| { | |
| "best_metric": 3.9596714973449707, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/pp-mod-subj2/transformer/2/checkpoints/checkpoint-305280", | |
| "epoch": 1.0250006060157382, | |
| "eval_steps": 10, | |
| "global_step": 305280, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999998362119627e-05, | |
| "loss": 11.0111, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999161405248948e-05, | |
| "loss": 6.8304, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.998322810497896e-05, | |
| "loss": 6.1934, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.997484215746844e-05, | |
| "loss": 5.9809, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.996645620995792e-05, | |
| "loss": 5.8098, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99580702624474e-05, | |
| "loss": 5.7038, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994968431493688e-05, | |
| "loss": 5.6068, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.994129836742636e-05, | |
| "loss": 5.5374, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993291241991584e-05, | |
| "loss": 5.4546, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992452647240532e-05, | |
| "loss": 5.4007, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.99161405248948e-05, | |
| "loss": 5.3464, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.990775457738428e-05, | |
| "loss": 5.3083, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989938500867749e-05, | |
| "loss": 5.2703, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989099906116697e-05, | |
| "loss": 5.2131, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.988261311365645e-05, | |
| "loss": 5.1814, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.987422716614593e-05, | |
| "loss": 5.1288, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986584121863541e-05, | |
| "loss": 5.1136, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.985745527112489e-05, | |
| "loss": 5.0782, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984906932361437e-05, | |
| "loss": 5.0411, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.984068337610385e-05, | |
| "loss": 5.0188, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.983229742859333e-05, | |
| "loss": 5.0187, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9823927859886547e-05, | |
| "loss": 4.9727, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9815541912376026e-05, | |
| "loss": 4.9591, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9807155964865506e-05, | |
| "loss": 4.9324, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9798770017354986e-05, | |
| "loss": 4.9072, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9790400448648195e-05, | |
| "loss": 4.8975, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9782014501137675e-05, | |
| "loss": 4.8631, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9773628553627155e-05, | |
| "loss": 4.859, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9765242606116635e-05, | |
| "loss": 4.8412, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9756856658606115e-05, | |
| "loss": 4.8142, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9748470711095595e-05, | |
| "loss": 4.8068, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.974008476358507e-05, | |
| "loss": 4.7844, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.973169881607455e-05, | |
| "loss": 4.7855, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9723329247367764e-05, | |
| "loss": 4.7661, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9714943299857244e-05, | |
| "loss": 4.7484, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.970657373115046e-05, | |
| "loss": 4.745, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.969818778363994e-05, | |
| "loss": 4.7463, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.968980183612942e-05, | |
| "loss": 4.7079, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96814158886189e-05, | |
| "loss": 4.7015, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.967302994110837e-05, | |
| "loss": 4.695, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.966464399359785e-05, | |
| "loss": 4.6749, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.965625804608733e-05, | |
| "loss": 4.6688, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.964787209857681e-05, | |
| "loss": 4.6714, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.963950252987002e-05, | |
| "loss": 4.6598, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96311165823595e-05, | |
| "loss": 4.6449, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.962273063484898e-05, | |
| "loss": 4.6366, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.961434468733847e-05, | |
| "loss": 4.6267, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.960595873982795e-05, | |
| "loss": 4.6274, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.959758917112116e-05, | |
| "loss": 4.5957, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9589219602414374e-05, | |
| "loss": 4.5957, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958083365490385e-05, | |
| "loss": 4.6022, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.957244770739333e-05, | |
| "loss": 4.5928, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.956406175988281e-05, | |
| "loss": 4.5793, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.955567581237229e-05, | |
| "loss": 4.5615, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.954728986486177e-05, | |
| "loss": 4.5695, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.953890391735125e-05, | |
| "loss": 4.5506, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9530517969840727e-05, | |
| "loss": 4.5676, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9522132022330207e-05, | |
| "loss": 4.5352, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951376245362342e-05, | |
| "loss": 4.5489, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.95053765061129e-05, | |
| "loss": 4.5302, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.949699055860238e-05, | |
| "loss": 4.5188, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948860461109186e-05, | |
| "loss": 4.5224, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.948021866358134e-05, | |
| "loss": 4.5084, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.947184909487455e-05, | |
| "loss": 4.5034, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.946346314736403e-05, | |
| "loss": 4.4952, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.945507719985351e-05, | |
| "loss": 4.505, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944669125234299e-05, | |
| "loss": 4.4809, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94383216836362e-05, | |
| "loss": 4.4763, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942993573612568e-05, | |
| "loss": 4.4645, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.942156616741889e-05, | |
| "loss": 4.4629, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.941318021990837e-05, | |
| "loss": 4.4826, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9404794272397856e-05, | |
| "loss": 4.4715, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9396408324887336e-05, | |
| "loss": 4.4523, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9388022377376816e-05, | |
| "loss": 4.4521, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9379636429866296e-05, | |
| "loss": 4.4569, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9371250482355776e-05, | |
| "loss": 4.4513, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9362880913648985e-05, | |
| "loss": 4.4344, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9354494966138465e-05, | |
| "loss": 4.439, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9346109018627945e-05, | |
| "loss": 4.4291, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9337723071117425e-05, | |
| "loss": 4.4267, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9329337123606905e-05, | |
| "loss": 4.4215, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.932095117609638e-05, | |
| "loss": 4.4195, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9312581607389594e-05, | |
| "loss": 4.4217, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9304195659879074e-05, | |
| "loss": 4.4137, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9295809712368554e-05, | |
| "loss": 4.393, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9287423764858034e-05, | |
| "loss": 4.4039, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9279037817347514e-05, | |
| "loss": 4.4025, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9270651869836994e-05, | |
| "loss": 4.3979, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9262265922326474e-05, | |
| "loss": 4.3908, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9253879974815954e-05, | |
| "loss": 4.3806, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.924551040610916e-05, | |
| "loss": 4.3867, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.923712445859864e-05, | |
| "loss": 4.3744, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922875488989185e-05, | |
| "loss": 4.3819, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.922036894238133e-05, | |
| "loss": 4.3697, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.921198299487081e-05, | |
| "loss": 4.3645, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.920359704736029e-05, | |
| "loss": 4.3834, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.919521109984978e-05, | |
| "loss": 4.3487, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.918682515233926e-05, | |
| "loss": 4.3473, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917843920482874e-05, | |
| "loss": 4.3556, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.917005325731822e-05, | |
| "loss": 4.3541, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.91616673098077e-05, | |
| "loss": 4.3581, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.915328136229718e-05, | |
| "loss": 4.3437, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.914491179359039e-05, | |
| "loss": 4.3372, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.913652584607987e-05, | |
| "loss": 4.3352, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.912813989856935e-05, | |
| "loss": 4.3373, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.911975395105883e-05, | |
| "loss": 4.3218, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9111384382352036e-05, | |
| "loss": 4.3192, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9102998434841516e-05, | |
| "loss": 4.3317, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9094612487330996e-05, | |
| "loss": 4.3218, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9086226539820476e-05, | |
| "loss": 4.3133, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.907785697111369e-05, | |
| "loss": 4.3101, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906947102360317e-05, | |
| "loss": 4.3068, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.906108507609265e-05, | |
| "loss": 4.3177, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.905269912858213e-05, | |
| "loss": 4.3039, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.904431318107161e-05, | |
| "loss": 4.3018, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.903592723356109e-05, | |
| "loss": 4.3053, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9027541286050565e-05, | |
| "loss": 4.3055, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901917171734378e-05, | |
| "loss": 4.2952, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.901078576983326e-05, | |
| "loss": 4.2981, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.900239982232274e-05, | |
| "loss": 4.2947, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8994013874812214e-05, | |
| "loss": 4.2937, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.89856279273017e-05, | |
| "loss": 4.2765, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.897724197979118e-05, | |
| "loss": 4.2885, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.896885603228066e-05, | |
| "loss": 4.2835, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.896048646357387e-05, | |
| "loss": 4.2888, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.895210051606335e-05, | |
| "loss": 4.271, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.894371456855283e-05, | |
| "loss": 4.2794, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.893532862104231e-05, | |
| "loss": 4.2694, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.892695905233552e-05, | |
| "loss": 4.2729, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8918573104825e-05, | |
| "loss": 4.2516, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.891018715731448e-05, | |
| "loss": 4.2786, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.890180120980396e-05, | |
| "loss": 4.2767, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.889341526229344e-05, | |
| "loss": 4.2601, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888502931478292e-05, | |
| "loss": 4.2503, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.88766433672724e-05, | |
| "loss": 4.2709, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.886825741976188e-05, | |
| "loss": 4.2564, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8859887851055094e-05, | |
| "loss": 4.2585, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8851501903544574e-05, | |
| "loss": 4.2507, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.884313233483778e-05, | |
| "loss": 4.2591, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.883474638732726e-05, | |
| "loss": 4.2503, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.882636043981674e-05, | |
| "loss": 4.2484, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881797449230622e-05, | |
| "loss": 4.235, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880960492359943e-05, | |
| "loss": 4.2468, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.880121897608891e-05, | |
| "loss": 4.2427, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.879283302857839e-05, | |
| "loss": 4.2355, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.878444708106787e-05, | |
| "loss": 4.2551, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.877606113355735e-05, | |
| "loss": 4.2296, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.876767518604683e-05, | |
| "loss": 4.2371, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875928923853632e-05, | |
| "loss": 4.2402, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.875091966982953e-05, | |
