| { | |
| "best_metric": 3.9028751850128174, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/re-irr-sv-agr/transformer/2/checkpoints/checkpoint-381600", | |
| "epoch": 1.0250006060157382, | |
| "eval_steps": 10, | |
| "global_step": 381600, | |
| "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": 10.9339, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 6.7961, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 6.1583, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 5.9557, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 5.7951, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 5.696, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 5.5911, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994129836742636e-05, | |
| "loss": 5.5362, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993291241991584e-05, | |
| "loss": 5.4416, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992452647240532e-05, | |
| "loss": 5.4026, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99161405248948e-05, | |
| "loss": 5.3441, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990775457738428e-05, | |
| "loss": 5.3218, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989936862987376e-05, | |
| "loss": 5.2749, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989099906116697e-05, | |
| "loss": 5.2218, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988261311365645e-05, | |
| "loss": 5.1892, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987422716614593e-05, | |
| "loss": 5.1384, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986584121863541e-05, | |
| "loss": 5.1165, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985745527112489e-05, | |
| "loss": 5.0927, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984906932361437e-05, | |
| "loss": 5.0637, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984068337610385e-05, | |
| "loss": 5.0354, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.0224, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9823927859886547e-05, | |
| "loss": 4.9923, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9815541912376026e-05, | |
| "loss": 4.9717, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9807155964865506e-05, | |
| "loss": 4.9355, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9798770017354986e-05, | |
| "loss": 4.9347, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9790400448648195e-05, | |
| "loss": 4.9029, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9782014501137675e-05, | |
| "loss": 4.8898, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773628553627155e-05, | |
| "loss": 4.8709, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9765242606116635e-05, | |
| "loss": 4.845, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9756873037409844e-05, | |
| "loss": 4.8283, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748487089899324e-05, | |
| "loss": 4.8102, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9740101142388804e-05, | |
| "loss": 4.8038, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9731715194878284e-05, | |
| "loss": 4.7855, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.97233456261715e-05, | |
| "loss": 4.7808, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.971495967866098e-05, | |
| "loss": 4.7658, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.970657373115046e-05, | |
| "loss": 4.7542, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.969820416244367e-05, | |
| "loss": 4.7417, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968981821493315e-05, | |
| "loss": 4.7058, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968143226742263e-05, | |
| "loss": 4.7041, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.967304631991211e-05, | |
| "loss": 4.693, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966466037240159e-05, | |
| "loss": 4.6758, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965627442489107e-05, | |
| "loss": 4.6793, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964790485618428e-05, | |
| "loss": 4.6739, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963951890867376e-05, | |
| "loss": 4.6516, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963113296116324e-05, | |
| "loss": 4.6505, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962274701365272e-05, | |
| "loss": 4.6436, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96143610661422e-05, | |
| "loss": 4.6353, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.960597511863168e-05, | |
| "loss": 4.6317, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.959758917112116e-05, | |
| "loss": 4.6123, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958920322361064e-05, | |
| "loss": 4.5983, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958081727610012e-05, | |
| "loss": 4.6005, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95724313285896e-05, | |
| "loss": 4.5863, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956406175988281e-05, | |
| "loss": 4.587, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955567581237229e-05, | |
| "loss": 4.5597, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9547306243665496e-05, | |
| "loss": 4.5737, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9538920296154976e-05, | |
| "loss": 4.5478, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9530534348644456e-05, | |
| "loss": 4.5683, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9522148401133936e-05, | |
| "loss": 4.543, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951376245362342e-05, | |
| "loss": 4.5395, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95053765061129e-05, | |
| "loss": 4.5312, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949699055860238e-05, | |
| "loss": 4.5263, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948860461109186e-05, | |
| "loss": 4.5128, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948023504238507e-05, | |
| "loss": 4.5184, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947184909487455e-05, | |
| "loss": 4.4938, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946346314736403e-05, | |
| "loss": 4.4876, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945509357865724e-05, | |
| "loss": 4.5015, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944670763114672e-05, | |
| "loss": 4.4921, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94383216836362e-05, | |
| "loss": 4.4802, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942993573612568e-05, | |
| "loss": 4.4651, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942154978861516e-05, | |
| "loss": 4.4673, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.941316384110464e-05, | |
| "loss": 4.4738, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.940477789359412e-05, | |
| "loss": 4.468, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9396408324887336e-05, | |
| "loss": 4.4601, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388022377376816e-05, | |
| "loss": 4.4587, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379636429866296e-05, | |
| "loss": 4.46, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371250482355776e-05, | |
| "loss": 4.4524, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362880913648985e-05, | |
| "loss": 4.4301, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9354494966138465e-05, | |
| "loss": 4.4314, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9346109018627945e-05, | |
| "loss": 4.4306, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9337723071117425e-05, | |
| "loss": 4.4217, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329337123606905e-05, | |
| "loss": 4.4218, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932095117609638e-05, | |
| "loss": 4.4141, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.931256522858586e-05, | |
| "loss": 4.4306, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9304195659879074e-05, | |
| "loss": 4.4069, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9295809712368554e-05, | |
| "loss": 4.3908, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9287423764858034e-05, | |
| "loss": 4.4015, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9279037817347514e-05, | |
| "loss": 4.3978, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9270651869836994e-05, | |
| "loss": 4.4028, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9262265922326474e-05, | |
| "loss": 4.3857, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.925389635361968e-05, | |
| "loss": 4.3858, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.924551040610916e-05, | |
| "loss": 4.384, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923712445859864e-05, | |
| "loss": 4.3776, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922873851108812e-05, | |
| "loss": 4.3665, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.92203525635776e-05, | |
| "loss": 4.3688, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921196661606708e-05, | |
| "loss": 4.3716, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920359704736029e-05, | |
| "loss": 4.3789, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919521109984978e-05, | |
| "loss": 4.3576, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918682515233926e-05, | |
