| { | |
| "best_metric": 4.046647071838379, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/existential-there-quantifier/transformer/4/checkpoints/checkpoint-152640", | |
| "epoch": 0.025000606015738065, | |
| "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": 10.9521, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 6.8262, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 6.1939, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 5.9899, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 5.8315, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 5.734, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 5.6212, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994129836742636e-05, | |
| "loss": 5.5598, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993291241991584e-05, | |
| "loss": 5.4784, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992452647240532e-05, | |
| "loss": 5.4148, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99161405248948e-05, | |
| "loss": 5.3678, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990775457738428e-05, | |
| "loss": 5.3295, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989936862987376e-05, | |
| "loss": 5.2839, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989099906116697e-05, | |
| "loss": 5.2291, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988261311365645e-05, | |
| "loss": 5.1923, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987422716614593e-05, | |
| "loss": 5.1557, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986584121863541e-05, | |
| "loss": 5.1189, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985745527112489e-05, | |
| "loss": 5.0961, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984906932361437e-05, | |
| "loss": 5.0691, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984068337610385e-05, | |
| "loss": 5.0371, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.0251, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.982391148108281e-05, | |
| "loss": 4.9935, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9815541912376026e-05, | |
| "loss": 4.9745, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9807155964865506e-05, | |
| "loss": 4.9472, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9798770017354986e-05, | |
| "loss": 4.9372, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9790384069844466e-05, | |
| "loss": 4.9066, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9782014501137675e-05, | |
| "loss": 4.8872, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773628553627155e-05, | |
| "loss": 4.8746, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9765242606116635e-05, | |
| "loss": 4.8532, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9756856658606115e-05, | |
| "loss": 4.8294, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748487089899324e-05, | |
| "loss": 4.824, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9740117521192533e-05, | |
| "loss": 4.8135, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9731731573682013e-05, | |
| "loss": 4.7994, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.97233456261715e-05, | |
| "loss": 4.7811, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.971495967866098e-05, | |
| "loss": 4.777, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.970657373115046e-05, | |
| "loss": 4.754, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.969820416244367e-05, | |
| "loss": 4.7444, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968981821493315e-05, | |
| "loss": 4.7168, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968143226742263e-05, | |
| "loss": 4.7181, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.967304631991211e-05, | |
| "loss": 4.7048, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966466037240159e-05, | |
| "loss": 4.7026, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965627442489107e-05, | |
| "loss": 4.6893, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964788847738054e-05, | |
| "loss": 4.6854, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963950252987002e-05, | |
| "loss": 4.6668, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963113296116324e-05, | |
| "loss": 4.6605, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962274701365272e-05, | |
| "loss": 4.6566, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96143610661422e-05, | |
| "loss": 4.6392, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.960597511863168e-05, | |
| "loss": 4.6453, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.959758917112116e-05, | |
| "loss": 4.6222, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95892359812181e-05, | |
| "loss": 4.6061, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958085003370758e-05, | |
| "loss": 4.6182, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.957246408619706e-05, | |
| "loss": 4.5989, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956407813868654e-05, | |
| "loss": 4.5903, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9555692191176016e-05, | |
| "loss": 4.5709, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9547306243665496e-05, | |
| "loss": 4.5828, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9538920296154976e-05, | |
| "loss": 4.5577, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9530534348644456e-05, | |
| "loss": 4.5893, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.952216477993767e-05, | |
| "loss": 4.5521, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951377883242715e-05, | |
| "loss": 4.5646, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.950539288491663e-05, | |
| "loss": 4.5548, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949700693740611e-05, | |
| "loss": 4.53, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948863736869932e-05, | |
| "loss": 4.5234, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94802514211888e-05, | |
| "loss": 4.528, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947186547367828e-05, | |
| "loss": 4.5125, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946347952616776e-05, | |
| "loss": 4.506, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945510995746097e-05, | |
| "loss": 4.5236, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944672400995045e-05, | |
| "loss": 4.507, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.943833806243993e-05, | |
| "loss": 4.4937, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942995211492941e-05, | |
| "loss": 4.4786, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9421582546222625e-05, | |
| "loss": 4.4766, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9413196598712105e-05, | |
| "loss": 4.487, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9404810651201585e-05, | |
| "loss": 4.489, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9396424703691065e-05, | |
| "loss": 4.4771, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388038756180545e-05, | |
| "loss": 4.4754, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379652808670025e-05, | |
| "loss": 4.4726, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371266861159505e-05, | |
| "loss": 4.4661, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362897292452714e-05, | |
| "loss": 4.454, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9354511344942194e-05, | |
| "loss": 4.4516, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9346125397431674e-05, | |
| "loss": 4.4477, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9337739449921154e-05, | |
| "loss": 4.4351, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329353502410634e-05, | |
| "loss": 4.4384, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9320967554900114e-05, | |
| "loss": 4.4397, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9312581607389594e-05, | |
| "loss": 4.4394, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9304195659879074e-05, | |
| "loss": 4.4221, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.929582609117229e-05, | |
| "loss": 4.4017, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.928744014366177e-05, | |
| "loss": 4.4219, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927905419615125e-05, | |
| "loss": 4.4163, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.927066824864073e-05, | |
| "loss": 4.4146, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.926229867993394e-05, | |
| "loss": 4.4065, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.925392911122715e-05, | |
| "loss": 4.4078, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.924554316371663e-05, | |
| "loss": 4.3966, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923715721620611e-05, | |
| "loss": 4.3884, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922877126869559e-05, | |
| "loss": 4.3916, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922038532118507e-05, | |
| "loss": 4.3893, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921201575247828e-05, | |
| "loss": 4.3966, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920362980496776e-05, | |
| "loss": 4.3941, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919524385745724e-05, | |
| "loss": 4.3859, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918685790994672e-05, | |
| "loss": 4.3678, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.91784719624362e-05, | |
| "loss": 4.3704, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9170086014925676e-05, | |
| "loss": 4.3724, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9161700067415156e-05, | |
| "loss": 4.3715, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9153314119904636e-05, | |
| "loss": 4.3604, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.914494455119785e-05, | |
| "loss": 4.3594, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9136558603687325e-05, | |
| "loss": 4.3481, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9128172656176805e-05, | |
| "loss": 4.3645, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9119786708666285e-05, | |
| "loss": 4.3332, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.911143351876323e-05, | |
| "loss": 4.3462, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.910304757125272e-05, | |
| "loss": 4.3458, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90946616237422e-05, | |
| "loss": 4.3349, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.908627567623168e-05, | |
| "loss": 4.3354, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907788972872115e-05, | |
| "loss": 4.3367, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906950378121063e-05, | |
| "loss": 4.3286, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906111783370011e-05, | |
| "loss": 4.3344, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905273188618959e-05, | |
| "loss": 4.3296, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90443623174828e-05, | |
| "loss": 4.3226, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903597636997228e-05, | |
| "loss": 4.3259, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.902759042246176e-05, | |
| "loss": 4.33, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9019220853754975e-05, | |
| "loss": 4.3205, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9010834906244455e-05, | |
| "loss": 4.3099, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9002448958733935e-05, | |
| "loss": 4.3134, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8994063011223415e-05, | |
| "loss": 4.314, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8985677063712895e-05, | |
| "loss": 4.3069, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8977291116202375e-05, | |
| "loss": 4.2962, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968905168691855e-05, | |
| "loss": 4.3048, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8960535599985064e-05, | |
| "loss": 4.3082, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8952149652474544e-05, | |
| "loss": 4.2993, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8943763704964024e-05, | |
| "loss": 4.2944, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8935377757453504e-05, | |
| "loss": 4.2866, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.892700818874671e-05, | |
| "loss": 4.2897, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.891862224123619e-05, | |
| "loss": 4.2842, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.891023629372567e-05, | |
| "loss": 4.2976, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890185034621515e-05, | |
| "loss": 4.2894, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889346439870464e-05, | |
| "loss": 4.2701, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888507845119412e-05, | |
| "loss": 4.2808, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88766925036836e-05, | |
| "loss": 4.2843, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.886830655617308e-05, | |
| "loss": 4.2759, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.885993698746629e-05, | |
| "loss": 4.2816, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88515674187595e-05, | |
| "loss": 4.2675, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884318147124898e-05, | |
| "loss": 4.2764, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883479552373846e-05, | |
| "loss": 4.2789, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882640957622794e-05, | |
| "loss": 4.2674, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8818040007521146e-05, | |
| "loss": 4.2642, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8809654060010626e-05, | |
| "loss": 4.2654, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8801268112500106e-05, | |
| "loss": 4.2532, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8792882164989586e-05, | |
| "loss": 4.2532, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.878449621747907e-05, | |
| "loss": 4.266, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877611026996855e-05, | |
| "loss": 4.2455, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.876772432245803e-05, | |
| "loss": 4.2504, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875933837494751e-05, | |
| "loss": 4.2546, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875096880624072e-05, | |
| "loss": 4.2507, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.215391635894775, | |
| "eval_runtime": 310.9131, | |
| "eval_samples_per_second": 1227.324, | |
| "eval_steps_per_second": 38.355, | |
| "step": 76320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.87425828587302e-05, | |
| "loss": 4.2277, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.873419691121968e-05, | |
| "loss": 4.2349, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.872581096370916e-05, | |
| "loss": 4.253, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8717425016198635e-05, | |
| "loss": 4.2354, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8709039068688115e-05, | |
| "loss": 4.2476, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8700653121177595e-05, | |
| "loss": 4.2215, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8692267173667075e-05, | |
| "loss": 4.2376, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8683881226156555e-05, | |
| "loss": 4.2233, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.867549527864604e-05, | |
| "loss": 4.2222, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.866710933113552e-05, | |
| "loss": 4.2242, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8658723383625e-05, | |
| "loss": 4.2245, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.865033743611448e-05, | |
| "loss": 4.2279, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.864198424621142e-05, | |
| "loss": 4.2123, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.86335982987009e-05, | |
| "loss": 4.2083, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.862521235119038e-05, | |
| "loss": 4.2065, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.861682640367986e-05, | |
| "loss": 4.2035, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.860844045616934e-05, | |
| "loss": 4.202, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.860005450865882e-05, | |
| "loss": 4.2046, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.85916685611483e-05, | |
| "loss": 4.2018, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.858328261363778e-05, | |
| "loss": 4.2275, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.857489666612726e-05, | |
| "loss": 4.2, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.856651071861674e-05, | |
| "loss": 4.2051, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8558124771106226e-05, | |
| "loss": 4.1988, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8549755202399435e-05, | |
| "loss": 4.2047, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8541385633692644e-05, | |
| "loss": 4.1937, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8532999686182124e-05, | |
| "loss": 4.1918, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8524613738671604e-05, | |
| "loss": 4.1903, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8516227791161084e-05, | |
| "loss": 4.1845, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8507841843650564e-05, | |
| "loss": 4.1698, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.849947227494377e-05, | |
| "loss": 4.1863, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.849108632743325e-05, | |
| "loss": 4.1834, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.848271675872646e-05, | |
| "loss": 4.1882, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.847433081121595e-05, | |
| "loss": 4.1787, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.846594486370543e-05, | |
| "loss": 4.1829, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.845755891619491e-05, | |
| "loss": 4.1718, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.844917296868439e-05, | |
| "loss": 4.1724, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.844078702117387e-05, | |
| "loss": 4.1591, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.843240107366335e-05, | |
| "loss": 4.1669, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.842401512615282e-05, | |
| "loss": 4.1633, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.84156291786423e-05, | |
| "loss": 4.1638, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.840725960993552e-05, | |
| "loss": 4.1644, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8398873662425e-05, | |
| "loss": 4.1699, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.839048771491447e-05, | |
| "loss": 4.1592, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.838210176740395e-05, | |
| "loss": 4.1607, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.837371581989343e-05, | |
| "loss": 4.1635, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.836532987238292e-05, | |
| "loss": 4.1495, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8356960303676126e-05, | |
| "loss": 4.1611, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8348574356165606e-05, | |
| "loss": 4.1478, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8340188408655086e-05, | |
| "loss": 4.1404, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8331802461144566e-05, | |
| "loss": 4.1568, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8323416513634046e-05, | |
| "loss": 4.1439, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8315046944927255e-05, | |
| "loss": 4.1408, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8306660997416735e-05, | |
| "loss": 4.1337, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8298275049906215e-05, | |
| "loss": 4.1415, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8289889102395695e-05, | |
| "loss": 4.1263, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8281519533688904e-05, | |
| "loss": 4.1564, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8273133586178384e-05, | |
| "loss": 4.1318, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.826474763866787e-05, | |
| "loss": 4.1453, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.825636169115735e-05, | |
| "loss": 4.1426, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.824797574364683e-05, | |
| "loss": 4.1245, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.823960617494004e-05, | |
| "loss": 4.1192, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.823122022742952e-05, | |
| "loss": 4.1306, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8222834279919e-05, | |
| "loss": 4.1206, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.821446471121221e-05, | |
| "loss": 4.1166, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.820607876370169e-05, | |
| "loss": 4.1337, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.819769281619117e-05, | |
| "loss": 4.1317, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.818930686868065e-05, | |
| "loss": 4.1154, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.818093729997386e-05, | |
| "loss": 4.1098, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.817255135246334e-05, | |
| "loss": 4.1121, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.816416540495282e-05, | |
| "loss": 4.1219, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8155779457442305e-05, | |
| "loss": 4.1236, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8147393509931785e-05, | |
| "loss": 4.1217, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8139023941224994e-05, | |
| "loss": 4.1226, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8130637993714474e-05, | |
| "loss": 4.1226, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8122252046203954e-05, | |
| "loss": 4.1236, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8113866098693434e-05, | |
| "loss": 4.1056, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.810549652998664e-05, | |
| "loss": 4.1138, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.809711058247612e-05, | |
| "loss": 4.1126, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.80887246349656e-05, | |
| "loss": 4.0995, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.808033868745508e-05, | |
| "loss": 4.1112, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.807196911874829e-05, | |
| "loss": 4.1137, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.806358317123777e-05, | |
| "loss": 4.1145, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.805519722372726e-05, | |
| "loss": 4.0972, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.804681127621674e-05, | |
| "loss": 4.0862, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.803844170750995e-05, | |
| "loss": 4.103, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.803005575999943e-05, | |
| "loss": 4.1014, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.802166981248891e-05, | |
| "loss": 4.1048, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.801328386497839e-05, | |
| "loss": 4.0976, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8004914296271596e-05, | |
| "loss": 4.0983, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7996528348761076e-05, | |
| "loss": 4.0983, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7988142401250556e-05, | |
| "loss": 4.0844, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7979756453740036e-05, | |
| "loss": 4.0928, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7971370506229516e-05, | |
| "loss": 4.0927, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7962984558718996e-05, | |
| "loss": 4.1017, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7954598611208476e-05, | |
| "loss": 4.1035, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7946212663697956e-05, | |
| "loss": 4.0972, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.793784309499117e-05, | |
| "loss": 4.0823, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792945714748065e-05, | |
| "loss": 4.0875, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792107119997013e-05, | |
| "loss": 4.0865, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7912685252459605e-05, | |
| "loss": 4.0953, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.790431568375282e-05, | |
| "loss": 4.0823, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.78959297362423e-05, | |
| "loss": 4.0804, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7887543788731774e-05, | |
| "loss": 4.0732, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7879157841221254e-05, | |
| "loss": 4.0938, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.787078827251447e-05, | |
| "loss": 4.0612, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.786240232500395e-05, | |
| "loss": 4.0773, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7854032756297166e-05, | |
| "loss": 4.0838, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7845646808786646e-05, | |
| "loss": 4.0714, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7837260861276126e-05, | |
| "loss": 4.0724, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7828874913765606e-05, | |
| "loss": 4.0809, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.782048896625508e-05, | |
| "loss": 4.071, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.781210301874456e-05, | |
| "loss": 4.0731, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.780371707123404e-05, | |
| "loss": 4.0769, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.779534750252725e-05, | |
| "loss": 4.0702, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.778696155501673e-05, | |
| "loss": 4.0772, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.777857560750621e-05, | |
| "loss": 4.0821, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7770206038799423e-05, | |
| "loss": 4.0731, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7761820091288903e-05, | |
| "loss": 4.0653, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7753434143778383e-05, | |
| "loss": 4.0699, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.774504819626786e-05, | |
| "loss": 4.0673, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.773666224875734e-05, | |
| "loss": 4.0636, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.772827630124682e-05, | |
| "loss": 4.0588, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.77198903537363e-05, | |
| "loss": 4.0619, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.771150440622578e-05, | |
| "loss": 4.0748, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.770313483751899e-05, | |
| "loss": 4.0638, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.769474889000847e-05, | |
| "loss": 4.0637, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.768636294249795e-05, | |
| "loss": 4.0532, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.767799337379116e-05, | |
| "loss": 4.0561, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.766960742628064e-05, | |
| "loss": 4.0539, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.766122147877012e-05, | |
| "loss": 4.0708, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.76528355312596e-05, | |
| "loss": 4.0662, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.764444958374909e-05, | |
| "loss": 4.044, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.76360800150423e-05, | |
| "loss": 4.0523, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.762769406753178e-05, | |
| "loss": 4.0622, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.761930812002126e-05, | |
| "loss": 4.0547, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.761092217251074e-05, | |
| "loss": 4.0627, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7602552603803946e-05, | |
| "loss": 4.0484, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7594166656293426e-05, | |
| "loss": 4.0548, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7585780708782906e-05, | |
| "loss": 4.0605, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7577394761272386e-05, | |
| "loss": 4.05, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7569025192565595e-05, | |
| "loss": 4.0513, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7560639245055075e-05, | |
| "loss": 4.0532, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7552253297544555e-05, | |
| "loss": 4.0434, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7543867350034035e-05, | |
| "loss": 4.042, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.753549778132725e-05, | |
| "loss": 4.0553, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.752711183381673e-05, | |
| "loss": 4.0371, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.751872588630621e-05, | |
| "loss": 4.0451, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.751035631759942e-05, | |
| "loss": 4.0469, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.75019703700889e-05, | |
| "loss": 4.0435, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.046647071838379, | |
| "eval_runtime": 311.1083, | |
| "eval_samples_per_second": 1226.554, | |
| "eval_steps_per_second": 38.331, | |
| "step": 152640 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 1.0565630839056384e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |