| { |
| "best_metric": 3.917680501937866, |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-case/transformer/3/checkpoints/checkpoint-305280", |
| "epoch": 1.0250006060157382, |
| "eval_steps": 10, |
| "global_step": 305280, |
| "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": 11.0367, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.999161405248948e-05, |
| "loss": 6.8185, |
| "step": 512 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.998322810497896e-05, |
| "loss": 6.184, |
| "step": 1024 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.997484215746844e-05, |
| "loss": 5.9671, |
| "step": 1536 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.996645620995792e-05, |
| "loss": 5.8129, |
| "step": 2048 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.99580702624474e-05, |
| "loss": 5.7096, |
| "step": 2560 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.994968431493688e-05, |
| "loss": 5.6119, |
| "step": 3072 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.994129836742636e-05, |
| "loss": 5.5405, |
| "step": 3584 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.993291241991584e-05, |
| "loss": 5.4592, |
| "step": 4096 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.992452647240532e-05, |
| "loss": 5.4127, |
| "step": 4608 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.99161405248948e-05, |
| "loss": 5.3528, |
| "step": 5120 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.990775457738428e-05, |
| "loss": 5.3116, |
| "step": 5632 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.989936862987376e-05, |
| "loss": 5.2666, |
| "step": 6144 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.989099906116697e-05, |
| "loss": 5.2216, |
| "step": 6656 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.988261311365645e-05, |
| "loss": 5.1839, |
| "step": 7168 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.987422716614593e-05, |
| "loss": 5.1431, |
| "step": 7680 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.986584121863541e-05, |
| "loss": 5.114, |
| "step": 8192 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.985745527112489e-05, |
| "loss": 5.0846, |
| "step": 8704 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.984906932361437e-05, |
| "loss": 5.0556, |
| "step": 9216 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.984068337610385e-05, |
| "loss": 5.0274, |
| "step": 9728 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.983229742859333e-05, |
| "loss": 5.0138, |
| "step": 10240 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.982391148108281e-05, |
| "loss": 4.9861, |
| "step": 10752 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9815541912376026e-05, |
| "loss": 4.9568, |
| "step": 11264 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9807155964865506e-05, |
| "loss": 4.9426, |
| "step": 11776 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9798770017354986e-05, |
| "loss": 4.9321, |
| "step": 12288 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9790384069844466e-05, |
| "loss": 4.8954, |
| "step": 12800 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9782014501137675e-05, |
| "loss": 4.8774, |
| "step": 13312 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9773628553627155e-05, |
| "loss": 4.8643, |
| "step": 13824 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9765242606116635e-05, |
| "loss": 4.8494, |
| "step": 14336 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9756856658606115e-05, |
| "loss": 4.8265, |
| "step": 14848 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9748487089899324e-05, |
| "loss": 4.8097, |
| "step": 15360 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9740101142388804e-05, |
| "loss": 4.807, |
| "step": 15872 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9731715194878284e-05, |
| "loss": 4.7914, |
| "step": 16384 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9723329247367764e-05, |
| "loss": 4.7853, |
| "step": 16896 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9714943299857244e-05, |
| "loss": 4.7692, |
| "step": 17408 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.970657373115046e-05, |
| "loss": 4.751, |
| "step": 17920 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.969818778363994e-05, |
| "loss": 4.7454, |
| "step": 18432 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.968981821493315e-05, |
| "loss": 4.711, |
| "step": 18944 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.968143226742263e-05, |
| "loss": 4.7171, |
| "step": 19456 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.967304631991211e-05, |
| "loss": 4.6882, |
| "step": 19968 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.966466037240159e-05, |
| "loss": 4.69, |
| "step": 20480 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.965627442489107e-05, |
| "loss": 4.6784, |
| "step": 20992 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.964788847738054e-05, |
| "loss": 4.6817, |
| "step": 21504 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.963950252987002e-05, |
| "loss": 4.6593, |
| "step": 22016 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.96311165823595e-05, |
| "loss": 4.6528, |
| "step": 22528 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.962274701365272e-05, |
| "loss": 4.6478, |
| "step": 23040 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.96143610661422e-05, |
| "loss": 4.6296, |
| "step": 23552 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.960597511863168e-05, |
| "loss": 4.6398, |
| "step": 24064 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.959758917112116e-05, |
| "loss": 4.6047, |
| "step": 24576 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.95892359812181e-05, |
| "loss": 4.5971, |
| "step": 25088 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.958085003370758e-05, |
| "loss": 4.61, |
| "step": 25600 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.957246408619706e-05, |
| "loss": 4.6042, |
| "step": 26112 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.956407813868654e-05, |
| "loss": 4.58, |
| "step": 26624 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9555692191176016e-05, |
| "loss": 4.5731, |
| "step": 27136 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.954732262246923e-05, |
| "loss": 4.5749, |
| "step": 27648 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.953893667495871e-05, |
| "loss": 4.5551, |
| "step": 28160 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.953055072744819e-05, |
| "loss": 4.5765, |
| "step": 28672 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.952216477993767e-05, |
| "loss": 4.5389, |
| "step": 29184 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.951377883242715e-05, |
| "loss": 4.5482, |
| "step": 29696 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.950539288491663e-05, |
| "loss": 4.5369, |
| "step": 30208 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.949700693740611e-05, |
| "loss": 4.5224, |
| "step": 30720 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.948862098989559e-05, |
| "loss": 4.5299, |
| "step": 31232 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.94802514211888e-05, |
| "loss": 4.5116, |
| "step": 31744 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.947186547367828e-05, |
| "loss": 4.5095, |
| "step": 32256 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.946347952616776e-05, |
| "loss": 4.4998, |
| "step": 32768 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.945509357865724e-05, |
| "loss": 4.5145, |
| "step": 33280 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.944672400995045e-05, |
| "loss": 4.4858, |
| "step": 33792 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.943833806243993e-05, |
| "loss": 4.4916, |
| "step": 34304 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.942995211492941e-05, |
| "loss": 4.4688, |
| "step": 34816 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.942156616741889e-05, |
| "loss": 4.4769, |
| "step": 35328 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9413196598712105e-05, |
| "loss": 4.4896, |
| "step": 35840 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9404810651201585e-05, |
| "loss": 4.4777, |
| "step": 36352 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9396424703691065e-05, |
| "loss": 4.4595, |
| "step": 36864 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9388038756180545e-05, |
| "loss": 4.4747, |
| "step": 37376 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9379669187473754e-05, |
| "loss": 4.4758, |
| "step": 37888 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9371283239963234e-05, |
| "loss": 4.4496, |
| "step": 38400 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.936291367125644e-05, |
| "loss": 4.4551, |
| "step": 38912 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.935452772374592e-05, |
| "loss": 4.4382, |
| "step": 39424 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.93461417762354e-05, |
| "loss": 4.4273, |
| "step": 39936 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.933775582872488e-05, |
| "loss": 4.4264, |
| "step": 40448 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.932936988121436e-05, |
| "loss": 4.4267, |
| "step": 40960 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.932098393370384e-05, |
| "loss": 4.4328, |
| "step": 41472 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.931259798619332e-05, |
| "loss": 4.4388, |
| "step": 41984 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.930422841748654e-05, |
| "loss": 4.4279, |
| "step": 42496 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.929584246997602e-05, |
| "loss": 4.403, |
| "step": 43008 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.92874565224655e-05, |
| "loss": 4.4092, |
| "step": 43520 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.927907057495498e-05, |
| "loss": 4.4114, |
| "step": 44032 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.927070100624819e-05, |
| "loss": 4.4034, |
| "step": 44544 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.926231505873767e-05, |
| "loss": 4.3951, |
| "step": 45056 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.925394549003088e-05, |
| "loss": 4.3913, |
| "step": 45568 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.924555954252036e-05, |
| "loss": 4.3853, |
| "step": 46080 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.923717359500984e-05, |
| "loss": 4.3936, |
| "step": 46592 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9228804026303046e-05, |
| "loss": 4.3807, |
| "step": 47104 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.922041807879253e-05, |
| "loss": 4.3876, |
| "step": 47616 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.921203213128201e-05, |
| "loss": 4.3756, |
| "step": 48128 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.920364618377149e-05, |
| "loss": 4.3756, |
| "step": 48640 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.919526023626097e-05, |
| "loss": 4.3617, |
| "step": 49152 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.918687428875045e-05, |
| "loss": 4.3639, |
| "step": 49664 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.917848834123993e-05, |
| "loss": 4.3601, |
| "step": 50176 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.917010239372941e-05, |
| "loss": 4.3655, |
| "step": 50688 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.916171644621889e-05, |
| "loss": 4.3654, |
| "step": 51200 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.915333049870837e-05, |
| "loss": 4.343, |
| "step": 51712 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.914494455119785e-05, |
| "loss": 4.3502, |
| "step": 52224 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9136558603687325e-05, |
| "loss": 4.3424, |
| "step": 52736 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.912818903498054e-05, |
| "loss": 4.3447, |
| "step": 53248 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.911980308747002e-05, |
| "loss": 4.3323, |
| "step": 53760 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.911143351876323e-05, |
| "loss": 4.3311, |
| "step": 54272 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.910304757125272e-05, |
| "loss": 4.3359, |
| "step": 54784 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.90946616237422e-05, |
| "loss": 4.3268, |
| "step": 55296 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.908627567623168e-05, |
| "loss": 4.3249, |
| "step": 55808 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.907788972872115e-05, |
| "loss": 4.3173, |
| "step": 56320 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.906950378121063e-05, |
| "loss": 4.3141, |
| "step": 56832 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9061134212503846e-05, |
| "loss": 4.3278, |
| "step": 57344 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9052748264993326e-05, |
| "loss": 4.3155, |
| "step": 57856 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.90443623174828e-05, |
| "loss": 4.3129, |
| "step": 58368 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.903597636997228e-05, |
| "loss": 4.3158, |
| "step": 58880 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9027606801265495e-05, |
| "loss": 4.3224, |
| "step": 59392 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9019220853754975e-05, |
| "loss": 4.3072, |
| "step": 59904 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9010834906244455e-05, |
| "loss": 4.2956, |
| "step": 60416 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9002448958733935e-05, |
| "loss": 4.3028, |
| "step": 60928 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8994063011223415e-05, |
| "loss": 4.3064, |
| "step": 61440 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8985693442516624e-05, |
| "loss": 4.294, |
| "step": 61952 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8977307495006104e-05, |
| "loss": 4.2913, |
| "step": 62464 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8968921547495584e-05, |
| "loss": 4.2907, |
| "step": 62976 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8960535599985064e-05, |
| "loss": 4.2907, |
| "step": 63488 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8952149652474544e-05, |
| "loss": 4.2851, |
| "step": 64000 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8943763704964024e-05, |
| "loss": 4.2884, |
| "step": 64512 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8935377757453504e-05, |
| "loss": 4.2803, |
| "step": 65024 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.892700818874671e-05, |
| "loss": 4.2821, |
| "step": 65536 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.891862224123619e-05, |
| "loss": 4.2742, |
| "step": 66048 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.891023629372567e-05, |
| "loss": 4.2915, |
| "step": 66560 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.890185034621515e-05, |
| "loss": 4.2755, |
| "step": 67072 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.889346439870464e-05, |
| "loss": 4.2678, |
| "step": 67584 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.888509482999785e-05, |
| "loss": 4.2627, |
| "step": 68096 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.887670888248733e-05, |
| "loss": 4.2739, |
| "step": 68608 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.886832293497681e-05, |
| "loss": 4.2682, |
| "step": 69120 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.885993698746629e-05, |
| "loss": 4.2735, |
| "step": 69632 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.88515674187595e-05, |
| "loss": 4.2606, |
| "step": 70144 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.884318147124898e-05, |
| "loss": 4.2664, |
| "step": 70656 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.883479552373846e-05, |
| "loss": 4.2599, |
| "step": 71168 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.882640957622794e-05, |
| "loss": 4.2602, |
| "step": 71680 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.881802362871742e-05, |
| "loss": 4.2496, |
| "step": 72192 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.88096376812069e-05, |
| "loss": 4.2601, |
| "step": 72704 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8801268112500106e-05, |
| "loss": 4.2458, |
| "step": 73216 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.879289854379332e-05, |
| "loss": 4.2367, |
| "step": 73728 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.87845125962828e-05, |
| "loss": 4.2577, |
| "step": 74240 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.877612664877228e-05, |
| "loss": 4.2432, |
| "step": 74752 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.876774070126176e-05, |
| "loss": 4.2427, |
| "step": 75264 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.875935475375124e-05, |
| "loss": 4.2483, |
| "step": 75776 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.875096880624072e-05, |
| "loss": 4.2314, |
| "step": 76288 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 4.204476833343506, |
| "eval_runtime": 303.8587, |
| "eval_samples_per_second": 1255.817, |
| "eval_steps_per_second": 39.245, |
| "step": 76320 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.87425828587302e-05, |
| "loss": 4.2212, |
| "step": 76800 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.873419691121968e-05, |
| "loss": 4.2257, |
| "step": 77312 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.872581096370916e-05, |
| "loss": 4.2327, |
| "step": 77824 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.871744139500237e-05, |
| "loss": 4.2252, |
| "step": 78336 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.870905544749185e-05, |
| "loss": 4.2341, |
| "step": 78848 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.870066949998133e-05, |
| "loss": 4.2187, |
| "step": 79360 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.869228355247081e-05, |
| "loss": 4.221, |
| "step": 79872 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.868389760496029e-05, |
| "loss": 4.2116, |
| "step": 80384 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8675528036253507e-05, |
| "loss": 4.2237, |
| "step": 80896 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8667142088742986e-05, |
| "loss": 4.206, |
| "step": 81408 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.865875614123246e-05, |
| "loss": 4.2066, |
| "step": 81920 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.865037019372194e-05, |
| "loss": 4.2093, |
| "step": 82432 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8642000625015155e-05, |
| "loss": 4.2057, |
| "step": 82944 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8633614677504635e-05, |
| "loss": 4.1971, |
| "step": 83456 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.862522872999411e-05, |
| "loss": 4.1982, |
| "step": 83968 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.861684278248359e-05, |
| "loss": 4.183, |
| "step": 84480 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.860845683497307e-05, |
| "loss": 4.1934, |
| "step": 84992 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.860007088746255e-05, |
| "loss": 4.186, |
| "step": 85504 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.859168493995203e-05, |
| "loss": 4.1796, |
| "step": 86016 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.858329899244151e-05, |
| "loss": 4.2064, |
| "step": 86528 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8574929423734724e-05, |
| "loss": 4.1841, |
| "step": 87040 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8566543476224204e-05, |
| "loss": 4.1877, |
| "step": 87552 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8558157528713684e-05, |
| "loss": 4.1838, |
| "step": 88064 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8549771581203164e-05, |
| "loss": 4.1972, |
| "step": 88576 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.854140201249637e-05, |
| "loss": 4.1709, |
| "step": 89088 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.853301606498585e-05, |
| "loss": 4.1732, |
| "step": 89600 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.852463011747533e-05, |
| "loss": 4.1776, |
| "step": 90112 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.851624416996481e-05, |
| "loss": 4.1772, |
| "step": 90624 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.850787460125802e-05, |
| "loss": 4.1682, |
| "step": 91136 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.84994886537475e-05, |
| "loss": 4.1651, |
| "step": 91648 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.849110270623698e-05, |
| "loss": 4.1736, |
| "step": 92160 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.848271675872646e-05, |
| "loss": 4.1744, |
| "step": 92672 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.847434719001968e-05, |
| "loss": 4.1755, |
| "step": 93184 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.846596124250916e-05, |
| "loss": 4.1656, |
| "step": 93696 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.845757529499864e-05, |
| "loss": 4.1605, |
| "step": 94208 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.844918934748812e-05, |
| "loss": 4.1713, |
| "step": 94720 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.844081977878133e-05, |
| "loss": 4.1418, |
| "step": 95232 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.843243383127081e-05, |
| "loss": 4.1626, |
| "step": 95744 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.842404788376029e-05, |
| "loss": 4.141, |
| "step": 96256 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.841566193624977e-05, |
| "loss": 4.1523, |
| "step": 96768 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8407292367542976e-05, |
| "loss": 4.147, |
| "step": 97280 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8398906420032456e-05, |
| "loss": 4.1567, |
| "step": 97792 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8390520472521936e-05, |
| "loss": 4.1495, |
| "step": 98304 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8382134525011416e-05, |
| "loss": 4.1426, |
| "step": 98816 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.837376495630463e-05, |
| "loss": 4.1497, |
| "step": 99328 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.836537900879411e-05, |
| "loss": 4.1407, |
| "step": 99840 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.835699306128359e-05, |
| "loss": 4.1534, |
| "step": 100352 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.834860711377307e-05, |
| "loss": 4.126, |
| "step": 100864 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.834023754506628e-05, |
| "loss": 4.1249, |
| "step": 101376 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.833185159755576e-05, |
| "loss": 4.1454, |
| "step": 101888 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.832346565004524e-05, |
| "loss": 4.1448, |
| "step": 102400 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.831507970253472e-05, |
| "loss": 4.1293, |
| "step": 102912 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.830671013382793e-05, |
| "loss": 4.1231, |
| "step": 103424 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.829832418631741e-05, |
| "loss": 4.1329, |
| "step": 103936 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.828993823880689e-05, |
| "loss": 4.1164, |
| "step": 104448 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.828155229129637e-05, |
| "loss": 4.1385, |
| "step": 104960 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8273182722589585e-05, |
| "loss": 4.1206, |
| "step": 105472 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8264796775079065e-05, |
| "loss": 4.1287, |
| "step": 105984 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8256410827568545e-05, |
| "loss": 4.1224, |
| "step": 106496 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8248024880058025e-05, |
| "loss": 4.1122, |
| "step": 107008 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8239655311351234e-05, |
| "loss": 4.1248, |
| "step": 107520 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8231269363840714e-05, |
| "loss": 4.112, |
| "step": 108032 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8222883416330194e-05, |
| "loss": 4.1103, |
| "step": 108544 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8214497468819674e-05, |
| "loss": 4.1075, |
| "step": 109056 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.820612790011288e-05, |
| "loss": 4.1269, |
| "step": 109568 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.819774195260236e-05, |
| "loss": 4.1075, |
| "step": 110080 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.818935600509184e-05, |
| "loss": 4.1093, |
| "step": 110592 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.818097005758132e-05, |
| "loss": 4.0987, |
| "step": 111104 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.817260048887454e-05, |
| "loss": 4.1054, |
| "step": 111616 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.816421454136402e-05, |
| "loss": 4.1221, |
| "step": 112128 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.81558285938535e-05, |
| "loss": 4.1158, |
| "step": 112640 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.814744264634298e-05, |
| "loss": 4.1035, |
| "step": 113152 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.813907307763619e-05, |
| "loss": 4.1145, |
| "step": 113664 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.813068713012567e-05, |
| "loss": 4.1172, |
| "step": 114176 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.812230118261515e-05, |
| "loss": 4.1019, |
| "step": 114688 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.811391523510463e-05, |
| "loss": 4.1083, |
| "step": 115200 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.810554566639784e-05, |
| "loss": 4.0993, |
| "step": 115712 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.809715971888732e-05, |
| "loss": 4.0869, |
| "step": 116224 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.80887737713768e-05, |
| "loss": 4.0918, |
| "step": 116736 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.808038782386628e-05, |
| "loss": 4.0947, |
| "step": 117248 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.807201825515949e-05, |
| "loss": 4.0994, |
| "step": 117760 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.806363230764897e-05, |
| "loss": 4.1152, |
| "step": 118272 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.805524636013845e-05, |
| "loss": 4.1032, |
| "step": 118784 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.804686041262793e-05, |
| "loss": 4.0796, |
| "step": 119296 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.803849084392114e-05, |
| "loss": 4.0897, |
| "step": 119808 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.803010489641062e-05, |
| "loss": 4.0943, |
| "step": 120320 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.80217189489001e-05, |
| "loss": 4.0909, |
| "step": 120832 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.801333300138958e-05, |
| "loss": 4.083, |
| "step": 121344 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.800496343268279e-05, |
| "loss": 4.0829, |
| "step": 121856 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.799657748517227e-05, |
| "loss": 4.0818, |
| "step": 122368 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.798819153766175e-05, |
| "loss": 4.0869, |
| "step": 122880 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.797980559015123e-05, |
| "loss": 4.0793, |
| "step": 123392 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7971436021444446e-05, |
| "loss": 4.0872, |
| "step": 123904 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7963050073933926e-05, |
| "loss": 4.0801, |
| "step": 124416 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7954680505227135e-05, |
| "loss": 4.0838, |
| "step": 124928 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7946294557716615e-05, |
| "loss": 4.0685, |
| "step": 125440 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7937908610206095e-05, |
| "loss": 4.0785, |
| "step": 125952 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7929522662695575e-05, |
| "loss": 4.0744, |
| "step": 126464 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7921136715185055e-05, |
| "loss": 4.0788, |
| "step": 126976 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7912750767674535e-05, |
| "loss": 4.0799, |
| "step": 127488 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.790436482016401e-05, |
| "loss": 4.0678, |
| "step": 128000 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7895995251457224e-05, |
| "loss": 4.0691, |
| "step": 128512 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7887609303946704e-05, |
| "loss": 4.0696, |
| "step": 129024 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7879223356436184e-05, |
| "loss": 4.0719, |
| "step": 129536 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7870837408925664e-05, |
| "loss": 4.0591, |
| "step": 130048 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.786246784021888e-05, |
| "loss": 4.0633, |
| "step": 130560 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.785408189270836e-05, |
| "loss": 4.0718, |
| "step": 131072 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.784569594519783e-05, |
| "loss": 4.0595, |
| "step": 131584 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.783730999768731e-05, |
| "loss": 4.0619, |
| "step": 132096 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.782892405017679e-05, |
| "loss": 4.0545, |
| "step": 132608 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.782053810266627e-05, |
| "loss": 4.0518, |
| "step": 133120 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.781215215515575e-05, |
| "loss": 4.0729, |
| "step": 133632 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.780376620764523e-05, |
| "loss": 4.0541, |
| "step": 134144 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.779539663893844e-05, |
| "loss": 4.0627, |
| "step": 134656 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.778701069142792e-05, |
| "loss": 4.0607, |
| "step": 135168 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.77786247439174e-05, |
| "loss": 4.0739, |
| "step": 135680 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.777023879640689e-05, |
| "loss": 4.0594, |
| "step": 136192 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.77618692277001e-05, |
| "loss": 4.0472, |
| "step": 136704 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.775348328018958e-05, |
| "loss": 4.0557, |
| "step": 137216 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.774509733267906e-05, |
| "loss": 4.0595, |
| "step": 137728 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.773672776397227e-05, |
| "loss": 4.0458, |
| "step": 138240 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7728341816461747e-05, |
| "loss": 4.0525, |
| "step": 138752 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7719955868951227e-05, |
| "loss": 4.0455, |
| "step": 139264 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7711569921440707e-05, |
| "loss": 4.0572, |
| "step": 139776 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7703183973930186e-05, |
| "loss": 4.0444, |
| "step": 140288 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7694798026419666e-05, |
| "loss": 4.0539, |
| "step": 140800 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7686412078909146e-05, |
| "loss": 4.0477, |
| "step": 141312 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7678026131398626e-05, |
| "loss": 4.0488, |
| "step": 141824 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.766965656269184e-05, |
| "loss": 4.0405, |
| "step": 142336 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.766128699398505e-05, |
| "loss": 4.0625, |
| "step": 142848 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.765290104647453e-05, |
| "loss": 4.0516, |
| "step": 143360 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.764451509896401e-05, |
| "loss": 4.036, |
| "step": 143872 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.763612915145349e-05, |
| "loss": 4.0382, |
| "step": 144384 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.762774320394297e-05, |
| "loss": 4.0458, |
| "step": 144896 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.761935725643245e-05, |
| "loss": 4.0474, |
| "step": 145408 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.761098768772566e-05, |
| "loss": 4.0546, |
| "step": 145920 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.760260174021514e-05, |
| "loss": 4.0373, |
| "step": 146432 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.759421579270462e-05, |
| "loss": 4.0456, |
| "step": 146944 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.75858298451941e-05, |
| "loss": 4.0414, |
| "step": 147456 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.757744389768358e-05, |
| "loss": 4.0463, |
| "step": 147968 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.756905795017306e-05, |
| "loss": 4.0364, |
| "step": 148480 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.756067200266254e-05, |
| "loss": 4.0407, |
| "step": 148992 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.755228605515202e-05, |
| "loss": 4.0346, |
| "step": 149504 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7543916486445236e-05, |
| "loss": 4.0272, |
| "step": 150016 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7535546917738445e-05, |
| "loss": 4.0477, |
| "step": 150528 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7527160970227925e-05, |
| "loss": 4.0385, |
| "step": 151040 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7518775022717405e-05, |
| "loss": 4.0327, |
| "step": 151552 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7510389075206885e-05, |
| "loss": 4.0355, |
| "step": 152064 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7502003127696365e-05, |
| "loss": 4.0255, |
| "step": 152576 |
| }, |
| { |
| "epoch": 1.03, |
| "eval_loss": 4.034289360046387, |
| "eval_runtime": 302.2571, |
| "eval_samples_per_second": 1262.471, |
| "eval_steps_per_second": 39.453, |
| "step": 152640 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7493617180185845e-05, |
| "loss": 4.0159, |
| "step": 153088 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.748523123267532e-05, |
| "loss": 4.0195, |
| "step": 153600 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.74768452851648e-05, |
| "loss": 4.0332, |
| "step": 154112 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.746845933765428e-05, |
| "loss": 4.0231, |
| "step": 154624 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7460073390143764e-05, |
| "loss": 4.038, |
| "step": 155136 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7451687442633244e-05, |
| "loss": 4.0201, |
| "step": 155648 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7443301495122724e-05, |
| "loss": 4.0222, |
| "step": 156160 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7434915547612204e-05, |
| "loss": 4.0153, |
| "step": 156672 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7426529600101684e-05, |
| "loss": 4.0312, |
| "step": 157184 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7418143652591164e-05, |
| "loss": 4.0147, |
| "step": 157696 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7409757705080644e-05, |
| "loss": 4.0134, |
| "step": 158208 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.740138813637385e-05, |
| "loss": 4.0155, |
| "step": 158720 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.739301856766706e-05, |
| "loss": 4.0164, |
| "step": 159232 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.738463262015654e-05, |
| "loss": 4.0082, |
| "step": 159744 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.737624667264602e-05, |
| "loss": 4.0111, |
| "step": 160256 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.73678607251355e-05, |
| "loss": 4.001, |
| "step": 160768 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.735947477762498e-05, |
| "loss": 4.0035, |
| "step": 161280 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.735108883011446e-05, |
| "loss": 4.0001, |
| "step": 161792 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.734271926140768e-05, |
| "loss": 3.9961, |
| "step": 162304 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.733433331389716e-05, |
| "loss": 4.0284, |
| "step": 162816 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.732594736638664e-05, |
| "loss": 4.0048, |
| "step": 163328 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.731756141887612e-05, |
| "loss": 4.0019, |
| "step": 163840 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.73091754713656e-05, |
| "loss": 4.0081, |
| "step": 164352 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.730078952385508e-05, |
| "loss": 4.0146, |
| "step": 164864 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.729241995514829e-05, |
| "loss": 3.9942, |
| "step": 165376 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.728403400763777e-05, |
| "loss": 3.9988, |
| "step": 165888 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.727564806012725e-05, |
| "loss": 4.0004, |
| "step": 166400 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.726726211261673e-05, |
| "loss": 4.0044, |
| "step": 166912 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.725887616510621e-05, |
| "loss": 3.9901, |
| "step": 167424 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7250506596399416e-05, |
| "loss": 3.9934, |
| "step": 167936 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7242120648888896e-05, |
| "loss": 3.9987, |
| "step": 168448 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.723373470137838e-05, |
| "loss": 4.0037, |
| "step": 168960 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7225348753867856e-05, |
| "loss": 4.0104, |
| "step": 169472 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7216962806357336e-05, |
| "loss": 3.9967, |
| "step": 169984 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7208576858846816e-05, |
| "loss": 3.9905, |
| "step": 170496 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.720020729014003e-05, |
| "loss": 4.0012, |
| "step": 171008 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7191821342629505e-05, |
| "loss": 3.9819, |
| "step": 171520 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7183435395118985e-05, |
| "loss": 3.9959, |
| "step": 172032 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7175049447608465e-05, |
| "loss": 3.9765, |
| "step": 172544 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.716667987890168e-05, |
| "loss": 3.9847, |
| "step": 173056 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7158293931391154e-05, |
| "loss": 3.9867, |
| "step": 173568 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7149907983880634e-05, |
| "loss": 3.997, |
| "step": 174080 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.714152203637012e-05, |
| "loss": 3.9897, |
| "step": 174592 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.71331360888596e-05, |
| "loss": 3.982, |
| "step": 175104 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.712475014134908e-05, |
| "loss": 3.9881, |
| "step": 175616 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.711636419383856e-05, |
| "loss": 3.9836, |
| "step": 176128 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.710797824632804e-05, |
| "loss": 3.9916, |
| "step": 176640 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.709960867762125e-05, |
| "loss": 3.9708, |
| "step": 177152 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.709123910891446e-05, |
| "loss": 3.9669, |
| "step": 177664 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.708285316140394e-05, |
| "loss": 3.9893, |
| "step": 178176 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.707446721389342e-05, |
| "loss": 3.9856, |
| "step": 178688 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.70660812663829e-05, |
| "loss": 3.9745, |
| "step": 179200 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.705771169767611e-05, |
| "loss": 3.9731, |
| "step": 179712 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.704932575016559e-05, |
| "loss": 3.9775, |
| "step": 180224 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7040939802655074e-05, |
| "loss": 3.9638, |
| "step": 180736 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7032553855144554e-05, |
| "loss": 3.9848, |
| "step": 181248 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7024167907634034e-05, |
| "loss": 3.9715, |
| "step": 181760 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.701579833892724e-05, |
| "loss": 3.9788, |
| "step": 182272 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.700741239141672e-05, |
| "loss": 3.9709, |
| "step": 182784 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.69990264439062e-05, |
| "loss": 3.9702, |
| "step": 183296 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.699064049639568e-05, |
| "loss": 3.9697, |
| "step": 183808 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.698227092768889e-05, |
| "loss": 3.9684, |
| "step": 184320 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.697388498017837e-05, |
| "loss": 3.9633, |
| "step": 184832 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.696549903266785e-05, |
| "loss": 3.9616, |
| "step": 185344 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.695711308515733e-05, |
| "loss": 3.9781, |
| "step": 185856 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.694872713764681e-05, |
| "loss": 3.9649, |
| "step": 186368 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.694034119013629e-05, |
| "loss": 3.965, |
| "step": 186880 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.693197162142951e-05, |
| "loss": 3.9546, |
| "step": 187392 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.692358567391899e-05, |
| "loss": 3.9618, |
| "step": 187904 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.691519972640847e-05, |
| "loss": 3.9789, |
| "step": 188416 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.690681377889795e-05, |
| "loss": 3.973, |
| "step": 188928 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6898444210191157e-05, |
| "loss": 3.9643, |
| "step": 189440 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6890058262680636e-05, |
| "loss": 3.9726, |
| "step": 189952 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6881672315170116e-05, |
| "loss": 3.9763, |
| "step": 190464 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6873286367659596e-05, |
| "loss": 3.9586, |
| "step": 190976 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6864900420149076e-05, |
| "loss": 3.968, |
| "step": 191488 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6856530851442285e-05, |
| "loss": 3.9611, |
| "step": 192000 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6848144903931765e-05, |
| "loss": 3.9505, |
| "step": 192512 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6839758956421245e-05, |
| "loss": 3.9606, |
| "step": 193024 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6831373008910725e-05, |
| "loss": 3.9528, |
| "step": 193536 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.682300344020394e-05, |
| "loss": 3.9629, |
| "step": 194048 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.681461749269342e-05, |
| "loss": 3.9716, |
| "step": 194560 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.68062315451829e-05, |
| "loss": 3.9699, |
| "step": 195072 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.679784559767238e-05, |
| "loss": 3.9414, |
| "step": 195584 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.678945965016186e-05, |
| "loss": 3.9591, |
| "step": 196096 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.678107370265134e-05, |
| "loss": 3.958, |
| "step": 196608 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6772687755140814e-05, |
| "loss": 3.9588, |
| "step": 197120 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.676431818643403e-05, |
| "loss": 3.9527, |
| "step": 197632 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.675593223892351e-05, |
| "loss": 3.9467, |
| "step": 198144 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.674754629141298e-05, |
| "loss": 3.9566, |
| "step": 198656 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.673916034390246e-05, |
| "loss": 3.9519, |
| "step": 199168 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.673079077519568e-05, |
| "loss": 3.9466, |
| "step": 199680 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.672240482768516e-05, |
| "loss": 3.955, |
| "step": 200192 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.671401888017464e-05, |
| "loss": 3.9486, |
| "step": 200704 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.670563293266412e-05, |
| "loss": 3.9575, |
| "step": 201216 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.66972469851536e-05, |
| "loss": 3.9427, |
| "step": 201728 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6688877416446815e-05, |
| "loss": 3.9452, |
| "step": 202240 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.668049146893629e-05, |
| "loss": 3.9467, |
| "step": 202752 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.667210552142577e-05, |
| "loss": 3.9559, |
| "step": 203264 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.666371957391525e-05, |
| "loss": 3.9532, |
| "step": 203776 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.665535000520846e-05, |
| "loss": 3.9429, |
| "step": 204288 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.664696405769794e-05, |
| "loss": 3.944, |
| "step": 204800 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.663857811018742e-05, |
| "loss": 3.9437, |
| "step": 205312 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.66301921626769e-05, |
| "loss": 3.9486, |
| "step": 205824 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6621806215166383e-05, |
| "loss": 3.9315, |
| "step": 206336 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.661343664645959e-05, |
| "loss": 3.9381, |
| "step": 206848 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.660505069894907e-05, |
| "loss": 3.9501, |
| "step": 207360 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.659666475143855e-05, |
| "loss": 3.9409, |
| "step": 207872 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.658827880392803e-05, |
| "loss": 3.9339, |
| "step": 208384 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.657989285641751e-05, |
| "loss": 3.9323, |
| "step": 208896 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.657152328771072e-05, |
| "loss": 3.9304, |
| "step": 209408 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.65631373402002e-05, |
| "loss": 3.9481, |
| "step": 209920 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.655475139268968e-05, |
| "loss": 3.9382, |
| "step": 210432 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.654636544517916e-05, |
| "loss": 3.9366, |
| "step": 210944 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.653797949766864e-05, |
| "loss": 3.9412, |
| "step": 211456 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.652959355015812e-05, |
| "loss": 3.9547, |
| "step": 211968 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.652122398145134e-05, |
| "loss": 3.9403, |
| "step": 212480 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.651283803394082e-05, |
| "loss": 3.9285, |
| "step": 212992 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.65044520864303e-05, |
| "loss": 3.9372, |
| "step": 213504 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.649606613891978e-05, |
| "loss": 3.9389, |
| "step": 214016 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6487696570212986e-05, |
| "loss": 3.9307, |
| "step": 214528 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6479310622702466e-05, |
| "loss": 3.9329, |
| "step": 215040 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6470924675191946e-05, |
| "loss": 3.9282, |
| "step": 215552 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6462538727681426e-05, |
| "loss": 3.9399, |
| "step": 216064 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6454169158974635e-05, |
| "loss": 3.9265, |
| "step": 216576 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6445783211464115e-05, |
| "loss": 3.9433, |
| "step": 217088 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6437397263953595e-05, |
| "loss": 3.9257, |
| "step": 217600 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6429011316443075e-05, |
| "loss": 3.9357, |
| "step": 218112 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.642064174773629e-05, |
| "loss": 3.9242, |
| "step": 218624 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.641225580022577e-05, |
| "loss": 3.9464, |
| "step": 219136 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.640386985271525e-05, |
| "loss": 3.9383, |
| "step": 219648 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.639548390520473e-05, |
| "loss": 3.9277, |
| "step": 220160 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.638711433649794e-05, |
| "loss": 3.9208, |
| "step": 220672 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.637872838898742e-05, |
| "loss": 3.9347, |
| "step": 221184 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.63703424414769e-05, |
| "loss": 3.93, |
| "step": 221696 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.636195649396638e-05, |
| "loss": 3.9424, |
| "step": 222208 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.635358692525959e-05, |
| "loss": 3.9229, |
| "step": 222720 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.634520097774907e-05, |
| "loss": 3.9357, |
| "step": 223232 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.633681503023855e-05, |
| "loss": 3.929, |
| "step": 223744 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.632842908272803e-05, |
| "loss": 3.9405, |
| "step": 224256 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.632004313521751e-05, |
| "loss": 3.9211, |
| "step": 224768 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6311673566510724e-05, |
| "loss": 3.9292, |
| "step": 225280 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6303287619000204e-05, |
| "loss": 3.9256, |
| "step": 225792 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6294901671489684e-05, |
| "loss": 3.9186, |
| "step": 226304 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6286532102782893e-05, |
| "loss": 3.9364, |
| "step": 226816 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6278146155272373e-05, |
| "loss": 3.9329, |
| "step": 227328 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.626976020776185e-05, |
| "loss": 3.9196, |
| "step": 227840 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.626137426025133e-05, |
| "loss": 3.928, |
| "step": 228352 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.625298831274081e-05, |
| "loss": 3.9162, |
| "step": 228864 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 3.959151268005371, |
| "eval_runtime": 435.6911, |
| "eval_samples_per_second": 875.829, |
| "eval_steps_per_second": 27.37, |
| "step": 228960 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.624460236523029e-05, |
| "loss": 3.9137, |
| "step": 229376 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6236216417719766e-05, |
| "loss": 3.9121, |
| "step": 229888 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6227830470209246e-05, |
| "loss": 3.9222, |
| "step": 230400 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6219444522698726e-05, |
| "loss": 3.9195, |
| "step": 230912 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.621107495399194e-05, |
| "loss": 3.9353, |
| "step": 231424 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.620268900648142e-05, |
| "loss": 3.9093, |
| "step": 231936 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.61943030589709e-05, |
| "loss": 3.9175, |
| "step": 232448 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.618591711146038e-05, |
| "loss": 3.9113, |
| "step": 232960 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.617754754275359e-05, |
| "loss": 3.9299, |
| "step": 233472 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.616916159524307e-05, |
| "loss": 3.9097, |
| "step": 233984 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.616077564773255e-05, |
| "loss": 3.9068, |
| "step": 234496 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.615238970022203e-05, |
| "loss": 3.9167, |
| "step": 235008 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.614402013151524e-05, |
| "loss": 3.909, |
| "step": 235520 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.613563418400472e-05, |
| "loss": 3.908, |
| "step": 236032 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.61272482364942e-05, |
| "loss": 3.9076, |
| "step": 236544 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.611886228898368e-05, |
| "loss": 3.9012, |
| "step": 237056 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.611047634147316e-05, |
| "loss": 3.8986, |
| "step": 237568 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.610209039396265e-05, |
| "loss": 3.9018, |
| "step": 238080 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.609370444645213e-05, |
| "loss": 3.897, |
| "step": 238592 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6085318498941607e-05, |
| "loss": 3.9256, |
| "step": 239104 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6076948930234816e-05, |
| "loss": 3.9026, |
| "step": 239616 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6068562982724296e-05, |
| "loss": 3.9039, |
| "step": 240128 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6060177035213776e-05, |
| "loss": 3.9078, |
| "step": 240640 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6051791087703256e-05, |
| "loss": 3.9123, |
| "step": 241152 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6043421518996465e-05, |
| "loss": 3.8975, |
| "step": 241664 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6035035571485945e-05, |
| "loss": 3.9021, |
| "step": 242176 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6026649623975425e-05, |
| "loss": 3.8985, |
| "step": 242688 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6018263676464904e-05, |
| "loss": 3.9043, |
| "step": 243200 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6009894107758114e-05, |
| "loss": 3.8924, |
| "step": 243712 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.60015081602476e-05, |
| "loss": 3.8964, |
| "step": 244224 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.599312221273708e-05, |
| "loss": 3.9025, |
| "step": 244736 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.598473626522656e-05, |
| "loss": 3.9053, |
| "step": 245248 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.597635031771604e-05, |
| "loss": 3.9123, |
| "step": 245760 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.596798074900925e-05, |
| "loss": 3.904, |
| "step": 246272 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.595959480149873e-05, |
| "loss": 3.8932, |
| "step": 246784 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.595120885398821e-05, |
| "loss": 3.9041, |
| "step": 247296 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.594282290647769e-05, |
| "loss": 3.889, |
| "step": 247808 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.59344533377709e-05, |
| "loss": 3.8982, |
| "step": 248320 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.592606739026038e-05, |
| "loss": 3.8835, |
| "step": 248832 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.591768144274986e-05, |
| "loss": 3.8891, |
| "step": 249344 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.590929549523934e-05, |
| "loss": 3.8879, |
| "step": 249856 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5900925926532554e-05, |
| "loss": 3.9053, |
| "step": 250368 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5892539979022034e-05, |
| "loss": 3.8988, |
| "step": 250880 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5884154031511514e-05, |
| "loss": 3.889, |
| "step": 251392 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5875768084000994e-05, |
| "loss": 3.8936, |
| "step": 251904 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5867382136490474e-05, |
| "loss": 3.8908, |
| "step": 252416 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.585901256778368e-05, |
| "loss": 3.898, |
| "step": 252928 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.585062662027316e-05, |
| "loss": 3.8806, |
| "step": 253440 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.584224067276264e-05, |
| "loss": 3.8722, |
| "step": 253952 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.583385472525212e-05, |
| "loss": 3.8963, |
| "step": 254464 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.582548515654533e-05, |
| "loss": 3.894, |
| "step": 254976 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.581709920903481e-05, |
| "loss": 3.8786, |
| "step": 255488 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.580871326152429e-05, |
| "loss": 3.8856, |
| "step": 256000 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.580032731401377e-05, |
| "loss": 3.8902, |
| "step": 256512 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.579195774530699e-05, |
| "loss": 3.8728, |
| "step": 257024 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.578357179779647e-05, |
| "loss": 3.8898, |
| "step": 257536 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.577518585028595e-05, |
| "loss": 3.8791, |
| "step": 258048 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.576679990277543e-05, |
| "loss": 3.8925, |
| "step": 258560 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5758430334068637e-05, |
| "loss": 3.8772, |
| "step": 259072 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5750044386558117e-05, |
| "loss": 3.8886, |
| "step": 259584 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5741658439047596e-05, |
| "loss": 3.8766, |
| "step": 260096 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5733272491537076e-05, |
| "loss": 3.8824, |
| "step": 260608 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5724902922830286e-05, |
| "loss": 3.8702, |
| "step": 261120 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5716516975319765e-05, |
| "loss": 3.8741, |
| "step": 261632 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5708131027809245e-05, |
| "loss": 3.8938, |
| "step": 262144 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5699745080298725e-05, |
| "loss": 3.8736, |
| "step": 262656 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.569137551159194e-05, |
| "loss": 3.8787, |
| "step": 263168 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.568298956408142e-05, |
| "loss": 3.8653, |
| "step": 263680 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.56746036165709e-05, |
| "loss": 3.8772, |
| "step": 264192 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5666217669060374e-05, |
| "loss": 3.889, |
| "step": 264704 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.565784810035359e-05, |
| "loss": 3.8847, |
| "step": 265216 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.564946215284307e-05, |
| "loss": 3.8814, |
| "step": 265728 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.564107620533255e-05, |
| "loss": 3.8862, |
| "step": 266240 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.563269025782202e-05, |
| "loss": 3.8859, |
| "step": 266752 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.562432068911524e-05, |
| "loss": 3.8758, |
| "step": 267264 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.561593474160472e-05, |
| "loss": 3.8811, |
| "step": 267776 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.56075487940942e-05, |
| "loss": 3.8784, |
| "step": 268288 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.559916284658368e-05, |
| "loss": 3.8687, |
| "step": 268800 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5590793277876895e-05, |
| "loss": 3.872, |
| "step": 269312 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5582407330366375e-05, |
| "loss": 3.8701, |
| "step": 269824 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.557402138285585e-05, |
| "loss": 3.8767, |
| "step": 270336 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.556563543534533e-05, |
| "loss": 3.885, |
| "step": 270848 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5557265866638544e-05, |
| "loss": 3.8831, |
| "step": 271360 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5548879919128024e-05, |
| "loss": 3.8573, |
| "step": 271872 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.55404939716175e-05, |
| "loss": 3.8805, |
| "step": 272384 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.553210802410698e-05, |
| "loss": 3.8746, |
| "step": 272896 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.552373845540019e-05, |
| "loss": 3.8753, |
| "step": 273408 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5515352507889666e-05, |
| "loss": 3.8701, |
| "step": 273920 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.550696656037915e-05, |
| "loss": 3.8625, |
| "step": 274432 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.549858061286863e-05, |
| "loss": 3.8735, |
| "step": 274944 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.549021104416185e-05, |
| "loss": 3.867, |
| "step": 275456 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.548182509665132e-05, |
| "loss": 3.8674, |
| "step": 275968 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.54734391491408e-05, |
| "loss": 3.8714, |
| "step": 276480 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.546505320163028e-05, |
| "loss": 3.8683, |
| "step": 276992 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.54566836329235e-05, |
| "loss": 3.877, |
| "step": 277504 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.544829768541297e-05, |
| "loss": 3.8635, |
| "step": 278016 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.543991173790245e-05, |
| "loss": 3.8618, |
| "step": 278528 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.543152579039193e-05, |
| "loss": 3.8645, |
| "step": 279040 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.542315622168514e-05, |
| "loss": 3.8776, |
| "step": 279552 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.541477027417462e-05, |
| "loss": 3.8677, |
| "step": 280064 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5406384326664106e-05, |
| "loss": 3.863, |
| "step": 280576 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5397998379153586e-05, |
| "loss": 3.8606, |
| "step": 281088 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5389628810446796e-05, |
| "loss": 3.8643, |
| "step": 281600 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5381242862936275e-05, |
| "loss": 3.8726, |
| "step": 282112 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5372856915425755e-05, |
| "loss": 3.8572, |
| "step": 282624 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5364470967915235e-05, |
| "loss": 3.8541, |
| "step": 283136 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5356101399208444e-05, |
| "loss": 3.8701, |
| "step": 283648 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5347715451697924e-05, |
| "loss": 3.859, |
| "step": 284160 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5339329504187404e-05, |
| "loss": 3.854, |
| "step": 284672 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5330943556676884e-05, |
| "loss": 3.8521, |
| "step": 285184 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5322573987970093e-05, |
| "loss": 3.8541, |
| "step": 285696 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5314188040459573e-05, |
| "loss": 3.8694, |
| "step": 286208 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.530580209294906e-05, |
| "loss": 3.8549, |
| "step": 286720 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.529741614543854e-05, |
| "loss": 3.8602, |
| "step": 287232 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.528903019792802e-05, |
| "loss": 3.8643, |
| "step": 287744 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.528066062922123e-05, |
| "loss": 3.8741, |
| "step": 288256 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.527227468171071e-05, |
| "loss": 3.8672, |
| "step": 288768 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.526388873420019e-05, |
| "loss": 3.8488, |
| "step": 289280 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.525550278668967e-05, |
| "loss": 3.8604, |
| "step": 289792 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.524711683917915e-05, |
| "loss": 3.8562, |
| "step": 290304 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.523874727047236e-05, |
| "loss": 3.858, |
| "step": 290816 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.523036132296184e-05, |
| "loss": 3.8564, |
| "step": 291328 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.522197537545132e-05, |
| "loss": 3.8498, |
| "step": 291840 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.52135894279408e-05, |
| "loss": 3.8636, |
| "step": 292352 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.520521985923401e-05, |
| "loss": 3.8512, |
| "step": 292864 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5196833911723494e-05, |
| "loss": 3.8624, |
| "step": 293376 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5188447964212974e-05, |
| "loss": 3.8525, |
| "step": 293888 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5180062016702454e-05, |
| "loss": 3.8587, |
| "step": 294400 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.517169244799566e-05, |
| "loss": 3.848, |
| "step": 294912 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.516330650048514e-05, |
| "loss": 3.8738, |
| "step": 295424 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.515492055297462e-05, |
| "loss": 3.8557, |
| "step": 295936 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.51465346054641e-05, |
| "loss": 3.8553, |
| "step": 296448 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.513816503675731e-05, |
| "loss": 3.845, |
| "step": 296960 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.512977908924679e-05, |
| "loss": 3.8613, |
| "step": 297472 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.512139314173627e-05, |
| "loss": 3.8539, |
| "step": 297984 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.511300719422575e-05, |
| "loss": 3.8669, |
| "step": 298496 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.510462124671523e-05, |
| "loss": 3.8497, |
| "step": 299008 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.509625167800845e-05, |
| "loss": 3.8588, |
| "step": 299520 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.508786573049793e-05, |
| "loss": 3.8568, |
| "step": 300032 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.507947978298741e-05, |
| "loss": 3.87, |
| "step": 300544 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.507109383547689e-05, |
| "loss": 3.8449, |
| "step": 301056 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5062724266770096e-05, |
| "loss": 3.8557, |
| "step": 301568 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5054338319259576e-05, |
| "loss": 3.8511, |
| "step": 302080 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5045952371749056e-05, |
| "loss": 3.8481, |
| "step": 302592 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5037582803042265e-05, |
| "loss": 3.8602, |
| "step": 303104 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5029196855531745e-05, |
| "loss": 3.8588, |
| "step": 303616 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5020810908021225e-05, |
| "loss": 3.8489, |
| "step": 304128 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5012424960510705e-05, |
| "loss": 3.8517, |
| "step": 304640 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5004039013000185e-05, |
| "loss": 3.8461, |
| "step": 305152 |
| }, |
| { |
| "epoch": 1.03, |
| "eval_loss": 3.917680501937866, |
| "eval_runtime": 442.5939, |
| "eval_samples_per_second": 862.17, |
| "eval_steps_per_second": 26.943, |
| "step": 305280 |
| } |
| ], |
| "logging_steps": 512, |
| "max_steps": 3052726, |
| "num_train_epochs": 9223372036854775807, |
| "save_steps": 10, |
| "total_flos": 2.10997743962923e+17, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|