| "loss": 4.2283, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 4.242136001586914, | |
| "eval_runtime": 304.3659, | |
| "eval_samples_per_second": 1253.725, | |
| "eval_steps_per_second": 39.18, | |
| "step": 76320 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.874253372231901e-05, | |
| "loss": 4.2164, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.873414777480849e-05, | |
| "loss": 4.2205, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.872576182729797e-05, | |
| "loss": 4.2306, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.871737587978745e-05, | |
| "loss": 4.2205, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.870898993227693e-05, | |
| "loss": 4.2211, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.87006039847664e-05, | |
| "loss": 4.2057, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8692234416059616e-05, | |
| "loss": 4.2073, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8683848468549096e-05, | |
| "loss": 4.2047, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8675462521038576e-05, | |
| "loss": 4.2127, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8667076573528056e-05, | |
| "loss": 4.1982, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865870700482127e-05, | |
| "loss": 4.2066, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.865033743611448e-05, | |
| "loss": 4.21, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.864195148860396e-05, | |
| "loss": 4.1932, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.863356554109344e-05, | |
| "loss": 4.191, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.862517959358292e-05, | |
| "loss": 4.1788, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.86167936460724e-05, | |
| "loss": 4.1883, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8608407698561874e-05, | |
| "loss": 4.1836, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8600021751051354e-05, | |
| "loss": 4.1786, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8591635803540834e-05, | |
| "loss": 4.1792, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8583249856030314e-05, | |
| "loss": 4.2071, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8574863908519794e-05, | |
| "loss": 4.1769, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8566477961009274e-05, | |
| "loss": 4.178, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8558092013498754e-05, | |
| "loss": 4.1857, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.854972244479197e-05, | |
| "loss": 4.1767, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.854133649728145e-05, | |
| "loss": 4.1722, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.853295054977093e-05, | |
| "loss": 4.1653, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.852456460226041e-05, | |
| "loss": 4.1699, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.851619503355362e-05, | |
| "loss": 4.1672, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.85078090860431e-05, | |
| "loss": 4.1534, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.849942313853258e-05, | |
| "loss": 4.1661, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.849105356982579e-05, | |
| "loss": 4.1531, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.848266762231527e-05, | |
| "loss": 4.1674, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.847428167480475e-05, | |
| "loss": 4.1631, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.846589572729423e-05, | |
| "loss": 4.1475, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8457526158587444e-05, | |
| "loss": 4.1639, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8449140211076924e-05, | |
| "loss": 4.169, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8440754263566404e-05, | |
| "loss": 4.1409, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8432368316055884e-05, | |
| "loss": 4.1518, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8423982368545363e-05, | |
| "loss": 4.1438, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8415596421034843e-05, | |
| "loss": 4.1371, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.840722685232805e-05, | |
| "loss": 4.1428, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839884090481753e-05, | |
| "loss": 4.151, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.839045495730701e-05, | |
| "loss": 4.1471, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.838206900979649e-05, | |
| "loss": 4.1387, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.83736994410897e-05, | |
| "loss": 4.1449, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.836531349357918e-05, | |
| "loss": 4.1333, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.835692754606866e-05, | |
| "loss": 4.1414, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834854159855814e-05, | |
| "loss": 4.1206, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.834017202985136e-05, | |
| "loss": 4.1247, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.833178608234084e-05, | |
| "loss": 4.1389, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.832340013483032e-05, | |
| "loss": 4.1385, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.83150141873198e-05, | |
| "loss": 4.1259, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.830662823980928e-05, | |
| "loss": 4.1163, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8298258671102486e-05, | |
| "loss": 4.1294, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8289872723591966e-05, | |
| "loss": 4.1106, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8281486776081446e-05, | |
| "loss": 4.1372, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8273100828570926e-05, | |
| "loss": 4.1185, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8264731259864135e-05, | |
| "loss": 4.1334, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8256345312353615e-05, | |
| "loss": 4.1184, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8247959364843095e-05, | |
| "loss": 4.1088, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823957341733258e-05, | |
| "loss": 4.1139, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.823118746982206e-05, | |
| "loss": 4.1173, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.822281790111527e-05, | |
| "loss": 4.1048, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.821443195360475e-05, | |
| "loss": 4.103, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.820604600609423e-05, | |
| "loss": 4.1186, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.819766005858371e-05, | |
| "loss": 4.1021, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.818929048987692e-05, | |
| "loss": 4.0946, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.81809045423664e-05, | |
| "loss": 4.0924, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.817251859485588e-05, | |
| "loss": 4.0935, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.816413264734535e-05, | |
| "loss": 4.1131, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.815576307863857e-05, | |
| "loss": 4.111, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.814737713112805e-05, | |
| "loss": 4.0958, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8138991183617535e-05, | |
| "loss": 4.0989, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.813060523610701e-05, | |
| "loss": 4.1028, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.812221928859649e-05, | |
| "loss": 4.1, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8113849719889704e-05, | |
| "loss": 4.0913, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8105463772379184e-05, | |
| "loss": 4.0992, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.809707782486866e-05, | |
| "loss": 4.0928, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808869187735814e-05, | |
| "loss": 4.0971, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.808032230865135e-05, | |
| "loss": 4.0878, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8071936361140827e-05, | |
| "loss": 4.0914, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8063550413630307e-05, | |
| "loss": 4.0925, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8055164466119786e-05, | |
| "loss": 4.0914, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8046794897413e-05, | |
| "loss": 4.0731, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803840894990248e-05, | |
| "loss": 4.0883, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.803002300239196e-05, | |
| "loss": 4.084, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.802163705488144e-05, | |
| "loss": 4.0861, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.801326748617466e-05, | |
| "loss": 4.0826, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.800488153866413e-05, | |
| "loss": 4.0697, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.799649559115361e-05, | |
| "loss": 4.083, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.798810964364309e-05, | |
| "loss": 4.0743, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.79797400749363e-05, | |
| "loss": 4.0839, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.797135412742578e-05, | |
| "loss": 4.0678, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.796296817991526e-05, | |
| "loss": 4.0737, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7954598611208476e-05, | |
| "loss": 4.085, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7946212663697956e-05, | |
| "loss": 4.0719, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7937826716187436e-05, | |
| "loss": 4.0545, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7929440768676916e-05, | |
| "loss": 4.0684, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7921054821166396e-05, | |
| "loss": 4.0729, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7912668873655876e-05, | |
| "loss": 4.074, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7904282926145356e-05, | |
| "loss": 4.0667, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7895896978634836e-05, | |
| "loss": 4.0613, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7887527409928045e-05, | |
| "loss": 4.0587, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7879141462417525e-05, | |
| "loss": 4.0689, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7870755514907005e-05, | |
| "loss": 4.0494, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7862385946200214e-05, | |
| "loss": 4.0546, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7853999998689694e-05, | |
| "loss": 4.0633, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.784561405117918e-05, | |
| "loss": 4.0634, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.783722810366866e-05, | |
| "loss": 4.0416, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782884215615814e-05, | |
| "loss": 4.0557, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.782045620864762e-05, | |
| "loss": 4.0444, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.78120702611371e-05, | |
| "loss": 4.0614, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.780368431362658e-05, | |
| "loss": 4.0484, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.779529836611606e-05, | |
| "loss": 4.043, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.778692879740927e-05, | |
| "loss": 4.0614, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777854284989875e-05, | |
| "loss": 4.0539, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.777017328119196e-05, | |
| "loss": 4.0473, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.776178733368144e-05, | |
| "loss": 4.0521, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.775340138617092e-05, | |
| "loss": 4.048, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.77450154386604e-05, | |
| "loss": 4.0502, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.773662949114988e-05, | |
| "loss": 4.0333, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.772824354363936e-05, | |
| "loss": 4.0525, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7719857596128845e-05, | |
| "loss": 4.0402, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.771147164861832e-05, | |
| "loss": 4.05, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7703102079911534e-05, | |
| "loss": 4.0403, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7694716132401014e-05, | |
| "loss": 4.0418, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7686330184890494e-05, | |
| "loss": 4.036, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.767794423737997e-05, | |
| "loss": 4.0399, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766957466867318e-05, | |
| "loss": 4.0225, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.766118872116266e-05, | |
| "loss": 4.0514, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7652802773652136e-05, | |
| "loss": 4.0432, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7644416826141616e-05, | |
| "loss": 4.0364, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.763604725743483e-05, | |
| "loss": 4.0277, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.762766130992431e-05, | |
| "loss": 4.0449, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761927536241379e-05, | |
| "loss": 4.04, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.761088941490327e-05, | |
| "loss": 4.0326, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.760251984619649e-05, | |
| "loss": 4.0287, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.759413389868596e-05, | |
| "loss": 4.0383, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.758574795117544e-05, | |
| "loss": 4.0326, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.757736200366492e-05, | |
| "loss": 4.0423, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7568992434958137e-05, | |
| "loss": 4.0168, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.756060648744761e-05, | |
| "loss": 4.0328, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.755222053993709e-05, | |
| "loss": 4.0293, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.754383459242657e-05, | |
| "loss": 4.0217, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7535465023719786e-05, | |
| "loss": 4.0447, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7527079076209266e-05, | |
| "loss": 4.0225, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7518693128698745e-05, | |
| "loss": 4.029, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7510307181188225e-05, | |
| "loss": 4.0274, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7501921233677705e-05, | |
| "loss": 4.0271, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 4.07383918762207, | |
| "eval_runtime": 304.4679, | |
| "eval_samples_per_second": 1253.304, | |
| "eval_steps_per_second": 39.167, | |
| "step": 152640 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7493535286167185e-05, | |
| "loss": 4.0117, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7485149338656665e-05, | |
| "loss": 4.0201, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7476763391146145e-05, | |
| "loss": 4.0274, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7468377443635625e-05, | |
| "loss": 4.0218, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7459991496125105e-05, | |
| "loss": 4.0283, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7451605548614585e-05, | |
| "loss": 4.0047, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7443219601104065e-05, | |
| "loss": 4.012, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7434833653593545e-05, | |
| "loss": 4.0094, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7426447706083025e-05, | |
| "loss": 4.02, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.7418061758572505e-05, | |
| "loss": 4.0096, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.740969218986572e-05, | |
| "loss": 4.0155, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.74013062423552e-05, | |
| "loss": 4.0195, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.739293667364841e-05, | |
| "loss": 4.0025, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.738455072613789e-05, | |
| "loss": 4.0061, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.737616477862737e-05, | |
| "loss": 3.9889, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.736777883111685e-05, | |
| "loss": 4.0012, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.735939288360632e-05, | |
| "loss": 4.0009, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.73510069360958e-05, | |
| "loss": 3.9965, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.734262098858528e-05, | |
| "loss": 3.9973, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.733423504107476e-05, | |
| "loss": 4.0216, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.732584909356424e-05, | |
| "loss": 3.9989, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.731746314605372e-05, | |
| "loss": 3.9976, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.73090771985432e-05, | |
| "loss": 4.005, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.730070762983642e-05, | |
| "loss": 4.002, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.72923216823259e-05, | |
| "loss": 3.9925, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.728393573481538e-05, | |
| "loss": 3.9884, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.727554978730486e-05, | |
| "loss": 3.9992, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.726718021859807e-05, | |
| "loss": 3.9885, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.725879427108755e-05, | |
| "loss": 3.9824, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.725040832357703e-05, | |
| "loss": 3.9946, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.724202237606651e-05, | |
| "loss": 3.9858, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7233652807359716e-05, | |
| "loss": 3.9962, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7225266859849196e-05, | |
| "loss": 3.9918, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7216880912338676e-05, | |
| "loss": 3.9797, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7208494964828156e-05, | |
| "loss": 3.9942, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.720012539612137e-05, | |
| "loss": 4.0048, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.719173944861085e-05, | |
| "loss": 3.9724, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.718335350110033e-05, | |
| "loss": 3.9881, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.717496755358981e-05, | |
| "loss": 3.9769, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.716659798488302e-05, | |
| "loss": 3.9782, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.71582120373725e-05, | |
| "loss": 3.9769, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714982608986198e-05, | |
| "loss": 3.9925, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.714144014235146e-05, | |
| "loss": 3.9858, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.713307057364467e-05, | |
| "loss": 3.9793, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.712468462613415e-05, | |
| "loss": 3.9843, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.711629867862363e-05, | |
| "loss": 3.9729, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.710791273111311e-05, | |
| "loss": 3.9861, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7099543162406326e-05, | |
| "loss": 3.9657, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7091157214895806e-05, | |
| "loss": 3.9698, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7082771267385286e-05, | |
| "loss": 3.982, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7074385319874766e-05, | |
| "loss": 3.9796, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7066015751167975e-05, | |
| "loss": 3.9719, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7057629803657455e-05, | |
| "loss": 3.9668, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7049243856146935e-05, | |
| "loss": 3.976, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7040857908636415e-05, | |
| "loss": 3.9563, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7032488339929624e-05, | |
| "loss": 3.9814, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7024102392419104e-05, | |
| "loss": 3.9718, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7015716444908584e-05, | |
| "loss": 3.9828, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7007330497398064e-05, | |
| "loss": 3.974, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699896092869128e-05, | |
| "loss": 3.9605, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.699057498118076e-05, | |
| "loss": 3.9639, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.698218903367024e-05, | |
| "loss": 3.9705, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.697380308615972e-05, | |
| "loss": 3.9572, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.696543351745293e-05, | |
| "loss": 3.9581, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.695704756994241e-05, | |
| "loss": 3.9704, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.694866162243189e-05, | |
| "loss": 3.9603, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.694027567492137e-05, | |
| "loss": 3.9496, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.693190610621458e-05, | |
| "loss": 3.9512, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.692352015870406e-05, | |
| "loss": 3.9499, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.691513421119354e-05, | |
| "loss": 3.9706, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.690674826368302e-05, | |
| "loss": 3.9663, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.689837869497623e-05, | |
| "loss": 3.9568, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688999274746571e-05, | |
| "loss": 3.9621, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.688160679995519e-05, | |
| "loss": 3.9582, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.687322085244467e-05, | |
| "loss": 3.9621, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.686485128373788e-05, | |