| "loss": 4.3496, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917843920482874e-05, | |
| "loss": 4.3499, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917005325731822e-05, | |
| "loss": 4.3495, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.91616673098077e-05, | |
| "loss": 4.3549, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.915328136229718e-05, | |
| "loss": 4.3403, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.914489541478666e-05, | |
| "loss": 4.3343, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.913652584607987e-05, | |
| "loss": 4.3277, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.912813989856935e-05, | |
| "loss": 4.3382, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.911975395105883e-05, | |
| "loss": 4.3062, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.911136800354831e-05, | |
| "loss": 4.3264, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9102998434841516e-05, | |
| "loss": 4.3179, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9094612487330996e-05, | |
| "loss": 4.3246, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9086226539820476e-05, | |
| "loss": 4.3134, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907784059230996e-05, | |
| "loss": 4.3177, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906945464479944e-05, | |
| "loss": 4.3067, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906106869728892e-05, | |
| "loss": 4.3072, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905269912858213e-05, | |
| "loss": 4.3117, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904431318107161e-05, | |
| "loss": 4.2927, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903592723356109e-05, | |
| "loss": 4.3113, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9027541286050565e-05, | |
| "loss": 4.3076, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901918809614751e-05, | |
| "loss": 4.3007, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901080214863699e-05, | |
| "loss": 4.2902, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.900241620112647e-05, | |
| "loss": 4.2891, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899403025361595e-05, | |
| "loss": 4.2905, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.898564430610543e-05, | |
| "loss": 4.2878, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.897725835859491e-05, | |
| "loss": 4.287, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8968872411084396e-05, | |
| "loss": 4.2768, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.896048646357387e-05, | |
| "loss": 4.2846, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.895210051606335e-05, | |
| "loss": 4.2669, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.894371456855283e-05, | |
| "loss": 4.2704, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.893532862104231e-05, | |
| "loss": 4.2597, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.892695905233552e-05, | |
| "loss": 4.2633, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918573104825e-05, | |
| "loss": 4.2529, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.891018715731448e-05, | |
| "loss": 4.2745, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890180120980396e-05, | |
| "loss": 4.2769, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889341526229344e-05, | |
| "loss": 4.2628, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888502931478292e-05, | |
| "loss": 4.2435, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88766433672724e-05, | |
| "loss": 4.2621, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.886825741976188e-05, | |
| "loss": 4.2561, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8859887851055094e-05, | |
| "loss": 4.253, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88515182823483e-05, | |
| "loss": 4.2448, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884313233483778e-05, | |
| "loss": 4.2504, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883474638732726e-05, | |
| "loss": 4.2454, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882636043981674e-05, | |
| "loss": 4.241, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881797449230622e-05, | |
| "loss": 4.2422, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88095885447957e-05, | |
| "loss": 4.2441, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880120259728518e-05, | |
| "loss": 4.2301, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879283302857839e-05, | |
| "loss": 4.2319, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.878444708106787e-05, | |
| "loss": 4.2429, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877606113355735e-05, | |
| "loss": 4.2291, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.876767518604683e-05, | |
| "loss": 4.2285, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875928923853632e-05, | |
| "loss": 4.2395, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.87509032910258e-05, | |
| "loss": 4.2207, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.2148919105529785, | |
| "eval_runtime": 304.5041, | |
| "eval_samples_per_second": 1253.155, | |
| "eval_steps_per_second": 39.162, | |
| "step": 76320 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.874251734351528e-05, | |
| "loss": 4.2011, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.873413139600475e-05, | |
| "loss": 4.212, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.872574544849423e-05, | |
| "loss": 4.2259, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.871735950098371e-05, | |
| "loss": 4.2144, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.870897355347319e-05, | |
| "loss": 4.2212, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.870058760596267e-05, | |
| "loss": 4.2, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.869220165845215e-05, | |
| "loss": 4.2117, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.868381571094163e-05, | |
| "loss": 4.1911, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.867542976343111e-05, | |
| "loss": 4.2081, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.866704381592059e-05, | |
| "loss": 4.192, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.865865786841007e-05, | |
| "loss": 4.2064, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.865027192089955e-05, | |
| "loss": 4.2017, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.864190235219277e-05, | |
| "loss": 4.1937, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.863351640468225e-05, | |
| "loss": 4.1842, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.862513045717173e-05, | |
| "loss": 4.1761, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.861674450966121e-05, | |
| "loss": 4.1766, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8608374940954416e-05, | |
| "loss": 4.1844, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8599988993443896e-05, | |
| "loss": 4.1834, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8591603045933376e-05, | |
| "loss": 4.1753, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8583217098422856e-05, | |
| "loss": 4.1977, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8574847529716065e-05, | |
| "loss": 4.1866, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8566461582205545e-05, | |
| "loss": 4.1784, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8558075634695025e-05, | |
| "loss": 4.1716, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8549689687184505e-05, | |
| "loss": 4.1891, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8541303739673985e-05, | |
| "loss": 4.162, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8532917792163465e-05, | |
| "loss": 4.175, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8524531844652945e-05, | |
| "loss": 4.1718, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.8516145897142425e-05, | |
| "loss": 4.1649, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.850777632843564e-05, | |
| "loss": 4.1566, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8499390380925114e-05, | |
| "loss": 4.1548, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.849102081221833e-05, | |
| "loss": 4.1611, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.848263486470781e-05, | |
| "loss": 4.1593, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.847424891719729e-05, | |
| "loss": 4.1606, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.846586296968676e-05, | |
| "loss": 4.1602, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.845747702217624e-05, | |
| "loss": 4.1594, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.844909107466572e-05, | |
| "loss": 4.1543, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.844072150595894e-05, | |
| "loss": 4.1362, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.843233555844842e-05, | |
| "loss": 4.1444, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.84239496109379e-05, | |
| "loss": 4.138, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.841556366342738e-05, | |
| "loss": 4.1334, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.840719409472059e-05, | |
| "loss": 4.1436, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.839880814721007e-05, | |
| "loss": 4.1456, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.839042219969955e-05, | |
| "loss": 4.1335, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.838203625218903e-05, | |
| "loss": 4.1326, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8373666683482236e-05, | |
| "loss": 4.1365, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8365280735971716e-05, | |
| "loss": 4.1362, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8356894788461196e-05, | |
| "loss": 4.1386, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8348508840950676e-05, | |
| "loss": 4.1276, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.834013927224389e-05, | |
| "loss": 4.1241, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.833175332473337e-05, | |
| "loss": 4.1293, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.832336737722285e-05, | |
| "loss": 4.1215, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.831498142971233e-05, | |
| "loss": 4.127, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.830659548220181e-05, | |
| "loss": 4.1133, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.829822591349502e-05, | |
| "loss": 4.1268, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.82898399659845e-05, | |
| "loss": 4.1025, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.828145401847398e-05, | |
| "loss": 4.1272, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.827306807096346e-05, | |
| "loss": 4.1137, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.826469850225667e-05, | |
| "loss": 4.1114, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.825631255474615e-05, | |
| "loss": 4.1201, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.824792660723563e-05, | |
| "loss": 4.1086, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.823954065972511e-05, | |
| "loss": 4.1014, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8231171091018326e-05, | |
| "loss": 4.1158, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8222785143507806e-05, | |
| "loss": 4.0902, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8214399195997286e-05, | |
| "loss": 4.0912, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8206013248486766e-05, | |
| "loss": 4.1074, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8197643679779975e-05, | |
| "loss": 4.1053, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8189257732269455e-05, | |
| "loss": 4.0985, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8180871784758935e-05, | |
| "loss": 4.0878, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8172485837248415e-05, | |
| "loss": 4.0884, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8164116268541624e-05, | |
| "loss": 4.0957, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8155730321031104e-05, | |
| "loss": 4.1038, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8147344373520584e-05, | |
| "loss": 4.0941, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8138958426010064e-05, | |
| "loss": 4.0994, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.813057247849955e-05, | |
| "loss": 4.1002, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.812218653098903e-05, | |
| "loss": 4.098, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.811381696228224e-05, | |
| "loss": 4.08, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.810543101477172e-05, | |
| "loss": 4.0841, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.80970450672612e-05, | |
| "loss": 4.091, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.808865911975068e-05, | |
| "loss": 4.0803, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.808028955104389e-05, | |
| "loss": 4.0854, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.807190360353337e-05, | |
| "loss": 4.0773, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.806351765602285e-05, | |
| "loss": 4.0987, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.805513170851233e-05, | |
| "loss": 4.0801, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.804676213980554e-05, | |
| "loss": 4.0679, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.803837619229502e-05, | |
| "loss": 4.0769, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8029990244784504e-05, | |
| "loss": 4.0793, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8021604297273984e-05, | |
| "loss": 4.085, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.801323472856719e-05, | |
| "loss": 4.0706, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.800484878105667e-05, | |
| "loss": 4.0712, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.799646283354615e-05, | |
| "loss": 4.0796, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.798807688603563e-05, | |
| "loss": 4.0681, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.797970731732884e-05, | |
| "loss": 4.0632, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.797132136981832e-05, | |
| "loss": 4.0679, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.79629354223078e-05, | |
| "loss": 4.074, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.795456585360101e-05, | |
| "loss": 4.0774, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.794617990609049e-05, | |
| "loss": 4.0706, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.793779395857997e-05, | |
| "loss": 4.0543, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792940801106946e-05, | |
| "loss": 4.0609, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.792102206355894e-05, | |
| "loss": 4.0635, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.791263611604842e-05, | |
| "loss": 4.0654, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.79042501685379e-05, | |
| "loss": 4.0621, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.789586422102737e-05, | |
| "loss": 4.0562, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7887494652320587e-05, | |
| "loss": 4.0488, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7879108704810067e-05, | |
| "loss": 4.0654, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7870722757299546e-05, | |
| "loss": 4.0299, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.786233680978902e-05, | |
| "loss": 4.0545, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7853967241082236e-05, | |
| "loss": 4.0469, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7845581293571715e-05, | |
| "loss": 4.0592, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7837195346061195e-05, | |
| "loss": 4.044, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7828809398550675e-05, | |
| "loss": 4.0537, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.782043982984389e-05, | |
| "loss": 4.0421, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.781205388233337e-05, | |
| "loss": 4.0446, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7803667934822844e-05, | |
| "loss": 4.0539, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7795281987312324e-05, | |
| "loss": 4.0393, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.778691241860554e-05, | |
| "loss": 4.0541, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.777852647109502e-05, | |
| "loss": 4.0553, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.777015690238823e-05, | |
| "loss": 4.0464, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.776177095487771e-05, | |
| "loss": 4.0398, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.775338500736719e-05, | |
| "loss": 4.0433, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.774499905985666e-05, | |
| "loss": 4.0452, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.773661311234615e-05, | |
| "loss": 4.0339, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.772822716483563e-05, | |
| "loss": 4.0476, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.771984121732511e-05, | |
| "loss": 4.0323, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.771145526981459e-05, | |
| "loss": 4.043, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.77030857011078e-05, | |
| "loss": 4.0338, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.769469975359728e-05, | |
| "loss": 4.0305, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.768631380608676e-05, | |
| "loss": 4.0214, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.767792785857624e-05, | |
| "loss": 4.0268, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.766955828986945e-05, | |
| "loss": 4.0191, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.766117234235893e-05, | |
| "loss": 4.042, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.765278639484841e-05, | |
| "loss": 4.0496, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.764440044733789e-05, | |
| "loss": 4.0332, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7636030878631096e-05, | |
| "loss": 4.0169, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.762764493112058e-05, | |
| "loss": 4.0364, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.761925898361006e-05, | |
| "loss": 4.0267, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.761087303609954e-05, | |
| "loss": 4.0298, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.760250346739275e-05, | |
| "loss": 4.0253, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.759411751988223e-05, | |
| "loss": 4.0276, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.758573157237171e-05, | |
| "loss": 4.0225, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.757734562486119e-05, | |
| "loss": 4.0272, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.75689760561544e-05, | |
| "loss": 4.0232, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.756059010864388e-05, | |
| "loss": 4.0284, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.755220416113336e-05, | |
| "loss": 4.0222, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.754381821362284e-05, | |
| "loss": 4.0141, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.753544864491605e-05, | |
| "loss": 4.0302, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7527062697405536e-05, | |
| "loss": 4.0231, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7518676749895016e-05, | |
| "loss": 4.0147, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7510290802384496e-05, | |
| "loss": 4.0301, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.7501921233677705e-05, | |
| "loss": 4.0118, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.044399738311768, | |
| "eval_runtime": 302.1723, | |
| "eval_samples_per_second": 1262.826, | |
| "eval_steps_per_second": 39.464, | |
| "step": 152640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7493535286167185e-05, | |
| "loss": 4.003, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7485149338656665e-05, | |
| "loss": 4.0051, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7476763391146145e-05, | |
| "loss": 4.022, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7468393822439354e-05, | |
| "loss": 4.0145, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7460007874928834e-05, | |
| "loss": 4.0197, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7451621927418314e-05, | |
| "loss": 4.001, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7443235979907794e-05, | |
| "loss": 4.0149, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7434866411201e-05, | |
| "loss": 3.9915, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.742648046369049e-05, | |
| "loss": 4.0149, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.741809451617997e-05, | |
| "loss": 3.9956, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.740970856866945e-05, | |
| "loss": 4.0134, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.740133899996266e-05, | |
| "loss": 4.0085, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.739296943125587e-05, | |
| "loss": 4.0023, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.738458348374535e-05, | |
| "loss": 3.9932, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.737619753623483e-05, | |
| "loss": 3.9905, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.736781158872431e-05, | |
| "loss": 3.9914, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.735942564121379e-05, | |
| "loss": 3.9955, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.735103969370327e-05, | |
| "loss": 3.9989, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.734265374619275e-05, | |
| "loss": 3.9951, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.733426779868223e-05, | |
| "loss": 4.011, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7325898229975444e-05, | |
| "loss": 4.0051, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7317512282464924e-05, | |
| "loss": 4.0008, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7309126334954404e-05, | |
| "loss": 3.9946, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7300740387443884e-05, | |
| "loss": 4.0035, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.729237081873709e-05, | |
| "loss": 3.9872, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.728398487122657e-05, | |
| "loss": 3.9961, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.727559892371605e-05, | |
| "loss": 3.9947, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.726722935500926e-05, | |
| "loss": 3.9909, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.725884340749874e-05, | |
| "loss": 3.9795, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.725045745998822e-05, | |
| "loss": 3.9834, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.72420715124777e-05, | |
| "loss": 3.9862, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.723370194377091e-05, | |
| "loss": 3.9909, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.72253159962604e-05, | |
| "loss": 3.994, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.721693004874988e-05, | |
| "loss": 3.9866, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.720854410123936e-05, | |
| "loss": 3.9871, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.720015815372884e-05, | |
| "loss": 3.9854, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.719177220621832e-05, | |
| "loss": 3.9711, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.71833862587078e-05, | |
| "loss": 3.9805, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7175016690001006e-05, | |
| "loss": 3.9702, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7166630742490486e-05, | |
| "loss": 3.9703, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7158244794979966e-05, | |
| "loss": 3.976, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7149858847469446e-05, | |
| "loss": 3.9855, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7141489278762655e-05, | |
| "loss": 3.9736, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7133103331252135e-05, | |
| "loss": 3.9691, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7124717383741615e-05, | |
| "loss": 3.9826, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7116331436231095e-05, | |
| "loss": 3.9718, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.710796186752431e-05, | |
| "loss": 3.9828, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.709957592001379e-05, | |
| "loss": 3.9664, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.709118997250327e-05, | |
| "loss": 3.9637, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.708282040379648e-05, | |
| "loss": 3.9772, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.707443445628596e-05, | |
| "loss": 3.9674, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.706604850877544e-05, | |
| "loss": 3.9725, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.705766256126492e-05, | |
| "loss": 3.9607, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.704927661375439e-05, | |
| "loss": 3.9682, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.704090704504761e-05, | |
| "loss": 3.9538, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.703252109753709e-05, | |
| "loss": 3.9687, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.702413515002657e-05, | |
| "loss": 3.9623, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.701574920251605e-05, | |
| "loss": 3.9639, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.700736325500553e-05, | |
| "loss": 3.9711, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.699897730749501e-05, | |
| "loss": 3.9553, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.699059135998449e-05, | |
| "loss": 3.9521, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.698220541247397e-05, | |
| "loss": 3.9717, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.697383584376718e-05, | |
| "loss": 3.9432, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.696544989625666e-05, | |
| "loss": 3.9437, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.695706394874614e-05, | |
| "loss": 3.9592, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.694867800123562e-05, | |
| "loss": 3.9651, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.694030843252883e-05, | |
| "loss": 3.9553, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6931922485018307e-05, | |
| "loss": 3.9435, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6923536537507787e-05, | |
| "loss": 3.9471, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6915166968801e-05, | |
| "loss": 3.9482, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.690678102129048e-05, | |
| "loss": 3.9621, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.689839507377996e-05, | |
| "loss": 3.9537, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.689000912626944e-05, | |
| "loss": 3.957, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.688163955756265e-05, | |
| "loss": 3.9571, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.687325361005213e-05, | |
| "loss": 3.9608, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.686488404134534e-05, | |
| "loss": 3.9424, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.685649809383482e-05, | |
| "loss": 3.9443, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.68481121463243e-05, | |
| "loss": 3.9544, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.683972619881378e-05, | |
| "loss": 3.9399, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.683134025130326e-05, | |
| "loss": 3.95, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.682295430379274e-05, | |
| "loss": 3.9412, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.681456835628222e-05, | |
| "loss": 3.9586, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.680618240877171e-05, | |
| "loss": 3.9457, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6797812840064916e-05, | |
| "loss": 3.9344, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6789426892554396e-05, | |
| "loss": 3.9392, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6781040945043876e-05, | |
| "loss": 3.9451, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6772654997533356e-05, | |
| "loss": 3.9552, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6764285428826565e-05, | |
| "loss": 3.9361, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6755899481316045e-05, | |
| "loss": 3.94, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6747513533805525e-05, | |
| "loss": 3.9379, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6739127586295005e-05, | |
| "loss": 3.9457, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6730758017588214e-05, | |
| "loss": 3.9322, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6722372070077694e-05, | |
| "loss": 3.9339, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6713986122567174e-05, | |
| "loss": 3.9446, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.670561655386039e-05, | |
| "loss": 3.9474, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.669723060634987e-05, | |
| "loss": 3.9439, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.668884465883935e-05, | |
| "loss": 3.9279, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.668045871132883e-05, | |
| "loss": 3.9318, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.667207276381831e-05, | |
| "loss": 3.9324, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.666368681630779e-05, | |
| "loss": 3.938, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.665530086879727e-05, | |
| "loss": 3.938, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.664691492128675e-05, | |
| "loss": 3.9294, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.663854535257996e-05, | |
| "loss": 3.9246, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.663015940506944e-05, | |
| "loss": 3.9388, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.662177345755892e-05, | |
| "loss": 3.9108, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.66133875100484e-05, | |
| "loss": 3.9271, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6605017941341614e-05, | |
| "loss": 3.9219, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6596631993831094e-05, | |
| "loss": 3.9357, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.65882624251243e-05, | |
| "loss": 3.9246, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.657987647761378e-05, | |
| "loss": 3.929, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.657149053010326e-05, | |
| "loss": 3.9196, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.656310458259274e-05, | |
| "loss": 3.9239, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.655471863508222e-05, | |
| "loss": 3.9323, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.65463326875717e-05, | |
| "loss": 3.9226, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6537946740061176e-05, | |
| "loss": 3.9322, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6529560792550656e-05, | |
| "loss": 3.9343, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.65212076026476e-05, | |
| "loss": 3.9302, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.651282165513708e-05, | |
| "loss": 3.9233, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.650443570762657e-05, | |
| "loss": 3.9192, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.649604976011605e-05, | |
| "loss": 3.9293, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.648766381260553e-05, | |
| "loss": 3.9148, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6479277865095e-05, | |
| "loss": 3.9291, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.647089191758448e-05, | |
| "loss": 3.9149, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.646250597007396e-05, | |
| "loss": 3.9253, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.645413640136718e-05, | |
| "loss": 3.9191, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.644575045385665e-05, | |
| "loss": 3.9196, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.643736450634613e-05, | |
| "loss": 3.9048, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6428994937639346e-05, | |
| "loss": 3.9153, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.642060899012882e-05, | |
| "loss": 3.908, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6412223042618306e-05, | |
| "loss": 3.9208, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6403837095107786e-05, | |
| "loss": 3.9345, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6395451147597266e-05, | |
| "loss": 3.926, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6387065200086746e-05, | |
| "loss": 3.8975, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6378679252576226e-05, | |
| "loss": 3.9278, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6370293305065705e-05, | |
| "loss": 3.916, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6361923736358915e-05, | |
| "loss": 3.9178, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6353537788848395e-05, | |
| "loss": 3.9062, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6345151841337874e-05, | |
| "loss": 3.9192, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6336765893827354e-05, | |
| "loss": 3.9109, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6328396325120564e-05, | |
| "loss": 3.9156, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6320010377610043e-05, | |
| "loss": 3.9144, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6311624430099523e-05, | |
| "loss": 3.9175, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6303238482589003e-05, | |
| "loss": 3.9137, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.629486891388222e-05, | |
| "loss": 3.9084, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.62864829663717e-05, | |
| "loss": 3.9175, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.627809701886118e-05, | |
| "loss": 3.9149, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.626971107135066e-05, | |
| "loss": 3.907, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.626134150264387e-05, | |
| "loss": 3.9194, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.625295555513335e-05, | |
| "loss": 3.905, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 3.9699602127075195, | |
| "eval_runtime": 302.1878, | |
| "eval_samples_per_second": 1262.761, | |
| "eval_steps_per_second": 39.462, | |
| "step": 228960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.624456960762283e-05, | |
| "loss": 3.8928, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.623618366011231e-05, | |
| "loss": 3.8991, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.622779771260179e-05, | |
| "loss": 3.914, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.621941176509127e-05, | |
| "loss": 3.9062, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.621102581758075e-05, | |
| "loss": 3.9192, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.620263987007023e-05, | |
| "loss": 3.8938, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.619425392255971e-05, | |
| "loss": 3.9125, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.618586797504919e-05, | |
| "loss": 3.8829, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.617748202753867e-05, | |
| "loss": 3.9144, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.616909608002815e-05, | |
| "loss": 3.896, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.616071013251763e-05, | |
| "loss": 3.9059, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.615234056381084e-05, | |
| "loss": 3.9047, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.614397099510405e-05, | |
| "loss": 3.8938, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.613558504759353e-05, | |
| "loss": 3.895, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6127199100083006e-05, | |
| "loss": 3.8889, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6118813152572486e-05, | |
| "loss": 3.8859, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6110427205061966e-05, | |
| "loss": 3.8987, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6102041257551446e-05, | |
| "loss": 3.8959, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6093655310040926e-05, | |
| "loss": 3.8952, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6085269362530406e-05, | |
| "loss": 3.909, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.607689979382362e-05, | |
| "loss": 3.9095, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.60685138463131e-05, | |
| "loss": 3.9002, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.606012789880258e-05, | |
| "loss": 3.8948, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.605175833009579e-05, | |
| "loss": 3.9029, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.604337238258527e-05, | |
| "loss": 3.8879, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.603498643507475e-05, | |
| "loss": 3.9013, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.602660048756423e-05, | |
| "loss": 3.8963, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.601821454005371e-05, | |
| "loss": 3.8889, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.600982859254319e-05, | |
| "loss": 3.883, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.600144264503267e-05, | |
| "loss": 3.8886, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.599307307632588e-05, | |
| "loss": 3.8879, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5984703507619095e-05, | |
| "loss": 3.8957, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5976317560108575e-05, | |
| "loss": 3.8975, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5967931612598055e-05, | |
| "loss": 3.8908, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5959545665087535e-05, | |
| "loss": 3.8931, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5951159717577015e-05, | |
| "loss": 3.8903, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5942790148870224e-05, | |
| "loss": 3.8761, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5934404201359704e-05, | |
| "loss": 3.8841, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5926018253849184e-05, | |
| "loss": 3.8758, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5917632306338664e-05, | |
| "loss": 3.8799, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5909246358828144e-05, | |
| "loss": 3.8819, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5900860411317624e-05, | |
| "loss": 3.888, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5892474463807104e-05, | |
| "loss": 3.8862, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5884088516296584e-05, | |
| "loss": 3.8721, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.58757189475898e-05, | |
| "loss": 3.8887, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.586733300007928e-05, | |
| "loss": 3.8868, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.585894705256876e-05, | |
| "loss": 3.8857, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.585057748386197e-05, | |
| "loss": 3.8782, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.584219153635145e-05, | |
| "loss": 3.8696, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.583382196764466e-05, | |
| "loss": 3.8868, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.582543602013414e-05, | |
| "loss": 3.8738, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.581705007262362e-05, | |
| "loss": 3.8798, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.58086641251131e-05, | |
| "loss": 3.8713, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.580027817760258e-05, | |
| "loss": 3.8739, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.579190860889579e-05, | |
| "loss": 3.8664, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5783522661385267e-05, | |
| "loss": 3.8759, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.577513671387475e-05, | |
| "loss": 3.8757, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.576675076636423e-05, | |
| "loss": 3.8763, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.575836481885371e-05, | |
| "loss": 3.8776, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.574997887134319e-05, | |
| "loss": 3.8664, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5741592923832666e-05, | |
| "loss": 3.8664, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5733206976322146e-05, | |
| "loss": 3.8857, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.572485378641909e-05, | |
| "loss": 3.8567, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.571646783890857e-05, | |
| "loss": 3.8553, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.570808189139805e-05, | |
| "loss": 3.8673, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.569969594388753e-05, | |
| "loss": 3.877, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.569130999637701e-05, | |
| "loss": 3.8654, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.568292404886649e-05, | |
| "loss": 3.8615, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.567453810135597e-05, | |
| "loss": 3.8559, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.566615215384545e-05, | |
| "loss": 3.862, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.565778258513867e-05, | |
| "loss": 3.8759, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5649413016431876e-05, | |
| "loss": 3.8657, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5641027068921356e-05, | |
| "loss": 3.8662, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5632641121410836e-05, | |
| "loss": 3.8762, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5624255173900316e-05, | |
| "loss": 3.8752, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.561586922638979e-05, | |
| "loss": 3.8578, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.560748327887927e-05, | |
| "loss": 3.8625, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.559909733136875e-05, | |
| "loss": 3.8711, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.559071138385823e-05, | |
| "loss": 3.8511, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5582358193955174e-05, | |
| "loss": 3.8679, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5573972246444654e-05, | |
| "loss": 3.8557, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.556558629893414e-05, | |
| "loss": 3.8773, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5557200351423614e-05, | |
| "loss": 3.8612, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5548814403913094e-05, | |
| "loss": 3.8528, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5540428456402574e-05, | |
| "loss": 3.8508, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.553205888769579e-05, | |
| "loss": 3.8631, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.552367294018526e-05, | |
| "loss": 3.874, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.551528699267474e-05, | |
| "loss": 3.857, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.550690104516422e-05, | |
| "loss": 3.8545, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.54985150976537e-05, | |
| "loss": 3.8541, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.549012915014318e-05, | |
| "loss": 3.8651, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.548174320263266e-05, | |
| "loss": 3.8492, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.547335725512214e-05, | |
| "loss": 3.8543, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.546500406521909e-05, | |
| "loss": 3.8625, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.54566344965123e-05, | |
| "loss": 3.8668, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.544824854900178e-05, | |
| "loss": 3.8615, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.543986260149126e-05, | |
| "loss": 3.8442, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5431476653980736e-05, | |
| "loss": 3.8554, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5423090706470216e-05, | |
| "loss": 3.8486, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5414704758959696e-05, | |
| "loss": 3.8578, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5406318811449176e-05, | |
| "loss": 3.8592, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5397932863938656e-05, | |
| "loss": 3.847, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5389546916428136e-05, | |
| "loss": 3.8479, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5381160968917616e-05, | |
| "loss": 3.856, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5372775021407096e-05, | |
| "loss": 3.8341, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.536440545270031e-05, | |
| "loss": 3.8505, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.535601950518979e-05, | |
| "loss": 3.8396, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.534763355767927e-05, | |
| "loss": 3.8578, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.533924761016875e-05, | |
| "loss": 3.8473, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.533087804146196e-05, | |
| "loss": 3.8477, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.532249209395144e-05, | |
| "loss": 3.8425, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.531410614644092e-05, | |
| "loss": 3.8507, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.53057201989304e-05, | |
| "loss": 3.8543, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.529733425141988e-05, | |
| "loss": 3.8437, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.528894830390936e-05, | |
| "loss": 3.8525, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.528057873520257e-05, | |
| "loss": 3.857, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.527219278769205e-05, | |
| "loss": 3.8492, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.526380684018153e-05, | |
| "loss": 3.8529, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5255420892671017e-05, | |
| "loss": 3.8377, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5247051323964226e-05, | |
| "loss": 3.8549, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5238665376453706e-05, | |
| "loss": 3.844, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5230279428943186e-05, | |
| "loss": 3.8492, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5221893481432665e-05, | |
| "loss": 3.8393, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5213523912725875e-05, | |
| "loss": 3.8481, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5205137965215355e-05, | |
| "loss": 3.8424, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5196752017704834e-05, | |
| "loss": 3.8409, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5188366070194314e-05, | |
| "loss": 3.8319, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5179996501487524e-05, | |
| "loss": 3.8358, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5171610553977003e-05, | |
| "loss": 3.8387, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5163224606466483e-05, | |
| "loss": 3.8438, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.515483865895597e-05, | |
| "loss": 3.8574, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.514645271144545e-05, | |
| "loss": 3.8515, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.513806676393492e-05, | |
| "loss": 3.8246, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.51296808164244e-05, | |
| "loss": 3.854, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.512129486891388e-05, | |
| "loss": 3.8406, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.51129253002071e-05, | |
| "loss": 3.8421, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.510453935269657e-05, | |
| "loss": 3.8337, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.509615340518605e-05, | |
| "loss": 3.8452, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.508778383647927e-05, | |
| "loss": 3.8362, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.507939788896874e-05, | |
| "loss": 3.8481, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.507101194145822e-05, | |
| "loss": 3.8395, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.506262599394771e-05, | |
| "loss": 3.8411, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5054256425240924e-05, | |
| "loss": 3.845, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.50458704777304e-05, | |
| "loss": 3.8337, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.503748453021988e-05, | |
| "loss": 3.8423, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.502909858270936e-05, | |
| "loss": 3.8473, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.502071263519884e-05, | |
| "loss": 3.8336, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.501232668768832e-05, | |
| "loss": 3.8419, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.5003957118981526e-05, | |
| "loss": 3.8377, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 3.9284117221832275, | |
| "eval_runtime": 330.261, | |
| "eval_samples_per_second": 1155.422, | |
| "eval_steps_per_second": 36.108, | |
| "step": 305280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4995571171471006e-05, | |
| "loss": 3.8281, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4987185223960486e-05, | |
| "loss": 3.8272, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4978799276449966e-05, | |
| "loss": 3.8388, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4970413328939446e-05, | |
| "loss": 3.8385, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.496204376023266e-05, | |
| "loss": 3.8515, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.495365781272214e-05, | |
| "loss": 3.8208, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.494527186521162e-05, | |
| "loss": 3.8376, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.49368859177011e-05, | |
| "loss": 3.8186, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.492849997019058e-05, | |
| "loss": 3.8393, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.492013040148379e-05, | |
| "loss": 3.8259, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.491174445397327e-05, | |
| "loss": 3.8321, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.490335850646275e-05, | |
| "loss": 3.8387, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.489498893775596e-05, | |
| "loss": 3.8213, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.488660299024544e-05, | |
| "loss": 3.8285, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.487821704273492e-05, | |
| "loss": 3.82, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.48698310952244e-05, | |
| "loss": 3.8174, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.486144514771388e-05, | |
| "loss": 3.8269, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.485305920020336e-05, | |
| "loss": 3.8286, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.484467325269284e-05, | |
| "loss": 3.8249, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4836287305182326e-05, | |
| "loss": 3.8407, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4827901357671806e-05, | |
| "loss": 3.8414, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4819531788965015e-05, | |
| "loss": 3.8322, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4811145841454495e-05, | |
| "loss": 3.8282, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4802759893943975e-05, | |
| "loss": 3.8364, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4794373946433455e-05, | |
| "loss": 3.8173, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4785987998922935e-05, | |
| "loss": 3.8334, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4777618430216144e-05, | |
| "loss": 3.8279, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4769232482705624e-05, | |
| "loss": 3.8199, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4760846535195104e-05, | |
| "loss": 3.817, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.475246058768458e-05, | |
| "loss": 3.8182, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4744074640174064e-05, | |
| "loss": 3.8239, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.473570507146728e-05, | |
| "loss": 3.8282, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.472731912395676e-05, | |
| "loss": 3.8315, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.471893317644623e-05, | |
| "loss": 3.8252, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.471054722893571e-05, | |
| "loss": 3.8253, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.470217766022893e-05, | |
| "loss": 3.8252, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.469379171271841e-05, | |
| "loss": 3.8087, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.468540576520788e-05, | |
| "loss": 3.8215, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.467701981769736e-05, | |
| "loss": 3.8098, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.466865024899058e-05, | |
| "loss": 3.8137, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.466026430148005e-05, | |
| "loss": 3.8136, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.465187835396953e-05, | |
| "loss": 3.8276, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.464349240645902e-05, | |
| "loss": 3.8187, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.463512283775223e-05, | |
| "loss": 3.807, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4626736890241707e-05, | |
| "loss": 3.8252, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4618350942731186e-05, | |
| "loss": 3.82, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4609964995220666e-05, | |
| "loss": 3.8192, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.460159542651388e-05, | |
| "loss": 3.8161, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4593209479003356e-05, | |
| "loss": 3.8037, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4584823531492835e-05, | |
| "loss": 3.8245, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4576437583982315e-05, | |
| "loss": 3.8082, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4568051636471795e-05, | |
| "loss": 3.819, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4559682067765004e-05, | |
| "loss": 3.8078, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4551296120254484e-05, | |
| "loss": 3.808, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.454291017274397e-05, | |
| "loss": 3.8047, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.453452422523345e-05, | |
| "loss": 3.8129, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.452615465652666e-05, | |
| "loss": 3.8111, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.451776870901614e-05, | |
| "loss": 3.8125, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.450938276150562e-05, | |
| "loss": 3.8176, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.45009968139951e-05, | |
| "loss": 3.8042, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.449261086648458e-05, | |
| "loss": 3.7997, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.448424129777779e-05, | |
| "loss": 3.8267, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.447585535026727e-05, | |
| "loss": 3.792, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.446746940275675e-05, | |
| "loss": 3.7934, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.445908345524623e-05, | |
| "loss": 3.8051, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.445071388653944e-05, | |
| "loss": 3.8193, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4442327939028925e-05, | |
| "loss": 3.798, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4433941991518405e-05, | |
| "loss": 3.803, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4425556044007885e-05, | |
| "loss": 3.7941, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4417170096497365e-05, | |
| "loss": 3.8001, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4408784148986845e-05, | |
| "loss": 3.8114, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4400414580280054e-05, | |
| "loss": 3.8082, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4392028632769534e-05, | |
| "loss": 3.8064, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4383642685259014e-05, | |
| "loss": 3.8118, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4375256737748494e-05, | |
| "loss": 3.8138, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.43668871690417e-05, | |
| "loss": 3.8, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.435850122153118e-05, | |
| "loss": 3.8011, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.435011527402066e-05, | |
| "loss": 3.8105, | |
| "step": 345088 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.434172932651014e-05, | |
| "loss": 3.7891, | |
| "step": 345600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.433335975780336e-05, | |
| "loss": 3.8061, | |
| "step": 346112 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.432497381029284e-05, | |
| "loss": 3.7963, | |
| "step": 346624 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.431658786278232e-05, | |
| "loss": 3.8172, | |
| "step": 347136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.43082019152718e-05, | |
| "loss": 3.801, | |
| "step": 347648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.429983234656501e-05, | |
| "loss": 3.795, | |
| "step": 348160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.429144639905449e-05, | |
| "loss": 3.7848, | |
| "step": 348672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.428306045154397e-05, | |
| "loss": 3.8071, | |
| "step": 349184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.427467450403345e-05, | |
| "loss": 3.8109, | |
| "step": 349696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4266304935326656e-05, | |
| "loss": 3.8021, | |
| "step": 350208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4257918987816136e-05, | |
| "loss": 3.7928, | |
| "step": 350720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4249533040305616e-05, | |
| "loss": 3.7966, | |
| "step": 351232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4241147092795096e-05, | |
| "loss": 3.8038, | |
| "step": 351744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.423277752408831e-05, | |
| "loss": 3.7912, | |
| "step": 352256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.422439157657779e-05, | |
| "loss": 3.7926, | |
| "step": 352768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.421600562906727e-05, | |
| "loss": 3.8046, | |
| "step": 353280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.420763606036048e-05, | |
| "loss": 3.8066, | |
| "step": 353792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.419925011284996e-05, | |
| "loss": 3.8022, | |
| "step": 354304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.419086416533944e-05, | |
| "loss": 3.7882, | |
| "step": 354816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.418247821782892e-05, | |
| "loss": 3.7973, | |
| "step": 355328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.41740922703184e-05, | |
| "loss": 3.789, | |
| "step": 355840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.416570632280788e-05, | |
| "loss": 3.7987, | |
| "step": 356352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.415732037529736e-05, | |
| "loss": 3.7984, | |
| "step": 356864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4148934427786834e-05, | |
| "loss": 3.7934, | |
| "step": 357376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4140548480276314e-05, | |
| "loss": 3.7899, | |
| "step": 357888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.413217891156953e-05, | |
| "loss": 3.7923, | |
| "step": 358400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.412379296405901e-05, | |
| "loss": 3.7857, | |
| "step": 358912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.411540701654849e-05, | |
| "loss": 3.7877, | |
| "step": 359424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.410702106903797e-05, | |
| "loss": 3.7818, | |
| "step": 359936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4098651500331186e-05, | |
| "loss": 3.8003, | |
| "step": 360448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.409026555282066e-05, | |
| "loss": 3.7915, | |
| "step": 360960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.408187960531014e-05, | |
| "loss": 3.788, | |
| "step": 361472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.407349365779962e-05, | |
| "loss": 3.7835, | |
| "step": 361984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4065124089092835e-05, | |
| "loss": 3.796, | |
| "step": 362496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.405673814158231e-05, | |
| "loss": 3.7966, | |
| "step": 363008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.404835219407179e-05, | |
| "loss": 3.7861, | |
| "step": 363520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.403996624656127e-05, | |
| "loss": 3.7961, | |
| "step": 364032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4031596677854484e-05, | |
| "loss": 3.8009, | |
| "step": 364544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4023210730343963e-05, | |
| "loss": 3.7927, | |
| "step": 365056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4014824782833443e-05, | |
| "loss": 3.7975, | |
| "step": 365568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4006438835322923e-05, | |
| "loss": 3.7829, | |
| "step": 366080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.399806926661613e-05, | |
| "loss": 3.7944, | |
| "step": 366592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.398968331910561e-05, | |
| "loss": 3.7902, | |
| "step": 367104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.398129737159509e-05, | |
| "loss": 3.7941, | |
| "step": 367616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.397291142408457e-05, | |
| "loss": 3.779, | |
| "step": 368128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.396454185537778e-05, | |
| "loss": 3.793, | |
| "step": 368640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.395615590786726e-05, | |
| "loss": 3.7872, | |
| "step": 369152 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.394776996035674e-05, | |
| "loss": 3.784, | |
| "step": 369664 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.393938401284622e-05, | |
| "loss": 3.7781, | |
| "step": 370176 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.393101444413944e-05, | |
| "loss": 3.7809, | |
| "step": 370688 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.392262849662892e-05, | |
| "loss": 3.7808, | |
| "step": 371200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.39142425491184e-05, | |
| "loss": 3.7922, | |
| "step": 371712 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.390585660160788e-05, | |
| "loss": 3.7981, | |
| "step": 372224 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3897487032901086e-05, | |
| "loss": 3.8003, | |
| "step": 372736 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3889101085390566e-05, | |
| "loss": 3.7699, | |
| "step": 373248 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3880715137880046e-05, | |
| "loss": 3.7962, | |
| "step": 373760 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3872329190369526e-05, | |
| "loss": 3.7888, | |
| "step": 374272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3863959621662735e-05, | |
| "loss": 3.7822, | |
| "step": 374784 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3855573674152215e-05, | |
| "loss": 3.7796, | |
| "step": 375296 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3847187726641695e-05, | |
| "loss": 3.7945, | |
| "step": 375808 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3838801779131175e-05, | |
| "loss": 3.7831, | |
| "step": 376320 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.383043221042439e-05, | |
| "loss": 3.7914, | |
| "step": 376832 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.382204626291387e-05, | |
| "loss": 3.7818, | |
| "step": 377344 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.381366031540335e-05, | |
| "loss": 3.7854, | |
| "step": 377856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.380527436789283e-05, | |
| "loss": 3.7908, | |
| "step": 378368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.379690479918604e-05, | |
| "loss": 3.7814, | |
| "step": 378880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.378851885167552e-05, | |
| "loss": 3.7862, | |
| "step": 379392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.3780132904165e-05, | |
| "loss": 3.7946, | |
| "step": 379904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.377174695665448e-05, | |
| "loss": 3.7794, | |
| "step": 380416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.376336100914396e-05, | |
| "loss": 3.7829, | |
| "step": 380928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.375499144043717e-05, | |
| "loss": 3.7866, | |
| "step": 381440 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 3.9028751850128174, | |
| "eval_runtime": 335.3025, | |
| "eval_samples_per_second": 1138.05, | |
| "eval_steps_per_second": 35.565, | |
| "step": 381600 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 2.5740123379117056e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |