| { |
| "best_metric": 3.843653917312622, |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/full2/transformer/1/checkpoints/checkpoint-763200", |
| "epoch": 0.025000606015738065, |
| "eval_steps": 10, |
| "global_step": 763200, |
| "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": 10.9454, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.999161405248948e-05, |
| "loss": 6.8259, |
| "step": 512 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.998322810497896e-05, |
| "loss": 6.1849, |
| "step": 1024 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.997484215746844e-05, |
| "loss": 5.9803, |
| "step": 1536 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.996645620995792e-05, |
| "loss": 5.8167, |
| "step": 2048 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.99580702624474e-05, |
| "loss": 5.7055, |
| "step": 2560 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.994968431493688e-05, |
| "loss": 5.6038, |
| "step": 3072 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.994129836742636e-05, |
| "loss": 5.5334, |
| "step": 3584 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.993291241991584e-05, |
| "loss": 5.4448, |
| "step": 4096 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.992452647240532e-05, |
| "loss": 5.3954, |
| "step": 4608 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.99161405248948e-05, |
| "loss": 5.3443, |
| "step": 5120 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.990775457738428e-05, |
| "loss": 5.3189, |
| "step": 5632 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.989938500867749e-05, |
| "loss": 5.2639, |
| "step": 6144 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.989099906116697e-05, |
| "loss": 5.1992, |
| "step": 6656 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.988261311365645e-05, |
| "loss": 5.1743, |
| "step": 7168 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.987422716614593e-05, |
| "loss": 5.137, |
| "step": 7680 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.986584121863541e-05, |
| "loss": 5.1066, |
| "step": 8192 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.985745527112489e-05, |
| "loss": 5.0804, |
| "step": 8704 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.984906932361437e-05, |
| "loss": 5.0551, |
| "step": 9216 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.984068337610385e-05, |
| "loss": 5.0135, |
| "step": 9728 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.983229742859333e-05, |
| "loss": 5.0066, |
| "step": 10240 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9823927859886547e-05, |
| "loss": 4.9713, |
| "step": 10752 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9815541912376026e-05, |
| "loss": 4.9507, |
| "step": 11264 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9807155964865506e-05, |
| "loss": 4.9237, |
| "step": 11776 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9798770017354986e-05, |
| "loss": 4.9215, |
| "step": 12288 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9790400448648195e-05, |
| "loss": 4.8944, |
| "step": 12800 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9782014501137675e-05, |
| "loss": 4.8692, |
| "step": 13312 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9773628553627155e-05, |
| "loss": 4.8503, |
| "step": 13824 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9765242606116635e-05, |
| "loss": 4.8324, |
| "step": 14336 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9756873037409844e-05, |
| "loss": 4.8222, |
| "step": 14848 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9748487089899324e-05, |
| "loss": 4.7998, |
| "step": 15360 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9740101142388804e-05, |
| "loss": 4.7909, |
| "step": 15872 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9731715194878284e-05, |
| "loss": 4.7796, |
| "step": 16384 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9723329247367764e-05, |
| "loss": 4.7645, |
| "step": 16896 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.971495967866098e-05, |
| "loss": 4.7622, |
| "step": 17408 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.970657373115046e-05, |
| "loss": 4.7476, |
| "step": 17920 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.969820416244367e-05, |
| "loss": 4.733, |
| "step": 18432 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.968981821493315e-05, |
| "loss": 4.7055, |
| "step": 18944 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.968143226742263e-05, |
| "loss": 4.7012, |
| "step": 19456 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.967304631991211e-05, |
| "loss": 4.6844, |
| "step": 19968 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.966466037240159e-05, |
| "loss": 4.6838, |
| "step": 20480 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.965627442489107e-05, |
| "loss": 4.6677, |
| "step": 20992 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.964788847738054e-05, |
| "loss": 4.6617, |
| "step": 21504 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.963950252987002e-05, |
| "loss": 4.6456, |
| "step": 22016 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.96311165823595e-05, |
| "loss": 4.6513, |
| "step": 22528 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.962274701365272e-05, |
| "loss": 4.6349, |
| "step": 23040 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.96143610661422e-05, |
| "loss": 4.6356, |
| "step": 23552 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.960597511863168e-05, |
| "loss": 4.6256, |
| "step": 24064 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.959758917112116e-05, |
| "loss": 4.5973, |
| "step": 24576 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9589219602414374e-05, |
| "loss": 4.5911, |
| "step": 25088 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.958083365490385e-05, |
| "loss": 4.6054, |
| "step": 25600 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.957244770739333e-05, |
| "loss": 4.5893, |
| "step": 26112 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.956406175988281e-05, |
| "loss": 4.5748, |
| "step": 26624 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9555692191176016e-05, |
| "loss": 4.5602, |
| "step": 27136 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9547306243665496e-05, |
| "loss": 4.5638, |
| "step": 27648 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9538920296154976e-05, |
| "loss": 4.5468, |
| "step": 28160 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9530534348644456e-05, |
| "loss": 4.5615, |
| "step": 28672 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.952216477993767e-05, |
| "loss": 4.525, |
| "step": 29184 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.951377883242715e-05, |
| "loss": 4.5379, |
| "step": 29696 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.950539288491663e-05, |
| "loss": 4.5354, |
| "step": 30208 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.949700693740611e-05, |
| "loss": 4.5131, |
| "step": 30720 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.948862098989559e-05, |
| "loss": 4.5289, |
| "step": 31232 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.94802514211888e-05, |
| "loss": 4.5001, |
| "step": 31744 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.947186547367828e-05, |
| "loss": 4.5001, |
| "step": 32256 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.946347952616776e-05, |
| "loss": 4.4881, |
| "step": 32768 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.945510995746097e-05, |
| "loss": 4.5029, |
| "step": 33280 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.944672400995045e-05, |
| "loss": 4.4753, |
| "step": 33792 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.943833806243993e-05, |
| "loss": 4.481, |
| "step": 34304 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.942995211492941e-05, |
| "loss": 4.4663, |
| "step": 34816 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.942156616741889e-05, |
| "loss": 4.4627, |
| "step": 35328 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9413196598712105e-05, |
| "loss": 4.4745, |
| "step": 35840 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.940482703000532e-05, |
| "loss": 4.4668, |
| "step": 36352 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9396441082494794e-05, |
| "loss": 4.4581, |
| "step": 36864 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9388055134984274e-05, |
| "loss": 4.4597, |
| "step": 37376 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9379669187473754e-05, |
| "loss": 4.4643, |
| "step": 37888 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9371283239963234e-05, |
| "loss": 4.4363, |
| "step": 38400 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.936291367125644e-05, |
| "loss": 4.4522, |
| "step": 38912 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.935452772374592e-05, |
| "loss": 4.4266, |
| "step": 39424 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.93461417762354e-05, |
| "loss": 4.4257, |
| "step": 39936 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.933775582872488e-05, |
| "loss": 4.424, |
| "step": 40448 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.932936988121436e-05, |
| "loss": 4.4271, |
| "step": 40960 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.932098393370384e-05, |
| "loss": 4.4194, |
| "step": 41472 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.931259798619332e-05, |
| "loss": 4.4262, |
| "step": 41984 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.930421203868281e-05, |
| "loss": 4.4109, |
| "step": 42496 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.929584246997602e-05, |
| "loss": 4.3903, |
| "step": 43008 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.92874565224655e-05, |
| "loss": 4.3918, |
| "step": 43520 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.927907057495498e-05, |
| "loss": 4.4035, |
| "step": 44032 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.927068462744446e-05, |
| "loss": 4.3888, |
| "step": 44544 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.92623314375414e-05, |
| "loss": 4.3881, |
| "step": 45056 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.925394549003088e-05, |
| "loss": 4.3925, |
| "step": 45568 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.924555954252036e-05, |
| "loss": 4.375, |
| "step": 46080 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.923717359500984e-05, |
| "loss": 4.3815, |
| "step": 46592 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.922878764749932e-05, |
| "loss": 4.3739, |
| "step": 47104 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.92204016999888e-05, |
| "loss": 4.3801, |
| "step": 47616 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.921203213128201e-05, |
| "loss": 4.3768, |
| "step": 48128 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.920364618377149e-05, |
| "loss": 4.3721, |
| "step": 48640 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.919526023626097e-05, |
| "loss": 4.3504, |
| "step": 49152 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.918687428875045e-05, |
| "loss": 4.3571, |
| "step": 49664 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.917848834123993e-05, |
| "loss": 4.3499, |
| "step": 50176 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.917010239372941e-05, |
| "loss": 4.3525, |
| "step": 50688 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.916171644621889e-05, |
| "loss": 4.3597, |
| "step": 51200 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.915333049870837e-05, |
| "loss": 4.3371, |
| "step": 51712 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.914496093000158e-05, |
| "loss": 4.3421, |
| "step": 52224 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.913657498249106e-05, |
| "loss": 4.3442, |
| "step": 52736 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.912818903498054e-05, |
| "loss": 4.3366, |
| "step": 53248 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.911980308747002e-05, |
| "loss": 4.3252, |
| "step": 53760 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.911143351876323e-05, |
| "loss": 4.3289, |
| "step": 54272 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.910304757125272e-05, |
| "loss": 4.3174, |
| "step": 54784 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.90946616237422e-05, |
| "loss": 4.3167, |
| "step": 55296 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9086292055035406e-05, |
| "loss": 4.3155, |
| "step": 55808 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9077906107524886e-05, |
| "loss": 4.3148, |
| "step": 56320 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9069520160014366e-05, |
| "loss": 4.3079, |
| "step": 56832 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9061134212503846e-05, |
| "loss": 4.3144, |
| "step": 57344 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9052748264993326e-05, |
| "loss": 4.3044, |
| "step": 57856 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.90443623174828e-05, |
| "loss": 4.3066, |
| "step": 58368 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.903597636997228e-05, |
| "loss": 4.3138, |
| "step": 58880 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9027606801265495e-05, |
| "loss": 4.31, |
| "step": 59392 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9019220853754975e-05, |
| "loss": 4.3043, |
| "step": 59904 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9010834906244455e-05, |
| "loss": 4.295, |
| "step": 60416 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9002448958733935e-05, |
| "loss": 4.2948, |
| "step": 60928 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8994063011223415e-05, |
| "loss": 4.2928, |
| "step": 61440 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8985677063712895e-05, |
| "loss": 4.2805, |
| "step": 61952 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8977307495006104e-05, |
| "loss": 4.2849, |
| "step": 62464 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8968921547495584e-05, |
| "loss": 4.2801, |
| "step": 62976 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8960535599985064e-05, |
| "loss": 4.284, |
| "step": 63488 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8952149652474544e-05, |
| "loss": 4.2741, |
| "step": 64000 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8943763704964024e-05, |
| "loss": 4.2821, |
| "step": 64512 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8935377757453504e-05, |
| "loss": 4.2762, |
| "step": 65024 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8926991809942984e-05, |
| "loss": 4.2694, |
| "step": 65536 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.891862224123619e-05, |
| "loss": 4.2626, |
| "step": 66048 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.891023629372567e-05, |
| "loss": 4.2799, |
| "step": 66560 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.890185034621515e-05, |
| "loss": 4.2705, |
| "step": 67072 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.889346439870464e-05, |
| "loss": 4.2632, |
| "step": 67584 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.888509482999785e-05, |
| "loss": 4.2629, |
| "step": 68096 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.887672526129106e-05, |
| "loss": 4.2671, |
| "step": 68608 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.886833931378054e-05, |
| "loss": 4.2628, |
| "step": 69120 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.885995336627002e-05, |
| "loss": 4.265, |
| "step": 69632 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.88515674187595e-05, |
| "loss": 4.2548, |
| "step": 70144 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.884318147124898e-05, |
| "loss": 4.2543, |
| "step": 70656 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.883479552373846e-05, |
| "loss": 4.2528, |
| "step": 71168 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.882640957622794e-05, |
| "loss": 4.2439, |
| "step": 71680 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.881802362871742e-05, |
| "loss": 4.2412, |
| "step": 72192 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8809654060010626e-05, |
| "loss": 4.2412, |
| "step": 72704 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8801268112500106e-05, |
| "loss": 4.2439, |
| "step": 73216 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8792882164989586e-05, |
| "loss": 4.2334, |
| "step": 73728 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.878449621747907e-05, |
| "loss": 4.2446, |
| "step": 74240 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.877614302757601e-05, |
| "loss": 4.2349, |
| "step": 74752 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.876775708006549e-05, |
| "loss": 4.2371, |
| "step": 75264 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.875937113255497e-05, |
| "loss": 4.2406, |
| "step": 75776 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.875098518504445e-05, |
| "loss": 4.2215, |
| "step": 76288 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 4.193474292755127, |
| "eval_runtime": 317.2956, |
| "eval_samples_per_second": 1202.636, |
| "eval_steps_per_second": 37.583, |
| "step": 76320 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.874259923753393e-05, |
| "loss": 4.2116, |
| "step": 76800 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.873421329002341e-05, |
| "loss": 4.21, |
| "step": 77312 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.872584372131662e-05, |
| "loss": 4.2335, |
| "step": 77824 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.87174577738061e-05, |
| "loss": 4.2158, |
| "step": 78336 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.870907182629558e-05, |
| "loss": 4.2208, |
| "step": 78848 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.870068587878506e-05, |
| "loss": 4.2133, |
| "step": 79360 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.869229993127454e-05, |
| "loss": 4.2059, |
| "step": 79872 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8683913983764027e-05, |
| "loss": 4.1979, |
| "step": 80384 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8675544415057236e-05, |
| "loss": 4.2071, |
| "step": 80896 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8667158467546716e-05, |
| "loss": 4.1993, |
| "step": 81408 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8658772520036196e-05, |
| "loss": 4.2226, |
| "step": 81920 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8650386572525676e-05, |
| "loss": 4.2066, |
| "step": 82432 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8642017003818885e-05, |
| "loss": 4.1887, |
| "step": 82944 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8633631056308365e-05, |
| "loss": 4.188, |
| "step": 83456 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8625245108797845e-05, |
| "loss": 4.1931, |
| "step": 83968 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8616859161287324e-05, |
| "loss": 4.1794, |
| "step": 84480 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8608473213776804e-05, |
| "loss": 4.1918, |
| "step": 84992 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.860008726626628e-05, |
| "loss": 4.189, |
| "step": 85504 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8591701318755764e-05, |
| "loss": 4.1751, |
| "step": 86016 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8583315371245244e-05, |
| "loss": 4.2017, |
| "step": 86528 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8574929423734724e-05, |
| "loss": 4.1808, |
| "step": 87040 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8566543476224204e-05, |
| "loss": 4.1821, |
| "step": 87552 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.855817390751741e-05, |
| "loss": 4.173, |
| "step": 88064 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.854978796000689e-05, |
| "loss": 4.1918, |
| "step": 88576 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.854140201249637e-05, |
| "loss": 4.1693, |
| "step": 89088 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.853301606498585e-05, |
| "loss": 4.1707, |
| "step": 89600 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.852463011747533e-05, |
| "loss": 4.166, |
| "step": 90112 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.851626054876854e-05, |
| "loss": 4.1657, |
| "step": 90624 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.850787460125802e-05, |
| "loss": 4.1628, |
| "step": 91136 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.84994886537475e-05, |
| "loss": 4.1599, |
| "step": 91648 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.849110270623698e-05, |
| "loss": 4.1631, |
| "step": 92160 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.848271675872646e-05, |
| "loss": 4.1636, |
| "step": 92672 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.847433081121595e-05, |
| "loss": 4.1601, |
| "step": 93184 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.846594486370543e-05, |
| "loss": 4.1662, |
| "step": 93696 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.845757529499864e-05, |
| "loss": 4.1608, |
| "step": 94208 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.844918934748812e-05, |
| "loss": 4.1597, |
| "step": 94720 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.84408033999776e-05, |
| "loss": 4.145, |
| "step": 95232 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.843241745246708e-05, |
| "loss": 4.1474, |
| "step": 95744 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.842403150495656e-05, |
| "loss": 4.1394, |
| "step": 96256 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.841564555744604e-05, |
| "loss": 4.1485, |
| "step": 96768 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.840727598873925e-05, |
| "loss": 4.1392, |
| "step": 97280 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.839889004122873e-05, |
| "loss": 4.1485, |
| "step": 97792 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.839050409371821e-05, |
| "loss": 4.1294, |
| "step": 98304 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8382118146207687e-05, |
| "loss": 4.1491, |
| "step": 98816 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8373732198697167e-05, |
| "loss": 4.1411, |
| "step": 99328 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.836536262999038e-05, |
| "loss": 4.1424, |
| "step": 99840 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.835697668247986e-05, |
| "loss": 4.1374, |
| "step": 100352 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.834859073496934e-05, |
| "loss": 4.1224, |
| "step": 100864 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.834022116626255e-05, |
| "loss": 4.1235, |
| "step": 101376 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.833183521875203e-05, |
| "loss": 4.1393, |
| "step": 101888 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.832344927124151e-05, |
| "loss": 4.1297, |
| "step": 102400 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.831506332373099e-05, |
| "loss": 4.1247, |
| "step": 102912 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.830667737622047e-05, |
| "loss": 4.1158, |
| "step": 103424 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8298291428709944e-05, |
| "loss": 4.1274, |
| "step": 103936 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8289905481199424e-05, |
| "loss": 4.1072, |
| "step": 104448 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8281519533688904e-05, |
| "loss": 4.1334, |
| "step": 104960 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8273133586178384e-05, |
| "loss": 4.103, |
| "step": 105472 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.82647640174716e-05, |
| "loss": 4.1229, |
| "step": 105984 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.825637806996108e-05, |
| "loss": 4.1192, |
| "step": 106496 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.824799212245056e-05, |
| "loss": 4.1066, |
| "step": 107008 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.823960617494004e-05, |
| "loss": 4.1188, |
| "step": 107520 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.823123660623325e-05, |
| "loss": 4.1073, |
| "step": 108032 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.822285065872273e-05, |
| "loss": 4.1064, |
| "step": 108544 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.821446471121221e-05, |
| "loss": 4.0946, |
| "step": 109056 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.820607876370169e-05, |
| "loss": 4.1185, |
| "step": 109568 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.81977091949949e-05, |
| "loss": 4.0927, |
| "step": 110080 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.818932324748438e-05, |
| "loss": 4.1058, |
| "step": 110592 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.818093729997386e-05, |
| "loss": 4.0837, |
| "step": 111104 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.817255135246334e-05, |
| "loss": 4.097, |
| "step": 111616 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8164181783756554e-05, |
| "loss": 4.1068, |
| "step": 112128 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8155795836246034e-05, |
| "loss": 4.1047, |
| "step": 112640 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8147409888735514e-05, |
| "loss": 4.0996, |
| "step": 113152 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8139023941224994e-05, |
| "loss": 4.1027, |
| "step": 113664 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.81306543725182e-05, |
| "loss": 4.1087, |
| "step": 114176 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.812226842500768e-05, |
| "loss": 4.0931, |
| "step": 114688 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.811388247749716e-05, |
| "loss": 4.1021, |
| "step": 115200 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.810549652998664e-05, |
| "loss": 4.0893, |
| "step": 115712 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.809712696127985e-05, |
| "loss": 4.0892, |
| "step": 116224 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.808874101376933e-05, |
| "loss": 4.0848, |
| "step": 116736 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.808035506625881e-05, |
| "loss": 4.0956, |
| "step": 117248 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.807196911874829e-05, |
| "loss": 4.0944, |
| "step": 117760 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.806358317123777e-05, |
| "loss": 4.0999, |
| "step": 118272 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.805519722372726e-05, |
| "loss": 4.0898, |
| "step": 118784 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.804682765502047e-05, |
| "loss": 4.0693, |
| "step": 119296 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.803844170750995e-05, |
| "loss": 4.0746, |
| "step": 119808 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.803005575999943e-05, |
| "loss": 4.085, |
| "step": 120320 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.802166981248891e-05, |
| "loss": 4.0808, |
| "step": 120832 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8013300243782116e-05, |
| "loss": 4.0758, |
| "step": 121344 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8004914296271596e-05, |
| "loss": 4.0856, |
| "step": 121856 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7996528348761076e-05, |
| "loss": 4.0698, |
| "step": 122368 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7988142401250556e-05, |
| "loss": 4.0733, |
| "step": 122880 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7979772832543765e-05, |
| "loss": 4.0757, |
| "step": 123392 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7971386885033245e-05, |
| "loss": 4.0839, |
| "step": 123904 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7963000937522725e-05, |
| "loss": 4.0795, |
| "step": 124416 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.795461499001221e-05, |
| "loss": 4.0863, |
| "step": 124928 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.794624542130542e-05, |
| "loss": 4.0585, |
| "step": 125440 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.79378594737949e-05, |
| "loss": 4.0702, |
| "step": 125952 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.792947352628438e-05, |
| "loss": 4.0669, |
| "step": 126464 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.792108757877386e-05, |
| "loss": 4.0652, |
| "step": 126976 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.791271801006707e-05, |
| "loss": 4.0818, |
| "step": 127488 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.790433206255655e-05, |
| "loss": 4.0596, |
| "step": 128000 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.789594611504603e-05, |
| "loss": 4.0654, |
| "step": 128512 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.788756016753551e-05, |
| "loss": 4.071, |
| "step": 129024 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.787919059882872e-05, |
| "loss": 4.0669, |
| "step": 129536 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.78708046513182e-05, |
| "loss": 4.0507, |
| "step": 130048 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.786241870380768e-05, |
| "loss": 4.0646, |
| "step": 130560 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7854032756297166e-05, |
| "loss": 4.0553, |
| "step": 131072 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7845663187590375e-05, |
| "loss": 4.0517, |
| "step": 131584 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7837293618883584e-05, |
| "loss": 4.0508, |
| "step": 132096 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7828907671373064e-05, |
| "loss": 4.0524, |
| "step": 132608 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7820521723862544e-05, |
| "loss": 4.0497, |
| "step": 133120 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7812135776352024e-05, |
| "loss": 4.0614, |
| "step": 133632 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7803749828841504e-05, |
| "loss": 4.0462, |
| "step": 134144 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7795363881330984e-05, |
| "loss": 4.0537, |
| "step": 134656 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7786977933820464e-05, |
| "loss": 4.0654, |
| "step": 135168 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7778591986309944e-05, |
| "loss": 4.0583, |
| "step": 135680 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.777022241760315e-05, |
| "loss": 4.0558, |
| "step": 136192 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.776183647009263e-05, |
| "loss": 4.0505, |
| "step": 136704 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.775345052258212e-05, |
| "loss": 4.0455, |
| "step": 137216 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.77450645750716e-05, |
| "loss": 4.0485, |
| "step": 137728 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.773669500636481e-05, |
| "loss": 4.0427, |
| "step": 138240 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.772832543765802e-05, |
| "loss": 4.0414, |
| "step": 138752 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.77199394901475e-05, |
| "loss": 4.0433, |
| "step": 139264 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.771155354263698e-05, |
| "loss": 4.0485, |
| "step": 139776 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.770316759512646e-05, |
| "loss": 4.0439, |
| "step": 140288 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.769478164761594e-05, |
| "loss": 4.0444, |
| "step": 140800 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.768639570010542e-05, |
| "loss": 4.045, |
| "step": 141312 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.76780097525949e-05, |
| "loss": 4.0345, |
| "step": 141824 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.766962380508437e-05, |
| "loss": 4.0331, |
| "step": 142336 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.766123785757386e-05, |
| "loss": 4.0517, |
| "step": 142848 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.765286828886707e-05, |
| "loss": 4.0464, |
| "step": 143360 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.764448234135655e-05, |
| "loss": 4.0413, |
| "step": 143872 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7636096393846026e-05, |
| "loss": 4.0357, |
| "step": 144384 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7627710446335506e-05, |
| "loss": 4.0427, |
| "step": 144896 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7619324498824986e-05, |
| "loss": 4.0382, |
| "step": 145408 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7610954930118195e-05, |
| "loss": 4.0468, |
| "step": 145920 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7602568982607675e-05, |
| "loss": 4.0383, |
| "step": 146432 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7594183035097155e-05, |
| "loss": 4.0336, |
| "step": 146944 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7585797087586635e-05, |
| "loss": 4.0348, |
| "step": 147456 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7577427518879844e-05, |
| "loss": 4.0317, |
| "step": 147968 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7569041571369324e-05, |
| "loss": 4.0245, |
| "step": 148480 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.756065562385881e-05, |
| "loss": 4.0304, |
| "step": 148992 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.755226967634829e-05, |
| "loss": 4.0347, |
| "step": 149504 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.754388372883777e-05, |
| "loss": 4.0212, |
| "step": 150016 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.753551416013098e-05, |
| "loss": 4.0394, |
| "step": 150528 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.752712821262046e-05, |
| "loss": 4.0288, |
| "step": 151040 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.751874226510994e-05, |
| "loss": 4.0277, |
| "step": 151552 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.751035631759942e-05, |
| "loss": 4.0313, |
| "step": 152064 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.750198674889263e-05, |
| "loss": 4.0184, |
| "step": 152576 |
| }, |
| { |
| "epoch": 1.03, |
| "eval_loss": 4.025022029876709, |
| "eval_runtime": 317.0746, |
| "eval_samples_per_second": 1203.474, |
| "eval_steps_per_second": 37.609, |
| "step": 152640 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.749360080138211e-05, |
| "loss": 4.0074, |
| "step": 153088 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.748521485387159e-05, |
| "loss": 4.0086, |
| "step": 153600 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.747682890636107e-05, |
| "loss": 4.0333, |
| "step": 154112 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.746844295885055e-05, |
| "loss": 4.0181, |
| "step": 154624 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.746005701134003e-05, |
| "loss": 4.0278, |
| "step": 155136 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.745167106382951e-05, |
| "loss": 4.0109, |
| "step": 155648 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.744328511631899e-05, |
| "loss": 4.0141, |
| "step": 156160 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7434899168808475e-05, |
| "loss": 4.0027, |
| "step": 156672 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7426529600101684e-05, |
| "loss": 4.0138, |
| "step": 157184 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7418143652591164e-05, |
| "loss": 4.0093, |
| "step": 157696 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7409757705080644e-05, |
| "loss": 4.0274, |
| "step": 158208 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.7401371757570124e-05, |
| "loss": 4.0181, |
| "step": 158720 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.739300218886333e-05, |
| "loss": 3.998, |
| "step": 159232 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.738461624135281e-05, |
| "loss": 4.0035, |
| "step": 159744 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.737623029384229e-05, |
| "loss": 4.0092, |
| "step": 160256 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.736784434633177e-05, |
| "loss": 3.992, |
| "step": 160768 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.735945839882125e-05, |
| "loss": 4.0043, |
| "step": 161280 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.735107245131073e-05, |
| "loss": 4.0069, |
| "step": 161792 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.734268650380021e-05, |
| "loss": 3.9951, |
| "step": 162304 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.733430055628969e-05, |
| "loss": 4.0229, |
| "step": 162816 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.732591460877917e-05, |
| "loss": 3.9992, |
| "step": 163328 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.731752866126865e-05, |
| "loss": 4.0041, |
| "step": 163840 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.730914271375813e-05, |
| "loss": 3.9967, |
| "step": 164352 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.730075676624761e-05, |
| "loss": 4.0149, |
| "step": 164864 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.729238719754082e-05, |
| "loss": 3.989, |
| "step": 165376 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.72840012500303e-05, |
| "loss": 3.9962, |
| "step": 165888 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.727561530251978e-05, |
| "loss": 3.9893, |
| "step": 166400 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.726722935500926e-05, |
| "loss": 3.9946, |
| "step": 166912 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.725885978630247e-05, |
| "loss": 3.9888, |
| "step": 167424 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.725047383879195e-05, |
| "loss": 3.9873, |
| "step": 167936 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.724210427008517e-05, |
| "loss": 3.9952, |
| "step": 168448 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7233718322574647e-05, |
| "loss": 3.996, |
| "step": 168960 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7225332375064127e-05, |
| "loss": 3.988, |
| "step": 169472 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7216946427553606e-05, |
| "loss": 4.0014, |
| "step": 169984 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7208560480043086e-05, |
| "loss": 3.9881, |
| "step": 170496 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7200174532532566e-05, |
| "loss": 3.9944, |
| "step": 171008 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7191788585022046e-05, |
| "loss": 3.9806, |
| "step": 171520 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7183402637511526e-05, |
| "loss": 3.9846, |
| "step": 172032 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7175033068804735e-05, |
| "loss": 3.9719, |
| "step": 172544 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7166647121294215e-05, |
| "loss": 3.9875, |
| "step": 173056 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7158261173783695e-05, |
| "loss": 3.9756, |
| "step": 173568 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7149875226273175e-05, |
| "loss": 3.9883, |
| "step": 174080 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7141505657566384e-05, |
| "loss": 3.9731, |
| "step": 174592 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7133119710055864e-05, |
| "loss": 3.9851, |
| "step": 175104 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.712473376254535e-05, |
| "loss": 3.9894, |
| "step": 175616 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.711634781503483e-05, |
| "loss": 3.9813, |
| "step": 176128 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.710797824632804e-05, |
| "loss": 3.984, |
| "step": 176640 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.709959229881752e-05, |
| "loss": 3.9663, |
| "step": 177152 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.709122273011073e-05, |
| "loss": 3.9671, |
| "step": 177664 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.708283678260021e-05, |
| "loss": 3.9843, |
| "step": 178176 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.707445083508969e-05, |
| "loss": 3.9758, |
| "step": 178688 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.706606488757917e-05, |
| "loss": 3.974, |
| "step": 179200 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.705767894006865e-05, |
| "loss": 3.9616, |
| "step": 179712 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.704929299255813e-05, |
| "loss": 3.9752, |
| "step": 180224 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.704090704504761e-05, |
| "loss": 3.9527, |
| "step": 180736 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.703252109753709e-05, |
| "loss": 3.986, |
| "step": 181248 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7024151528830305e-05, |
| "loss": 3.9538, |
| "step": 181760 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7015765581319785e-05, |
| "loss": 3.9739, |
| "step": 182272 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7007379633809265e-05, |
| "loss": 3.9729, |
| "step": 182784 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6998993686298745e-05, |
| "loss": 3.9587, |
| "step": 183296 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6990624117591954e-05, |
| "loss": 3.9746, |
| "step": 183808 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6982238170081434e-05, |
| "loss": 3.9588, |
| "step": 184320 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6973852222570914e-05, |
| "loss": 3.9598, |
| "step": 184832 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6965466275060394e-05, |
| "loss": 3.9478, |
| "step": 185344 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.69570967063536e-05, |
| "loss": 3.9715, |
| "step": 185856 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.694871075884308e-05, |
| "loss": 3.9525, |
| "step": 186368 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.694032481133256e-05, |
| "loss": 3.9603, |
| "step": 186880 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.693193886382204e-05, |
| "loss": 3.9448, |
| "step": 187392 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.692356929511526e-05, |
| "loss": 3.9528, |
| "step": 187904 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.691518334760474e-05, |
| "loss": 3.9644, |
| "step": 188416 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.690679740009422e-05, |
| "loss": 3.9676, |
| "step": 188928 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.689841145258369e-05, |
| "loss": 3.9551, |
| "step": 189440 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.689004188387691e-05, |
| "loss": 3.9669, |
| "step": 189952 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.688165593636639e-05, |
| "loss": 3.9676, |
| "step": 190464 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.687326998885587e-05, |
| "loss": 3.9573, |
| "step": 190976 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.686488404134534e-05, |
| "loss": 3.958, |
| "step": 191488 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6856514472638556e-05, |
| "loss": 3.9583, |
| "step": 192000 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6848128525128036e-05, |
| "loss": 3.9522, |
| "step": 192512 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.683974257761751e-05, |
| "loss": 3.9499, |
| "step": 193024 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6831356630106996e-05, |
| "loss": 3.959, |
| "step": 193536 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6822987061400205e-05, |
| "loss": 3.9605, |
| "step": 194048 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.681460111388969e-05, |
| "loss": 3.9668, |
| "step": 194560 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6806215166379165e-05, |
| "loss": 3.952, |
| "step": 195072 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6797829218868645e-05, |
| "loss": 3.9326, |
| "step": 195584 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.678945965016186e-05, |
| "loss": 3.9416, |
| "step": 196096 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.678107370265134e-05, |
| "loss": 3.9539, |
| "step": 196608 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6772687755140814e-05, |
| "loss": 3.948, |
| "step": 197120 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6764301807630294e-05, |
| "loss": 3.9429, |
| "step": 197632 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.675593223892351e-05, |
| "loss": 3.9503, |
| "step": 198144 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.674754629141298e-05, |
| "loss": 3.9454, |
| "step": 198656 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.673916034390246e-05, |
| "loss": 3.9373, |
| "step": 199168 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.673077439639194e-05, |
| "loss": 3.945, |
| "step": 199680 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.672240482768516e-05, |
| "loss": 3.9548, |
| "step": 200192 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.671401888017464e-05, |
| "loss": 3.9505, |
| "step": 200704 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.670563293266412e-05, |
| "loss": 3.9554, |
| "step": 201216 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.66972469851536e-05, |
| "loss": 3.9388, |
| "step": 201728 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6688877416446815e-05, |
| "loss": 3.9423, |
| "step": 202240 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.668049146893629e-05, |
| "loss": 3.9405, |
| "step": 202752 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.667210552142577e-05, |
| "loss": 3.9429, |
| "step": 203264 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.666371957391525e-05, |
| "loss": 3.9524, |
| "step": 203776 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.665535000520846e-05, |
| "loss": 3.937, |
| "step": 204288 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.664696405769794e-05, |
| "loss": 3.9386, |
| "step": 204800 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.663857811018742e-05, |
| "loss": 3.9455, |
| "step": 205312 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.66301921626769e-05, |
| "loss": 3.9468, |
| "step": 205824 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.662182259397011e-05, |
| "loss": 3.9272, |
| "step": 206336 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.661343664645959e-05, |
| "loss": 3.9357, |
| "step": 206848 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.660505069894907e-05, |
| "loss": 3.9376, |
| "step": 207360 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.659668113024229e-05, |
| "loss": 3.9292, |
| "step": 207872 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.65883115615355e-05, |
| "loss": 3.9297, |
| "step": 208384 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.657992561402498e-05, |
| "loss": 3.9309, |
| "step": 208896 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.657153966651446e-05, |
| "loss": 3.933, |
| "step": 209408 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.656315371900393e-05, |
| "loss": 3.9375, |
| "step": 209920 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.655476777149341e-05, |
| "loss": 3.9263, |
| "step": 210432 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.654638182398289e-05, |
| "loss": 3.9327, |
| "step": 210944 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.653799587647237e-05, |
| "loss": 3.9452, |
| "step": 211456 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.652960992896185e-05, |
| "loss": 3.9463, |
| "step": 211968 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6521240360255066e-05, |
| "loss": 3.9346, |
| "step": 212480 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6512854412744546e-05, |
| "loss": 3.9348, |
| "step": 212992 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6504468465234026e-05, |
| "loss": 3.9303, |
| "step": 213504 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6496082517723506e-05, |
| "loss": 3.9311, |
| "step": 214016 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6487712949016715e-05, |
| "loss": 3.9221, |
| "step": 214528 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.647934338030993e-05, |
| "loss": 3.9235, |
| "step": 215040 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6470957432799404e-05, |
| "loss": 3.9275, |
| "step": 215552 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6462571485288884e-05, |
| "loss": 3.9355, |
| "step": 216064 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6454185537778364e-05, |
| "loss": 3.9269, |
| "step": 216576 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6445799590267844e-05, |
| "loss": 3.9308, |
| "step": 217088 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6437413642757324e-05, |
| "loss": 3.9288, |
| "step": 217600 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6429027695246804e-05, |
| "loss": 3.9202, |
| "step": 218112 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.642064174773629e-05, |
| "loss": 3.9177, |
| "step": 218624 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.64122721790295e-05, |
| "loss": 3.9377, |
| "step": 219136 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.640388623151898e-05, |
| "loss": 3.9347, |
| "step": 219648 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.639550028400846e-05, |
| "loss": 3.9259, |
| "step": 220160 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.638713071530167e-05, |
| "loss": 3.9231, |
| "step": 220672 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.637874476779115e-05, |
| "loss": 3.9261, |
| "step": 221184 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.637035882028063e-05, |
| "loss": 3.9294, |
| "step": 221696 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.636197287277011e-05, |
| "loss": 3.9395, |
| "step": 222208 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.635360330406332e-05, |
| "loss": 3.9267, |
| "step": 222720 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.63452173565528e-05, |
| "loss": 3.9211, |
| "step": 223232 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.633683140904228e-05, |
| "loss": 3.9246, |
| "step": 223744 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.632844546153176e-05, |
| "loss": 3.9214, |
| "step": 224256 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6320059514021244e-05, |
| "loss": 3.9185, |
| "step": 224768 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6311673566510724e-05, |
| "loss": 3.9207, |
| "step": 225280 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6303287619000204e-05, |
| "loss": 3.925, |
| "step": 225792 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6294901671489684e-05, |
| "loss": 3.9151, |
| "step": 226304 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.628654848158662e-05, |
| "loss": 3.928, |
| "step": 226816 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.62781625340761e-05, |
| "loss": 3.9243, |
| "step": 227328 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.626977658656558e-05, |
| "loss": 3.9191, |
| "step": 227840 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.626139063905506e-05, |
| "loss": 3.9233, |
| "step": 228352 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.625300469154454e-05, |
| "loss": 3.9089, |
| "step": 228864 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 3.9513087272644043, |
| "eval_runtime": 301.973, |
| "eval_samples_per_second": 1263.659, |
| "eval_steps_per_second": 39.49, |
| "step": 228960 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.624461874403402e-05, |
| "loss": 3.9125, |
| "step": 229376 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.62362327965235e-05, |
| "loss": 3.9009, |
| "step": 229888 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.622786322781671e-05, |
| "loss": 3.9276, |
| "step": 230400 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.62194772803062e-05, |
| "loss": 3.9126, |
| "step": 230912 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.621109133279568e-05, |
| "loss": 3.9262, |
| "step": 231424 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.620270538528516e-05, |
| "loss": 3.9052, |
| "step": 231936 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.619431943777464e-05, |
| "loss": 3.9158, |
| "step": 232448 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.618593349026412e-05, |
| "loss": 3.8953, |
| "step": 232960 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.617756392155733e-05, |
| "loss": 3.9142, |
| "step": 233472 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.616917797404681e-05, |
| "loss": 3.9035, |
| "step": 233984 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.616079202653629e-05, |
| "loss": 3.9261, |
| "step": 234496 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.615240607902577e-05, |
| "loss": 3.9134, |
| "step": 235008 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6144036510318976e-05, |
| "loss": 3.8943, |
| "step": 235520 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6135650562808456e-05, |
| "loss": 3.901, |
| "step": 236032 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6127264615297936e-05, |
| "loss": 3.909, |
| "step": 236544 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6118878667787416e-05, |
| "loss": 3.8948, |
| "step": 237056 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6110492720276896e-05, |
| "loss": 3.9025, |
| "step": 237568 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6102106772766376e-05, |
| "loss": 3.9059, |
| "step": 238080 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6093720825255856e-05, |
| "loss": 3.8965, |
| "step": 238592 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6085334877745336e-05, |
| "loss": 3.9278, |
| "step": 239104 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6076948930234816e-05, |
| "loss": 3.8983, |
| "step": 239616 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6068562982724296e-05, |
| "loss": 3.9022, |
| "step": 240128 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6060177035213776e-05, |
| "loss": 3.9017, |
| "step": 240640 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6051807466506985e-05, |
| "loss": 3.9094, |
| "step": 241152 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6043421518996465e-05, |
| "loss": 3.8932, |
| "step": 241664 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6035035571485945e-05, |
| "loss": 3.8994, |
| "step": 242176 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6026649623975425e-05, |
| "loss": 3.8938, |
| "step": 242688 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6018280055268634e-05, |
| "loss": 3.8967, |
| "step": 243200 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6009894107758114e-05, |
| "loss": 3.8913, |
| "step": 243712 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.60015081602476e-05, |
| "loss": 3.8909, |
| "step": 244224 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.599312221273708e-05, |
| "loss": 3.895, |
| "step": 244736 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.598473626522656e-05, |
| "loss": 3.9013, |
| "step": 245248 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.597635031771604e-05, |
| "loss": 3.8996, |
| "step": 245760 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.596796437020552e-05, |
| "loss": 3.9016, |
| "step": 246272 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.595959480149873e-05, |
| "loss": 3.8928, |
| "step": 246784 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.595120885398821e-05, |
| "loss": 3.8981, |
| "step": 247296 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.594282290647769e-05, |
| "loss": 3.8839, |
| "step": 247808 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.593443695896717e-05, |
| "loss": 3.8972, |
| "step": 248320 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.592605101145665e-05, |
| "loss": 3.8781, |
| "step": 248832 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.591768144274986e-05, |
| "loss": 3.8905, |
| "step": 249344 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.590929549523934e-05, |
| "loss": 3.8828, |
| "step": 249856 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.590090954772882e-05, |
| "loss": 3.898, |
| "step": 250368 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.58925236002183e-05, |
| "loss": 3.8818, |
| "step": 250880 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5884137652707785e-05, |
| "loss": 3.8869, |
| "step": 251392 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.587575170519726e-05, |
| "loss": 3.8972, |
| "step": 251904 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.586736575768674e-05, |
| "loss": 3.8911, |
| "step": 252416 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.585897981017622e-05, |
| "loss": 3.8929, |
| "step": 252928 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.585061024146943e-05, |
| "loss": 3.877, |
| "step": 253440 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.584224067276264e-05, |
| "loss": 3.8754, |
| "step": 253952 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.583385472525212e-05, |
| "loss": 3.8892, |
| "step": 254464 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.58254687777416e-05, |
| "loss": 3.8863, |
| "step": 254976 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5817082830231076e-05, |
| "loss": 3.8807, |
| "step": 255488 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5808696882720556e-05, |
| "loss": 3.8751, |
| "step": 256000 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.580032731401377e-05, |
| "loss": 3.8864, |
| "step": 256512 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.579194136650325e-05, |
| "loss": 3.8654, |
| "step": 257024 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.578355541899273e-05, |
| "loss": 3.8862, |
| "step": 257536 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.577516947148221e-05, |
| "loss": 3.8675, |
| "step": 258048 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.576678352397169e-05, |
| "loss": 3.8814, |
| "step": 258560 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.57584139552649e-05, |
| "loss": 3.8863, |
| "step": 259072 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.575002800775438e-05, |
| "loss": 3.8715, |
| "step": 259584 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.574164206024386e-05, |
| "loss": 3.8817, |
| "step": 260096 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.573325611273334e-05, |
| "loss": 3.8735, |
| "step": 260608 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.572487016522282e-05, |
| "loss": 3.8714, |
| "step": 261120 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.571650059651603e-05, |
| "loss": 3.8657, |
| "step": 261632 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.570811464900551e-05, |
| "loss": 3.8805, |
| "step": 262144 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.569972870149499e-05, |
| "loss": 3.8672, |
| "step": 262656 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5691342753984476e-05, |
| "loss": 3.8752, |
| "step": 263168 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5682956806473956e-05, |
| "loss": 3.8592, |
| "step": 263680 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5674587237767165e-05, |
| "loss": 3.8654, |
| "step": 264192 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5666201290256645e-05, |
| "loss": 3.8766, |
| "step": 264704 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5657815342746125e-05, |
| "loss": 3.8832, |
| "step": 265216 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5649429395235605e-05, |
| "loss": 3.869, |
| "step": 265728 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5641043447725085e-05, |
| "loss": 3.883, |
| "step": 266240 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5632657500214565e-05, |
| "loss": 3.8808, |
| "step": 266752 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5624271552704045e-05, |
| "loss": 3.8709, |
| "step": 267264 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5615885605193525e-05, |
| "loss": 3.8756, |
| "step": 267776 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5607516036486734e-05, |
| "loss": 3.8727, |
| "step": 268288 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5599130088976214e-05, |
| "loss": 3.8685, |
| "step": 268800 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5590744141465694e-05, |
| "loss": 3.8669, |
| "step": 269312 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5582358193955174e-05, |
| "loss": 3.8741, |
| "step": 269824 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.557398862524839e-05, |
| "loss": 3.8781, |
| "step": 270336 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.556560267773787e-05, |
| "loss": 3.8743, |
| "step": 270848 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.555721673022735e-05, |
| "loss": 3.8724, |
| "step": 271360 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.554883078271683e-05, |
| "loss": 3.853, |
| "step": 271872 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.554044483520631e-05, |
| "loss": 3.8575, |
| "step": 272384 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.553207526649952e-05, |
| "loss": 3.8682, |
| "step": 272896 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5523689318989e-05, |
| "loss": 3.8678, |
| "step": 273408 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.551530337147848e-05, |
| "loss": 3.8623, |
| "step": 273920 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.550691742396796e-05, |
| "loss": 3.8668, |
| "step": 274432 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.549854785526117e-05, |
| "loss": 3.8631, |
| "step": 274944 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.549016190775065e-05, |
| "loss": 3.8563, |
| "step": 275456 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.548177596024013e-05, |
| "loss": 3.8641, |
| "step": 275968 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.547339001272961e-05, |
| "loss": 3.8712, |
| "step": 276480 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.546500406521909e-05, |
| "loss": 3.8682, |
| "step": 276992 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.545661811770857e-05, |
| "loss": 3.8783, |
| "step": 277504 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.544823217019805e-05, |
| "loss": 3.8612, |
| "step": 278016 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.543984622268753e-05, |
| "loss": 3.859, |
| "step": 278528 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.543149303278447e-05, |
| "loss": 3.8615, |
| "step": 279040 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.542310708527395e-05, |
| "loss": 3.8634, |
| "step": 279552 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.541472113776343e-05, |
| "loss": 3.8735, |
| "step": 280064 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.540633519025291e-05, |
| "loss": 3.8565, |
| "step": 280576 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.539796562154612e-05, |
| "loss": 3.86, |
| "step": 281088 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.53895796740356e-05, |
| "loss": 3.864, |
| "step": 281600 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.538119372652508e-05, |
| "loss": 3.8692, |
| "step": 282112 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.537280777901456e-05, |
| "loss": 3.8491, |
| "step": 282624 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.536442183150404e-05, |
| "loss": 3.8591, |
| "step": 283136 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.535605226279726e-05, |
| "loss": 3.8593, |
| "step": 283648 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.534766631528674e-05, |
| "loss": 3.8511, |
| "step": 284160 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.533928036777621e-05, |
| "loss": 3.8475, |
| "step": 284672 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.533089442026569e-05, |
| "loss": 3.8569, |
| "step": 285184 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.532250847275517e-05, |
| "loss": 3.8564, |
| "step": 285696 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.531412252524465e-05, |
| "loss": 3.8581, |
| "step": 286208 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.530575295653786e-05, |
| "loss": 3.8508, |
| "step": 286720 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.529736700902734e-05, |
| "loss": 3.851, |
| "step": 287232 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.528898106151682e-05, |
| "loss": 3.8702, |
| "step": 287744 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.52805951140063e-05, |
| "loss": 3.8668, |
| "step": 288256 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5272209166495786e-05, |
| "loss": 3.8609, |
| "step": 288768 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5263839597788995e-05, |
| "loss": 3.8579, |
| "step": 289280 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5255453650278475e-05, |
| "loss": 3.855, |
| "step": 289792 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5247067702767955e-05, |
| "loss": 3.8528, |
| "step": 290304 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5238681755257435e-05, |
| "loss": 3.8471, |
| "step": 290816 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5230312186550644e-05, |
| "loss": 3.847, |
| "step": 291328 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5221926239040124e-05, |
| "loss": 3.8507, |
| "step": 291840 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5213540291529604e-05, |
| "loss": 3.8596, |
| "step": 292352 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5205154344019084e-05, |
| "loss": 3.8497, |
| "step": 292864 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.519678477531229e-05, |
| "loss": 3.8584, |
| "step": 293376 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.518839882780177e-05, |
| "loss": 3.8505, |
| "step": 293888 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.518001288029125e-05, |
| "loss": 3.8466, |
| "step": 294400 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.517162693278074e-05, |
| "loss": 3.843, |
| "step": 294912 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.516325736407395e-05, |
| "loss": 3.8639, |
| "step": 295424 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.515487141656343e-05, |
| "loss": 3.8618, |
| "step": 295936 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.514648546905291e-05, |
| "loss": 3.8516, |
| "step": 296448 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.513809952154239e-05, |
| "loss": 3.8481, |
| "step": 296960 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.512971357403187e-05, |
| "loss": 3.8549, |
| "step": 297472 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.512134400532508e-05, |
| "loss": 3.8514, |
| "step": 297984 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.511295805781456e-05, |
| "loss": 3.8633, |
| "step": 298496 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.510457211030404e-05, |
| "loss": 3.8484, |
| "step": 299008 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.509618616279352e-05, |
| "loss": 3.8522, |
| "step": 299520 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5087816594086726e-05, |
| "loss": 3.8489, |
| "step": 300032 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5079430646576206e-05, |
| "loss": 3.8544, |
| "step": 300544 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.507104469906569e-05, |
| "loss": 3.8425, |
| "step": 301056 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.506265875155517e-05, |
| "loss": 3.8469, |
| "step": 301568 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.505428918284838e-05, |
| "loss": 3.8517, |
| "step": 302080 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.504590323533786e-05, |
| "loss": 3.844, |
| "step": 302592 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.503751728782734e-05, |
| "loss": 3.8518, |
| "step": 303104 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.502913134031682e-05, |
| "loss": 3.8557, |
| "step": 303616 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.502076177161003e-05, |
| "loss": 3.8473, |
| "step": 304128 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.501237582409951e-05, |
| "loss": 3.8488, |
| "step": 304640 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.500398987658899e-05, |
| "loss": 3.8437, |
| "step": 305152 |
| }, |
| { |
| "epoch": 1.03, |
| "eval_loss": 3.9104487895965576, |
| "eval_runtime": 302.9202, |
| "eval_samples_per_second": 1259.708, |
| "eval_steps_per_second": 39.367, |
| "step": 305280 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.499560392907847e-05, |
| "loss": 3.8326, |
| "step": 305664 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.498721798156795e-05, |
| "loss": 3.8302, |
| "step": 306176 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.497883203405743e-05, |
| "loss": 3.8531, |
| "step": 306688 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.497044608654691e-05, |
| "loss": 3.8444, |
| "step": 307200 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.496206013903639e-05, |
| "loss": 3.8557, |
| "step": 307712 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.495367419152587e-05, |
| "loss": 3.8337, |
| "step": 308224 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.494528824401535e-05, |
| "loss": 3.8457, |
| "step": 308736 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.493690229650483e-05, |
| "loss": 3.8244, |
| "step": 309248 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4928532727798047e-05, |
| "loss": 3.8468, |
| "step": 309760 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.492014678028752e-05, |
| "loss": 3.8338, |
| "step": 310272 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4911760832777e-05, |
| "loss": 3.8514, |
| "step": 310784 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.490337488526648e-05, |
| "loss": 3.8463, |
| "step": 311296 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.489500531655969e-05, |
| "loss": 3.8216, |
| "step": 311808 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.488661936904917e-05, |
| "loss": 3.8368, |
| "step": 312320 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.487823342153865e-05, |
| "loss": 3.8387, |
| "step": 312832 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.486984747402813e-05, |
| "loss": 3.826, |
| "step": 313344 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.486146152651761e-05, |
| "loss": 3.8334, |
| "step": 313856 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4853075579007095e-05, |
| "loss": 3.8362, |
| "step": 314368 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4844689631496575e-05, |
| "loss": 3.8322, |
| "step": 314880 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4836303683986055e-05, |
| "loss": 3.8521, |
| "step": 315392 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4827917736475535e-05, |
| "loss": 3.8287, |
| "step": 315904 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4819548167768744e-05, |
| "loss": 3.8341, |
| "step": 316416 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4811162220258224e-05, |
| "loss": 3.8359, |
| "step": 316928 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4802776272747704e-05, |
| "loss": 3.8415, |
| "step": 317440 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.4794390325237184e-05, |
| "loss": 3.826, |
| "step": 317952 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.478602075653039e-05, |
| "loss": 3.8372, |
| "step": 318464 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.477763480901987e-05, |
| "loss": 3.8217, |
| "step": 318976 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.476924886150935e-05, |
| "loss": 3.8282, |
| "step": 319488 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.476086291399883e-05, |
| "loss": 3.8237, |
| "step": 320000 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.475249334529205e-05, |
| "loss": 3.8253, |
| "step": 320512 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.474410739778153e-05, |
| "loss": 3.8293, |
| "step": 321024 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.473572145027101e-05, |
| "loss": 3.8338, |
| "step": 321536 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.472733550276049e-05, |
| "loss": 3.8334, |
| "step": 322048 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.471894955524997e-05, |
| "loss": 3.8353, |
| "step": 322560 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.471057998654318e-05, |
| "loss": 3.8236, |
| "step": 323072 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.470219403903266e-05, |
| "loss": 3.8344, |
| "step": 323584 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.469380809152214e-05, |
| "loss": 3.8238, |
| "step": 324096 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.468542214401162e-05, |
| "loss": 3.8244, |
| "step": 324608 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.467705257530483e-05, |
| "loss": 3.8124, |
| "step": 325120 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.466866662779431e-05, |
| "loss": 3.8236, |
| "step": 325632 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.466028068028379e-05, |
| "loss": 3.8196, |
| "step": 326144 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.465189473277327e-05, |
| "loss": 3.8326, |
| "step": 326656 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.464350878526275e-05, |
| "loss": 3.8179, |
| "step": 327168 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.463513921655596e-05, |
| "loss": 3.8209, |
| "step": 327680 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.462675326904544e-05, |
| "loss": 3.8344, |
| "step": 328192 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.461836732153492e-05, |
| "loss": 3.8255, |
| "step": 328704 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.46099813740244e-05, |
| "loss": 3.8256, |
| "step": 329216 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.460161180531761e-05, |
| "loss": 3.8168, |
| "step": 329728 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.459322585780709e-05, |
| "loss": 3.8048, |
| "step": 330240 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.458483991029657e-05, |
| "loss": 3.8274, |
| "step": 330752 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.457645396278605e-05, |
| "loss": 3.8211, |
| "step": 331264 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4568068015275525e-05, |
| "loss": 3.8153, |
| "step": 331776 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.455969844656874e-05, |
| "loss": 3.8152, |
| "step": 332288 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.455131249905822e-05, |
| "loss": 3.8246, |
| "step": 332800 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.45429265515477e-05, |
| "loss": 3.8019, |
| "step": 333312 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.453454060403718e-05, |
| "loss": 3.8263, |
| "step": 333824 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4526171035330396e-05, |
| "loss": 3.8046, |
| "step": 334336 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4517785087819876e-05, |
| "loss": 3.8188, |
| "step": 334848 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4509399140309356e-05, |
| "loss": 3.8206, |
| "step": 335360 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.450101319279883e-05, |
| "loss": 3.8113, |
| "step": 335872 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4492643624092045e-05, |
| "loss": 3.8145, |
| "step": 336384 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4484257676581525e-05, |
| "loss": 3.8137, |
| "step": 336896 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4475871729071e-05, |
| "loss": 3.8072, |
| "step": 337408 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.446748578156048e-05, |
| "loss": 3.8098, |
| "step": 337920 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4459116212853694e-05, |
| "loss": 3.8134, |
| "step": 338432 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4450730265343174e-05, |
| "loss": 3.8094, |
| "step": 338944 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4442344317832654e-05, |
| "loss": 3.8108, |
| "step": 339456 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4433958370322134e-05, |
| "loss": 3.799, |
| "step": 339968 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.442558880161535e-05, |
| "loss": 3.8004, |
| "step": 340480 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.441720285410482e-05, |
| "loss": 3.8135, |
| "step": 340992 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.44088169065943e-05, |
| "loss": 3.8243, |
| "step": 341504 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.440043095908378e-05, |
| "loss": 3.8074, |
| "step": 342016 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.439204501157326e-05, |
| "loss": 3.824, |
| "step": 342528 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.438367544286647e-05, |
| "loss": 3.82, |
| "step": 343040 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.437528949535595e-05, |
| "loss": 3.8086, |
| "step": 343552 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.436690354784543e-05, |
| "loss": 3.8113, |
| "step": 344064 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.435851760033491e-05, |
| "loss": 3.817, |
| "step": 344576 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.435014803162813e-05, |
| "loss": 3.8074, |
| "step": 345088 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.434176208411761e-05, |
| "loss": 3.8072, |
| "step": 345600 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.433337613660709e-05, |
| "loss": 3.8149, |
| "step": 346112 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.432499018909657e-05, |
| "loss": 3.815, |
| "step": 346624 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.431662062038978e-05, |
| "loss": 3.8161, |
| "step": 347136 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4308234672879257e-05, |
| "loss": 3.8103, |
| "step": 347648 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4299848725368737e-05, |
| "loss": 3.795, |
| "step": 348160 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4291462777858217e-05, |
| "loss": 3.7946, |
| "step": 348672 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4283093209151426e-05, |
| "loss": 3.8101, |
| "step": 349184 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4274707261640906e-05, |
| "loss": 3.8107, |
| "step": 349696 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4266321314130386e-05, |
| "loss": 3.8036, |
| "step": 350208 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.4257935366619865e-05, |
| "loss": 3.8064, |
| "step": 350720 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.424956579791308e-05, |
| "loss": 3.8023, |
| "step": 351232 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.424117985040256e-05, |
| "loss": 3.8009, |
| "step": 351744 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.423279390289204e-05, |
| "loss": 3.8053, |
| "step": 352256 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.422440795538152e-05, |
| "loss": 3.8069, |
| "step": 352768 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.421603838667473e-05, |
| "loss": 3.8117, |
| "step": 353280 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.420765243916421e-05, |
| "loss": 3.8146, |
| "step": 353792 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.419926649165369e-05, |
| "loss": 3.8086, |
| "step": 354304 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.419088054414317e-05, |
| "loss": 3.7985, |
| "step": 354816 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.418251097543638e-05, |
| "loss": 3.8088, |
| "step": 355328 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.417412502792586e-05, |
| "loss": 3.8027, |
| "step": 355840 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.416573908041534e-05, |
| "loss": 3.8149, |
| "step": 356352 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.4157369511708555e-05, |
| "loss": 3.8012, |
| "step": 356864 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.4148983564198035e-05, |
| "loss": 3.7959, |
| "step": 357376 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.4140597616687515e-05, |
| "loss": 3.8125, |
| "step": 357888 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.4132211669176995e-05, |
| "loss": 3.8078, |
| "step": 358400 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.4123825721666475e-05, |
| "loss": 3.7977, |
| "step": 358912 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.4115456152959684e-05, |
| "loss": 3.8004, |
| "step": 359424 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.4107070205449164e-05, |
| "loss": 3.8027, |
| "step": 359936 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.4098684257938644e-05, |
| "loss": 3.7872, |
| "step": 360448 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.409031468923185e-05, |
| "loss": 3.791, |
| "step": 360960 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.408192874172133e-05, |
| "loss": 3.8037, |
| "step": 361472 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.407354279421081e-05, |
| "loss": 3.7968, |
| "step": 361984 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.406515684670029e-05, |
| "loss": 3.799, |
| "step": 362496 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.405677089918977e-05, |
| "loss": 3.7937, |
| "step": 363008 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.404838495167925e-05, |
| "loss": 3.795, |
| "step": 363520 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.403999900416873e-05, |
| "loss": 3.8141, |
| "step": 364032 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.403161305665822e-05, |
| "loss": 3.8132, |
| "step": 364544 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.40232271091477e-05, |
| "loss": 3.8046, |
| "step": 365056 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.401485754044091e-05, |
| "loss": 3.8024, |
| "step": 365568 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.400647159293039e-05, |
| "loss": 3.7955, |
| "step": 366080 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.399808564541987e-05, |
| "loss": 3.7941, |
| "step": 366592 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.398971607671308e-05, |
| "loss": 3.792, |
| "step": 367104 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.398133012920256e-05, |
| "loss": 3.7953, |
| "step": 367616 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.397294418169204e-05, |
| "loss": 3.7937, |
| "step": 368128 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.396455823418152e-05, |
| "loss": 3.8018, |
| "step": 368640 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3956188665474726e-05, |
| "loss": 3.7946, |
| "step": 369152 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3947802717964206e-05, |
| "loss": 3.7994, |
| "step": 369664 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3939416770453686e-05, |
| "loss": 3.7934, |
| "step": 370176 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.393103082294317e-05, |
| "loss": 3.7907, |
| "step": 370688 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.392266125423638e-05, |
| "loss": 3.7905, |
| "step": 371200 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.391427530672586e-05, |
| "loss": 3.8036, |
| "step": 371712 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.390588935921534e-05, |
| "loss": 3.8077, |
| "step": 372224 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.389750341170482e-05, |
| "loss": 3.7997, |
| "step": 372736 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.388913384299803e-05, |
| "loss": 3.7913, |
| "step": 373248 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.388074789548751e-05, |
| "loss": 3.8013, |
| "step": 373760 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.387236194797699e-05, |
| "loss": 3.7952, |
| "step": 374272 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.386397600046647e-05, |
| "loss": 3.8062, |
| "step": 374784 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.385559005295595e-05, |
| "loss": 3.7953, |
| "step": 375296 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.384722048424916e-05, |
| "loss": 3.7957, |
| "step": 375808 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.383883453673864e-05, |
| "loss": 3.7949, |
| "step": 376320 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.383044858922813e-05, |
| "loss": 3.8032, |
| "step": 376832 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3822079020521336e-05, |
| "loss": 3.7868, |
| "step": 377344 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3813693073010816e-05, |
| "loss": 3.7917, |
| "step": 377856 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3805307125500296e-05, |
| "loss": 3.7962, |
| "step": 378368 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3796921177989776e-05, |
| "loss": 3.7922, |
| "step": 378880 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3788551609282985e-05, |
| "loss": 3.7994, |
| "step": 379392 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3780165661772465e-05, |
| "loss": 3.7994, |
| "step": 379904 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3771779714261945e-05, |
| "loss": 3.7953, |
| "step": 380416 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.3763393766751425e-05, |
| "loss": 3.7909, |
| "step": 380928 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.37550078192409e-05, |
| "loss": 3.7912, |
| "step": 381440 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 3.8856112957000732, |
| "eval_runtime": 304.7445, |
| "eval_samples_per_second": 1252.167, |
| "eval_steps_per_second": 39.131, |
| "step": 381600 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3746638250534114e-05, |
| "loss": 3.7817, |
| "step": 381952 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3738252303023594e-05, |
| "loss": 3.7742, |
| "step": 382464 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.372986635551308e-05, |
| "loss": 3.801, |
| "step": 382976 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3721480408002554e-05, |
| "loss": 3.7922, |
| "step": 383488 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3713094460492034e-05, |
| "loss": 3.7994, |
| "step": 384000 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3704708512981514e-05, |
| "loss": 3.7849, |
| "step": 384512 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3696322565470993e-05, |
| "loss": 3.788, |
| "step": 385024 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3687936617960473e-05, |
| "loss": 3.7763, |
| "step": 385536 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.367956704925368e-05, |
| "loss": 3.7882, |
| "step": 386048 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.367118110174316e-05, |
| "loss": 3.7857, |
| "step": 386560 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.366279515423264e-05, |
| "loss": 3.7963, |
| "step": 387072 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.365440920672212e-05, |
| "loss": 3.794, |
| "step": 387584 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.364605601681907e-05, |
| "loss": 3.7688, |
| "step": 388096 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.363767006930855e-05, |
| "loss": 3.7848, |
| "step": 388608 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.362928412179803e-05, |
| "loss": 3.7876, |
| "step": 389120 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.362089817428751e-05, |
| "loss": 3.775, |
| "step": 389632 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.361251222677699e-05, |
| "loss": 3.78, |
| "step": 390144 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.360412627926647e-05, |
| "loss": 3.7831, |
| "step": 390656 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.359574033175595e-05, |
| "loss": 3.7803, |
| "step": 391168 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.358735438424543e-05, |
| "loss": 3.7982, |
| "step": 391680 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3578984815538636e-05, |
| "loss": 3.7806, |
| "step": 392192 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3570598868028116e-05, |
| "loss": 3.7849, |
| "step": 392704 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3562212920517596e-05, |
| "loss": 3.7824, |
| "step": 393216 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3553826973007076e-05, |
| "loss": 3.7878, |
| "step": 393728 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.3545457404300285e-05, |
| "loss": 3.7744, |
| "step": 394240 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.353707145678977e-05, |
| "loss": 3.7837, |
| "step": 394752 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.352868550927925e-05, |
| "loss": 3.7766, |
| "step": 395264 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.352031594057246e-05, |
| "loss": 3.7742, |
| "step": 395776 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.351192999306194e-05, |
| "loss": 3.7737, |
| "step": 396288 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.350354404555142e-05, |
| "loss": 3.7761, |
| "step": 396800 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.34951580980409e-05, |
| "loss": 3.777, |
| "step": 397312 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.348677215053038e-05, |
| "loss": 3.7816, |
| "step": 397824 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.347840258182359e-05, |
| "loss": 3.7854, |
| "step": 398336 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.347001663431307e-05, |
| "loss": 3.7859, |
| "step": 398848 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.346163068680255e-05, |
| "loss": 3.7728, |
| "step": 399360 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.345324473929203e-05, |
| "loss": 3.7821, |
| "step": 399872 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.344487517058524e-05, |
| "loss": 3.7768, |
| "step": 400384 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3436489223074726e-05, |
| "loss": 3.7763, |
| "step": 400896 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3428103275564206e-05, |
| "loss": 3.7634, |
| "step": 401408 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3419717328053685e-05, |
| "loss": 3.7709, |
| "step": 401920 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3411347759346895e-05, |
| "loss": 3.7722, |
| "step": 402432 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3402961811836375e-05, |
| "loss": 3.7797, |
| "step": 402944 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3394575864325854e-05, |
| "loss": 3.7691, |
| "step": 403456 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3386189916815334e-05, |
| "loss": 3.771, |
| "step": 403968 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3377820348108544e-05, |
| "loss": 3.7868, |
| "step": 404480 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3369434400598024e-05, |
| "loss": 3.7749, |
| "step": 404992 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3361048453087503e-05, |
| "loss": 3.7769, |
| "step": 405504 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3352662505576983e-05, |
| "loss": 3.7696, |
| "step": 406016 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.334429293687019e-05, |
| "loss": 3.7554, |
| "step": 406528 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.333590698935968e-05, |
| "loss": 3.7761, |
| "step": 407040 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.332752104184916e-05, |
| "loss": 3.775, |
| "step": 407552 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.331915147314237e-05, |
| "loss": 3.7645, |
| "step": 408064 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.331076552563185e-05, |
| "loss": 3.7679, |
| "step": 408576 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.330239595692506e-05, |
| "loss": 3.7773, |
| "step": 409088 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.329401000941454e-05, |
| "loss": 3.7475, |
| "step": 409600 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.328562406190402e-05, |
| "loss": 3.7771, |
| "step": 410112 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.32772381143935e-05, |
| "loss": 3.7592, |
| "step": 410624 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.326885216688298e-05, |
| "loss": 3.7669, |
| "step": 411136 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.326046621937246e-05, |
| "loss": 3.775, |
| "step": 411648 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.325208027186194e-05, |
| "loss": 3.7645, |
| "step": 412160 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.324369432435142e-05, |
| "loss": 3.7627, |
| "step": 412672 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3235324755644626e-05, |
| "loss": 3.7691, |
| "step": 413184 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.322693880813411e-05, |
| "loss": 3.7588, |
| "step": 413696 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.321855286062359e-05, |
| "loss": 3.7566, |
| "step": 414208 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.321016691311307e-05, |
| "loss": 3.7669, |
| "step": 414720 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.320179734440628e-05, |
| "loss": 3.7648, |
| "step": 415232 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.319341139689576e-05, |
| "loss": 3.7607, |
| "step": 415744 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.318502544938524e-05, |
| "loss": 3.753, |
| "step": 416256 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.317663950187472e-05, |
| "loss": 3.754, |
| "step": 416768 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.316826993316793e-05, |
| "loss": 3.7629, |
| "step": 417280 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.315988398565741e-05, |
| "loss": 3.7742, |
| "step": 417792 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.315149803814689e-05, |
| "loss": 3.7601, |
| "step": 418304 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.314311209063637e-05, |
| "loss": 3.7771, |
| "step": 418816 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.313474252192958e-05, |
| "loss": 3.7746, |
| "step": 419328 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3126356574419067e-05, |
| "loss": 3.7643, |
| "step": 419840 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3117970626908546e-05, |
| "loss": 3.7583, |
| "step": 420352 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3109584679398026e-05, |
| "loss": 3.7723, |
| "step": 420864 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3101215110691236e-05, |
| "loss": 3.7624, |
| "step": 421376 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3092845541984445e-05, |
| "loss": 3.7586, |
| "step": 421888 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3084459594473925e-05, |
| "loss": 3.771, |
| "step": 422400 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3076073646963405e-05, |
| "loss": 3.7655, |
| "step": 422912 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3067687699452885e-05, |
| "loss": 3.7698, |
| "step": 423424 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3059301751942364e-05, |
| "loss": 3.7641, |
| "step": 423936 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3050915804431844e-05, |
| "loss": 3.7507, |
| "step": 424448 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3042529856921324e-05, |
| "loss": 3.7447, |
| "step": 424960 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.3034143909410804e-05, |
| "loss": 3.7611, |
| "step": 425472 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.302577434070402e-05, |
| "loss": 3.7659, |
| "step": 425984 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.30173883931935e-05, |
| "loss": 3.757, |
| "step": 426496 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.300900244568298e-05, |
| "loss": 3.7577, |
| "step": 427008 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.300061649817245e-05, |
| "loss": 3.7589, |
| "step": 427520 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.299224692946567e-05, |
| "loss": 3.751, |
| "step": 428032 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.298386098195515e-05, |
| "loss": 3.763, |
| "step": 428544 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.297547503444463e-05, |
| "loss": 3.7563, |
| "step": 429056 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.29670890869341e-05, |
| "loss": 3.7683, |
| "step": 429568 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.295871951822732e-05, |
| "loss": 3.7725, |
| "step": 430080 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.29503335707168e-05, |
| "loss": 3.7631, |
| "step": 430592 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.294194762320628e-05, |
| "loss": 3.755, |
| "step": 431104 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.293356167569576e-05, |
| "loss": 3.759, |
| "step": 431616 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2925192106988974e-05, |
| "loss": 3.7565, |
| "step": 432128 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2916806159478454e-05, |
| "loss": 3.77, |
| "step": 432640 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.290842021196793e-05, |
| "loss": 3.7534, |
| "step": 433152 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.290003426445741e-05, |
| "loss": 3.7522, |
| "step": 433664 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.289166469575062e-05, |
| "loss": 3.7654, |
| "step": 434176 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.28832787482401e-05, |
| "loss": 3.7595, |
| "step": 434688 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2874892800729576e-05, |
| "loss": 3.755, |
| "step": 435200 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2866506853219056e-05, |
| "loss": 3.7558, |
| "step": 435712 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.285813728451227e-05, |
| "loss": 3.7534, |
| "step": 436224 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.284975133700175e-05, |
| "loss": 3.7503, |
| "step": 436736 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.284138176829496e-05, |
| "loss": 3.7413, |
| "step": 437248 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.283299582078444e-05, |
| "loss": 3.7595, |
| "step": 437760 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.282460987327393e-05, |
| "loss": 3.757, |
| "step": 438272 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.28162239257634e-05, |
| "loss": 3.7538, |
| "step": 438784 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.280783797825288e-05, |
| "loss": 3.7517, |
| "step": 439296 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.279945203074236e-05, |
| "loss": 3.7466, |
| "step": 439808 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.279106608323184e-05, |
| "loss": 3.7695, |
| "step": 440320 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.278268013572132e-05, |
| "loss": 3.7682, |
| "step": 440832 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.277431056701453e-05, |
| "loss": 3.7613, |
| "step": 441344 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.276592461950401e-05, |
| "loss": 3.7601, |
| "step": 441856 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.275753867199349e-05, |
| "loss": 3.7498, |
| "step": 442368 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.274915272448297e-05, |
| "loss": 3.7516, |
| "step": 442880 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.274078315577618e-05, |
| "loss": 3.7457, |
| "step": 443392 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2732413587069394e-05, |
| "loss": 3.7498, |
| "step": 443904 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2724027639558874e-05, |
| "loss": 3.7545, |
| "step": 444416 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2715641692048354e-05, |
| "loss": 3.7532, |
| "step": 444928 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2707255744537834e-05, |
| "loss": 3.7543, |
| "step": 445440 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2698869797027314e-05, |
| "loss": 3.7517, |
| "step": 445952 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2690483849516794e-05, |
| "loss": 3.7522, |
| "step": 446464 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2682097902006274e-05, |
| "loss": 3.7487, |
| "step": 446976 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2673711954495754e-05, |
| "loss": 3.749, |
| "step": 447488 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.266534238578896e-05, |
| "loss": 3.7541, |
| "step": 448000 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.265695643827844e-05, |
| "loss": 3.7625, |
| "step": 448512 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.264857049076792e-05, |
| "loss": 3.761, |
| "step": 449024 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.26401845432574e-05, |
| "loss": 3.7473, |
| "step": 449536 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.263181497455062e-05, |
| "loss": 3.7585, |
| "step": 450048 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.26234290270401e-05, |
| "loss": 3.7532, |
| "step": 450560 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.261504307952958e-05, |
| "loss": 3.759, |
| "step": 451072 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.260665713201906e-05, |
| "loss": 3.753, |
| "step": 451584 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.259828756331227e-05, |
| "loss": 3.7562, |
| "step": 452096 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.258990161580175e-05, |
| "loss": 3.7523, |
| "step": 452608 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.258151566829123e-05, |
| "loss": 3.7575, |
| "step": 453120 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.257312972078071e-05, |
| "loss": 3.7436, |
| "step": 453632 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2564776530877646e-05, |
| "loss": 3.7517, |
| "step": 454144 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2556390583367126e-05, |
| "loss": 3.7498, |
| "step": 454656 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2548004635856606e-05, |
| "loss": 3.7529, |
| "step": 455168 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.2539618688346086e-05, |
| "loss": 3.7544, |
| "step": 455680 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.253123274083557e-05, |
| "loss": 3.7519, |
| "step": 456192 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.252284679332505e-05, |
| "loss": 3.7603, |
| "step": 456704 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.251446084581453e-05, |
| "loss": 3.7425, |
| "step": 457216 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.250609127710774e-05, |
| "loss": 3.7513, |
| "step": 457728 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 3.8687496185302734, |
| "eval_runtime": 310.0094, |
| "eval_samples_per_second": 1230.902, |
| "eval_steps_per_second": 38.467, |
| "step": 457920 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.249770532959722e-05, |
| "loss": 3.7488, |
| "step": 458240 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.248933576089043e-05, |
| "loss": 3.7293, |
| "step": 458752 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.248094981337991e-05, |
| "loss": 3.7597, |
| "step": 459264 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.247256386586939e-05, |
| "loss": 3.75, |
| "step": 459776 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.246417791835887e-05, |
| "loss": 3.7586, |
| "step": 460288 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.245580834965208e-05, |
| "loss": 3.7413, |
| "step": 460800 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.244742240214156e-05, |
| "loss": 3.7407, |
| "step": 461312 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.243903645463104e-05, |
| "loss": 3.7393, |
| "step": 461824 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.243065050712052e-05, |
| "loss": 3.7441, |
| "step": 462336 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2422264559610006e-05, |
| "loss": 3.7429, |
| "step": 462848 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2413878612099486e-05, |
| "loss": 3.7523, |
| "step": 463360 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2405492664588966e-05, |
| "loss": 3.754, |
| "step": 463872 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2397123095882175e-05, |
| "loss": 3.7306, |
| "step": 464384 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2388737148371655e-05, |
| "loss": 3.7397, |
| "step": 464896 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2380351200861135e-05, |
| "loss": 3.7471, |
| "step": 465408 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2371965253350615e-05, |
| "loss": 3.7319, |
| "step": 465920 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2363579305840095e-05, |
| "loss": 3.7379, |
| "step": 466432 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2355193358329575e-05, |
| "loss": 3.7385, |
| "step": 466944 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2346807410819055e-05, |
| "loss": 3.7389, |
| "step": 467456 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.233842146330853e-05, |
| "loss": 3.7558, |
| "step": 467968 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2330051894601744e-05, |
| "loss": 3.7423, |
| "step": 468480 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2321665947091224e-05, |
| "loss": 3.7427, |
| "step": 468992 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2313279999580704e-05, |
| "loss": 3.7403, |
| "step": 469504 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.2304894052070184e-05, |
| "loss": 3.746, |
| "step": 470016 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.22965244833634e-05, |
| "loss": 3.7356, |
| "step": 470528 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.228813853585288e-05, |
| "loss": 3.7425, |
| "step": 471040 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.227975258834236e-05, |
| "loss": 3.7341, |
| "step": 471552 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.227136664083183e-05, |
| "loss": 3.7326, |
| "step": 472064 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.226299707212505e-05, |
| "loss": 3.735, |
| "step": 472576 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.225461112461453e-05, |
| "loss": 3.7342, |
| "step": 473088 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2246225177104e-05, |
| "loss": 3.7386, |
| "step": 473600 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.223783922959348e-05, |
| "loss": 3.7394, |
| "step": 474112 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.22294696608867e-05, |
| "loss": 3.747, |
| "step": 474624 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.222108371337618e-05, |
| "loss": 3.7465, |
| "step": 475136 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.221269776586566e-05, |
| "loss": 3.7313, |
| "step": 475648 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2204328197158874e-05, |
| "loss": 3.741, |
| "step": 476160 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2195942249648353e-05, |
| "loss": 3.7395, |
| "step": 476672 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2187556302137833e-05, |
| "loss": 3.7331, |
| "step": 477184 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.217917035462731e-05, |
| "loss": 3.7223, |
| "step": 477696 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.217081716472425e-05, |
| "loss": 3.7306, |
| "step": 478208 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.216243121721373e-05, |
| "loss": 3.7342, |
| "step": 478720 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.215404526970321e-05, |
| "loss": 3.7394, |
| "step": 479232 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.214565932219269e-05, |
| "loss": 3.7304, |
| "step": 479744 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.213727337468217e-05, |
| "loss": 3.731, |
| "step": 480256 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.212888742717165e-05, |
| "loss": 3.7497, |
| "step": 480768 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.212050147966113e-05, |
| "loss": 3.7344, |
| "step": 481280 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.211211553215061e-05, |
| "loss": 3.7391, |
| "step": 481792 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.210374596344383e-05, |
| "loss": 3.7257, |
| "step": 482304 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.209536001593331e-05, |
| "loss": 3.7174, |
| "step": 482816 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2086990447226516e-05, |
| "loss": 3.741, |
| "step": 483328 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2078604499715996e-05, |
| "loss": 3.7315, |
| "step": 483840 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2070218552205476e-05, |
| "loss": 3.7273, |
| "step": 484352 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.206183260469495e-05, |
| "loss": 3.7291, |
| "step": 484864 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.205344665718443e-05, |
| "loss": 3.7353, |
| "step": 485376 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.204506070967391e-05, |
| "loss": 3.7106, |
| "step": 485888 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2036691140967125e-05, |
| "loss": 3.7321, |
| "step": 486400 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2028305193456605e-05, |
| "loss": 3.7229, |
| "step": 486912 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2019919245946085e-05, |
| "loss": 3.7209, |
| "step": 487424 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2011533298435565e-05, |
| "loss": 3.7409, |
| "step": 487936 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.2003147350925045e-05, |
| "loss": 3.725, |
| "step": 488448 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.1994761403414525e-05, |
| "loss": 3.7227, |
| "step": 488960 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.1986375455904005e-05, |
| "loss": 3.7304, |
| "step": 489472 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.1977989508393485e-05, |
| "loss": 3.7212, |
| "step": 489984 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.1969619939686694e-05, |
| "loss": 3.7196, |
| "step": 490496 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.1961233992176174e-05, |
| "loss": 3.7259, |
| "step": 491008 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.195286442346938e-05, |
| "loss": 3.7244, |
| "step": 491520 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.194447847595886e-05, |
| "loss": 3.7262, |
| "step": 492032 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.193609252844834e-05, |
| "loss": 3.7171, |
| "step": 492544 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.192770658093782e-05, |
| "loss": 3.709, |
| "step": 493056 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.19193206334273e-05, |
| "loss": 3.7228, |
| "step": 493568 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.191093468591679e-05, |
| "loss": 3.7368, |
| "step": 494080 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.190254873840627e-05, |
| "loss": 3.7203, |
| "step": 494592 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.189416279089575e-05, |
| "loss": 3.741, |
| "step": 495104 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.188579322218896e-05, |
| "loss": 3.7347, |
| "step": 495616 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.187740727467844e-05, |
| "loss": 3.7305, |
| "step": 496128 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.186902132716792e-05, |
| "loss": 3.7153, |
| "step": 496640 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.186065175846113e-05, |
| "loss": 3.7351, |
| "step": 497152 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.185226581095061e-05, |
| "loss": 3.7245, |
| "step": 497664 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.184387986344009e-05, |
| "loss": 3.7205, |
| "step": 498176 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.183549391592957e-05, |
| "loss": 3.7352, |
| "step": 498688 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.182710796841905e-05, |
| "loss": 3.7274, |
| "step": 499200 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.181872202090853e-05, |
| "loss": 3.7326, |
| "step": 499712 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.181033607339801e-05, |
| "loss": 3.7275, |
| "step": 500224 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.180195012588749e-05, |
| "loss": 3.7111, |
| "step": 500736 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.17935805571807e-05, |
| "loss": 3.7093, |
| "step": 501248 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.178519460967018e-05, |
| "loss": 3.7198, |
| "step": 501760 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.177682504096339e-05, |
| "loss": 3.7286, |
| "step": 502272 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.176843909345287e-05, |
| "loss": 3.7216, |
| "step": 502784 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.176005314594235e-05, |
| "loss": 3.7163, |
| "step": 503296 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.175166719843183e-05, |
| "loss": 3.7241, |
| "step": 503808 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.174329762972504e-05, |
| "loss": 3.7137, |
| "step": 504320 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.173491168221452e-05, |
| "loss": 3.7213, |
| "step": 504832 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1726525734704e-05, |
| "loss": 3.7171, |
| "step": 505344 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.171813978719348e-05, |
| "loss": 3.735, |
| "step": 505856 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.170975383968296e-05, |
| "loss": 3.7356, |
| "step": 506368 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.170136789217244e-05, |
| "loss": 3.7243, |
| "step": 506880 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.169298194466192e-05, |
| "loss": 3.7168, |
| "step": 507392 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.16845959971514e-05, |
| "loss": 3.7246, |
| "step": 507904 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.167622642844461e-05, |
| "loss": 3.7159, |
| "step": 508416 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.166784048093409e-05, |
| "loss": 3.7347, |
| "step": 508928 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.165945453342357e-05, |
| "loss": 3.7155, |
| "step": 509440 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.165106858591305e-05, |
| "loss": 3.719, |
| "step": 509952 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.164269901720626e-05, |
| "loss": 3.7243, |
| "step": 510464 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.163431306969574e-05, |
| "loss": 3.7264, |
| "step": 510976 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.162592712218522e-05, |
| "loss": 3.7135, |
| "step": 511488 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.16175411746747e-05, |
| "loss": 3.7207, |
| "step": 512000 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1609171605967915e-05, |
| "loss": 3.7156, |
| "step": 512512 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1600785658457395e-05, |
| "loss": 3.7113, |
| "step": 513024 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.159241608975061e-05, |
| "loss": 3.7059, |
| "step": 513536 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1584030142240084e-05, |
| "loss": 3.7203, |
| "step": 514048 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1575644194729564e-05, |
| "loss": 3.7212, |
| "step": 514560 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1567258247219044e-05, |
| "loss": 3.7138, |
| "step": 515072 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1558872299708523e-05, |
| "loss": 3.718, |
| "step": 515584 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1550486352198003e-05, |
| "loss": 3.7121, |
| "step": 516096 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.154211678349121e-05, |
| "loss": 3.7287, |
| "step": 516608 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.153373083598069e-05, |
| "loss": 3.734, |
| "step": 517120 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.152534488847017e-05, |
| "loss": 3.7258, |
| "step": 517632 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.151695894095965e-05, |
| "loss": 3.722, |
| "step": 518144 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.150857299344913e-05, |
| "loss": 3.7118, |
| "step": 518656 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.150020342474235e-05, |
| "loss": 3.7185, |
| "step": 519168 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.149181747723183e-05, |
| "loss": 3.705, |
| "step": 519680 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.148343152972131e-05, |
| "loss": 3.7156, |
| "step": 520192 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.147504558221079e-05, |
| "loss": 3.7165, |
| "step": 520704 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.146665963470027e-05, |
| "loss": 3.7155, |
| "step": 521216 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.145827368718975e-05, |
| "loss": 3.7194, |
| "step": 521728 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.144990411848296e-05, |
| "loss": 3.7155, |
| "step": 522240 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.144151817097244e-05, |
| "loss": 3.7199, |
| "step": 522752 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.143313222346192e-05, |
| "loss": 3.7124, |
| "step": 523264 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.14247462759514e-05, |
| "loss": 3.7149, |
| "step": 523776 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.141636032844088e-05, |
| "loss": 3.7176, |
| "step": 524288 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.140797438093036e-05, |
| "loss": 3.722, |
| "step": 524800 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.139958843341984e-05, |
| "loss": 3.7285, |
| "step": 525312 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.139120248590932e-05, |
| "loss": 3.7107, |
| "step": 525824 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.138283291720253e-05, |
| "loss": 3.7237, |
| "step": 526336 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.137444696969201e-05, |
| "loss": 3.7183, |
| "step": 526848 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.136606102218149e-05, |
| "loss": 3.7228, |
| "step": 527360 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.13576914534747e-05, |
| "loss": 3.7165, |
| "step": 527872 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.134930550596418e-05, |
| "loss": 3.722, |
| "step": 528384 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.134091955845366e-05, |
| "loss": 3.7117, |
| "step": 528896 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.133253361094314e-05, |
| "loss": 3.725, |
| "step": 529408 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.132416404223635e-05, |
| "loss": 3.7055, |
| "step": 529920 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.131579447352956e-05, |
| "loss": 3.7154, |
| "step": 530432 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.130740852601904e-05, |
| "loss": 3.7141, |
| "step": 530944 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.129902257850852e-05, |
| "loss": 3.7176, |
| "step": 531456 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1290636630998006e-05, |
| "loss": 3.7147, |
| "step": 531968 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1282250683487486e-05, |
| "loss": 3.7192, |
| "step": 532480 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1273881114780695e-05, |
| "loss": 3.7222, |
| "step": 532992 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1265495167270175e-05, |
| "loss": 3.7067, |
| "step": 533504 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.1257109219759655e-05, |
| "loss": 3.722, |
| "step": 534016 |
| }, |
| { |
| "epoch": 1.03, |
| "eval_loss": 3.8589980602264404, |
| "eval_runtime": 310.3836, |
| "eval_samples_per_second": 1229.417, |
| "eval_steps_per_second": 38.42, |
| "step": 534240 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1248723272249135e-05, |
| "loss": 3.7093, |
| "step": 534528 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1240337324738615e-05, |
| "loss": 3.6977, |
| "step": 535040 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1231951377228095e-05, |
| "loss": 3.7189, |
| "step": 535552 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.122356542971757e-05, |
| "loss": 3.7173, |
| "step": 536064 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.121517948220705e-05, |
| "loss": 3.7199, |
| "step": 536576 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1206809913500264e-05, |
| "loss": 3.7075, |
| "step": 537088 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1198423965989744e-05, |
| "loss": 3.7079, |
| "step": 537600 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1190038018479224e-05, |
| "loss": 3.7052, |
| "step": 538112 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1181652070968704e-05, |
| "loss": 3.7096, |
| "step": 538624 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1173266123458184e-05, |
| "loss": 3.7045, |
| "step": 539136 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.116489655475139e-05, |
| "loss": 3.7181, |
| "step": 539648 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.115651060724087e-05, |
| "loss": 3.7156, |
| "step": 540160 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.114814103853409e-05, |
| "loss": 3.7042, |
| "step": 540672 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.113975509102357e-05, |
| "loss": 3.7003, |
| "step": 541184 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.113136914351304e-05, |
| "loss": 3.7129, |
| "step": 541696 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.112298319600252e-05, |
| "loss": 3.6944, |
| "step": 542208 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1114597248492e-05, |
| "loss": 3.7065, |
| "step": 542720 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.110621130098148e-05, |
| "loss": 3.7052, |
| "step": 543232 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.109782535347096e-05, |
| "loss": 3.7061, |
| "step": 543744 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.108943940596044e-05, |
| "loss": 3.7164, |
| "step": 544256 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.108105345844992e-05, |
| "loss": 3.7115, |
| "step": 544768 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.107266751093941e-05, |
| "loss": 3.7064, |
| "step": 545280 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.106428156342889e-05, |
| "loss": 3.7051, |
| "step": 545792 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.105589561591837e-05, |
| "loss": 3.7087, |
| "step": 546304 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.104752604721158e-05, |
| "loss": 3.6999, |
| "step": 546816 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.103914009970106e-05, |
| "loss": 3.7105, |
| "step": 547328 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.103075415219054e-05, |
| "loss": 3.7027, |
| "step": 547840 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.102236820468002e-05, |
| "loss": 3.6972, |
| "step": 548352 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1013998635973227e-05, |
| "loss": 3.6997, |
| "step": 548864 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.1005612688462706e-05, |
| "loss": 3.7046, |
| "step": 549376 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0997243119755916e-05, |
| "loss": 3.6985, |
| "step": 549888 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0988857172245396e-05, |
| "loss": 3.709, |
| "step": 550400 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0980471224734875e-05, |
| "loss": 3.7144, |
| "step": 550912 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.097208527722436e-05, |
| "loss": 3.7094, |
| "step": 551424 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.096369932971384e-05, |
| "loss": 3.6953, |
| "step": 551936 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.095532976100705e-05, |
| "loss": 3.7074, |
| "step": 552448 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.094694381349653e-05, |
| "loss": 3.7081, |
| "step": 552960 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.093855786598601e-05, |
| "loss": 3.6987, |
| "step": 553472 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.093017191847549e-05, |
| "loss": 3.6898, |
| "step": 553984 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.092178597096497e-05, |
| "loss": 3.6932, |
| "step": 554496 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.091341640225818e-05, |
| "loss": 3.7018, |
| "step": 555008 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.090503045474766e-05, |
| "loss": 3.7052, |
| "step": 555520 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.089664450723714e-05, |
| "loss": 3.6934, |
| "step": 556032 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.088825855972662e-05, |
| "loss": 3.6963, |
| "step": 556544 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.08798726122161e-05, |
| "loss": 3.7152, |
| "step": 557056 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.087148666470558e-05, |
| "loss": 3.7011, |
| "step": 557568 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.086310071719506e-05, |
| "loss": 3.7066, |
| "step": 558080 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0854731148488276e-05, |
| "loss": 3.6943, |
| "step": 558592 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0846361579781485e-05, |
| "loss": 3.6812, |
| "step": 559104 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0837975632270965e-05, |
| "loss": 3.711, |
| "step": 559616 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0829589684760445e-05, |
| "loss": 3.6972, |
| "step": 560128 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0821203737249925e-05, |
| "loss": 3.6938, |
| "step": 560640 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0812817789739405e-05, |
| "loss": 3.7018, |
| "step": 561152 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0804448221032614e-05, |
| "loss": 3.6962, |
| "step": 561664 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0796062273522094e-05, |
| "loss": 3.6867, |
| "step": 562176 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0787676326011574e-05, |
| "loss": 3.6922, |
| "step": 562688 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0779290378501054e-05, |
| "loss": 3.6898, |
| "step": 563200 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0770904430990534e-05, |
| "loss": 3.694, |
| "step": 563712 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.076253486228375e-05, |
| "loss": 3.7043, |
| "step": 564224 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.075414891477323e-05, |
| "loss": 3.6926, |
| "step": 564736 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.07457629672627e-05, |
| "loss": 3.6908, |
| "step": 565248 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.073737701975218e-05, |
| "loss": 3.6997, |
| "step": 565760 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.072899107224166e-05, |
| "loss": 3.6852, |
| "step": 566272 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.072062150353488e-05, |
| "loss": 3.689, |
| "step": 566784 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.071223555602435e-05, |
| "loss": 3.69, |
| "step": 567296 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.070384960851383e-05, |
| "loss": 3.6948, |
| "step": 567808 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.069546366100331e-05, |
| "loss": 3.6899, |
| "step": 568320 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.068707771349279e-05, |
| "loss": 3.6892, |
| "step": 568832 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.067869176598227e-05, |
| "loss": 3.6728, |
| "step": 569344 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.067032219727549e-05, |
| "loss": 3.6908, |
| "step": 569856 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.066193624976497e-05, |
| "loss": 3.7039, |
| "step": 570368 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.065355030225445e-05, |
| "loss": 3.6861, |
| "step": 570880 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.064516435474393e-05, |
| "loss": 3.7075, |
| "step": 571392 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.063677840723341e-05, |
| "loss": 3.6981, |
| "step": 571904 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.062839245972289e-05, |
| "loss": 3.7013, |
| "step": 572416 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.062000651221237e-05, |
| "loss": 3.6855, |
| "step": 572928 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0611636943505576e-05, |
| "loss": 3.7009, |
| "step": 573440 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0603250995995056e-05, |
| "loss": 3.6927, |
| "step": 573952 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0594865048484536e-05, |
| "loss": 3.6875, |
| "step": 574464 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0586479100974016e-05, |
| "loss": 3.7031, |
| "step": 574976 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0578093153463496e-05, |
| "loss": 3.6932, |
| "step": 575488 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.0569723584756705e-05, |
| "loss": 3.6998, |
| "step": 576000 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.056133763724619e-05, |
| "loss": 3.6967, |
| "step": 576512 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.055295168973567e-05, |
| "loss": 3.6733, |
| "step": 577024 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.054456574222515e-05, |
| "loss": 3.6833, |
| "step": 577536 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.053619617351836e-05, |
| "loss": 3.6838, |
| "step": 578048 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.052781022600784e-05, |
| "loss": 3.6975, |
| "step": 578560 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.051942427849732e-05, |
| "loss": 3.6912, |
| "step": 579072 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.05110383309868e-05, |
| "loss": 3.6845, |
| "step": 579584 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.050265238347628e-05, |
| "loss": 3.6886, |
| "step": 580096 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.049428281476949e-05, |
| "loss": 3.6845, |
| "step": 580608 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.048589686725897e-05, |
| "loss": 3.6873, |
| "step": 581120 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.047751091974845e-05, |
| "loss": 3.6861, |
| "step": 581632 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.046912497223793e-05, |
| "loss": 3.7027, |
| "step": 582144 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.046073902472741e-05, |
| "loss": 3.7086, |
| "step": 582656 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0452369456020625e-05, |
| "loss": 3.6883, |
| "step": 583168 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0443983508510105e-05, |
| "loss": 3.6854, |
| "step": 583680 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0435597560999585e-05, |
| "loss": 3.6937, |
| "step": 584192 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0427211613489065e-05, |
| "loss": 3.6849, |
| "step": 584704 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0418842044782274e-05, |
| "loss": 3.7009, |
| "step": 585216 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0410456097271754e-05, |
| "loss": 3.6869, |
| "step": 585728 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0402070149761234e-05, |
| "loss": 3.6884, |
| "step": 586240 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.039368420225071e-05, |
| "loss": 3.6935, |
| "step": 586752 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.038529825474019e-05, |
| "loss": 3.6951, |
| "step": 587264 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.037691230722967e-05, |
| "loss": 3.6812, |
| "step": 587776 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.036852635971915e-05, |
| "loss": 3.6883, |
| "step": 588288 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.036015679101236e-05, |
| "loss": 3.6865, |
| "step": 588800 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.035177084350184e-05, |
| "loss": 3.6797, |
| "step": 589312 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.034338489599132e-05, |
| "loss": 3.6773, |
| "step": 589824 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.033501532728454e-05, |
| "loss": 3.6881, |
| "step": 590336 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.032662937977401e-05, |
| "loss": 3.6891, |
| "step": 590848 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.031824343226349e-05, |
| "loss": 3.6839, |
| "step": 591360 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.030985748475297e-05, |
| "loss": 3.6885, |
| "step": 591872 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.030147153724245e-05, |
| "loss": 3.6811, |
| "step": 592384 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.029310196853566e-05, |
| "loss": 3.6993, |
| "step": 592896 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.028471602102514e-05, |
| "loss": 3.702, |
| "step": 593408 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.027633007351462e-05, |
| "loss": 3.6972, |
| "step": 593920 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.02679441260041e-05, |
| "loss": 3.69, |
| "step": 594432 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.025955817849358e-05, |
| "loss": 3.6775, |
| "step": 594944 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.02511886097868e-05, |
| "loss": 3.69, |
| "step": 595456 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.024280266227628e-05, |
| "loss": 3.6765, |
| "step": 595968 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.023441671476576e-05, |
| "loss": 3.6856, |
| "step": 596480 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.022603076725524e-05, |
| "loss": 3.682, |
| "step": 596992 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.021764481974472e-05, |
| "loss": 3.6873, |
| "step": 597504 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0209258872234197e-05, |
| "loss": 3.6859, |
| "step": 598016 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0200889303527406e-05, |
| "loss": 3.6871, |
| "step": 598528 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0192503356016886e-05, |
| "loss": 3.6914, |
| "step": 599040 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0184117408506366e-05, |
| "loss": 3.6825, |
| "step": 599552 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0175731460995846e-05, |
| "loss": 3.6828, |
| "step": 600064 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0167361892289055e-05, |
| "loss": 3.6902, |
| "step": 600576 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0158975944778535e-05, |
| "loss": 3.6878, |
| "step": 601088 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0150589997268015e-05, |
| "loss": 3.697, |
| "step": 601600 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.01422040497575e-05, |
| "loss": 3.6815, |
| "step": 602112 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.013381810224698e-05, |
| "loss": 3.6951, |
| "step": 602624 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.012544853354019e-05, |
| "loss": 3.6878, |
| "step": 603136 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.011706258602967e-05, |
| "loss": 3.6913, |
| "step": 603648 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.010867663851915e-05, |
| "loss": 3.6853, |
| "step": 604160 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.010029069100863e-05, |
| "loss": 3.69, |
| "step": 604672 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.009192112230184e-05, |
| "loss": 3.6819, |
| "step": 605184 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.008353517479132e-05, |
| "loss": 3.695, |
| "step": 605696 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.00751492272808e-05, |
| "loss": 3.6739, |
| "step": 606208 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.006676327977028e-05, |
| "loss": 3.6856, |
| "step": 606720 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.005837733225976e-05, |
| "loss": 3.6897, |
| "step": 607232 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.005000776355297e-05, |
| "loss": 3.6875, |
| "step": 607744 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0041621816042455e-05, |
| "loss": 3.6824, |
| "step": 608256 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0033235868531935e-05, |
| "loss": 3.6852, |
| "step": 608768 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0024849921021415e-05, |
| "loss": 3.6961, |
| "step": 609280 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0016463973510895e-05, |
| "loss": 3.676, |
| "step": 609792 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.0008094404804104e-05, |
| "loss": 3.6915, |
| "step": 610304 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 3.851436138153076, |
| "eval_runtime": 305.4421, |
| "eval_samples_per_second": 1249.307, |
| "eval_steps_per_second": 39.042, |
| "step": 610560 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.9999708457293584e-05, |
| "loss": 3.6763, |
| "step": 610816 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.9991322509783064e-05, |
| "loss": 3.6727, |
| "step": 611328 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.9982936562272544e-05, |
| "loss": 3.6866, |
| "step": 611840 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.997455061476202e-05, |
| "loss": 3.6862, |
| "step": 612352 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.996618104605523e-05, |
| "loss": 3.6942, |
| "step": 612864 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.995779509854471e-05, |
| "loss": 3.6763, |
| "step": 613376 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.994940915103419e-05, |
| "loss": 3.6824, |
| "step": 613888 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.994102320352367e-05, |
| "loss": 3.6779, |
| "step": 614400 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.993265363481689e-05, |
| "loss": 3.677, |
| "step": 614912 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.992426768730637e-05, |
| "loss": 3.6774, |
| "step": 615424 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.991588173979584e-05, |
| "loss": 3.6879, |
| "step": 615936 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.990749579228532e-05, |
| "loss": 3.6862, |
| "step": 616448 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.989912622357854e-05, |
| "loss": 3.672, |
| "step": 616960 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.989074027606802e-05, |
| "loss": 3.6708, |
| "step": 617472 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.988235432855749e-05, |
| "loss": 3.6849, |
| "step": 617984 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.987396838104697e-05, |
| "loss": 3.6646, |
| "step": 618496 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.986558243353645e-05, |
| "loss": 3.677, |
| "step": 619008 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.985719648602593e-05, |
| "loss": 3.675, |
| "step": 619520 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.984881053851541e-05, |
| "loss": 3.6789, |
| "step": 620032 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.984042459100489e-05, |
| "loss": 3.6822, |
| "step": 620544 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.983203864349437e-05, |
| "loss": 3.6853, |
| "step": 621056 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.982365269598386e-05, |
| "loss": 3.6797, |
| "step": 621568 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.981526674847334e-05, |
| "loss": 3.675, |
| "step": 622080 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.980688080096282e-05, |
| "loss": 3.6764, |
| "step": 622592 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.9798511232256026e-05, |
| "loss": 3.6739, |
| "step": 623104 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.9790141663549235e-05, |
| "loss": 3.6816, |
| "step": 623616 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.9781755716038715e-05, |
| "loss": 3.6727, |
| "step": 624128 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.9773369768528195e-05, |
| "loss": 3.6689, |
| "step": 624640 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.9764983821017675e-05, |
| "loss": 3.6754, |
| "step": 625152 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.9756614252310884e-05, |
| "loss": 3.6712, |
| "step": 625664 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9748228304800364e-05, |
| "loss": 3.6685, |
| "step": 626176 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9739842357289844e-05, |
| "loss": 3.6838, |
| "step": 626688 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9731456409779324e-05, |
| "loss": 3.6823, |
| "step": 627200 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.972307046226881e-05, |
| "loss": 3.6783, |
| "step": 627712 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.971468451475829e-05, |
| "loss": 3.6688, |
| "step": 628224 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.97063149460515e-05, |
| "loss": 3.6812, |
| "step": 628736 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.969792899854098e-05, |
| "loss": 3.6779, |
| "step": 629248 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.968954305103046e-05, |
| "loss": 3.6732, |
| "step": 629760 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.968115710351994e-05, |
| "loss": 3.6604, |
| "step": 630272 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.967277115600942e-05, |
| "loss": 3.6645, |
| "step": 630784 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.966440158730263e-05, |
| "loss": 3.6683, |
| "step": 631296 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.965601563979211e-05, |
| "loss": 3.6783, |
| "step": 631808 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.964762969228159e-05, |
| "loss": 3.6682, |
| "step": 632320 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.963924374477107e-05, |
| "loss": 3.6685, |
| "step": 632832 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.963085779726055e-05, |
| "loss": 3.6839, |
| "step": 633344 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.962247184975003e-05, |
| "loss": 3.6721, |
| "step": 633856 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.961408590223951e-05, |
| "loss": 3.68, |
| "step": 634368 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9605716333532724e-05, |
| "loss": 3.666, |
| "step": 634880 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9597346764825933e-05, |
| "loss": 3.6496, |
| "step": 635392 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9588960817315413e-05, |
| "loss": 3.6838, |
| "step": 635904 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9580574869804893e-05, |
| "loss": 3.6709, |
| "step": 636416 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.957218892229437e-05, |
| "loss": 3.668, |
| "step": 636928 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.956380297478385e-05, |
| "loss": 3.6689, |
| "step": 637440 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.955543340607706e-05, |
| "loss": 3.6714, |
| "step": 637952 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.954704745856654e-05, |
| "loss": 3.6549, |
| "step": 638464 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.953866151105602e-05, |
| "loss": 3.6624, |
| "step": 638976 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.95302755635455e-05, |
| "loss": 3.6627, |
| "step": 639488 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.952190599483872e-05, |
| "loss": 3.6638, |
| "step": 640000 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.95135200473282e-05, |
| "loss": 3.6768, |
| "step": 640512 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.950513409981768e-05, |
| "loss": 3.6635, |
| "step": 641024 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.949674815230715e-05, |
| "loss": 3.6618, |
| "step": 641536 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.948836220479663e-05, |
| "loss": 3.6769, |
| "step": 642048 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.947999263608985e-05, |
| "loss": 3.6529, |
| "step": 642560 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.947160668857933e-05, |
| "loss": 3.6587, |
| "step": 643072 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.94632207410688e-05, |
| "loss": 3.6625, |
| "step": 643584 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.945483479355828e-05, |
| "loss": 3.6663, |
| "step": 644096 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9446465224851496e-05, |
| "loss": 3.6608, |
| "step": 644608 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.943807927734097e-05, |
| "loss": 3.6621, |
| "step": 645120 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9429693329830456e-05, |
| "loss": 3.6473, |
| "step": 645632 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9421307382319936e-05, |
| "loss": 3.6608, |
| "step": 646144 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9412921434809416e-05, |
| "loss": 3.6749, |
| "step": 646656 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9404551866102625e-05, |
| "loss": 3.6599, |
| "step": 647168 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9396165918592105e-05, |
| "loss": 3.6799, |
| "step": 647680 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9387779971081585e-05, |
| "loss": 3.6713, |
| "step": 648192 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9379394023571065e-05, |
| "loss": 3.6766, |
| "step": 648704 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9371008076060545e-05, |
| "loss": 3.6539, |
| "step": 649216 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9362638507353754e-05, |
| "loss": 3.6729, |
| "step": 649728 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9354252559843234e-05, |
| "loss": 3.6673, |
| "step": 650240 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9345866612332714e-05, |
| "loss": 3.6565, |
| "step": 650752 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9337480664822194e-05, |
| "loss": 3.6752, |
| "step": 651264 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.9329094717311674e-05, |
| "loss": 3.6663, |
| "step": 651776 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.932072514860489e-05, |
| "loss": 3.6686, |
| "step": 652288 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.931233920109437e-05, |
| "loss": 3.6689, |
| "step": 652800 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.930395325358385e-05, |
| "loss": 3.6497, |
| "step": 653312 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.929556730607333e-05, |
| "loss": 3.6538, |
| "step": 653824 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.928719773736654e-05, |
| "loss": 3.6574, |
| "step": 654336 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.927881178985602e-05, |
| "loss": 3.6665, |
| "step": 654848 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.92704258423455e-05, |
| "loss": 3.6616, |
| "step": 655360 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.926203989483498e-05, |
| "loss": 3.6569, |
| "step": 655872 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.925367032612819e-05, |
| "loss": 3.6599, |
| "step": 656384 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.924528437861767e-05, |
| "loss": 3.6618, |
| "step": 656896 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.923689843110715e-05, |
| "loss": 3.6591, |
| "step": 657408 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.922851248359663e-05, |
| "loss": 3.6564, |
| "step": 657920 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.922014291488984e-05, |
| "loss": 3.6783, |
| "step": 658432 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.921175696737932e-05, |
| "loss": 3.6759, |
| "step": 658944 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.92033710198688e-05, |
| "loss": 3.6661, |
| "step": 659456 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.919498507235828e-05, |
| "loss": 3.6552, |
| "step": 659968 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.918659912484776e-05, |
| "loss": 3.6649, |
| "step": 660480 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.917822955614097e-05, |
| "loss": 3.657, |
| "step": 660992 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.916984360863045e-05, |
| "loss": 3.6733, |
| "step": 661504 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.916145766111993e-05, |
| "loss": 3.6623, |
| "step": 662016 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.915307171360941e-05, |
| "loss": 3.659, |
| "step": 662528 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.914468576609889e-05, |
| "loss": 3.6665, |
| "step": 663040 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.913629981858837e-05, |
| "loss": 3.6695, |
| "step": 663552 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.912793024988158e-05, |
| "loss": 3.6542, |
| "step": 664064 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.911954430237106e-05, |
| "loss": 3.661, |
| "step": 664576 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.911115835486054e-05, |
| "loss": 3.6553, |
| "step": 665088 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.910277240735003e-05, |
| "loss": 3.6564, |
| "step": 665600 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.909440283864324e-05, |
| "loss": 3.6498, |
| "step": 666112 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.908601689113272e-05, |
| "loss": 3.66, |
| "step": 666624 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.90776309436222e-05, |
| "loss": 3.6593, |
| "step": 667136 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.906924499611168e-05, |
| "loss": 3.6614, |
| "step": 667648 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.9060875427404886e-05, |
| "loss": 3.661, |
| "step": 668160 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.9052489479894366e-05, |
| "loss": 3.6508, |
| "step": 668672 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.9044103532383846e-05, |
| "loss": 3.6685, |
| "step": 669184 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.9035717584873326e-05, |
| "loss": 3.6783, |
| "step": 669696 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.9027331637362806e-05, |
| "loss": 3.6639, |
| "step": 670208 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.901894568985228e-05, |
| "loss": 3.6705, |
| "step": 670720 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.9010576121145495e-05, |
| "loss": 3.6473, |
| "step": 671232 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.900219017363498e-05, |
| "loss": 3.6643, |
| "step": 671744 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.899380422612446e-05, |
| "loss": 3.6532, |
| "step": 672256 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.8985418278613934e-05, |
| "loss": 3.6558, |
| "step": 672768 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.8977032331103414e-05, |
| "loss": 3.6539, |
| "step": 673280 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.8968646383592894e-05, |
| "loss": 3.6591, |
| "step": 673792 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.8960260436082374e-05, |
| "loss": 3.6602, |
| "step": 674304 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.8951890867375583e-05, |
| "loss": 3.6587, |
| "step": 674816 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.8943504919865063e-05, |
| "loss": 3.6634, |
| "step": 675328 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.893511897235454e-05, |
| "loss": 3.6579, |
| "step": 675840 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.892673302484402e-05, |
| "loss": 3.6507, |
| "step": 676352 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.891836345613723e-05, |
| "loss": 3.6635, |
| "step": 676864 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.890997750862672e-05, |
| "loss": 3.6628, |
| "step": 677376 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.89015915611162e-05, |
| "loss": 3.6708, |
| "step": 677888 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.889320561360568e-05, |
| "loss": 3.653, |
| "step": 678400 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.888481966609516e-05, |
| "loss": 3.6688, |
| "step": 678912 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.887645009738837e-05, |
| "loss": 3.662, |
| "step": 679424 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.886806414987785e-05, |
| "loss": 3.6615, |
| "step": 679936 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.885967820236733e-05, |
| "loss": 3.6606, |
| "step": 680448 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.885130863366054e-05, |
| "loss": 3.6617, |
| "step": 680960 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.884292268615002e-05, |
| "loss": 3.6567, |
| "step": 681472 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.88345367386395e-05, |
| "loss": 3.6688, |
| "step": 681984 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.882615079112898e-05, |
| "loss": 3.6498, |
| "step": 682496 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.881776484361846e-05, |
| "loss": 3.6574, |
| "step": 683008 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.880939527491167e-05, |
| "loss": 3.6621, |
| "step": 683520 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.880100932740115e-05, |
| "loss": 3.6657, |
| "step": 684032 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.879262337989063e-05, |
| "loss": 3.6512, |
| "step": 684544 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.878423743238011e-05, |
| "loss": 3.6604, |
| "step": 685056 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.877586786367332e-05, |
| "loss": 3.6659, |
| "step": 685568 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.87674819161628e-05, |
| "loss": 3.6551, |
| "step": 686080 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.875909596865228e-05, |
| "loss": 3.6647, |
| "step": 686592 |
| }, |
| { |
| "epoch": 1.03, |
| "eval_loss": 3.8468892574310303, |
| "eval_runtime": 304.9955, |
| "eval_samples_per_second": 1251.136, |
| "eval_steps_per_second": 39.099, |
| "step": 686880 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.875071002114176e-05, |
| "loss": 3.6385, |
| "step": 687104 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.874232407363124e-05, |
| "loss": 3.6486, |
| "step": 687616 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.873393812612072e-05, |
| "loss": 3.6597, |
| "step": 688128 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.87255521786102e-05, |
| "loss": 3.6606, |
| "step": 688640 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.871718260990341e-05, |
| "loss": 3.6656, |
| "step": 689152 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.870879666239289e-05, |
| "loss": 3.6488, |
| "step": 689664 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.870041071488237e-05, |
| "loss": 3.6554, |
| "step": 690176 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.869202476737186e-05, |
| "loss": 3.6541, |
| "step": 690688 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8683655198665066e-05, |
| "loss": 3.6481, |
| "step": 691200 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8675269251154546e-05, |
| "loss": 3.6526, |
| "step": 691712 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8666883303644026e-05, |
| "loss": 3.6589, |
| "step": 692224 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8658497356133506e-05, |
| "loss": 3.6562, |
| "step": 692736 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8650127787426715e-05, |
| "loss": 3.657, |
| "step": 693248 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8641741839916195e-05, |
| "loss": 3.6382, |
| "step": 693760 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8633355892405675e-05, |
| "loss": 3.6583, |
| "step": 694272 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8624969944895155e-05, |
| "loss": 3.6401, |
| "step": 694784 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8616583997384635e-05, |
| "loss": 3.6496, |
| "step": 695296 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8608198049874115e-05, |
| "loss": 3.6526, |
| "step": 695808 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8599812102363595e-05, |
| "loss": 3.6497, |
| "step": 696320 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8591426154853075e-05, |
| "loss": 3.6573, |
| "step": 696832 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8583040207342555e-05, |
| "loss": 3.6586, |
| "step": 697344 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8574654259832035e-05, |
| "loss": 3.6568, |
| "step": 697856 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8566284691125244e-05, |
| "loss": 3.6532, |
| "step": 698368 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8557898743614724e-05, |
| "loss": 3.6463, |
| "step": 698880 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8549512796104204e-05, |
| "loss": 3.6514, |
| "step": 699392 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.8541126848593684e-05, |
| "loss": 3.6539, |
| "step": 699904 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.853275727988689e-05, |
| "loss": 3.6507, |
| "step": 700416 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.852437133237637e-05, |
| "loss": 3.6459, |
| "step": 700928 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.851598538486585e-05, |
| "loss": 3.6453, |
| "step": 701440 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.850759943735533e-05, |
| "loss": 3.6487, |
| "step": 701952 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.849922986864854e-05, |
| "loss": 3.639, |
| "step": 702464 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.849084392113803e-05, |
| "loss": 3.6574, |
| "step": 702976 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.848245797362751e-05, |
| "loss": 3.6553, |
| "step": 703488 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.847407202611699e-05, |
| "loss": 3.6541, |
| "step": 704000 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.84657024574102e-05, |
| "loss": 3.6443, |
| "step": 704512 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.845731650989968e-05, |
| "loss": 3.6519, |
| "step": 705024 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.844893056238916e-05, |
| "loss": 3.6539, |
| "step": 705536 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.844054461487864e-05, |
| "loss": 3.6469, |
| "step": 706048 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.843217504617185e-05, |
| "loss": 3.6377, |
| "step": 706560 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8423789098661327e-05, |
| "loss": 3.6391, |
| "step": 707072 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8415403151150807e-05, |
| "loss": 3.6421, |
| "step": 707584 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8407017203640286e-05, |
| "loss": 3.6528, |
| "step": 708096 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8398631256129766e-05, |
| "loss": 3.6415, |
| "step": 708608 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.839026168742298e-05, |
| "loss": 3.6421, |
| "step": 709120 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.838187573991246e-05, |
| "loss": 3.6562, |
| "step": 709632 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.837348979240194e-05, |
| "loss": 3.6475, |
| "step": 710144 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.836510384489142e-05, |
| "loss": 3.6534, |
| "step": 710656 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.83567178973809e-05, |
| "loss": 3.6421, |
| "step": 711168 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.834834832867411e-05, |
| "loss": 3.6272, |
| "step": 711680 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.833996238116359e-05, |
| "loss": 3.656, |
| "step": 712192 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.833157643365307e-05, |
| "loss": 3.6424, |
| "step": 712704 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.832319048614255e-05, |
| "loss": 3.6464, |
| "step": 713216 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.831480453863203e-05, |
| "loss": 3.6445, |
| "step": 713728 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.830643496992524e-05, |
| "loss": 3.6444, |
| "step": 714240 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.829804902241472e-05, |
| "loss": 3.6283, |
| "step": 714752 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.82896630749042e-05, |
| "loss": 3.6414, |
| "step": 715264 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.828127712739368e-05, |
| "loss": 3.6368, |
| "step": 715776 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.827289117988317e-05, |
| "loss": 3.6395, |
| "step": 716288 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8264521611176376e-05, |
| "loss": 3.655, |
| "step": 716800 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8256135663665856e-05, |
| "loss": 3.638, |
| "step": 717312 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8247749716155336e-05, |
| "loss": 3.636, |
| "step": 717824 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8239363768644816e-05, |
| "loss": 3.6545, |
| "step": 718336 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8230994199938025e-05, |
| "loss": 3.6276, |
| "step": 718848 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8222608252427505e-05, |
| "loss": 3.6296, |
| "step": 719360 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8214222304916985e-05, |
| "loss": 3.6433, |
| "step": 719872 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8205836357406465e-05, |
| "loss": 3.6418, |
| "step": 720384 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8197466788699674e-05, |
| "loss": 3.6339, |
| "step": 720896 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8189080841189154e-05, |
| "loss": 3.6405, |
| "step": 721408 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8180694893678634e-05, |
| "loss": 3.6221, |
| "step": 721920 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.817230894616812e-05, |
| "loss": 3.6354, |
| "step": 722432 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.81639229986576e-05, |
| "loss": 3.6518, |
| "step": 722944 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.815555342995081e-05, |
| "loss": 3.638, |
| "step": 723456 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.814716748244029e-05, |
| "loss": 3.6505, |
| "step": 723968 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.813878153492977e-05, |
| "loss": 3.6478, |
| "step": 724480 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.813039558741925e-05, |
| "loss": 3.6576, |
| "step": 724992 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.812200963990872e-05, |
| "loss": 3.6303, |
| "step": 725504 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.811364007120194e-05, |
| "loss": 3.6465, |
| "step": 726016 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.810525412369142e-05, |
| "loss": 3.6444, |
| "step": 726528 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.80968681761809e-05, |
| "loss": 3.6282, |
| "step": 727040 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.808848222867037e-05, |
| "loss": 3.6539, |
| "step": 727552 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.808011265996359e-05, |
| "loss": 3.6411, |
| "step": 728064 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8071726712453074e-05, |
| "loss": 3.6445, |
| "step": 728576 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.806334076494255e-05, |
| "loss": 3.6468, |
| "step": 729088 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.805495481743203e-05, |
| "loss": 3.6269, |
| "step": 729600 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.804656886992151e-05, |
| "loss": 3.6256, |
| "step": 730112 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.803819930121472e-05, |
| "loss": 3.6346, |
| "step": 730624 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8029813353704196e-05, |
| "loss": 3.6431, |
| "step": 731136 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8021427406193676e-05, |
| "loss": 3.6368, |
| "step": 731648 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.8013041458683156e-05, |
| "loss": 3.6323, |
| "step": 732160 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.800467188997637e-05, |
| "loss": 3.6331, |
| "step": 732672 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7996285942465845e-05, |
| "loss": 3.6409, |
| "step": 733184 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7987899994955325e-05, |
| "loss": 3.6338, |
| "step": 733696 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.797951404744481e-05, |
| "loss": 3.6295, |
| "step": 734208 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.797114447873802e-05, |
| "loss": 3.6548, |
| "step": 734720 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.79627585312275e-05, |
| "loss": 3.6525, |
| "step": 735232 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.795437258371698e-05, |
| "loss": 3.6448, |
| "step": 735744 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.794598663620646e-05, |
| "loss": 3.6292, |
| "step": 736256 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.793761706749967e-05, |
| "loss": 3.6416, |
| "step": 736768 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.792923111998915e-05, |
| "loss": 3.6357, |
| "step": 737280 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.792084517247863e-05, |
| "loss": 3.6461, |
| "step": 737792 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.791245922496811e-05, |
| "loss": 3.6376, |
| "step": 738304 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.790408965626132e-05, |
| "loss": 3.6312, |
| "step": 738816 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.78957037087508e-05, |
| "loss": 3.641, |
| "step": 739328 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.788731776124028e-05, |
| "loss": 3.6487, |
| "step": 739840 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7878931813729766e-05, |
| "loss": 3.6305, |
| "step": 740352 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7870545866219246e-05, |
| "loss": 3.6409, |
| "step": 740864 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7862176297512455e-05, |
| "loss": 3.6276, |
| "step": 741376 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7853790350001935e-05, |
| "loss": 3.6352, |
| "step": 741888 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7845404402491415e-05, |
| "loss": 3.6297, |
| "step": 742400 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7837018454980894e-05, |
| "loss": 3.6294, |
| "step": 742912 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7828648886274104e-05, |
| "loss": 3.6375, |
| "step": 743424 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7820262938763584e-05, |
| "loss": 3.638, |
| "step": 743936 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7811876991253063e-05, |
| "loss": 3.6379, |
| "step": 744448 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7803491043742543e-05, |
| "loss": 3.6278, |
| "step": 744960 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.779512147503575e-05, |
| "loss": 3.6467, |
| "step": 745472 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.778673552752523e-05, |
| "loss": 3.6513, |
| "step": 745984 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.777834958001471e-05, |
| "loss": 3.6412, |
| "step": 746496 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.77699636325042e-05, |
| "loss": 3.6483, |
| "step": 747008 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.776159406379741e-05, |
| "loss": 3.6262, |
| "step": 747520 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.775320811628689e-05, |
| "loss": 3.6375, |
| "step": 748032 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.774482216877637e-05, |
| "loss": 3.6329, |
| "step": 748544 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.773643622126585e-05, |
| "loss": 3.6293, |
| "step": 749056 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.772806665255906e-05, |
| "loss": 3.6311, |
| "step": 749568 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.771968070504854e-05, |
| "loss": 3.6343, |
| "step": 750080 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.771129475753802e-05, |
| "loss": 3.6366, |
| "step": 750592 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.77029088100275e-05, |
| "loss": 3.634, |
| "step": 751104 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.769452286251698e-05, |
| "loss": 3.6393, |
| "step": 751616 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7686153293810186e-05, |
| "loss": 3.638, |
| "step": 752128 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7677767346299666e-05, |
| "loss": 3.6257, |
| "step": 752640 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.766938139878915e-05, |
| "loss": 3.64, |
| "step": 753152 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.766099545127863e-05, |
| "loss": 3.6416, |
| "step": 753664 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.765262588257184e-05, |
| "loss": 3.645, |
| "step": 754176 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.764423993506132e-05, |
| "loss": 3.6312, |
| "step": 754688 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.76358539875508e-05, |
| "loss": 3.6468, |
| "step": 755200 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.762746804004028e-05, |
| "loss": 3.638, |
| "step": 755712 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.761908209252976e-05, |
| "loss": 3.6379, |
| "step": 756224 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.761071252382297e-05, |
| "loss": 3.6379, |
| "step": 756736 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.760232657631245e-05, |
| "loss": 3.6342, |
| "step": 757248 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.759394062880193e-05, |
| "loss": 3.6332, |
| "step": 757760 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.758555468129141e-05, |
| "loss": 3.6496, |
| "step": 758272 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.757718511258462e-05, |
| "loss": 3.621, |
| "step": 758784 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7568799165074107e-05, |
| "loss": 3.6364, |
| "step": 759296 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7560413217563586e-05, |
| "loss": 3.6402, |
| "step": 759808 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7552027270053066e-05, |
| "loss": 3.6362, |
| "step": 760320 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7543657701346276e-05, |
| "loss": 3.6313, |
| "step": 760832 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7535271753835755e-05, |
| "loss": 3.6369, |
| "step": 761344 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7526885806325235e-05, |
| "loss": 3.6397, |
| "step": 761856 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7518499858814715e-05, |
| "loss": 3.636, |
| "step": 762368 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.7510130290107924e-05, |
| "loss": 3.6384, |
| "step": 762880 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 3.843653917312622, |
| "eval_runtime": 301.8904, |
| "eval_samples_per_second": 1264.005, |
| "eval_steps_per_second": 39.501, |
| "step": 763200 |
| } |
| ], |
| "logging_steps": 512, |
| "max_steps": 3052726, |
| "num_train_epochs": 9223372036854775807, |
| "save_steps": 10, |
| "total_flos": 5.311679277613363e+17, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|