| "loss": 3.9527, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.685648171503109e-05, | |
| "loss": 3.9625, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.684809576752057e-05, | |
| "loss": 3.9574, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683970982001005e-05, | |
| "loss": 3.9559, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.683132387249953e-05, | |
| "loss": 3.9513, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.682293792498901e-05, | |
| "loss": 3.9577, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.681455197747849e-05, | |
| "loss": 3.9568, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.680616602996797e-05, | |
| "loss": 3.9568, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.679779646126119e-05, | |
| "loss": 3.9437, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678941051375067e-05, | |
| "loss": 3.9516, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.678102456624015e-05, | |
| "loss": 3.9498, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.677263861872963e-05, | |
| "loss": 3.9508, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6764269050022836e-05, | |
| "loss": 3.9521, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6755883102512316e-05, | |
| "loss": 3.9366, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6747497155001796e-05, | |
| "loss": 3.9485, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6739111207491276e-05, | |
| "loss": 3.9459, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6730741638784485e-05, | |
| "loss": 3.9517, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6722355691273965e-05, | |
| "loss": 3.9352, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6713969743763445e-05, | |
| "loss": 3.9445, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6705583796252925e-05, | |
| "loss": 3.9573, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6697197848742405e-05, | |
| "loss": 3.9435, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6688811901231885e-05, | |
| "loss": 3.9273, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6680425953721365e-05, | |
| "loss": 3.9407, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6672040006210844e-05, | |
| "loss": 3.9464, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6663670437504054e-05, | |
| "loss": 3.9479, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6655284489993534e-05, | |
| "loss": 3.9421, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6646898542483013e-05, | |
| "loss": 3.9321, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6638512594972493e-05, | |
| "loss": 3.9336, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.663015940506944e-05, | |
| "loss": 3.9445, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.662177345755892e-05, | |
| "loss": 3.9269, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.66133875100484e-05, | |
| "loss": 3.9315, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.660500156253788e-05, | |
| "loss": 3.9388, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.659661561502736e-05, | |
| "loss": 3.9394, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6588246046320574e-05, | |
| "loss": 3.9179, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6579860098810054e-05, | |
| "loss": 3.9342, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.657147415129953e-05, | |
| "loss": 3.9222, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.656308820378901e-05, | |
| "loss": 3.9355, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.655470225627849e-05, | |
| "loss": 3.9327, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.654631630876797e-05, | |
| "loss": 3.9249, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.653793036125745e-05, | |
| "loss": 3.9387, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.652954441374693e-05, | |
| "loss": 3.9374, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6521174845040136e-05, | |
| "loss": 3.9265, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6512788897529616e-05, | |
| "loss": 3.933, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6504402950019096e-05, | |
| "loss": 3.9316, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.649601700250858e-05, | |
| "loss": 3.9299, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.648763105499806e-05, | |
| "loss": 3.9179, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647924510748754e-05, | |
| "loss": 3.9327, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.647085915997702e-05, | |
| "loss": 3.9249, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.646248959127023e-05, | |
| "loss": 3.9311, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.645412002256344e-05, | |
| "loss": 3.9276, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.644573407505292e-05, | |
| "loss": 3.9299, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.64373481275424e-05, | |
| "loss": 3.9173, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642896218003188e-05, | |
| "loss": 3.9233, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.642057623252136e-05, | |
| "loss": 3.9102, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.641220666381457e-05, | |
| "loss": 3.9367, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.640382071630405e-05, | |
| "loss": 3.9255, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6395434768793536e-05, | |
| "loss": 3.9249, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6387048821283016e-05, | |
| "loss": 3.9156, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6378662873772496e-05, | |
| "loss": 3.9314, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6370276926261976e-05, | |
| "loss": 3.9291, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6361890978751456e-05, | |
| "loss": 3.9171, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6353505031240936e-05, | |
| "loss": 3.9192, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6345151841337874e-05, | |
| "loss": 3.926, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6336765893827354e-05, | |
| "loss": 3.9219, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6328379946316834e-05, | |
| "loss": 3.9305, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6319993998806314e-05, | |
| "loss": 3.9125, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6311624430099523e-05, | |
| "loss": 3.9214, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6303238482589003e-05, | |
| "loss": 3.917, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.629485253507848e-05, | |
| "loss": 3.9114, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.628646658756797e-05, | |
| "loss": 3.9352, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.627808064005745e-05, | |
| "loss": 3.9133, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.626971107135066e-05, | |
| "loss": 3.9207, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.626132512384014e-05, | |
| "loss": 3.9189, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.625293917632962e-05, | |
| "loss": 3.9174, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 3.9986412525177, | |
| "eval_runtime": 302.0543, | |
| "eval_samples_per_second": 1263.319, | |
| "eval_steps_per_second": 39.48, | |
| "step": 228960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.62445532288191e-05, | |
| "loss": 3.9113, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.623616728130858e-05, | |
| "loss": 3.9116, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.622778133379806e-05, | |
| "loss": 3.9206, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.621939538628753e-05, | |
| "loss": 3.9163, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.621102581758075e-05, | |
| "loss": 3.9216, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.620263987007023e-05, | |
| "loss": 3.9032, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.619427030136344e-05, | |
| "loss": 3.9092, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6185884353852924e-05, | |
| "loss": 3.8974, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6177498406342404e-05, | |
| "loss": 3.9199, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6169112458831884e-05, | |
| "loss": 3.904, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.616074289012509e-05, | |
| "loss": 3.9119, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.615235694261457e-05, | |
| "loss": 3.9149, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.614398737390778e-05, | |
| "loss": 3.9037, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.613560142639726e-05, | |
| "loss": 3.8959, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.612721547888674e-05, | |
| "loss": 3.893, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.611882953137622e-05, | |
| "loss": 3.8932, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.61104435838657e-05, | |
| "loss": 3.9012, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.610205763635518e-05, | |
| "loss": 3.8978, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.609367168884466e-05, | |
| "loss": 3.8959, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.608528574133414e-05, | |
| "loss": 3.9198, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.607691617262736e-05, | |
| "loss": 3.9016, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606853022511684e-05, | |
| "loss": 3.8988, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.606014427760631e-05, | |
| "loss": 3.9048, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.605175833009579e-05, | |
| "loss": 3.9034, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.604337238258527e-05, | |
| "loss": 3.8952, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.603500281387848e-05, | |
| "loss": 3.8898, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.602661686636796e-05, | |
| "loss": 3.9044, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.601823091885744e-05, | |
| "loss": 3.8854, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.600984497134692e-05, | |
| "loss": 3.8878, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.60014590238364e-05, | |
| "loss": 3.8933, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5993089455129615e-05, | |
| "loss": 3.8874, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5984703507619095e-05, | |
| "loss": 3.9007, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5976317560108575e-05, | |
| "loss": 3.8932, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5967931612598055e-05, | |
| "loss": 3.8845, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5959545665087535e-05, | |
| "loss": 3.8975, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5951176096380744e-05, | |
| "loss": 3.9056, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5942790148870224e-05, | |
| "loss": 3.8818, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5934404201359704e-05, | |
| "loss": 3.893, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5926018253849184e-05, | |
| "loss": 3.8787, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5917632306338664e-05, | |
| "loss": 3.8928, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.590926273763187e-05, | |
| "loss": 3.8783, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.590087679012135e-05, | |
| "loss": 3.9012, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.589249084261083e-05, | |
| "loss": 3.891, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.588410489510031e-05, | |
| "loss": 3.8868, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.587573532639353e-05, | |
| "loss": 3.8917, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.586734937888301e-05, | |
| "loss": 3.88, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.585896343137249e-05, | |
| "loss": 3.8932, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.585057748386197e-05, | |
| "loss": 3.8773, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.584220791515518e-05, | |
| "loss": 3.8752, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.583382196764466e-05, | |
| "loss": 3.8905, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.582543602013414e-05, | |
| "loss": 3.8856, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.581705007262362e-05, | |
| "loss": 3.8859, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.580868050391683e-05, | |
| "loss": 3.8744, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.580029455640631e-05, | |
| "loss": 3.8835, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.579190860889579e-05, | |
| "loss": 3.8674, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5783522661385267e-05, | |
| "loss": 3.8887, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.577513671387475e-05, | |
| "loss": 3.882, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.576676714516796e-05, | |
| "loss": 3.895, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.575838119765744e-05, | |
| "loss": 3.8829, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.574999525014692e-05, | |
| "loss": 3.8757, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.57416093026364e-05, | |
| "loss": 3.8713, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.573323973392961e-05, | |
| "loss": 3.8823, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.572485378641909e-05, | |
| "loss": 3.8708, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.571646783890857e-05, | |
| "loss": 3.8688, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.570808189139805e-05, | |
| "loss": 3.8791, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.569969594388753e-05, | |
| "loss": 3.8758, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.569130999637701e-05, | |
| "loss": 3.8591, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.568294042767022e-05, | |
| "loss": 3.8646, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.567455448015971e-05, | |
| "loss": 3.8623, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.566616853264919e-05, | |
| "loss": 3.8871, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.565778258513867e-05, | |
| "loss": 3.8772, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5649413016431876e-05, | |
| "loss": 3.8706, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5641027068921356e-05, | |
| "loss": 3.8766, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5632641121410836e-05, | |
| "loss": 3.87, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5624255173900316e-05, | |
| "loss": 3.8824, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.561586922638979e-05, | |
| "loss": 3.8625, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5607499657683005e-05, | |
| "loss": 3.8831, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5599113710172485e-05, | |
| "loss": 3.8671, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5590727762661965e-05, | |
| "loss": 3.8738, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5582341815151445e-05, | |
| "loss": 3.8719, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5573972246444654e-05, | |
| "loss": 3.8697, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.556558629893414e-05, | |
| "loss": 3.8779, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5557200351423614e-05, | |
| "loss": 3.8694, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.5548814403913094e-05, | |
| "loss": 3.8588, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.554044483520631e-05, | |
| "loss": 3.8688, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.553205888769579e-05, | |
| "loss": 3.8629, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.552367294018526e-05, | |
| "loss": 3.8733, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.551528699267474e-05, | |
| "loss": 3.8681, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.550690104516422e-05, | |
| "loss": 3.8537, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.549853147645744e-05, | |
| "loss": 3.8678, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.549014552894691e-05, | |
| "loss": 3.8594, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.548175958143639e-05, | |
| "loss": 3.8687, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.547337363392588e-05, | |
| "loss": 3.8555, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.546500406521909e-05, | |
| "loss": 3.8651, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.545661811770857e-05, | |
| "loss": 3.875, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.544823217019805e-05, | |
| "loss": 3.8596, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.543984622268753e-05, | |
| "loss": 3.8504, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5431476653980736e-05, | |
| "loss": 3.8606, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5423090706470216e-05, | |
| "loss": 3.8573, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5414704758959696e-05, | |
| "loss": 3.8718, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5406318811449176e-05, | |
| "loss": 3.8623, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5397949242742385e-05, | |
| "loss": 3.854, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5389563295231865e-05, | |
| "loss": 3.8553, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5381177347721345e-05, | |
| "loss": 3.8657, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.537279140021083e-05, | |
| "loss": 3.8444, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.536442183150404e-05, | |
| "loss": 3.856, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.535603588399352e-05, | |
| "loss": 3.858, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5347649936483e-05, | |
| "loss": 3.8623, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533926398897248e-05, | |
| "loss": 3.8344, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.533089442026569e-05, | |
| "loss": 3.8563, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.532250847275517e-05, | |
| "loss": 3.8508, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.531412252524465e-05, | |
| "loss": 3.8495, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.530573657773413e-05, | |
| "loss": 3.8555, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.529735063022361e-05, | |
| "loss": 3.8444, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.528898106151682e-05, | |
| "loss": 3.8642, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.52805951140063e-05, | |
| "loss": 3.8557, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5272209166495786e-05, | |
| "loss": 3.8489, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5263823218985266e-05, | |
| "loss": 3.8557, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5255453650278475e-05, | |
| "loss": 3.8537, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5247067702767955e-05, | |
| "loss": 3.8543, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5238681755257435e-05, | |
| "loss": 3.836, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5230295807746915e-05, | |
| "loss": 3.8591, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5221926239040124e-05, | |
| "loss": 3.8451, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5213540291529604e-05, | |
| "loss": 3.852, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5205154344019084e-05, | |
| "loss": 3.8504, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5196768396508564e-05, | |
| "loss": 3.8577, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.518839882780177e-05, | |
| "loss": 3.8463, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.518001288029125e-05, | |
| "loss": 3.8443, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.517162693278074e-05, | |
| "loss": 3.8369, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.516324098527022e-05, | |
| "loss": 3.8591, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.515487141656343e-05, | |
| "loss": 3.8519, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.514648546905291e-05, | |
| "loss": 3.851, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.513809952154239e-05, | |
| "loss": 3.8413, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512971357403187e-05, | |
| "loss": 3.8562, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.512134400532508e-05, | |
| "loss": 3.8515, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.511295805781456e-05, | |
| "loss": 3.8423, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.510457211030404e-05, | |
| "loss": 3.8449, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.509618616279352e-05, | |
| "loss": 3.851, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5087800215283e-05, | |
| "loss": 3.8502, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5079430646576206e-05, | |
| "loss": 3.856, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.507104469906569e-05, | |
| "loss": 3.8377, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.506265875155517e-05, | |
| "loss": 3.848, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.505427280404465e-05, | |
| "loss": 3.8435, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.504590323533786e-05, | |
| "loss": 3.8421, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.503753366663107e-05, | |
| "loss": 3.8588, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.502914771912055e-05, | |
| "loss": 3.8428, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.502076177161003e-05, | |
| "loss": 3.8474, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.501237582409951e-05, | |
| "loss": 3.8413, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.500398987658899e-05, | |
| "loss": 3.8489, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 3.9596714973449707, | |
| "eval_runtime": 361.5509, | |
| "eval_samples_per_second": 1055.428, | |
| "eval_steps_per_second": 32.983, | |
| "step": 305280 | |
| } | |
| ], | |
| "logging_steps": 512, | |
| "max_steps": 3052726, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 10, | |
| "total_flos": 1.9807281118550016e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |