| { |
| "best_metric": 3.9311435222625732, |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/existential-there-quantifier/transformer/2/checkpoints/checkpoint-305280", |
| "epoch": 1.0250006060157382, |
| "eval_steps": 10, |
| "global_step": 305280, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.999998362119627e-05, |
| "loss": 10.9166, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.999161405248948e-05, |
| "loss": 6.8283, |
| "step": 512 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.998322810497896e-05, |
| "loss": 6.2048, |
| "step": 1024 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.997484215746844e-05, |
| "loss": 5.9964, |
| "step": 1536 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.996645620995792e-05, |
| "loss": 5.8327, |
| "step": 2048 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.99580702624474e-05, |
| "loss": 5.7348, |
| "step": 2560 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.994968431493688e-05, |
| "loss": 5.6227, |
| "step": 3072 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.994129836742636e-05, |
| "loss": 5.5631, |
| "step": 3584 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.993291241991584e-05, |
| "loss": 5.4813, |
| "step": 4096 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.992452647240532e-05, |
| "loss": 5.4183, |
| "step": 4608 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.99161405248948e-05, |
| "loss": 5.3724, |
| "step": 5120 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.990775457738428e-05, |
| "loss": 5.3348, |
| "step": 5632 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.989936862987376e-05, |
| "loss": 5.2877, |
| "step": 6144 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.989099906116697e-05, |
| "loss": 5.2345, |
| "step": 6656 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.988261311365645e-05, |
| "loss": 5.1989, |
| "step": 7168 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.987422716614593e-05, |
| "loss": 5.1625, |
| "step": 7680 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.986584121863541e-05, |
| "loss": 5.1257, |
| "step": 8192 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.985745527112489e-05, |
| "loss": 5.1041, |
| "step": 8704 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.984906932361437e-05, |
| "loss": 5.0766, |
| "step": 9216 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.984068337610385e-05, |
| "loss": 5.0445, |
| "step": 9728 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.983229742859333e-05, |
| "loss": 5.0345, |
| "step": 10240 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9823927859886547e-05, |
| "loss": 5.0011, |
| "step": 10752 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9815541912376026e-05, |
| "loss": 4.9832, |
| "step": 11264 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9807155964865506e-05, |
| "loss": 4.956, |
| "step": 11776 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9798770017354986e-05, |
| "loss": 4.9435, |
| "step": 12288 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9790400448648195e-05, |
| "loss": 4.9146, |
| "step": 12800 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9782014501137675e-05, |
| "loss": 4.8953, |
| "step": 13312 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9773628553627155e-05, |
| "loss": 4.8822, |
| "step": 13824 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9765242606116635e-05, |
| "loss": 4.8622, |
| "step": 14336 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9756873037409844e-05, |
| "loss": 4.8349, |
| "step": 14848 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9748487089899324e-05, |
| "loss": 4.8325, |
| "step": 15360 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9740101142388804e-05, |
| "loss": 4.8213, |
| "step": 15872 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9731715194878284e-05, |
| "loss": 4.8074, |
| "step": 16384 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.97233456261715e-05, |
| "loss": 4.7883, |
| "step": 16896 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.971495967866098e-05, |
| "loss": 4.7844, |
| "step": 17408 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.970657373115046e-05, |
| "loss": 4.7602, |
| "step": 17920 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.969818778363994e-05, |
| "loss": 4.7506, |
| "step": 18432 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.968981821493315e-05, |
| "loss": 4.7242, |
| "step": 18944 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.968143226742263e-05, |
| "loss": 4.725, |
| "step": 19456 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.967304631991211e-05, |
| "loss": 4.7118, |
| "step": 19968 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.966466037240159e-05, |
| "loss": 4.7087, |
| "step": 20480 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.965627442489107e-05, |
| "loss": 4.6969, |
| "step": 20992 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.964788847738054e-05, |
| "loss": 4.6931, |
| "step": 21504 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.963950252987002e-05, |
| "loss": 4.6733, |
| "step": 22016 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.96311165823595e-05, |
| "loss": 4.6683, |
| "step": 22528 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.962273063484898e-05, |
| "loss": 4.661, |
| "step": 23040 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.96143610661422e-05, |
| "loss": 4.6458, |
| "step": 23552 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.960597511863168e-05, |
| "loss": 4.6523, |
| "step": 24064 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.959758917112116e-05, |
| "loss": 4.6296, |
| "step": 24576 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.958920322361064e-05, |
| "loss": 4.6138, |
| "step": 25088 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.958081727610012e-05, |
| "loss": 4.6249, |
| "step": 25600 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.957244770739333e-05, |
| "loss": 4.6042, |
| "step": 26112 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.956406175988281e-05, |
| "loss": 4.5984, |
| "step": 26624 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.955567581237229e-05, |
| "loss": 4.5776, |
| "step": 27136 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9547306243665496e-05, |
| "loss": 4.5866, |
| "step": 27648 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9538920296154976e-05, |
| "loss": 4.5636, |
| "step": 28160 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9530534348644456e-05, |
| "loss": 4.5956, |
| "step": 28672 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9522148401133936e-05, |
| "loss": 4.5581, |
| "step": 29184 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.951376245362342e-05, |
| "loss": 4.5709, |
| "step": 29696 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.95053765061129e-05, |
| "loss": 4.5619, |
| "step": 30208 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.949699055860238e-05, |
| "loss": 4.5366, |
| "step": 30720 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.948860461109186e-05, |
| "loss": 4.5299, |
| "step": 31232 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.948023504238507e-05, |
| "loss": 4.535, |
| "step": 31744 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.947184909487455e-05, |
| "loss": 4.518, |
| "step": 32256 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.946346314736403e-05, |
| "loss": 4.5114, |
| "step": 32768 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.945507719985351e-05, |
| "loss": 4.5302, |
| "step": 33280 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.944670763114672e-05, |
| "loss": 4.5107, |
| "step": 33792 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.94383216836362e-05, |
| "loss": 4.4995, |
| "step": 34304 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.942993573612568e-05, |
| "loss": 4.4835, |
| "step": 34816 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.942156616741889e-05, |
| "loss": 4.4833, |
| "step": 35328 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.941318021990837e-05, |
| "loss": 4.4932, |
| "step": 35840 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9404794272397856e-05, |
| "loss": 4.494, |
| "step": 36352 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9396408324887336e-05, |
| "loss": 4.4816, |
| "step": 36864 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9388022377376816e-05, |
| "loss": 4.4806, |
| "step": 37376 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9379636429866296e-05, |
| "loss": 4.4788, |
| "step": 37888 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9371250482355776e-05, |
| "loss": 4.4716, |
| "step": 38400 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9362880913648985e-05, |
| "loss": 4.4609, |
| "step": 38912 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9354494966138465e-05, |
| "loss": 4.4556, |
| "step": 39424 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9346109018627945e-05, |
| "loss": 4.4518, |
| "step": 39936 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9337723071117425e-05, |
| "loss": 4.4405, |
| "step": 40448 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9329337123606905e-05, |
| "loss": 4.4415, |
| "step": 40960 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.932095117609638e-05, |
| "loss": 4.4445, |
| "step": 41472 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.931256522858586e-05, |
| "loss": 4.4426, |
| "step": 41984 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.930417928107534e-05, |
| "loss": 4.4273, |
| "step": 42496 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9295809712368554e-05, |
| "loss": 4.4064, |
| "step": 43008 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9287423764858034e-05, |
| "loss": 4.4256, |
| "step": 43520 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9279037817347514e-05, |
| "loss": 4.4207, |
| "step": 44032 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9270651869836994e-05, |
| "loss": 4.4195, |
| "step": 44544 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.92622823011302e-05, |
| "loss": 4.4117, |
| "step": 45056 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.925391273242342e-05, |
| "loss": 4.4124, |
| "step": 45568 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.92455267849129e-05, |
| "loss": 4.4016, |
| "step": 46080 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.923714083740238e-05, |
| "loss": 4.3898, |
| "step": 46592 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.922875488989185e-05, |
| "loss": 4.3966, |
| "step": 47104 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.922036894238133e-05, |
| "loss": 4.3928, |
| "step": 47616 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.921198299487081e-05, |
| "loss": 4.4008, |
| "step": 48128 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.920359704736029e-05, |
| "loss": 4.3973, |
| "step": 48640 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.919521109984978e-05, |
| "loss": 4.3903, |
| "step": 49152 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.918684153114299e-05, |
| "loss": 4.3711, |
| "step": 49664 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.917845558363247e-05, |
| "loss": 4.3732, |
| "step": 50176 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.917006963612195e-05, |
| "loss": 4.376, |
| "step": 50688 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9161700067415156e-05, |
| "loss": 4.3736, |
| "step": 51200 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9153314119904636e-05, |
| "loss": 4.3631, |
| "step": 51712 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9144928172394116e-05, |
| "loss": 4.3643, |
| "step": 52224 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9136542224883596e-05, |
| "loss": 4.3523, |
| "step": 52736 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9128156277373076e-05, |
| "loss": 4.3675, |
| "step": 53248 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9119786708666285e-05, |
| "loss": 4.3378, |
| "step": 53760 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9111400761155765e-05, |
| "loss": 4.3497, |
| "step": 54272 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9103014813645245e-05, |
| "loss": 4.35, |
| "step": 54784 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.909462886613473e-05, |
| "loss": 4.3402, |
| "step": 55296 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.908624291862421e-05, |
| "loss": 4.3381, |
| "step": 55808 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.907785697111369e-05, |
| "loss": 4.3407, |
| "step": 56320 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.906947102360317e-05, |
| "loss": 4.3324, |
| "step": 56832 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.906108507609265e-05, |
| "loss": 4.3383, |
| "step": 57344 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.905271550738586e-05, |
| "loss": 4.3341, |
| "step": 57856 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.904432955987534e-05, |
| "loss": 4.3254, |
| "step": 58368 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.903594361236482e-05, |
| "loss": 4.3291, |
| "step": 58880 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.902757404365803e-05, |
| "loss": 4.3337, |
| "step": 59392 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.901920447495124e-05, |
| "loss": 4.3227, |
| "step": 59904 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.901081852744072e-05, |
| "loss": 4.3126, |
| "step": 60416 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.90024325799302e-05, |
| "loss": 4.3157, |
| "step": 60928 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8994046632419686e-05, |
| "loss": 4.3178, |
| "step": 61440 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8985660684909166e-05, |
| "loss": 4.3104, |
| "step": 61952 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8977274737398646e-05, |
| "loss": 4.2997, |
| "step": 62464 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8968888789888125e-05, |
| "loss": 4.3077, |
| "step": 62976 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8960519221181335e-05, |
| "loss": 4.3121, |
| "step": 63488 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8952133273670815e-05, |
| "loss": 4.3025, |
| "step": 64000 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8943747326160294e-05, |
| "loss": 4.2989, |
| "step": 64512 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8935377757453504e-05, |
| "loss": 4.2894, |
| "step": 65024 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8926991809942984e-05, |
| "loss": 4.2931, |
| "step": 65536 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8918605862432463e-05, |
| "loss": 4.2871, |
| "step": 66048 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8910219914921943e-05, |
| "loss": 4.3012, |
| "step": 66560 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8901833967411423e-05, |
| "loss": 4.2928, |
| "step": 67072 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.88934480199009e-05, |
| "loss": 4.2736, |
| "step": 67584 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.888506207239038e-05, |
| "loss": 4.2811, |
| "step": 68096 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.887667612487986e-05, |
| "loss": 4.2871, |
| "step": 68608 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.886830655617308e-05, |
| "loss": 4.28, |
| "step": 69120 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.885993698746629e-05, |
| "loss": 4.2854, |
| "step": 69632 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.885155103995577e-05, |
| "loss": 4.2714, |
| "step": 70144 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.884316509244525e-05, |
| "loss": 4.279, |
| "step": 70656 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.883477914493473e-05, |
| "loss": 4.2809, |
| "step": 71168 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.882639319742421e-05, |
| "loss": 4.2697, |
| "step": 71680 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.881800724991369e-05, |
| "loss": 4.2672, |
| "step": 72192 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.880962130240316e-05, |
| "loss": 4.268, |
| "step": 72704 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.880123535489264e-05, |
| "loss": 4.2566, |
| "step": 73216 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.879286578618586e-05, |
| "loss": 4.2551, |
| "step": 73728 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.878447983867534e-05, |
| "loss": 4.2694, |
| "step": 74240 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.877609389116482e-05, |
| "loss": 4.2469, |
| "step": 74752 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.87677079436543e-05, |
| "loss": 4.253, |
| "step": 75264 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.875933837494751e-05, |
| "loss": 4.2581, |
| "step": 75776 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8750952427436986e-05, |
| "loss": 4.2526, |
| "step": 76288 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 4.215147495269775, |
| "eval_runtime": 304.8841, |
| "eval_samples_per_second": 1251.593, |
| "eval_steps_per_second": 39.113, |
| "step": 76320 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8742566479926466e-05, |
| "loss": 4.2344, |
| "step": 76800 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8734180532415946e-05, |
| "loss": 4.236, |
| "step": 77312 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.872581096370916e-05, |
| "loss": 4.255, |
| "step": 77824 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8717425016198635e-05, |
| "loss": 4.2389, |
| "step": 78336 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8709039068688115e-05, |
| "loss": 4.2493, |
| "step": 78848 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8700653121177595e-05, |
| "loss": 4.2222, |
| "step": 79360 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.869228355247081e-05, |
| "loss": 4.2406, |
| "step": 79872 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.868389760496029e-05, |
| "loss": 4.2264, |
| "step": 80384 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8675528036253507e-05, |
| "loss": 4.2254, |
| "step": 80896 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8667142088742986e-05, |
| "loss": 4.2283, |
| "step": 81408 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.865875614123246e-05, |
| "loss": 4.2277, |
| "step": 81920 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.865037019372194e-05, |
| "loss": 4.2317, |
| "step": 82432 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8642017003818885e-05, |
| "loss": 4.2154, |
| "step": 82944 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8633631056308365e-05, |
| "loss": 4.2123, |
| "step": 83456 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8625245108797845e-05, |
| "loss": 4.2101, |
| "step": 83968 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8616859161287324e-05, |
| "loss": 4.2067, |
| "step": 84480 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8608473213776804e-05, |
| "loss": 4.2051, |
| "step": 84992 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.860008726626628e-05, |
| "loss": 4.2084, |
| "step": 85504 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8591701318755764e-05, |
| "loss": 4.2063, |
| "step": 86016 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8583315371245244e-05, |
| "loss": 4.229, |
| "step": 86528 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8574929423734724e-05, |
| "loss": 4.2041, |
| "step": 87040 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8566543476224204e-05, |
| "loss": 4.2086, |
| "step": 87552 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8558157528713684e-05, |
| "loss": 4.2011, |
| "step": 88064 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.8549771581203164e-05, |
| "loss": 4.2066, |
| "step": 88576 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.854140201249637e-05, |
| "loss": 4.1972, |
| "step": 89088 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.853301606498585e-05, |
| "loss": 4.1946, |
| "step": 89600 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.852463011747533e-05, |
| "loss": 4.1939, |
| "step": 90112 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.851624416996481e-05, |
| "loss": 4.1859, |
| "step": 90624 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.850785822245429e-05, |
| "loss": 4.1722, |
| "step": 91136 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.849947227494377e-05, |
| "loss": 4.1868, |
| "step": 91648 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.849108632743325e-05, |
| "loss": 4.1856, |
| "step": 92160 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.848271675872646e-05, |
| "loss": 4.1907, |
| "step": 92672 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.847433081121595e-05, |
| "loss": 4.1807, |
| "step": 93184 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.846594486370543e-05, |
| "loss": 4.1859, |
| "step": 93696 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.845755891619491e-05, |
| "loss": 4.175, |
| "step": 94208 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.844918934748812e-05, |
| "loss": 4.1747, |
| "step": 94720 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.84408033999776e-05, |
| "loss": 4.1618, |
| "step": 95232 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.843241745246708e-05, |
| "loss": 4.1688, |
| "step": 95744 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.842403150495656e-05, |
| "loss": 4.166, |
| "step": 96256 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.841564555744604e-05, |
| "loss": 4.1645, |
| "step": 96768 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.840725960993552e-05, |
| "loss": 4.1661, |
| "step": 97280 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8398873662425e-05, |
| "loss": 4.173, |
| "step": 97792 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.839048771491447e-05, |
| "loss": 4.1625, |
| "step": 98304 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8382118146207687e-05, |
| "loss": 4.1639, |
| "step": 98816 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8373732198697167e-05, |
| "loss": 4.1662, |
| "step": 99328 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8365346251186646e-05, |
| "loss": 4.1513, |
| "step": 99840 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8356960303676126e-05, |
| "loss": 4.1621, |
| "step": 100352 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.834859073496934e-05, |
| "loss": 4.1495, |
| "step": 100864 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.834020478745882e-05, |
| "loss": 4.1445, |
| "step": 101376 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.833183521875203e-05, |
| "loss": 4.1601, |
| "step": 101888 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.832344927124151e-05, |
| "loss": 4.1463, |
| "step": 102400 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.831506332373099e-05, |
| "loss": 4.1454, |
| "step": 102912 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.830667737622047e-05, |
| "loss": 4.1358, |
| "step": 103424 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.829830780751368e-05, |
| "loss": 4.1425, |
| "step": 103936 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.828992186000316e-05, |
| "loss": 4.1296, |
| "step": 104448 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.828153591249264e-05, |
| "loss": 4.1575, |
| "step": 104960 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.827314996498212e-05, |
| "loss": 4.1355, |
| "step": 105472 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8264780396275336e-05, |
| "loss": 4.1476, |
| "step": 105984 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8256394448764816e-05, |
| "loss": 4.1466, |
| "step": 106496 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8248008501254296e-05, |
| "loss": 4.1275, |
| "step": 107008 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.823962255374377e-05, |
| "loss": 4.1215, |
| "step": 107520 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.823123660623325e-05, |
| "loss": 4.1332, |
| "step": 108032 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.822285065872273e-05, |
| "loss": 4.1233, |
| "step": 108544 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8214481090015945e-05, |
| "loss": 4.1178, |
| "step": 109056 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.820609514250542e-05, |
| "loss": 4.1365, |
| "step": 109568 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.81977091949949e-05, |
| "loss": 4.1314, |
| "step": 110080 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.818932324748438e-05, |
| "loss": 4.1191, |
| "step": 110592 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.818093729997386e-05, |
| "loss": 4.1123, |
| "step": 111104 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.817255135246334e-05, |
| "loss": 4.1137, |
| "step": 111616 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8164181783756554e-05, |
| "loss": 4.1262, |
| "step": 112128 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8155795836246034e-05, |
| "loss": 4.1256, |
| "step": 112640 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8147409888735514e-05, |
| "loss": 4.123, |
| "step": 113152 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8139023941224994e-05, |
| "loss": 4.1258, |
| "step": 113664 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8130637993714474e-05, |
| "loss": 4.125, |
| "step": 114176 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8122252046203954e-05, |
| "loss": 4.1249, |
| "step": 114688 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.8113866098693434e-05, |
| "loss": 4.1091, |
| "step": 115200 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.810549652998664e-05, |
| "loss": 4.118, |
| "step": 115712 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.809711058247612e-05, |
| "loss": 4.1149, |
| "step": 116224 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.80887246349656e-05, |
| "loss": 4.1022, |
| "step": 116736 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.808033868745508e-05, |
| "loss": 4.1144, |
| "step": 117248 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.807195273994456e-05, |
| "loss": 4.1136, |
| "step": 117760 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.806358317123777e-05, |
| "loss": 4.1179, |
| "step": 118272 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.805519722372726e-05, |
| "loss": 4.099, |
| "step": 118784 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.804681127621674e-05, |
| "loss": 4.088, |
| "step": 119296 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.803842532870622e-05, |
| "loss": 4.1055, |
| "step": 119808 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.80300393811957e-05, |
| "loss": 4.105, |
| "step": 120320 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.802165343368518e-05, |
| "loss": 4.1073, |
| "step": 120832 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.801326748617466e-05, |
| "loss": 4.1008, |
| "step": 121344 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.800489791746787e-05, |
| "loss": 4.1016, |
| "step": 121856 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.799651196995735e-05, |
| "loss": 4.0987, |
| "step": 122368 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.798812602244683e-05, |
| "loss": 4.0874, |
| "step": 122880 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.79797400749363e-05, |
| "loss": 4.0959, |
| "step": 123392 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7971370506229516e-05, |
| "loss": 4.0942, |
| "step": 123904 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7962984558718996e-05, |
| "loss": 4.1015, |
| "step": 124416 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7954598611208476e-05, |
| "loss": 4.1071, |
| "step": 124928 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7946212663697956e-05, |
| "loss": 4.0982, |
| "step": 125440 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.793784309499117e-05, |
| "loss": 4.086, |
| "step": 125952 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.792945714748065e-05, |
| "loss": 4.088, |
| "step": 126464 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.792107119997013e-05, |
| "loss": 4.0892, |
| "step": 126976 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7912685252459605e-05, |
| "loss": 4.0973, |
| "step": 127488 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7904299304949085e-05, |
| "loss": 4.0829, |
| "step": 128000 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.78959297362423e-05, |
| "loss": 4.0843, |
| "step": 128512 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7887543788731774e-05, |
| "loss": 4.077, |
| "step": 129024 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7879157841221254e-05, |
| "loss": 4.0957, |
| "step": 129536 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7870771893710734e-05, |
| "loss": 4.0656, |
| "step": 130048 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7862385946200214e-05, |
| "loss": 4.0803, |
| "step": 130560 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.785401637749343e-05, |
| "loss": 4.0852, |
| "step": 131072 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.784563042998291e-05, |
| "loss": 4.0736, |
| "step": 131584 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.783724448247239e-05, |
| "loss": 4.0738, |
| "step": 132096 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.782885853496187e-05, |
| "loss": 4.083, |
| "step": 132608 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.782048896625508e-05, |
| "loss": 4.073, |
| "step": 133120 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.781210301874456e-05, |
| "loss": 4.0778, |
| "step": 133632 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.780371707123404e-05, |
| "loss": 4.0805, |
| "step": 134144 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.779533112372352e-05, |
| "loss": 4.0713, |
| "step": 134656 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.778696155501673e-05, |
| "loss": 4.079, |
| "step": 135168 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.777857560750621e-05, |
| "loss": 4.0847, |
| "step": 135680 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7770206038799423e-05, |
| "loss": 4.0737, |
| "step": 136192 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7761820091288903e-05, |
| "loss": 4.0665, |
| "step": 136704 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7753434143778383e-05, |
| "loss": 4.0716, |
| "step": 137216 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.774504819626786e-05, |
| "loss": 4.0716, |
| "step": 137728 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.773666224875734e-05, |
| "loss": 4.0648, |
| "step": 138240 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.772827630124682e-05, |
| "loss": 4.0603, |
| "step": 138752 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.77198903537363e-05, |
| "loss": 4.063, |
| "step": 139264 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.771150440622578e-05, |
| "loss": 4.0758, |
| "step": 139776 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.770313483751899e-05, |
| "loss": 4.0663, |
| "step": 140288 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.769474889000847e-05, |
| "loss": 4.0652, |
| "step": 140800 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.768636294249795e-05, |
| "loss": 4.0562, |
| "step": 141312 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.767797699498743e-05, |
| "loss": 4.0592, |
| "step": 141824 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.766960742628064e-05, |
| "loss": 4.056, |
| "step": 142336 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.766122147877012e-05, |
| "loss": 4.073, |
| "step": 142848 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.76528355312596e-05, |
| "loss": 4.0694, |
| "step": 143360 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.764444958374909e-05, |
| "loss": 4.0453, |
| "step": 143872 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.76360800150423e-05, |
| "loss": 4.0531, |
| "step": 144384 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.762769406753178e-05, |
| "loss": 4.0651, |
| "step": 144896 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.761930812002126e-05, |
| "loss": 4.0597, |
| "step": 145408 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.761092217251074e-05, |
| "loss": 4.0652, |
| "step": 145920 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7602552603803946e-05, |
| "loss": 4.0512, |
| "step": 146432 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7594166656293426e-05, |
| "loss": 4.0585, |
| "step": 146944 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7585780708782906e-05, |
| "loss": 4.0613, |
| "step": 147456 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7577394761272386e-05, |
| "loss": 4.0534, |
| "step": 147968 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7569008813761866e-05, |
| "loss": 4.0516, |
| "step": 148480 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7560639245055075e-05, |
| "loss": 4.0562, |
| "step": 148992 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7552253297544555e-05, |
| "loss": 4.0462, |
| "step": 149504 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.7543867350034035e-05, |
| "loss": 4.0457, |
| "step": 150016 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.753548140252352e-05, |
| "loss": 4.06, |
| "step": 150528 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.752711183381673e-05, |
| "loss": 4.0409, |
| "step": 151040 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.751874226510994e-05, |
| "loss": 4.0463, |
| "step": 151552 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.751035631759942e-05, |
| "loss": 4.0508, |
| "step": 152064 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.75019703700889e-05, |
| "loss": 4.0455, |
| "step": 152576 |
| }, |
| { |
| "epoch": 1.03, |
| "eval_loss": 4.04611349105835, |
| "eval_runtime": 306.1519, |
| "eval_samples_per_second": 1246.411, |
| "eval_steps_per_second": 38.951, |
| "step": 152640 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.749358442257838e-05, |
| "loss": 4.0283, |
| "step": 153088 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.748519847506786e-05, |
| "loss": 4.0347, |
| "step": 153600 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.747681252755734e-05, |
| "loss": 4.0529, |
| "step": 154112 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.746842658004682e-05, |
| "loss": 4.0345, |
| "step": 154624 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.74600406325363e-05, |
| "loss": 4.0531, |
| "step": 155136 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.745165468502578e-05, |
| "loss": 4.0278, |
| "step": 155648 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.744326873751526e-05, |
| "loss": 4.0451, |
| "step": 156160 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.743488279000474e-05, |
| "loss": 4.0256, |
| "step": 156672 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.742649684249422e-05, |
| "loss": 4.0388, |
| "step": 157184 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.74181108949837e-05, |
| "loss": 4.0354, |
| "step": 157696 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.740972494747318e-05, |
| "loss": 4.0332, |
| "step": 158208 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.740135537876639e-05, |
| "loss": 4.0432, |
| "step": 158720 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.739300218886333e-05, |
| "loss": 4.0248, |
| "step": 159232 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.738461624135281e-05, |
| "loss": 4.0242, |
| "step": 159744 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.737623029384229e-05, |
| "loss": 4.0223, |
| "step": 160256 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.736784434633177e-05, |
| "loss": 4.0183, |
| "step": 160768 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.735945839882125e-05, |
| "loss": 4.021, |
| "step": 161280 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.735107245131073e-05, |
| "loss": 4.0269, |
| "step": 161792 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.734268650380021e-05, |
| "loss": 4.0204, |
| "step": 162304 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.733430055628969e-05, |
| "loss": 4.0457, |
| "step": 162816 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.732591460877917e-05, |
| "loss": 4.0241, |
| "step": 163328 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.731752866126865e-05, |
| "loss": 4.0303, |
| "step": 163840 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.730914271375813e-05, |
| "loss": 4.0238, |
| "step": 164352 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.730077314505134e-05, |
| "loss": 4.0295, |
| "step": 164864 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.729238719754082e-05, |
| "loss": 4.0212, |
| "step": 165376 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.72840012500303e-05, |
| "loss": 4.0171, |
| "step": 165888 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.727561530251978e-05, |
| "loss": 4.0213, |
| "step": 166400 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.726722935500926e-05, |
| "loss": 4.0112, |
| "step": 166912 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.725885978630247e-05, |
| "loss": 3.9993, |
| "step": 167424 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.725047383879195e-05, |
| "loss": 4.0125, |
| "step": 167936 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.724208789128143e-05, |
| "loss": 4.0211, |
| "step": 168448 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7233718322574647e-05, |
| "loss": 4.0172, |
| "step": 168960 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7225332375064127e-05, |
| "loss": 4.0157, |
| "step": 169472 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7216946427553606e-05, |
| "loss": 4.0187, |
| "step": 169984 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7208560480043086e-05, |
| "loss": 4.0067, |
| "step": 170496 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7200190911336296e-05, |
| "loss": 4.006, |
| "step": 171008 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7191804963825775e-05, |
| "loss": 3.9989, |
| "step": 171520 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7183419016315255e-05, |
| "loss": 4.0015, |
| "step": 172032 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7175033068804735e-05, |
| "loss": 3.998, |
| "step": 172544 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7166663500097945e-05, |
| "loss": 4.0005, |
| "step": 173056 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7158277552587424e-05, |
| "loss": 4.0049, |
| "step": 173568 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7149891605076904e-05, |
| "loss": 4.0124, |
| "step": 174080 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7141505657566384e-05, |
| "loss": 4.0008, |
| "step": 174592 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.71331360888596e-05, |
| "loss": 4.0052, |
| "step": 175104 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.712475014134908e-05, |
| "loss": 4.0087, |
| "step": 175616 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.711636419383856e-05, |
| "loss": 3.9909, |
| "step": 176128 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.710797824632804e-05, |
| "loss": 4.0026, |
| "step": 176640 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.709960867762125e-05, |
| "loss": 3.9968, |
| "step": 177152 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.709122273011073e-05, |
| "loss": 3.9862, |
| "step": 177664 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.708283678260021e-05, |
| "loss": 4.0021, |
| "step": 178176 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.707445083508969e-05, |
| "loss": 3.9938, |
| "step": 178688 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.706609764518663e-05, |
| "loss": 3.9906, |
| "step": 179200 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.705771169767611e-05, |
| "loss": 3.9861, |
| "step": 179712 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.704932575016559e-05, |
| "loss": 3.9891, |
| "step": 180224 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7040939802655074e-05, |
| "loss": 3.9741, |
| "step": 180736 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7032553855144554e-05, |
| "loss": 4.0014, |
| "step": 181248 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7024167907634034e-05, |
| "loss": 3.9855, |
| "step": 181760 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7015781960123514e-05, |
| "loss": 3.9965, |
| "step": 182272 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.7007396012612994e-05, |
| "loss": 3.9991, |
| "step": 182784 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.69990264439062e-05, |
| "loss": 3.9802, |
| "step": 183296 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.699064049639568e-05, |
| "loss": 3.9723, |
| "step": 183808 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.698225454888516e-05, |
| "loss": 3.9901, |
| "step": 184320 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.697386860137464e-05, |
| "loss": 3.9767, |
| "step": 184832 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.696549903266785e-05, |
| "loss": 3.9713, |
| "step": 185344 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.695711308515733e-05, |
| "loss": 3.9921, |
| "step": 185856 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.694872713764681e-05, |
| "loss": 3.9876, |
| "step": 186368 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.694034119013629e-05, |
| "loss": 3.9735, |
| "step": 186880 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.693197162142951e-05, |
| "loss": 3.9691, |
| "step": 187392 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.692358567391899e-05, |
| "loss": 3.9721, |
| "step": 187904 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.691519972640847e-05, |
| "loss": 3.979, |
| "step": 188416 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.690681377889795e-05, |
| "loss": 3.9864, |
| "step": 188928 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6898444210191157e-05, |
| "loss": 3.9797, |
| "step": 189440 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6890058262680636e-05, |
| "loss": 3.988, |
| "step": 189952 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6881672315170116e-05, |
| "loss": 3.9815, |
| "step": 190464 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6873286367659596e-05, |
| "loss": 3.9879, |
| "step": 190976 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6864916798952806e-05, |
| "loss": 3.9702, |
| "step": 191488 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6856530851442285e-05, |
| "loss": 3.9785, |
| "step": 192000 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6848144903931765e-05, |
| "loss": 3.9801, |
| "step": 192512 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.6839758956421245e-05, |
| "loss": 3.9659, |
| "step": 193024 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.683138938771446e-05, |
| "loss": 3.9765, |
| "step": 193536 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.682300344020394e-05, |
| "loss": 3.9825, |
| "step": 194048 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.681461749269342e-05, |
| "loss": 3.9821, |
| "step": 194560 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.68062315451829e-05, |
| "loss": 3.9655, |
| "step": 195072 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.679786197647611e-05, |
| "loss": 3.9529, |
| "step": 195584 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.678947602896559e-05, |
| "loss": 3.9726, |
| "step": 196096 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.678109008145507e-05, |
| "loss": 3.9654, |
| "step": 196608 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.677270413394455e-05, |
| "loss": 3.9769, |
| "step": 197120 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.676433456523776e-05, |
| "loss": 3.9685, |
| "step": 197632 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.675596499653097e-05, |
| "loss": 3.9687, |
| "step": 198144 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.674757904902045e-05, |
| "loss": 3.9671, |
| "step": 198656 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6739193101509935e-05, |
| "loss": 3.956, |
| "step": 199168 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6730807153999415e-05, |
| "loss": 3.9638, |
| "step": 199680 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6722421206488895e-05, |
| "loss": 3.9624, |
| "step": 200192 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6714035258978375e-05, |
| "loss": 3.9756, |
| "step": 200704 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6705649311467855e-05, |
| "loss": 3.9805, |
| "step": 201216 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6697263363957335e-05, |
| "loss": 3.9688, |
| "step": 201728 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6688893795250544e-05, |
| "loss": 3.9541, |
| "step": 202240 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6680507847740024e-05, |
| "loss": 3.963, |
| "step": 202752 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6672121900229504e-05, |
| "loss": 3.9609, |
| "step": 203264 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6663735952718984e-05, |
| "loss": 3.9698, |
| "step": 203776 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.665536638401219e-05, |
| "loss": 3.9569, |
| "step": 204288 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.664698043650167e-05, |
| "loss": 3.9606, |
| "step": 204800 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.663859448899115e-05, |
| "loss": 3.9511, |
| "step": 205312 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.663020854148063e-05, |
| "loss": 3.971, |
| "step": 205824 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.662183897277385e-05, |
| "loss": 3.9422, |
| "step": 206336 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.661345302526333e-05, |
| "loss": 3.9585, |
| "step": 206848 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.660508345655654e-05, |
| "loss": 3.9545, |
| "step": 207360 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.659669750904602e-05, |
| "loss": 3.9569, |
| "step": 207872 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.65883115615355e-05, |
| "loss": 3.9498, |
| "step": 208384 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.657992561402498e-05, |
| "loss": 3.9613, |
| "step": 208896 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.657153966651446e-05, |
| "loss": 3.9568, |
| "step": 209408 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.656315371900393e-05, |
| "loss": 3.9521, |
| "step": 209920 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.655476777149341e-05, |
| "loss": 3.9605, |
| "step": 210432 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6546398202786626e-05, |
| "loss": 3.9479, |
| "step": 210944 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6538012255276106e-05, |
| "loss": 3.9643, |
| "step": 211456 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6529626307765586e-05, |
| "loss": 3.9694, |
| "step": 211968 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.65212567390588e-05, |
| "loss": 3.9522, |
| "step": 212480 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.651287079154828e-05, |
| "loss": 3.9504, |
| "step": 212992 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.650448484403776e-05, |
| "loss": 3.9506, |
| "step": 213504 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6496098896527235e-05, |
| "loss": 3.9528, |
| "step": 214016 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6487712949016715e-05, |
| "loss": 3.9464, |
| "step": 214528 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6479327001506195e-05, |
| "loss": 3.9465, |
| "step": 215040 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6470941053995675e-05, |
| "loss": 3.943, |
| "step": 215552 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6462555106485155e-05, |
| "loss": 3.9605, |
| "step": 216064 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6454185537778364e-05, |
| "loss": 3.9515, |
| "step": 216576 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6445799590267844e-05, |
| "loss": 3.9489, |
| "step": 217088 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6437413642757324e-05, |
| "loss": 3.941, |
| "step": 217600 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6429027695246804e-05, |
| "loss": 3.9455, |
| "step": 218112 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.642065812654002e-05, |
| "loss": 3.9401, |
| "step": 218624 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.64122721790295e-05, |
| "loss": 3.9599, |
| "step": 219136 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.640390261032271e-05, |
| "loss": 3.9559, |
| "step": 219648 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.639551666281219e-05, |
| "loss": 3.9361, |
| "step": 220160 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.638713071530167e-05, |
| "loss": 3.9362, |
| "step": 220672 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.637874476779115e-05, |
| "loss": 3.9511, |
| "step": 221184 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.637035882028063e-05, |
| "loss": 3.9445, |
| "step": 221696 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.636197287277011e-05, |
| "loss": 3.9571, |
| "step": 222208 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.635358692525959e-05, |
| "loss": 3.9359, |
| "step": 222720 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.634520097774907e-05, |
| "loss": 3.9516, |
| "step": 223232 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.633683140904228e-05, |
| "loss": 3.945, |
| "step": 223744 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.632844546153176e-05, |
| "loss": 3.9465, |
| "step": 224256 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6320059514021244e-05, |
| "loss": 3.9431, |
| "step": 224768 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6311673566510724e-05, |
| "loss": 3.9447, |
| "step": 225280 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6303303997803934e-05, |
| "loss": 3.9362, |
| "step": 225792 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6294918050293413e-05, |
| "loss": 3.9393, |
| "step": 226304 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6286532102782893e-05, |
| "loss": 3.9497, |
| "step": 226816 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.6278146155272373e-05, |
| "loss": 3.9389, |
| "step": 227328 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.626977658656558e-05, |
| "loss": 3.9349, |
| "step": 227840 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.626139063905506e-05, |
| "loss": 3.9408, |
| "step": 228352 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.625300469154454e-05, |
| "loss": 3.9381, |
| "step": 228864 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 3.9720582962036133, |
| "eval_runtime": 303.5227, |
| "eval_samples_per_second": 1257.207, |
| "eval_steps_per_second": 39.289, |
| "step": 228960 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.624461874403402e-05, |
| "loss": 3.9291, |
| "step": 229376 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.623624917532723e-05, |
| "loss": 3.9265, |
| "step": 229888 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.622786322781671e-05, |
| "loss": 3.947, |
| "step": 230400 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.62194772803062e-05, |
| "loss": 3.9296, |
| "step": 230912 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.621109133279568e-05, |
| "loss": 3.9511, |
| "step": 231424 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.620272176408889e-05, |
| "loss": 3.9187, |
| "step": 231936 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.619433581657837e-05, |
| "loss": 3.9413, |
| "step": 232448 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.618594986906785e-05, |
| "loss": 3.9247, |
| "step": 232960 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.617756392155733e-05, |
| "loss": 3.9298, |
| "step": 233472 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6169194352850536e-05, |
| "loss": 3.9343, |
| "step": 233984 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6160808405340016e-05, |
| "loss": 3.9274, |
| "step": 234496 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6152422457829496e-05, |
| "loss": 3.9388, |
| "step": 235008 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6144069267926434e-05, |
| "loss": 3.9238, |
| "step": 235520 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.613568332041592e-05, |
| "loss": 3.9195, |
| "step": 236032 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.61272973729054e-05, |
| "loss": 3.9241, |
| "step": 236544 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.611891142539488e-05, |
| "loss": 3.9122, |
| "step": 237056 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.611052547788436e-05, |
| "loss": 3.922, |
| "step": 237568 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.610213953037384e-05, |
| "loss": 3.9251, |
| "step": 238080 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.609375358286332e-05, |
| "loss": 3.9186, |
| "step": 238592 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.60853676353528e-05, |
| "loss": 3.9435, |
| "step": 239104 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.607698168784228e-05, |
| "loss": 3.9275, |
| "step": 239616 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.606859574033176e-05, |
| "loss": 3.9341, |
| "step": 240128 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.606020979282124e-05, |
| "loss": 3.9226, |
| "step": 240640 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6051823845310714e-05, |
| "loss": 3.9275, |
| "step": 241152 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.604345427660393e-05, |
| "loss": 3.9205, |
| "step": 241664 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.603506832909341e-05, |
| "loss": 3.9201, |
| "step": 242176 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.602668238158289e-05, |
| "loss": 3.9226, |
| "step": 242688 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.60183128128761e-05, |
| "loss": 3.9122, |
| "step": 243200 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.6009926865365585e-05, |
| "loss": 3.9083, |
| "step": 243712 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.6001540917855065e-05, |
| "loss": 3.9116, |
| "step": 244224 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.599315497034454e-05, |
| "loss": 3.9243, |
| "step": 244736 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5984785401637754e-05, |
| "loss": 3.9214, |
| "step": 245248 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5976399454127234e-05, |
| "loss": 3.922, |
| "step": 245760 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5968013506616714e-05, |
| "loss": 3.9253, |
| "step": 246272 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5959643937909923e-05, |
| "loss": 3.9054, |
| "step": 246784 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5951257990399403e-05, |
| "loss": 3.9136, |
| "step": 247296 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.594287204288888e-05, |
| "loss": 3.9089, |
| "step": 247808 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.593448609537836e-05, |
| "loss": 3.9012, |
| "step": 248320 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5926100147867836e-05, |
| "loss": 3.9058, |
| "step": 248832 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.591771420035732e-05, |
| "loss": 3.9048, |
| "step": 249344 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.590934463165054e-05, |
| "loss": 3.9135, |
| "step": 249856 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.590095868414001e-05, |
| "loss": 3.9151, |
| "step": 250368 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.589257273662949e-05, |
| "loss": 3.9127, |
| "step": 250880 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.588418678911897e-05, |
| "loss": 3.9061, |
| "step": 251392 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.587580084160845e-05, |
| "loss": 3.9179, |
| "step": 251904 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.586741489409793e-05, |
| "loss": 3.8995, |
| "step": 252416 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.585904532539114e-05, |
| "loss": 3.9107, |
| "step": 252928 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.585065937788062e-05, |
| "loss": 3.9055, |
| "step": 253440 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.58422734303701e-05, |
| "loss": 3.8928, |
| "step": 253952 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.583388748285958e-05, |
| "loss": 3.9142, |
| "step": 254464 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.582550153534906e-05, |
| "loss": 3.902, |
| "step": 254976 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.581711558783854e-05, |
| "loss": 3.9021, |
| "step": 255488 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.580874601913176e-05, |
| "loss": 3.893, |
| "step": 256000 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.580036007162124e-05, |
| "loss": 3.8982, |
| "step": 256512 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.579197412411072e-05, |
| "loss": 3.8843, |
| "step": 257024 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.57835881766002e-05, |
| "loss": 3.9066, |
| "step": 257536 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.577520222908968e-05, |
| "loss": 3.8999, |
| "step": 258048 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.576681628157916e-05, |
| "loss": 3.9031, |
| "step": 258560 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5758430334068637e-05, |
| "loss": 3.9092, |
| "step": 259072 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5750060765361846e-05, |
| "loss": 3.8931, |
| "step": 259584 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5741691196655055e-05, |
| "loss": 3.882, |
| "step": 260096 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5733305249144535e-05, |
| "loss": 3.904, |
| "step": 260608 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5724919301634015e-05, |
| "loss": 3.8882, |
| "step": 261120 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5716533354123495e-05, |
| "loss": 3.8844, |
| "step": 261632 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5708147406612975e-05, |
| "loss": 3.9049, |
| "step": 262144 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.569976145910246e-05, |
| "loss": 3.8985, |
| "step": 262656 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.569137551159194e-05, |
| "loss": 3.8881, |
| "step": 263168 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.568298956408142e-05, |
| "loss": 3.8861, |
| "step": 263680 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.567461999537463e-05, |
| "loss": 3.8798, |
| "step": 264192 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.566623404786411e-05, |
| "loss": 3.8905, |
| "step": 264704 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.565784810035359e-05, |
| "loss": 3.9019, |
| "step": 265216 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.56494785316468e-05, |
| "loss": 3.8974, |
| "step": 265728 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.564109258413628e-05, |
| "loss": 3.9009, |
| "step": 266240 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.563270663662576e-05, |
| "loss": 3.8929, |
| "step": 266752 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.562432068911524e-05, |
| "loss": 3.9086, |
| "step": 267264 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.561593474160472e-05, |
| "loss": 3.8838, |
| "step": 267776 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.56075487940942e-05, |
| "loss": 3.8955, |
| "step": 268288 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.559916284658368e-05, |
| "loss": 3.8933, |
| "step": 268800 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5590793277876895e-05, |
| "loss": 3.8775, |
| "step": 269312 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.5582407330366375e-05, |
| "loss": 3.8993, |
| "step": 269824 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.557402138285585e-05, |
| "loss": 3.8966, |
| "step": 270336 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.556563543534533e-05, |
| "loss": 3.8931, |
| "step": 270848 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.555724948783481e-05, |
| "loss": 3.8829, |
| "step": 271360 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.554886354032429e-05, |
| "loss": 3.8743, |
| "step": 271872 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.55404939716175e-05, |
| "loss": 3.8874, |
| "step": 272384 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.553210802410698e-05, |
| "loss": 3.888, |
| "step": 272896 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.552372207659646e-05, |
| "loss": 3.8909, |
| "step": 273408 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.551533612908594e-05, |
| "loss": 3.8807, |
| "step": 273920 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 4.550695018157542e-05, |
| "loss": 3.8897, |
| "step": 274432 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.549858061286863e-05, |
| "loss": 3.8825, |
| "step": 274944 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.549019466535811e-05, |
| "loss": 3.8793, |
| "step": 275456 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.548180871784759e-05, |
| "loss": 3.8804, |
| "step": 275968 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.547342277033707e-05, |
| "loss": 3.8775, |
| "step": 276480 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.546505320163028e-05, |
| "loss": 3.8967, |
| "step": 276992 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.545666725411976e-05, |
| "loss": 3.8953, |
| "step": 277504 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.544828130660924e-05, |
| "loss": 3.8939, |
| "step": 278016 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.543989535909872e-05, |
| "loss": 3.8717, |
| "step": 278528 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.54315094115882e-05, |
| "loss": 3.8851, |
| "step": 279040 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.542313984288141e-05, |
| "loss": 3.8799, |
| "step": 279552 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.541475389537089e-05, |
| "loss": 3.8857, |
| "step": 280064 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.540636794786037e-05, |
| "loss": 3.8774, |
| "step": 280576 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.539798200034985e-05, |
| "loss": 3.8769, |
| "step": 281088 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5389612431643066e-05, |
| "loss": 3.8699, |
| "step": 281600 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5381226484132546e-05, |
| "loss": 3.8907, |
| "step": 282112 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5372840536622026e-05, |
| "loss": 3.8632, |
| "step": 282624 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5364454589111506e-05, |
| "loss": 3.879, |
| "step": 283136 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5356068641600986e-05, |
| "loss": 3.8781, |
| "step": 283648 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5347699072894195e-05, |
| "loss": 3.8771, |
| "step": 284160 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5339313125383675e-05, |
| "loss": 3.8704, |
| "step": 284672 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5330927177873155e-05, |
| "loss": 3.887, |
| "step": 285184 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5322541230362635e-05, |
| "loss": 3.8765, |
| "step": 285696 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5314171661655844e-05, |
| "loss": 3.8749, |
| "step": 286208 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5305785714145324e-05, |
| "loss": 3.8868, |
| "step": 286720 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5297399766634804e-05, |
| "loss": 3.8723, |
| "step": 287232 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5289013819124284e-05, |
| "loss": 3.8844, |
| "step": 287744 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.52806442504175e-05, |
| "loss": 3.891, |
| "step": 288256 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.527225830290698e-05, |
| "loss": 3.878, |
| "step": 288768 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.526387235539646e-05, |
| "loss": 3.8731, |
| "step": 289280 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.525548640788594e-05, |
| "loss": 3.8714, |
| "step": 289792 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.524711683917915e-05, |
| "loss": 3.8734, |
| "step": 290304 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.523873089166863e-05, |
| "loss": 3.8704, |
| "step": 290816 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.523034494415811e-05, |
| "loss": 3.8772, |
| "step": 291328 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.522195899664759e-05, |
| "loss": 3.8629, |
| "step": 291840 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.521357304913707e-05, |
| "loss": 3.8834, |
| "step": 292352 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.520520348043028e-05, |
| "loss": 3.8785, |
| "step": 292864 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.519681753291976e-05, |
| "loss": 3.8753, |
| "step": 293376 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.518843158540924e-05, |
| "loss": 3.8618, |
| "step": 293888 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5180045637898725e-05, |
| "loss": 3.8704, |
| "step": 294400 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5171676069191934e-05, |
| "loss": 3.8687, |
| "step": 294912 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5163290121681414e-05, |
| "loss": 3.88, |
| "step": 295424 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5154904174170894e-05, |
| "loss": 3.8827, |
| "step": 295936 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5146518226660373e-05, |
| "loss": 3.8678, |
| "step": 296448 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.513814865795358e-05, |
| "loss": 3.8562, |
| "step": 296960 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.512976271044306e-05, |
| "loss": 3.8768, |
| "step": 297472 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.512137676293254e-05, |
| "loss": 3.868, |
| "step": 297984 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.511299081542202e-05, |
| "loss": 3.8832, |
| "step": 298496 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.510462124671523e-05, |
| "loss": 3.8621, |
| "step": 299008 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.509623529920471e-05, |
| "loss": 3.8745, |
| "step": 299520 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.508784935169419e-05, |
| "loss": 3.8729, |
| "step": 300032 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.507946340418368e-05, |
| "loss": 3.8761, |
| "step": 300544 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.507107745667316e-05, |
| "loss": 3.8702, |
| "step": 301056 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.506270788796637e-05, |
| "loss": 3.8704, |
| "step": 301568 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.505432194045585e-05, |
| "loss": 3.8622, |
| "step": 302080 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.504593599294533e-05, |
| "loss": 3.8658, |
| "step": 302592 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.50375500454348e-05, |
| "loss": 3.8732, |
| "step": 303104 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5029180476728016e-05, |
| "loss": 3.8697, |
| "step": 303616 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5020794529217496e-05, |
| "loss": 3.8607, |
| "step": 304128 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.5012408581706976e-05, |
| "loss": 3.87, |
| "step": 304640 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 4.500402263419645e-05, |
| "loss": 3.8628, |
| "step": 305152 |
| }, |
| { |
| "epoch": 1.03, |
| "eval_loss": 3.9311435222625732, |
| "eval_runtime": 301.8823, |
| "eval_samples_per_second": 1264.039, |
| "eval_steps_per_second": 39.502, |
| "step": 305280 |
| } |
| ], |
| "logging_steps": 512, |
| "max_steps": 3052726, |
| "num_train_epochs": 9223372036854775807, |
| "save_steps": 10, |
| "total_flos": 2.1131261678112768e+17, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|