4.0505, | |
| "step": 786432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.711535198376795e-05, | |
| "loss": 4.0521, | |
| "step": 786944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.710696603625744e-05, | |
| "loss": 4.0419, | |
| "step": 787456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.709858008874691e-05, | |
| "loss": 4.0336, | |
| "step": 787968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.709019414123639e-05, | |
| "loss": 4.0577, | |
| "step": 788480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.708180819372587e-05, | |
| "loss": 4.0591, | |
| "step": 788992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.707342224621535e-05, | |
| "loss": 4.0673, | |
| "step": 789504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.706503629870483e-05, | |
| "loss": 4.0451, | |
| "step": 790016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.705665035119431e-05, | |
| "loss": 4.046, | |
| "step": 790528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.704826440368379e-05, | |
| "loss": 4.0506, | |
| "step": 791040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.703987845617327e-05, | |
| "loss": 4.0455, | |
| "step": 791552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.703150888746648e-05, | |
| "loss": 4.0605, | |
| "step": 792064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.702312293995596e-05, | |
| "loss": 4.0312, | |
| "step": 792576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7014753371249176e-05, | |
| "loss": 4.0746, | |
| "step": 793088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7006367423738655e-05, | |
| "loss": 4.0509, | |
| "step": 793600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6997981476228135e-05, | |
| "loss": 4.0284, | |
| "step": 794112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6989595528717615e-05, | |
| "loss": 4.0537, | |
| "step": 794624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6981209581207095e-05, | |
| "loss": 4.0414, | |
| "step": 795136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6972823633696575e-05, | |
| "loss": 4.0296, | |
| "step": 795648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6964437686186055e-05, | |
| "loss": 4.0556, | |
| "step": 796160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6956051738675535e-05, | |
| "loss": 4.0552, | |
| "step": 796672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6947665791165015e-05, | |
| "loss": 4.04, | |
| "step": 797184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6939296222458224e-05, | |
| "loss": 4.0535, | |
| "step": 797696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.693092665375143e-05, | |
| "loss": 4.0244, | |
| "step": 798208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.692254070624091e-05, | |
| "loss": 4.0423, | |
| "step": 798720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.691415475873039e-05, | |
| "loss": 4.0409, | |
| "step": 799232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.690576881121987e-05, | |
| "loss": 4.0455, | |
| "step": 799744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.689738286370936e-05, | |
| "loss": 4.0484, | |
| "step": 800256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.688899691619884e-05, | |
| "loss": 4.042, | |
| "step": 800768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.688062734749205e-05, | |
| "loss": 4.0664, | |
| "step": 801280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.687224139998153e-05, | |
| "loss": 4.045, | |
| "step": 801792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.686385545247101e-05, | |
| "loss": 4.0499, | |
| "step": 802304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.685546950496049e-05, | |
| "loss": 4.0439, | |
| "step": 802816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.684708355744997e-05, | |
| "loss": 4.0397, | |
| "step": 803328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.683869760993945e-05, | |
| "loss": 4.0485, | |
| "step": 803840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.683031166242892e-05, | |
| "loss": 4.0416, | |
| "step": 804352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.682194209372214e-05, | |
| "loss": 4.0455, | |
| "step": 804864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.681355614621162e-05, | |
| "loss": 4.0563, | |
| "step": 805376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.68051701987011e-05, | |
| "loss": 4.0415, | |
| "step": 805888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.679678425119058e-05, | |
| "loss": 4.0315, | |
| "step": 806400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6788414682483794e-05, | |
| "loss": 4.0426, | |
| "step": 806912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6780028734973274e-05, | |
| "loss": 4.0528, | |
| "step": 807424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.677164278746275e-05, | |
| "loss": 4.0333, | |
| "step": 807936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.676325683995223e-05, | |
| "loss": 4.0502, | |
| "step": 808448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.675488727124544e-05, | |
| "loss": 4.0396, | |
| "step": 808960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.674650132373492e-05, | |
| "loss": 4.0349, | |
| "step": 809472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6738115376224396e-05, | |
| "loss": 4.0381, | |
| "step": 809984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6729729428713876e-05, | |
| "loss": 4.038, | |
| "step": 810496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6721343481203356e-05, | |
| "loss": 4.0596, | |
| "step": 811008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6712957533692836e-05, | |
| "loss": 4.0416, | |
| "step": 811520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6704587964986045e-05, | |
| "loss": 4.0474, | |
| "step": 812032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.669620201747553e-05, | |
| "loss": 4.0398, | |
| "step": 812544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.668781606996501e-05, | |
| "loss": 4.0383, | |
| "step": 813056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.667943012245449e-05, | |
| "loss": 4.0389, | |
| "step": 813568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.667104417494397e-05, | |
| "loss": 4.0501, | |
| "step": 814080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.666265822743345e-05, | |
| "loss": 4.0498, | |
| "step": 814592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.665428865872666e-05, | |
| "loss": 4.046, | |
| "step": 815104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.664590271121614e-05, | |
| "loss": 4.0458, | |
| "step": 815616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.663751676370562e-05, | |
| "loss": 4.0356, | |
| "step": 816128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.66291308161951e-05, | |
| "loss": 4.047, | |
| "step": 816640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.662074486868458e-05, | |
| "loss": 4.0389, | |
| "step": 817152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.661239167878152e-05, | |
| "loss": 4.0347, | |
| "step": 817664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6604005731271e-05, | |
| "loss": 4.0396, | |
| "step": 818176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6595619783760485e-05, | |
| "loss": 4.0391, | |
| "step": 818688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6587233836249965e-05, | |
| "loss": 4.0394, | |
| "step": 819200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6578847888739445e-05, | |
| "loss": 4.032, | |
| "step": 819712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6570461941228925e-05, | |
| "loss": 4.0314, | |
| "step": 820224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6562075993718405e-05, | |
| "loss": 4.0409, | |
| "step": 820736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6553690046207885e-05, | |
| "loss": 4.0423, | |
| "step": 821248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6545304098697365e-05, | |
| "loss": 4.0443, | |
| "step": 821760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6536918151186845e-05, | |
| "loss": 4.049, | |
| "step": 822272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6528532203676325e-05, | |
| "loss": 4.0564, | |
| "step": 822784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6520162634969534e-05, | |
| "loss": 4.0396, | |
| "step": 823296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6511776687459014e-05, | |
| "loss": 4.0397, | |
| "step": 823808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6503390739948494e-05, | |
| "loss": 4.0321, | |
| "step": 824320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6495004792437974e-05, | |
| "loss": 4.045, | |
| "step": 824832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.648663522373118e-05, | |
| "loss": 4.0438, | |
| "step": 825344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.647824927622067e-05, | |
| "loss": 4.0329, | |
| "step": 825856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.646987970751388e-05, | |
| "loss": 4.0342, | |
| "step": 826368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.646149376000336e-05, | |
| "loss": 4.0465, | |
| "step": 826880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.645310781249284e-05, | |
| "loss": 4.0424, | |
| "step": 827392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.644472186498232e-05, | |
| "loss": 4.0328, | |
| "step": 827904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.643635229627553e-05, | |
| "loss": 4.0377, | |
| "step": 828416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.642796634876501e-05, | |
| "loss": 4.0368, | |
| "step": 828928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.641958040125449e-05, | |
| "loss": 4.0469, | |
| "step": 829440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.641119445374397e-05, | |
| "loss": 4.0441, | |
| "step": 829952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.640280850623345e-05, | |
| "loss": 4.0492, | |
| "step": 830464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.639442255872293e-05, | |
| "loss": 4.0339, | |
| "step": 830976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.638603661121241e-05, | |
| "loss": 4.0457, | |
| "step": 831488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.637765066370189e-05, | |
| "loss": 4.0484, | |
| "step": 832000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.636926471619137e-05, | |
| "loss": 4.031, | |
| "step": 832512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.636087876868085e-05, | |
| "loss": 4.0408, | |
| "step": 833024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.635249282117033e-05, | |
| "loss": 4.0431, | |
| "step": 833536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6344123252463536e-05, | |
| "loss": 4.0477, | |
| "step": 834048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6335737304953016e-05, | |
| "loss": 4.0499, | |
| "step": 834560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6327351357442496e-05, | |
| "loss": 4.038, | |
| "step": 835072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6318965409931976e-05, | |
| "loss": 4.0421, | |
| "step": 835584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6310579462421456e-05, | |
| "loss": 4.0439, | |
| "step": 836096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6302193514910936e-05, | |
| "loss": 4.0379, | |
| "step": 836608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6293807567400416e-05, | |
| "loss": 4.0327, | |
| "step": 837120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6285421619889896e-05, | |
| "loss": 4.0336, | |
| "step": 837632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6277052051183105e-05, | |
| "loss": 4.0306, | |
| "step": 838144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.626866610367259e-05, | |
| "loss": 4.0391, | |
| "step": 838656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.626028015616207e-05, | |
| "loss": 4.0397, | |
| "step": 839168 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.069751739501953, | |
| "eval_runtime": 290.1959, | |
| "eval_samples_per_second": 1314.943, | |
| "eval_steps_per_second": 41.093, | |
| "step": 839520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.625189420865155e-05, | |
| "loss": 4.0365, | |
| "step": 839680 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.624350826114103e-05, | |
| "loss": 4.0351, | |
| "step": 840192 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.623513869243424e-05, | |
| "loss": 4.0383, | |
| "step": 840704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.622675274492372e-05, | |
| "loss": 4.0417, | |
| "step": 841216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.62183667974132e-05, | |
| "loss": 4.047, | |
| "step": 841728 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.620998084990268e-05, | |
| "loss": 4.039, | |
| "step": 842240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.620161128119589e-05, | |
| "loss": 4.0431, | |
| "step": 842752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.619322533368537e-05, | |
| "loss": 4.0346, | |
| "step": 843264 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.618483938617485e-05, | |
| "loss": 4.0272, | |
| "step": 843776 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.617646981746806e-05, | |
| "loss": 4.0357, | |
| "step": 844288 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.616808386995754e-05, | |
| "loss": 4.049, | |
| "step": 844800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6159697922447025e-05, | |
| "loss": 4.0313, | |
| "step": 845312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6151328353740234e-05, | |
| "loss": 4.0465, | |
| "step": 845824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6142942406229714e-05, | |
| "loss": 4.0262, | |
| "step": 846336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6134556458719194e-05, | |
| "loss": 4.0338, | |
| "step": 846848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6126170511208674e-05, | |
| "loss": 4.0166, | |
| "step": 847360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6117784563698154e-05, | |
| "loss": 4.0405, | |
| "step": 847872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6109398616187634e-05, | |
| "loss": 4.0367, | |
| "step": 848384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.6101012668677114e-05, | |
| "loss": 4.0312, | |
| "step": 848896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.609262672116659e-05, | |
| "loss": 4.0364, | |
| "step": 849408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.608424077365607e-05, | |
| "loss": 4.0504, | |
| "step": 849920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.607585482614555e-05, | |
| "loss": 4.042, | |
| "step": 850432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.606746887863503e-05, | |
| "loss": 4.0381, | |
| "step": 850944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.605908293112451e-05, | |
| "loss": 4.0347, | |
| "step": 851456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.605072974122146e-05, | |
| "loss": 4.0436, | |
| "step": 851968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.604234379371094e-05, | |
| "loss": 4.0378, | |
| "step": 852480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.603395784620042e-05, | |
| "loss": 4.0272, | |
| "step": 852992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.602557189868989e-05, | |
| "loss": 4.0341, | |
| "step": 853504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.601718595117937e-05, | |
| "loss": 4.0285, | |
| "step": 854016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.600881638247259e-05, | |
| "loss": 4.0272, | |
| "step": 854528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.600043043496206e-05, | |
| "loss": 4.04, | |
| "step": 855040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.599204448745154e-05, | |
| "loss": 4.033, | |
| "step": 855552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.598365853994102e-05, | |
| "loss": 4.0427, | |
| "step": 856064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.59752725924305e-05, | |
| "loss": 4.0437, | |
| "step": 856576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.596688664491998e-05, | |
| "loss": 4.0313, | |
| "step": 857088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.59585170762132e-05, | |
| "loss": 4.0378, | |
| "step": 857600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.595013112870268e-05, | |
| "loss": 4.0406, | |
| "step": 858112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.594174518119216e-05, | |
| "loss": 4.0228, | |
| "step": 858624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5933359233681637e-05, | |
| "loss": 4.0273, | |
| "step": 859136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5924973286171117e-05, | |
| "loss": 4.0215, | |
| "step": 859648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5916587338660597e-05, | |
| "loss": 4.0217, | |
| "step": 860160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5908201391150076e-05, | |
| "loss": 4.032, | |
| "step": 860672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5899831822443286e-05, | |
| "loss": 4.0416, | |
| "step": 861184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5891445874932766e-05, | |
| "loss": 4.0251, | |
| "step": 861696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5883059927422245e-05, | |
| "loss": 4.0269, | |
| "step": 862208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5874673979911725e-05, | |
| "loss": 4.0346, | |
| "step": 862720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5866288032401205e-05, | |
| "loss": 4.0326, | |
| "step": 863232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5857918463694414e-05, | |
| "loss": 4.0244, | |
| "step": 863744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.58495325161839e-05, | |
| "loss": 4.0151, | |
| "step": 864256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.584114656867338e-05, | |
| "loss": 4.0362, | |
| "step": 864768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.583276062116286e-05, | |
| "loss": 4.0419, | |
| "step": 865280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.582437467365234e-05, | |
| "loss": 4.0423, | |
| "step": 865792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.581598872614182e-05, | |
| "loss": 4.0263, | |
| "step": 866304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.58076027786313e-05, | |
| "loss": 4.027, | |
| "step": 866816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.579923320992451e-05, | |
| "loss": 4.0299, | |
| "step": 867328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.579084726241399e-05, | |
| "loss": 4.0239, | |
| "step": 867840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.578246131490347e-05, | |
| "loss": 4.0433, | |
| "step": 868352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.577407536739295e-05, | |
| "loss": 4.0147, | |
| "step": 868864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.576568941988242e-05, | |
| "loss": 4.0513, | |
| "step": 869376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.57573034723719e-05, | |
| "loss": 4.0316, | |
| "step": 869888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.574891752486138e-05, | |
| "loss": 4.0116, | |
| "step": 870400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.57405479561546e-05, | |
| "loss": 4.0308, | |
| "step": 870912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.573216200864408e-05, | |
| "loss": 4.0233, | |
| "step": 871424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.572377606113356e-05, | |
| "loss": 4.0129, | |
| "step": 871936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.571539011362304e-05, | |
| "loss": 4.0313, | |
| "step": 872448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.570700416611252e-05, | |
| "loss": 4.0355, | |
| "step": 872960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5698618218602e-05, | |
| "loss": 4.0225, | |
| "step": 873472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.569024864989521e-05, | |
| "loss": 4.0344, | |
| "step": 873984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.568186270238469e-05, | |
| "loss": 4.0056, | |
| "step": 874496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.567347675487417e-05, | |
| "loss": 4.0241, | |
| "step": 875008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.566509080736365e-05, | |
| "loss": 4.023, | |
| "step": 875520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.565670485985313e-05, | |
| "loss": 4.0283, | |
| "step": 876032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.564833529114634e-05, | |
| "loss": 4.0263, | |
| "step": 876544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5639949343635823e-05, | |
| "loss": 4.024, | |
| "step": 877056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.5631563396125303e-05, | |
| "loss": 4.0444, | |
| "step": 877568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.562317744861478e-05, | |
| "loss": 4.0289, | |
| "step": 878080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.561479150110426e-05, | |
| "loss": 4.0283, | |
| "step": 878592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.560640555359374e-05, | |
| "loss": 4.0264, | |
| "step": 879104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.559803598488695e-05, | |
| "loss": 4.0218, | |
| "step": 879616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.558965003737643e-05, | |
| "loss": 4.0319, | |
| "step": 880128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.558126408986591e-05, | |
| "loss": 4.0234, | |
| "step": 880640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.557287814235539e-05, | |
| "loss": 4.0232, | |
| "step": 881152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.556449219484487e-05, | |
| "loss": 4.0352, | |
| "step": 881664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.555610624733435e-05, | |
| "loss": 4.0264, | |
| "step": 882176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.554773667862756e-05, | |
| "loss": 4.0095, | |
| "step": 882688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.553935073111704e-05, | |
| "loss": 4.0254, | |
| "step": 883200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.553096478360652e-05, | |
| "loss": 4.0317, | |
| "step": 883712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.552259521489974e-05, | |
| "loss": 4.0165, | |
| "step": 884224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.551420926738922e-05, | |
| "loss": 4.0273, | |
| "step": 884736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.55058233198787e-05, | |
| "loss": 4.024, | |
| "step": 885248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.549743737236818e-05, | |
| "loss": 4.0143, | |
| "step": 885760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.548905142485766e-05, | |
| "loss": 4.0231, | |
| "step": 886272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.548066547734714e-05, | |
| "loss": 4.0177, | |
| "step": 886784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.547227952983661e-05, | |
| "loss": 4.0382, | |
| "step": 887296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.546389358232609e-05, | |
| "loss": 4.0256, | |
| "step": 887808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5455524013619306e-05, | |
| "loss": 4.0304, | |
| "step": 888320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5447138066108786e-05, | |
| "loss": 4.0171, | |
| "step": 888832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.543875211859826e-05, | |
| "loss": 4.0181, | |
| "step": 889344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.543036617108774e-05, | |
| "loss": 4.0256, | |
| "step": 889856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5421980223577226e-05, | |
| "loss": 4.0302, | |
| "step": 890368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5413594276066706e-05, | |
| "loss": 4.0296, | |
| "step": 890880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5405224707359915e-05, | |
| "loss": 4.0269, | |
| "step": 891392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5396838759849395e-05, | |
| "loss": 4.0244, | |
| "step": 891904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5388452812338875e-05, | |
| "loss": 4.0155, | |
| "step": 892416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5380066864828355e-05, | |
| "loss": 4.0293, | |
| "step": 892928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5371680917317835e-05, | |
| "loss": 4.0197, | |
| "step": 893440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5363311348611044e-05, | |
| "loss": 4.0221, | |
| "step": 893952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5354925401100524e-05, | |
| "loss": 4.0192, | |
| "step": 894464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5346539453590004e-05, | |
| "loss": 4.0198, | |
| "step": 894976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5338153506079483e-05, | |
| "loss": 4.0213, | |
| "step": 895488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5329767558568963e-05, | |
| "loss": 4.0114, | |
| "step": 896000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.532139798986218e-05, | |
| "loss": 4.0121, | |
| "step": 896512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.531301204235166e-05, | |
| "loss": 4.0244, | |
| "step": 897024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.530464247364487e-05, | |
| "loss": 4.0222, | |
| "step": 897536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.529625652613435e-05, | |
| "loss": 4.0236, | |
| "step": 898048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.528787057862383e-05, | |
| "loss": 4.0294, | |
| "step": 898560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.527948463111331e-05, | |
| "loss": 4.0381, | |
| "step": 899072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.527109868360279e-05, | |
| "loss": 4.0208, | |
| "step": 899584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.526271273609227e-05, | |
| "loss": 4.0171, | |
| "step": 900096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.525432678858175e-05, | |
| "loss": 4.0183, | |
| "step": 900608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.524594084107123e-05, | |
| "loss": 4.0243, | |
| "step": 901120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.523755489356071e-05, | |
| "loss": 4.0244, | |
| "step": 901632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.522916894605019e-05, | |
| "loss": 4.0176, | |
| "step": 902144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.522078299853967e-05, | |
| "loss": 4.0169, | |
| "step": 902656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.521239705102915e-05, | |
| "loss": 4.0268, | |
| "step": 903168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5204027482322364e-05, | |
| "loss": 4.0184, | |
| "step": 903680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5195641534811844e-05, | |
| "loss": 4.0191, | |
| "step": 904192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5187255587301324e-05, | |
| "loss": 4.0253, | |
| "step": 904704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.51788696397908e-05, | |
| "loss": 4.0201, | |
| "step": 905216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.517050007108401e-05, | |
| "loss": 4.0223, | |
| "step": 905728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.516211412357349e-05, | |
| "loss": 4.029, | |
| "step": 906240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.515372817606297e-05, | |
| "loss": 4.0295, | |
| "step": 906752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5145342228552446e-05, | |
| "loss": 4.0179, | |
| "step": 907264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5136956281041926e-05, | |
| "loss": 4.0279, | |
| "step": 907776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5128570333531406e-05, | |
| "loss": 4.031, | |
| "step": 908288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.5120184386020886e-05, | |
| "loss": 4.0169, | |
| "step": 908800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.51118148173141e-05, | |
| "loss": 4.0211, | |
| "step": 909312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.510344524860731e-05, | |
| "loss": 4.0256, | |
| "step": 909824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.50950593010968e-05, | |
| "loss": 4.031, | |
| "step": 910336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.508667335358627e-05, | |
| "loss": 4.0335, | |
| "step": 910848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.507828740607575e-05, | |
| "loss": 4.0251, | |
| "step": 911360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.506990145856523e-05, | |
| "loss": 4.0212, | |
| "step": 911872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.506151551105471e-05, | |
| "loss": 4.0256, | |
| "step": 912384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.505312956354419e-05, | |
| "loss": 4.0244, | |
| "step": 912896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.504474361603367e-05, | |
| "loss": 4.0127, | |
| "step": 913408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.503637404732688e-05, | |
| "loss": 4.0211, | |
| "step": 913920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.502798809981636e-05, | |
| "loss": 4.0084, | |
| "step": 914432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.501960215230584e-05, | |
| "loss": 4.0197, | |
| "step": 914944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.501123258359905e-05, | |
| "loss": 4.0201, | |
| "step": 915456 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.058034420013428, | |
| "eval_runtime": 291.6524, | |
| "eval_samples_per_second": 1308.376, | |
| "eval_steps_per_second": 40.888, | |
| "step": 915840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.5002846636088535e-05, | |
| "loss": 4.0278, | |
| "step": 915968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4994460688578015e-05, | |
| "loss": 4.0125, | |
| "step": 916480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4986074741067495e-05, | |
| "loss": 4.0176, | |
| "step": 916992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4977688793556975e-05, | |
| "loss": 4.0322, | |
| "step": 917504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4969302846046455e-05, | |
| "loss": 4.0252, | |
| "step": 918016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4960916898535935e-05, | |
| "loss": 4.0267, | |
| "step": 918528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4952530951025415e-05, | |
| "loss": 4.0241, | |
| "step": 919040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4944145003514895e-05, | |
| "loss": 4.0166, | |
| "step": 919552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4935775434808104e-05, | |
| "loss": 4.0086, | |
| "step": 920064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4927389487297584e-05, | |
| "loss": 4.0251, | |
| "step": 920576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4919003539787064e-05, | |
| "loss": 4.0219, | |
| "step": 921088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4910617592276544e-05, | |
| "loss": 4.0211, | |
| "step": 921600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.490224802356975e-05, | |
| "loss": 4.0274, | |
| "step": 922112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.489386207605923e-05, | |
| "loss": 4.0081, | |
| "step": 922624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.488547612854872e-05, | |
| "loss": 4.0177, | |
| "step": 923136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.48770901810382e-05, | |
| "loss": 3.9988, | |
| "step": 923648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.486870423352768e-05, | |
| "loss": 4.0251, | |
| "step": 924160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.486031828601716e-05, | |
| "loss": 4.016, | |
| "step": 924672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.485193233850663e-05, | |
| "loss": 4.0185, | |
| "step": 925184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.484354639099611e-05, | |
| "loss": 4.0223, | |
| "step": 925696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.483516044348559e-05, | |
| "loss": 4.0296, | |
| "step": 926208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.482679087477881e-05, | |
| "loss": 4.0246, | |
| "step": 926720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.481840492726828e-05, | |
| "loss": 4.0178, | |
| "step": 927232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.481001897975776e-05, | |
| "loss": 4.0169, | |
| "step": 927744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.480163303224724e-05, | |
| "loss": 4.0303, | |
| "step": 928256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.479324708473672e-05, | |
| "loss": 4.0203, | |
| "step": 928768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.47848611372262e-05, | |
| "loss": 4.0093, | |
| "step": 929280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.477649156851942e-05, | |
| "loss": 4.0201, | |
| "step": 929792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.47681056210089e-05, | |
| "loss": 4.0074, | |
| "step": 930304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.475971967349838e-05, | |
| "loss": 4.0155, | |
| "step": 930816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.475133372598786e-05, | |
| "loss": 4.0234, | |
| "step": 931328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.474294777847734e-05, | |
| "loss": 4.0154, | |
| "step": 931840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.473456183096682e-05, | |
| "loss": 4.0228, | |
| "step": 932352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.47261758834563e-05, | |
| "loss": 4.0277, | |
| "step": 932864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.471778993594578e-05, | |
| "loss": 4.0162, | |
| "step": 933376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4709420367238986e-05, | |
| "loss": 4.0227, | |
| "step": 933888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4701034419728466e-05, | |
| "loss": 4.0245, | |
| "step": 934400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4692648472217946e-05, | |
| "loss": 4.0044, | |
| "step": 934912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4684262524707426e-05, | |
| "loss": 4.0126, | |
| "step": 935424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4675876577196906e-05, | |
| "loss": 4.0035, | |
| "step": 935936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4667490629686386e-05, | |
| "loss": 4.0071, | |
| "step": 936448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.46591210609796e-05, | |
| "loss": 4.0166, | |
| "step": 936960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.465073511346908e-05, | |
| "loss": 4.0221, | |
| "step": 937472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.464234916595856e-05, | |
| "loss": 4.0071, | |
| "step": 937984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.463396321844804e-05, | |
| "loss": 4.0129, | |
| "step": 938496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.462557727093752e-05, | |
| "loss": 4.0142, | |
| "step": 939008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4617191323426995e-05, | |
| "loss": 4.0192, | |
| "step": 939520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4608805375916475e-05, | |
| "loss": 4.0083, | |
| "step": 940032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4600419428405955e-05, | |
| "loss": 3.9969, | |
| "step": 940544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.459204985969917e-05, | |
| "loss": 4.0161, | |
| "step": 941056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4583663912188644e-05, | |
| "loss": 4.0305, | |
| "step": 941568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4575277964678124e-05, | |
| "loss": 4.0267, | |
| "step": 942080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.456690839597134e-05, | |
| "loss": 4.0059, | |
| "step": 942592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.455852244846082e-05, | |
| "loss": 4.0124, | |
| "step": 943104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.45501365009503e-05, | |
| "loss": 4.0145, | |
| "step": 943616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4541783311047244e-05, | |
| "loss": 4.0012, | |
| "step": 944128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4533397363536724e-05, | |
| "loss": 4.0317, | |
| "step": 944640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4525011416026204e-05, | |
| "loss": 3.9953, | |
| "step": 945152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4516625468515684e-05, | |
| "loss": 4.0341, | |
| "step": 945664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4508239521005164e-05, | |
| "loss": 4.0135, | |
| "step": 946176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4499853573494644e-05, | |
| "loss": 3.9987, | |
| "step": 946688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.449146762598412e-05, | |
| "loss": 4.0133, | |
| "step": 947200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.44830816784736e-05, | |
| "loss": 4.0098, | |
| "step": 947712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.447469573096308e-05, | |
| "loss": 3.9943, | |
| "step": 948224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.446630978345256e-05, | |
| "loss": 4.0153, | |
| "step": 948736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.445794021474577e-05, | |
| "loss": 4.02, | |
| "step": 949248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.444955426723525e-05, | |
| "loss": 4.0069, | |
| "step": 949760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.444116831972473e-05, | |
| "loss": 4.0151, | |
| "step": 950272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.443278237221421e-05, | |
| "loss": 3.9902, | |
| "step": 950784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.442439642470369e-05, | |
| "loss": 4.0081, | |
| "step": 951296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.441601047719317e-05, | |
| "loss": 4.0046, | |
| "step": 951808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.440762452968265e-05, | |
| "loss": 4.0143, | |
| "step": 952320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.439923858217213e-05, | |
| "loss": 4.0069, | |
| "step": 952832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.439085263466161e-05, | |
| "loss": 4.01, | |
| "step": 953344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.438246668715109e-05, | |
| "loss": 4.0256, | |
| "step": 953856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.43740971184443e-05, | |
| "loss": 4.0128, | |
| "step": 954368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.436571117093378e-05, | |
| "loss": 4.0109, | |
| "step": 954880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.435732522342326e-05, | |
| "loss": 4.0096, | |
| "step": 955392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.434893927591274e-05, | |
| "loss": 4.0075, | |
| "step": 955904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.434055332840223e-05, | |
| "loss": 4.0151, | |
| "step": 956416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.433218375969544e-05, | |
| "loss": 4.0057, | |
| "step": 956928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.432379781218492e-05, | |
| "loss": 4.0132, | |
| "step": 957440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.43154118646744e-05, | |
| "loss": 4.0145, | |
| "step": 957952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.430702591716388e-05, | |
| "loss": 4.0143, | |
| "step": 958464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.429863996965336e-05, | |
| "loss": 3.9905, | |
| "step": 958976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.429025402214283e-05, | |
| "loss": 4.0122, | |
| "step": 959488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.428186807463231e-05, | |
| "loss": 4.0122, | |
| "step": 960000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.427348212712179e-05, | |
| "loss": 4.0023, | |
| "step": 960512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4265128937218735e-05, | |
| "loss": 4.011, | |
| "step": 961024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4256742989708215e-05, | |
| "loss": 4.0061, | |
| "step": 961536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4248357042197695e-05, | |
| "loss": 4.0014, | |
| "step": 962048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.423998747349091e-05, | |
| "loss": 4.0034, | |
| "step": 962560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.423160152598039e-05, | |
| "loss": 4.0055, | |
| "step": 963072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.422321557846987e-05, | |
| "loss": 4.0173, | |
| "step": 963584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.421482963095935e-05, | |
| "loss": 4.0097, | |
| "step": 964096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.420644368344883e-05, | |
| "loss": 4.0136, | |
| "step": 964608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4198057735938304e-05, | |
| "loss": 4.0036, | |
| "step": 965120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4189671788427784e-05, | |
| "loss": 4.0037, | |
| "step": 965632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4181285840917264e-05, | |
| "loss": 4.0074, | |
| "step": 966144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4172899893406744e-05, | |
| "loss": 4.0157, | |
| "step": 966656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4164513945896224e-05, | |
| "loss": 4.0123, | |
| "step": 967168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4156127998385704e-05, | |
| "loss": 4.0114, | |
| "step": 967680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4147742050875184e-05, | |
| "loss": 4.0105, | |
| "step": 968192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.41393724821684e-05, | |
| "loss": 4.0007, | |
| "step": 968704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.413098653465788e-05, | |
| "loss": 4.0126, | |
| "step": 969216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.412260058714736e-05, | |
| "loss": 4.0061, | |
| "step": 969728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.411423101844057e-05, | |
| "loss": 4.007, | |
| "step": 970240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.410584507093005e-05, | |
| "loss": 4.0018, | |
| "step": 970752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.409745912341953e-05, | |
| "loss": 4.0046, | |
| "step": 971264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.408907317590901e-05, | |
| "loss": 4.0043, | |
| "step": 971776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.408068722839849e-05, | |
| "loss": 4.0002, | |
| "step": 972288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.407230128088797e-05, | |
| "loss": 3.9934, | |
| "step": 972800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.406391533337745e-05, | |
| "loss": 4.0073, | |
| "step": 973312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.405552938586693e-05, | |
| "loss": 4.0133, | |
| "step": 973824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.404715981716014e-05, | |
| "loss": 4.0022, | |
| "step": 974336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.403877386964962e-05, | |
| "loss": 4.0135, | |
| "step": 974848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.40303879221391e-05, | |
| "loss": 4.0227, | |
| "step": 975360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4022001974628584e-05, | |
| "loss": 4.0048, | |
| "step": 975872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4013616027118064e-05, | |
| "loss": 4.0022, | |
| "step": 976384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4005230079607544e-05, | |
| "loss": 4.0035, | |
| "step": 976896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.399684413209702e-05, | |
| "loss": 4.0068, | |
| "step": 977408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.398847456339023e-05, | |
| "loss": 4.0066, | |
| "step": 977920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.398008861587971e-05, | |
| "loss": 4.0025, | |
| "step": 978432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.397170266836919e-05, | |
| "loss": 4.0039, | |
| "step": 978944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3963316720858666e-05, | |
| "loss": 4.011, | |
| "step": 979456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3954930773348146e-05, | |
| "loss": 4.0033, | |
| "step": 979968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3946544825837626e-05, | |
| "loss": 4.0088, | |
| "step": 980480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3938158878327106e-05, | |
| "loss": 4.0052, | |
| "step": 980992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.392978930962032e-05, | |
| "loss": 4.004, | |
| "step": 981504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.39214033621098e-05, | |
| "loss": 4.0008, | |
| "step": 982016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.391301741459928e-05, | |
| "loss": 4.0169, | |
| "step": 982528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.390463146708876e-05, | |
| "loss": 4.0112, | |
| "step": 983040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.389624551957824e-05, | |
| "loss": 4.0023, | |
| "step": 983552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.388785957206772e-05, | |
| "loss": 4.0079, | |
| "step": 984064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.38794736245572e-05, | |
| "loss": 4.0171, | |
| "step": 984576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.387108767704668e-05, | |
| "loss": 3.9982, | |
| "step": 985088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.386270172953616e-05, | |
| "loss": 4.0085, | |
| "step": 985600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.385433216082937e-05, | |
| "loss": 4.0096, | |
| "step": 986112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.384594621331885e-05, | |
| "loss": 4.0162, | |
| "step": 986624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.383756026580833e-05, | |
| "loss": 4.0144, | |
| "step": 987136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.382917431829781e-05, | |
| "loss": 4.0116, | |
| "step": 987648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.382080474959102e-05, | |
| "loss": 4.005, | |
| "step": 988160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3812418802080507e-05, | |
| "loss": 4.0175, | |
| "step": 988672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3804032854569986e-05, | |
| "loss": 4.0046, | |
| "step": 989184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3795646907059466e-05, | |
| "loss": 3.9957, | |
| "step": 989696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3787260959548946e-05, | |
| "loss": 4.0057, | |
| "step": 990208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3778907769645885e-05, | |
| "loss": 3.9948, | |
| "step": 990720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3770521822135365e-05, | |
| "loss": 4.0065, | |
| "step": 991232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3762135874624845e-05, | |
| "loss": 4.002, | |
| "step": 991744 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.048644065856934, | |
| "eval_runtime": 294.6646, | |
| "eval_samples_per_second": 1295.001, | |
| "eval_steps_per_second": 40.47, | |
| "step": 992160 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 4.0218938426440384e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |