| { | |
| "best_metric": 4.025022029876709, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/full2/transformer/1/checkpoints/checkpoint-152640", | |
| "epoch": 1.0250006060157382, | |
| "eval_steps": 10, | |
| "global_step": 152640, | |
| "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 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 1.0623358555226726e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |