diff --git "a/checkpoint-2900145/trainer_state.json" "b/checkpoint-2900145/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-2900145/trainer_state.json" @@ -0,0 +1,34313 @@ +{ + "best_metric": 3.979407548904419, + "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl/lstm/2/checkpoints/checkpoint-2900145", + "epoch": 0.025000606015738065, + "eval_steps": 10, + "global_step": 2900145, + "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": 10.8203, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 4.999161405248948e-05, + "loss": 7.5687, + "step": 512 + }, + { + "epoch": 0.0, + "learning_rate": 4.998322810497896e-05, + "loss": 7.0534, + "step": 1024 + }, + { + "epoch": 0.0, + "learning_rate": 4.997484215746844e-05, + "loss": 6.9801, + "step": 1536 + }, + { + "epoch": 0.0, + "learning_rate": 4.996645620995792e-05, + "loss": 6.9528, + "step": 2048 + }, + { + "epoch": 0.0, + "learning_rate": 4.99580702624474e-05, + "loss": 6.8928, + "step": 2560 + }, + { + "epoch": 0.0, + "learning_rate": 4.994970069374061e-05, + "loss": 6.7305, + "step": 3072 + }, + { + "epoch": 0.0, + "learning_rate": 4.994131474623009e-05, + "loss": 6.6212, + "step": 3584 + }, + { + "epoch": 0.0, + "learning_rate": 4.993292879871958e-05, + "loss": 6.531, + "step": 4096 + }, + { + "epoch": 0.0, + "learning_rate": 4.992454285120906e-05, + "loss": 6.4382, + "step": 4608 + }, + { + "epoch": 0.0, + "learning_rate": 4.991615690369854e-05, + "loss": 6.3957, + "step": 5120 + }, + { + "epoch": 0.0, + "learning_rate": 4.990777095618801e-05, + "loss": 6.3302, + "step": 5632 + }, + { + "epoch": 0.0, + "learning_rate": 4.989938500867749e-05, + "loss": 6.261, + "step": 6144 + }, + { + "epoch": 0.0, + "learning_rate": 4.989099906116697e-05, + "loss": 6.1908, + "step": 6656 + }, + { + "epoch": 0.0, + "learning_rate": 4.988261311365645e-05, + "loss": 6.1396, + "step": 7168 + }, + { + "epoch": 0.0, + "learning_rate": 4.987422716614593e-05, + "loss": 6.0984, + "step": 7680 + }, + { + "epoch": 0.0, + "learning_rate": 4.986584121863541e-05, + "loss": 6.0549, + "step": 8192 + }, + { + "epoch": 0.0, + "learning_rate": 4.985745527112489e-05, + "loss": 5.997, + "step": 8704 + }, + { + "epoch": 0.0, + "learning_rate": 4.984906932361437e-05, + "loss": 5.9588, + "step": 9216 + }, + { + "epoch": 0.0, + "learning_rate": 4.984068337610385e-05, + "loss": 5.9206, + "step": 9728 + }, + { + "epoch": 0.0, + "learning_rate": 4.983229742859333e-05, + "loss": 5.8828, + "step": 10240 + }, + { + "epoch": 0.0, + "learning_rate": 4.982391148108281e-05, + "loss": 5.8553, + "step": 10752 + }, + { + "epoch": 0.0, + "learning_rate": 4.981552553357229e-05, + "loss": 5.8258, + "step": 11264 + }, + { + "epoch": 0.0, + "learning_rate": 4.980713958606178e-05, + "loss": 5.7928, + "step": 11776 + }, + { + "epoch": 0.0, + "learning_rate": 4.979875363855125e-05, + "loss": 5.7648, + "step": 12288 + }, + { + "epoch": 0.0, + "learning_rate": 4.979036769104073e-05, + "loss": 5.7474, + "step": 12800 + }, + { + "epoch": 0.0, + "learning_rate": 4.9781998122333946e-05, + "loss": 5.7141, + "step": 13312 + }, + { + "epoch": 0.0, + "learning_rate": 4.9773628553627155e-05, + "loss": 5.6924, + "step": 13824 + }, + { + "epoch": 0.0, + "learning_rate": 4.9765242606116635e-05, + "loss": 5.6582, + "step": 14336 + }, + { + "epoch": 0.0, + "learning_rate": 4.9756856658606115e-05, + "loss": 5.6537, + "step": 14848 + }, + { + "epoch": 0.01, + "learning_rate": 4.9748470711095595e-05, + "loss": 5.6206, + "step": 15360 + }, + { + "epoch": 0.01, + "learning_rate": 4.9740101142388804e-05, + "loss": 5.604, + "step": 15872 + }, + { + "epoch": 0.01, + "learning_rate": 4.9731715194878284e-05, + "loss": 5.5936, + "step": 16384 + }, + { + "epoch": 0.01, + "learning_rate": 4.9723329247367764e-05, + "loss": 5.5759, + "step": 16896 + }, + { + "epoch": 0.01, + "learning_rate": 4.9714943299857244e-05, + "loss": 5.5452, + "step": 17408 + }, + { + "epoch": 0.01, + "learning_rate": 4.970657373115046e-05, + "loss": 5.5326, + "step": 17920 + }, + { + "epoch": 0.01, + "learning_rate": 4.969820416244367e-05, + "loss": 5.5245, + "step": 18432 + }, + { + "epoch": 0.01, + "learning_rate": 4.968981821493315e-05, + "loss": 5.5017, + "step": 18944 + }, + { + "epoch": 0.01, + "learning_rate": 4.968143226742263e-05, + "loss": 5.4895, + "step": 19456 + }, + { + "epoch": 0.01, + "learning_rate": 4.967304631991211e-05, + "loss": 5.4689, + "step": 19968 + }, + { + "epoch": 0.01, + "learning_rate": 4.966466037240159e-05, + "loss": 5.4454, + "step": 20480 + }, + { + "epoch": 0.01, + "learning_rate": 4.965627442489107e-05, + "loss": 5.4333, + "step": 20992 + }, + { + "epoch": 0.01, + "learning_rate": 4.964788847738054e-05, + "loss": 5.4194, + "step": 21504 + }, + { + "epoch": 0.01, + "learning_rate": 4.963950252987002e-05, + "loss": 5.4124, + "step": 22016 + }, + { + "epoch": 0.01, + "learning_rate": 4.963113296116324e-05, + "loss": 5.4004, + "step": 22528 + }, + { + "epoch": 0.01, + "learning_rate": 4.962274701365272e-05, + "loss": 5.3951, + "step": 23040 + }, + { + "epoch": 0.01, + "learning_rate": 4.9614377444945934e-05, + "loss": 5.3712, + "step": 23552 + }, + { + "epoch": 0.01, + "learning_rate": 4.9605991497435414e-05, + "loss": 5.3707, + "step": 24064 + }, + { + "epoch": 0.01, + "learning_rate": 4.9597605549924894e-05, + "loss": 5.3748, + "step": 24576 + }, + { + "epoch": 0.01, + "learning_rate": 4.9589219602414374e-05, + "loss": 5.347, + "step": 25088 + }, + { + "epoch": 0.01, + "learning_rate": 4.958083365490385e-05, + "loss": 5.3368, + "step": 25600 + }, + { + "epoch": 0.01, + "learning_rate": 4.957244770739333e-05, + "loss": 5.3234, + "step": 26112 + }, + { + "epoch": 0.01, + "learning_rate": 4.956406175988281e-05, + "loss": 5.3324, + "step": 26624 + }, + { + "epoch": 0.01, + "learning_rate": 4.955567581237229e-05, + "loss": 5.295, + "step": 27136 + }, + { + "epoch": 0.01, + "learning_rate": 4.9547306243665496e-05, + "loss": 5.3127, + "step": 27648 + }, + { + "epoch": 0.01, + "learning_rate": 4.953893667495871e-05, + "loss": 5.2827, + "step": 28160 + }, + { + "epoch": 0.01, + "learning_rate": 4.953055072744819e-05, + "loss": 5.2793, + "step": 28672 + }, + { + "epoch": 0.01, + "learning_rate": 4.952216477993767e-05, + "loss": 5.2744, + "step": 29184 + }, + { + "epoch": 0.01, + "learning_rate": 4.951377883242715e-05, + "loss": 5.2637, + "step": 29696 + }, + { + "epoch": 0.01, + "learning_rate": 4.950539288491663e-05, + "loss": 5.2516, + "step": 30208 + }, + { + "epoch": 0.01, + "learning_rate": 4.949700693740611e-05, + "loss": 5.2388, + "step": 30720 + }, + { + "epoch": 0.01, + "learning_rate": 4.948862098989559e-05, + "loss": 5.2241, + "step": 31232 + }, + { + "epoch": 0.01, + "learning_rate": 4.948023504238507e-05, + "loss": 5.2312, + "step": 31744 + }, + { + "epoch": 0.01, + "learning_rate": 4.947184909487455e-05, + "loss": 5.22, + "step": 32256 + }, + { + "epoch": 0.01, + "learning_rate": 4.946346314736403e-05, + "loss": 5.2093, + "step": 32768 + }, + { + "epoch": 0.01, + "learning_rate": 4.945509357865724e-05, + "loss": 5.2011, + "step": 33280 + }, + { + "epoch": 0.01, + "learning_rate": 4.944670763114672e-05, + "loss": 5.1917, + "step": 33792 + }, + { + "epoch": 0.01, + "learning_rate": 4.94383216836362e-05, + "loss": 5.1736, + "step": 34304 + }, + { + "epoch": 0.01, + "learning_rate": 4.942993573612568e-05, + "loss": 5.1913, + "step": 34816 + }, + { + "epoch": 0.01, + "learning_rate": 4.942156616741889e-05, + "loss": 5.1683, + "step": 35328 + }, + { + "epoch": 0.01, + "learning_rate": 4.941318021990837e-05, + "loss": 5.1615, + "step": 35840 + }, + { + "epoch": 0.01, + "learning_rate": 4.9404794272397856e-05, + "loss": 5.1706, + "step": 36352 + }, + { + "epoch": 0.01, + "learning_rate": 4.9396408324887336e-05, + "loss": 5.1465, + "step": 36864 + }, + { + "epoch": 0.01, + "learning_rate": 4.9388022377376816e-05, + "loss": 5.1592, + "step": 37376 + }, + { + "epoch": 0.01, + "learning_rate": 4.9379652808670025e-05, + "loss": 5.1496, + "step": 37888 + }, + { + "epoch": 0.01, + "learning_rate": 4.9371266861159505e-05, + "loss": 5.139, + "step": 38400 + }, + { + "epoch": 0.01, + "learning_rate": 4.9362897292452714e-05, + "loss": 5.137, + "step": 38912 + }, + { + "epoch": 0.01, + "learning_rate": 4.9354511344942194e-05, + "loss": 5.1119, + "step": 39424 + }, + { + "epoch": 0.01, + "learning_rate": 4.93461417762354e-05, + "loss": 5.108, + "step": 39936 + }, + { + "epoch": 0.01, + "learning_rate": 4.933775582872488e-05, + "loss": 5.1074, + "step": 40448 + }, + { + "epoch": 0.01, + "learning_rate": 4.932936988121436e-05, + "loss": 5.1066, + "step": 40960 + }, + { + "epoch": 0.01, + "learning_rate": 4.932098393370384e-05, + "loss": 5.0962, + "step": 41472 + }, + { + "epoch": 0.01, + "learning_rate": 4.931259798619332e-05, + "loss": 5.0821, + "step": 41984 + }, + { + "epoch": 0.01, + "learning_rate": 4.930421203868281e-05, + "loss": 5.0683, + "step": 42496 + }, + { + "epoch": 0.01, + "learning_rate": 4.929582609117229e-05, + "loss": 5.0755, + "step": 43008 + }, + { + "epoch": 0.01, + "learning_rate": 4.928744014366177e-05, + "loss": 5.0744, + "step": 43520 + }, + { + "epoch": 0.01, + "learning_rate": 4.927905419615125e-05, + "loss": 5.0703, + "step": 44032 + }, + { + "epoch": 0.01, + "learning_rate": 4.927066824864073e-05, + "loss": 5.0674, + "step": 44544 + }, + { + "epoch": 0.01, + "learning_rate": 4.92622823011302e-05, + "loss": 5.0616, + "step": 45056 + }, + { + "epoch": 0.01, + "learning_rate": 4.925389635361968e-05, + "loss": 5.0502, + "step": 45568 + }, + { + "epoch": 0.02, + "learning_rate": 4.92455267849129e-05, + "loss": 5.0315, + "step": 46080 + }, + { + "epoch": 0.02, + "learning_rate": 4.923714083740238e-05, + "loss": 5.0285, + "step": 46592 + }, + { + "epoch": 0.02, + "learning_rate": 4.922875488989185e-05, + "loss": 5.0339, + "step": 47104 + }, + { + "epoch": 0.02, + "learning_rate": 4.922038532118507e-05, + "loss": 5.0307, + "step": 47616 + }, + { + "epoch": 0.02, + "learning_rate": 4.921199937367455e-05, + "loss": 5.0084, + "step": 48128 + }, + { + "epoch": 0.02, + "learning_rate": 4.920361342616403e-05, + "loss": 5.0133, + "step": 48640 + }, + { + "epoch": 0.02, + "learning_rate": 4.919522747865351e-05, + "loss": 5.0087, + "step": 49152 + }, + { + "epoch": 0.02, + "learning_rate": 4.918684153114299e-05, + "loss": 5.0097, + "step": 49664 + }, + { + "epoch": 0.02, + "learning_rate": 4.91784719624362e-05, + "loss": 5.0005, + "step": 50176 + }, + { + "epoch": 0.02, + "learning_rate": 4.9170086014925676e-05, + "loss": 5.0025, + "step": 50688 + }, + { + "epoch": 0.02, + "learning_rate": 4.9161700067415156e-05, + "loss": 4.9942, + "step": 51200 + }, + { + "epoch": 0.02, + "learning_rate": 4.9153314119904636e-05, + "loss": 4.9846, + "step": 51712 + }, + { + "epoch": 0.02, + "learning_rate": 4.9144928172394116e-05, + "loss": 4.9828, + "step": 52224 + }, + { + "epoch": 0.02, + "learning_rate": 4.9136542224883596e-05, + "loss": 4.9844, + "step": 52736 + }, + { + "epoch": 0.02, + "learning_rate": 4.9128156277373076e-05, + "loss": 4.9684, + "step": 53248 + }, + { + "epoch": 0.02, + "learning_rate": 4.9119770329862556e-05, + "loss": 4.9716, + "step": 53760 + }, + { + "epoch": 0.02, + "learning_rate": 4.9111400761155765e-05, + "loss": 4.9488, + "step": 54272 + }, + { + "epoch": 0.02, + "learning_rate": 4.9103014813645245e-05, + "loss": 4.9452, + "step": 54784 + }, + { + "epoch": 0.02, + "learning_rate": 4.909462886613473e-05, + "loss": 4.9431, + "step": 55296 + }, + { + "epoch": 0.02, + "learning_rate": 4.908624291862421e-05, + "loss": 4.9489, + "step": 55808 + }, + { + "epoch": 0.02, + "learning_rate": 4.907785697111369e-05, + "loss": 4.941, + "step": 56320 + }, + { + "epoch": 0.02, + "learning_rate": 4.906947102360317e-05, + "loss": 4.9176, + "step": 56832 + }, + { + "epoch": 0.02, + "learning_rate": 4.906108507609265e-05, + "loss": 4.9233, + "step": 57344 + }, + { + "epoch": 0.02, + "learning_rate": 4.905271550738586e-05, + "loss": 4.9261, + "step": 57856 + }, + { + "epoch": 0.02, + "learning_rate": 4.904432955987534e-05, + "loss": 4.9335, + "step": 58368 + }, + { + "epoch": 0.02, + "learning_rate": 4.903594361236482e-05, + "loss": 4.9202, + "step": 58880 + }, + { + "epoch": 0.02, + "learning_rate": 4.90275576648543e-05, + "loss": 4.9187, + "step": 59392 + }, + { + "epoch": 0.02, + "learning_rate": 4.901917171734378e-05, + "loss": 4.915, + "step": 59904 + }, + { + "epoch": 0.02, + "learning_rate": 4.901078576983326e-05, + "loss": 4.9177, + "step": 60416 + }, + { + "epoch": 0.02, + "learning_rate": 4.900239982232274e-05, + "loss": 4.9009, + "step": 60928 + }, + { + "epoch": 0.02, + "learning_rate": 4.899403025361595e-05, + "loss": 4.9012, + "step": 61440 + }, + { + "epoch": 0.02, + "learning_rate": 4.898564430610543e-05, + "loss": 4.8918, + "step": 61952 + }, + { + "epoch": 0.02, + "learning_rate": 4.897725835859491e-05, + "loss": 4.8958, + "step": 62464 + }, + { + "epoch": 0.02, + "learning_rate": 4.8968872411084396e-05, + "loss": 4.8777, + "step": 62976 + }, + { + "epoch": 0.02, + "learning_rate": 4.896048646357387e-05, + "loss": 4.8901, + "step": 63488 + }, + { + "epoch": 0.02, + "learning_rate": 4.895210051606335e-05, + "loss": 4.881, + "step": 64000 + }, + { + "epoch": 0.02, + "learning_rate": 4.894371456855283e-05, + "loss": 4.8882, + "step": 64512 + }, + { + "epoch": 0.02, + "learning_rate": 4.893532862104231e-05, + "loss": 4.8592, + "step": 65024 + }, + { + "epoch": 0.02, + "learning_rate": 4.892695905233552e-05, + "loss": 4.8706, + "step": 65536 + }, + { + "epoch": 0.02, + "learning_rate": 4.8918573104825e-05, + "loss": 4.8804, + "step": 66048 + }, + { + "epoch": 0.02, + "learning_rate": 4.891018715731448e-05, + "loss": 4.853, + "step": 66560 + }, + { + "epoch": 0.02, + "learning_rate": 4.890180120980396e-05, + "loss": 4.8728, + "step": 67072 + }, + { + "epoch": 0.02, + "learning_rate": 4.889341526229344e-05, + "loss": 4.8565, + "step": 67584 + }, + { + "epoch": 0.02, + "learning_rate": 4.888504569358665e-05, + "loss": 4.8597, + "step": 68096 + }, + { + "epoch": 0.02, + "learning_rate": 4.8876659746076134e-05, + "loss": 4.8493, + "step": 68608 + }, + { + "epoch": 0.02, + "learning_rate": 4.8868273798565614e-05, + "loss": 4.8443, + "step": 69120 + }, + { + "epoch": 0.02, + "learning_rate": 4.8859887851055094e-05, + "loss": 4.8456, + "step": 69632 + }, + { + "epoch": 0.02, + "learning_rate": 4.88515182823483e-05, + "loss": 4.844, + "step": 70144 + }, + { + "epoch": 0.02, + "learning_rate": 4.884313233483778e-05, + "loss": 4.845, + "step": 70656 + }, + { + "epoch": 0.02, + "learning_rate": 4.883474638732726e-05, + "loss": 4.8369, + "step": 71168 + }, + { + "epoch": 0.02, + "learning_rate": 4.882636043981674e-05, + "loss": 4.8388, + "step": 71680 + }, + { + "epoch": 0.02, + "learning_rate": 4.881797449230622e-05, + "loss": 4.8373, + "step": 72192 + }, + { + "epoch": 0.02, + "learning_rate": 4.880960492359943e-05, + "loss": 4.8183, + "step": 72704 + }, + { + "epoch": 0.02, + "learning_rate": 4.880123535489264e-05, + "loss": 4.8198, + "step": 73216 + }, + { + "epoch": 0.02, + "learning_rate": 4.879284940738212e-05, + "loss": 4.8097, + "step": 73728 + }, + { + "epoch": 0.02, + "learning_rate": 4.87844634598716e-05, + "loss": 4.8113, + "step": 74240 + }, + { + "epoch": 0.02, + "learning_rate": 4.877607751236109e-05, + "loss": 4.8115, + "step": 74752 + }, + { + "epoch": 0.02, + "learning_rate": 4.876769156485057e-05, + "loss": 4.8039, + "step": 75264 + }, + { + "epoch": 0.02, + "learning_rate": 4.875930561734005e-05, + "loss": 4.8016, + "step": 75776 + }, + { + "epoch": 0.02, + "learning_rate": 4.875093604863326e-05, + "loss": 4.8008, + "step": 76288 + }, + { + "epoch": 0.03, + "eval_loss": 4.762693881988525, + "eval_runtime": 547.5599, + "eval_samples_per_second": 696.894, + "eval_steps_per_second": 21.778, + "step": 76319 + }, + { + "epoch": 0.0, + "learning_rate": 4.874255010112274e-05, + "loss": 4.8123, + "step": 76800 + }, + { + "epoch": 0.0, + "learning_rate": 4.873416415361222e-05, + "loss": 4.8018, + "step": 77312 + }, + { + "epoch": 0.0, + "learning_rate": 4.87257782061017e-05, + "loss": 4.7914, + "step": 77824 + }, + { + "epoch": 0.0, + "learning_rate": 4.871739225859118e-05, + "loss": 4.7917, + "step": 78336 + }, + { + "epoch": 0.0, + "learning_rate": 4.8709006311080657e-05, + "loss": 4.7865, + "step": 78848 + }, + { + "epoch": 0.0, + "learning_rate": 4.8700620363570137e-05, + "loss": 4.7638, + "step": 79360 + }, + { + "epoch": 0.0, + "learning_rate": 4.8692234416059616e-05, + "loss": 4.778, + "step": 79872 + }, + { + "epoch": 0.0, + "learning_rate": 4.8683848468549096e-05, + "loss": 4.7801, + "step": 80384 + }, + { + "epoch": 0.0, + "learning_rate": 4.8675462521038576e-05, + "loss": 4.7616, + "step": 80896 + }, + { + "epoch": 0.0, + "learning_rate": 4.8667092952331785e-05, + "loss": 4.7768, + "step": 81408 + }, + { + "epoch": 0.0, + "learning_rate": 4.865870700482127e-05, + "loss": 4.7818, + "step": 81920 + }, + { + "epoch": 0.0, + "learning_rate": 4.865032105731075e-05, + "loss": 4.7651, + "step": 82432 + }, + { + "epoch": 0.0, + "learning_rate": 4.8641935109800225e-05, + "loss": 4.7583, + "step": 82944 + }, + { + "epoch": 0.0, + "learning_rate": 4.8633549162289705e-05, + "loss": 4.7605, + "step": 83456 + }, + { + "epoch": 0.0, + "learning_rate": 4.8625163214779185e-05, + "loss": 4.7607, + "step": 83968 + }, + { + "epoch": 0.0, + "learning_rate": 4.8616777267268665e-05, + "loss": 4.7545, + "step": 84480 + }, + { + "epoch": 0.0, + "learning_rate": 4.8608391319758145e-05, + "loss": 4.75, + "step": 84992 + }, + { + "epoch": 0.0, + "learning_rate": 4.8600005372247625e-05, + "loss": 4.7488, + "step": 85504 + }, + { + "epoch": 0.0, + "learning_rate": 4.8591619424737105e-05, + "loss": 4.7441, + "step": 86016 + }, + { + "epoch": 0.0, + "learning_rate": 4.8583233477226585e-05, + "loss": 4.7474, + "step": 86528 + }, + { + "epoch": 0.0, + "learning_rate": 4.8574847529716065e-05, + "loss": 4.7475, + "step": 87040 + }, + { + "epoch": 0.0, + "learning_rate": 4.8566461582205545e-05, + "loss": 4.747, + "step": 87552 + }, + { + "epoch": 0.0, + "learning_rate": 4.8558092013498754e-05, + "loss": 4.7408, + "step": 88064 + }, + { + "epoch": 0.0, + "learning_rate": 4.854970606598824e-05, + "loss": 4.7411, + "step": 88576 + }, + { + "epoch": 0.0, + "learning_rate": 4.854132011847772e-05, + "loss": 4.7342, + "step": 89088 + }, + { + "epoch": 0.0, + "learning_rate": 4.85329341709672e-05, + "loss": 4.7323, + "step": 89600 + }, + { + "epoch": 0.0, + "learning_rate": 4.852456460226041e-05, + "loss": 4.7334, + "step": 90112 + }, + { + "epoch": 0.0, + "learning_rate": 4.851617865474989e-05, + "loss": 4.716, + "step": 90624 + }, + { + "epoch": 0.0, + "learning_rate": 4.850779270723937e-05, + "loss": 4.7295, + "step": 91136 + }, + { + "epoch": 0.01, + "learning_rate": 4.849942313853258e-05, + "loss": 4.7101, + "step": 91648 + }, + { + "epoch": 0.01, + "learning_rate": 4.849103719102206e-05, + "loss": 4.7153, + "step": 92160 + }, + { + "epoch": 0.01, + "learning_rate": 4.848265124351154e-05, + "loss": 4.7248, + "step": 92672 + }, + { + "epoch": 0.01, + "learning_rate": 4.847426529600102e-05, + "loss": 4.7218, + "step": 93184 + }, + { + "epoch": 0.01, + "learning_rate": 4.84658793484905e-05, + "loss": 4.708, + "step": 93696 + }, + { + "epoch": 0.01, + "learning_rate": 4.845749340097998e-05, + "loss": 4.7001, + "step": 94208 + }, + { + "epoch": 0.01, + "learning_rate": 4.844910745346946e-05, + "loss": 4.718, + "step": 94720 + }, + { + "epoch": 0.01, + "learning_rate": 4.844072150595894e-05, + "loss": 4.7053, + "step": 95232 + }, + { + "epoch": 0.01, + "learning_rate": 4.843233555844842e-05, + "loss": 4.7041, + "step": 95744 + }, + { + "epoch": 0.01, + "learning_rate": 4.84239496109379e-05, + "loss": 4.7015, + "step": 96256 + }, + { + "epoch": 0.01, + "learning_rate": 4.8415580042231114e-05, + "loss": 4.6859, + "step": 96768 + }, + { + "epoch": 0.01, + "learning_rate": 4.840719409472059e-05, + "loss": 4.683, + "step": 97280 + }, + { + "epoch": 0.01, + "learning_rate": 4.839880814721007e-05, + "loss": 4.6846, + "step": 97792 + }, + { + "epoch": 0.01, + "learning_rate": 4.839042219969955e-05, + "loss": 4.6891, + "step": 98304 + }, + { + "epoch": 0.01, + "learning_rate": 4.838203625218903e-05, + "loss": 4.6821, + "step": 98816 + }, + { + "epoch": 0.01, + "learning_rate": 4.837365030467851e-05, + "loss": 4.6854, + "step": 99328 + }, + { + "epoch": 0.01, + "learning_rate": 4.836526435716799e-05, + "loss": 4.6781, + "step": 99840 + }, + { + "epoch": 0.01, + "learning_rate": 4.8356894788461196e-05, + "loss": 4.6801, + "step": 100352 + }, + { + "epoch": 0.01, + "learning_rate": 4.8348508840950676e-05, + "loss": 4.6916, + "step": 100864 + }, + { + "epoch": 0.01, + "learning_rate": 4.8340122893440156e-05, + "loss": 4.6734, + "step": 101376 + }, + { + "epoch": 0.01, + "learning_rate": 4.833173694592964e-05, + "loss": 4.6745, + "step": 101888 + }, + { + "epoch": 0.01, + "learning_rate": 4.832335099841912e-05, + "loss": 4.6637, + "step": 102400 + }, + { + "epoch": 0.01, + "learning_rate": 4.831498142971233e-05, + "loss": 4.6853, + "step": 102912 + }, + { + "epoch": 0.01, + "learning_rate": 4.830659548220181e-05, + "loss": 4.6546, + "step": 103424 + }, + { + "epoch": 0.01, + "learning_rate": 4.829820953469129e-05, + "loss": 4.6767, + "step": 103936 + }, + { + "epoch": 0.01, + "learning_rate": 4.828982358718077e-05, + "loss": 4.6536, + "step": 104448 + }, + { + "epoch": 0.01, + "learning_rate": 4.828143763967025e-05, + "loss": 4.662, + "step": 104960 + }, + { + "epoch": 0.01, + "learning_rate": 4.827305169215973e-05, + "loss": 4.6628, + "step": 105472 + }, + { + "epoch": 0.01, + "learning_rate": 4.826468212345294e-05, + "loss": 4.6577, + "step": 105984 + }, + { + "epoch": 0.01, + "learning_rate": 4.825629617594242e-05, + "loss": 4.6493, + "step": 106496 + }, + { + "epoch": 0.01, + "learning_rate": 4.82479102284319e-05, + "loss": 4.6494, + "step": 107008 + }, + { + "epoch": 0.01, + "learning_rate": 4.823952428092138e-05, + "loss": 4.6365, + "step": 107520 + }, + { + "epoch": 0.01, + "learning_rate": 4.823113833341086e-05, + "loss": 4.6537, + "step": 108032 + }, + { + "epoch": 0.01, + "learning_rate": 4.822275238590034e-05, + "loss": 4.6452, + "step": 108544 + }, + { + "epoch": 0.01, + "learning_rate": 4.8214382817193557e-05, + "loss": 4.6394, + "step": 109056 + }, + { + "epoch": 0.01, + "learning_rate": 4.8205996869683037e-05, + "loss": 4.6377, + "step": 109568 + }, + { + "epoch": 0.01, + "learning_rate": 4.8197610922172516e-05, + "loss": 4.6367, + "step": 110080 + }, + { + "epoch": 0.01, + "learning_rate": 4.8189224974661996e-05, + "loss": 4.6197, + "step": 110592 + }, + { + "epoch": 0.01, + "learning_rate": 4.8180839027151476e-05, + "loss": 4.6439, + "step": 111104 + }, + { + "epoch": 0.01, + "learning_rate": 4.817245307964095e-05, + "loss": 4.6283, + "step": 111616 + }, + { + "epoch": 0.01, + "learning_rate": 4.8164083510934165e-05, + "loss": 4.6296, + "step": 112128 + }, + { + "epoch": 0.01, + "learning_rate": 4.8155697563423645e-05, + "loss": 4.6379, + "step": 112640 + }, + { + "epoch": 0.01, + "learning_rate": 4.8147311615913125e-05, + "loss": 4.6233, + "step": 113152 + }, + { + "epoch": 0.01, + "learning_rate": 4.8138942047206334e-05, + "loss": 4.6383, + "step": 113664 + }, + { + "epoch": 0.01, + "learning_rate": 4.8130556099695814e-05, + "loss": 4.6384, + "step": 114176 + }, + { + "epoch": 0.01, + "learning_rate": 4.8122170152185294e-05, + "loss": 4.6325, + "step": 114688 + }, + { + "epoch": 0.01, + "learning_rate": 4.8113784204674774e-05, + "loss": 4.6309, + "step": 115200 + }, + { + "epoch": 0.01, + "learning_rate": 4.8105398257164254e-05, + "loss": 4.6124, + "step": 115712 + }, + { + "epoch": 0.01, + "learning_rate": 4.8097012309653734e-05, + "loss": 4.6112, + "step": 116224 + }, + { + "epoch": 0.01, + "learning_rate": 4.8088626362143214e-05, + "loss": 4.6203, + "step": 116736 + }, + { + "epoch": 0.01, + "learning_rate": 4.8080240414632694e-05, + "loss": 4.6231, + "step": 117248 + }, + { + "epoch": 0.01, + "learning_rate": 4.8071854467122174e-05, + "loss": 4.6138, + "step": 117760 + }, + { + "epoch": 0.01, + "learning_rate": 4.8063468519611654e-05, + "loss": 4.6085, + "step": 118272 + }, + { + "epoch": 0.01, + "learning_rate": 4.805509895090486e-05, + "loss": 4.5931, + "step": 118784 + }, + { + "epoch": 0.01, + "learning_rate": 4.804671300339434e-05, + "loss": 4.6024, + "step": 119296 + }, + { + "epoch": 0.01, + "learning_rate": 4.803832705588382e-05, + "loss": 4.6073, + "step": 119808 + }, + { + "epoch": 0.01, + "learning_rate": 4.80299411083733e-05, + "loss": 4.6061, + "step": 120320 + }, + { + "epoch": 0.01, + "learning_rate": 4.802155516086278e-05, + "loss": 4.6146, + "step": 120832 + }, + { + "epoch": 0.01, + "learning_rate": 4.8013185592156e-05, + "loss": 4.6105, + "step": 121344 + }, + { + "epoch": 0.01, + "learning_rate": 4.800479964464548e-05, + "loss": 4.5996, + "step": 121856 + }, + { + "epoch": 0.02, + "learning_rate": 4.799643007593869e-05, + "loss": 4.5937, + "step": 122368 + }, + { + "epoch": 0.02, + "learning_rate": 4.798804412842817e-05, + "loss": 4.5871, + "step": 122880 + }, + { + "epoch": 0.02, + "learning_rate": 4.797965818091765e-05, + "loss": 4.5926, + "step": 123392 + }, + { + "epoch": 0.02, + "learning_rate": 4.797127223340713e-05, + "loss": 4.5957, + "step": 123904 + }, + { + "epoch": 0.02, + "learning_rate": 4.796288628589661e-05, + "loss": 4.5838, + "step": 124416 + }, + { + "epoch": 0.02, + "learning_rate": 4.795450033838609e-05, + "loss": 4.5833, + "step": 124928 + }, + { + "epoch": 0.02, + "learning_rate": 4.794611439087557e-05, + "loss": 4.5923, + "step": 125440 + }, + { + "epoch": 0.02, + "learning_rate": 4.793772844336505e-05, + "loss": 4.5951, + "step": 125952 + }, + { + "epoch": 0.02, + "learning_rate": 4.792934249585453e-05, + "loss": 4.5842, + "step": 126464 + }, + { + "epoch": 0.02, + "learning_rate": 4.792097292714774e-05, + "loss": 4.59, + "step": 126976 + }, + { + "epoch": 0.02, + "learning_rate": 4.7912586979637217e-05, + "loss": 4.5911, + "step": 127488 + }, + { + "epoch": 0.02, + "learning_rate": 4.790421741093043e-05, + "loss": 4.5834, + "step": 128000 + }, + { + "epoch": 0.02, + "learning_rate": 4.789583146341991e-05, + "loss": 4.579, + "step": 128512 + }, + { + "epoch": 0.02, + "learning_rate": 4.788744551590939e-05, + "loss": 4.5886, + "step": 129024 + }, + { + "epoch": 0.02, + "learning_rate": 4.787905956839887e-05, + "loss": 4.5753, + "step": 129536 + }, + { + "epoch": 0.02, + "learning_rate": 4.787067362088835e-05, + "loss": 4.5868, + "step": 130048 + }, + { + "epoch": 0.02, + "learning_rate": 4.786228767337783e-05, + "loss": 4.5671, + "step": 130560 + }, + { + "epoch": 0.02, + "learning_rate": 4.785390172586731e-05, + "loss": 4.5559, + "step": 131072 + }, + { + "epoch": 0.02, + "learning_rate": 4.7845515778356785e-05, + "loss": 4.5642, + "step": 131584 + }, + { + "epoch": 0.02, + "learning_rate": 4.7837129830846265e-05, + "loss": 4.5728, + "step": 132096 + }, + { + "epoch": 0.02, + "learning_rate": 4.7828743883335745e-05, + "loss": 4.5648, + "step": 132608 + }, + { + "epoch": 0.02, + "learning_rate": 4.7820357935825225e-05, + "loss": 4.5506, + "step": 133120 + }, + { + "epoch": 0.02, + "learning_rate": 4.7811971988314705e-05, + "loss": 4.5516, + "step": 133632 + }, + { + "epoch": 0.02, + "learning_rate": 4.780360241960792e-05, + "loss": 4.5602, + "step": 134144 + }, + { + "epoch": 0.02, + "learning_rate": 4.779523285090114e-05, + "loss": 4.5708, + "step": 134656 + }, + { + "epoch": 0.02, + "learning_rate": 4.778684690339061e-05, + "loss": 4.5587, + "step": 135168 + }, + { + "epoch": 0.02, + "learning_rate": 4.777846095588009e-05, + "loss": 4.5675, + "step": 135680 + }, + { + "epoch": 0.02, + "learning_rate": 4.777007500836957e-05, + "loss": 4.5612, + "step": 136192 + }, + { + "epoch": 0.02, + "learning_rate": 4.776168906085905e-05, + "loss": 4.5589, + "step": 136704 + }, + { + "epoch": 0.02, + "learning_rate": 4.7753335870955995e-05, + "loss": 4.5543, + "step": 137216 + }, + { + "epoch": 0.02, + "learning_rate": 4.7744949923445475e-05, + "loss": 4.5512, + "step": 137728 + }, + { + "epoch": 0.02, + "learning_rate": 4.7736563975934955e-05, + "loss": 4.5459, + "step": 138240 + }, + { + "epoch": 0.02, + "learning_rate": 4.7728178028424435e-05, + "loss": 4.5565, + "step": 138752 + }, + { + "epoch": 0.02, + "learning_rate": 4.771979208091391e-05, + "loss": 4.5422, + "step": 139264 + }, + { + "epoch": 0.02, + "learning_rate": 4.771140613340339e-05, + "loss": 4.5447, + "step": 139776 + }, + { + "epoch": 0.02, + "learning_rate": 4.7703020185892875e-05, + "loss": 4.5482, + "step": 140288 + }, + { + "epoch": 0.02, + "learning_rate": 4.7694650617186084e-05, + "loss": 4.5548, + "step": 140800 + }, + { + "epoch": 0.02, + "learning_rate": 4.7686264669675564e-05, + "loss": 4.5325, + "step": 141312 + }, + { + "epoch": 0.02, + "learning_rate": 4.7677878722165044e-05, + "loss": 4.5379, + "step": 141824 + }, + { + "epoch": 0.02, + "learning_rate": 4.7669492774654524e-05, + "loss": 4.5592, + "step": 142336 + }, + { + "epoch": 0.02, + "learning_rate": 4.7661106827144004e-05, + "loss": 4.5323, + "step": 142848 + }, + { + "epoch": 0.02, + "learning_rate": 4.765273725843721e-05, + "loss": 4.5536, + "step": 143360 + }, + { + "epoch": 0.02, + "learning_rate": 4.764435131092669e-05, + "loss": 4.5361, + "step": 143872 + }, + { + "epoch": 0.02, + "learning_rate": 4.763596536341617e-05, + "loss": 4.5473, + "step": 144384 + }, + { + "epoch": 0.02, + "learning_rate": 4.762757941590565e-05, + "loss": 4.5339, + "step": 144896 + }, + { + "epoch": 0.02, + "learning_rate": 4.761919346839513e-05, + "loss": 4.5286, + "step": 145408 + }, + { + "epoch": 0.02, + "learning_rate": 4.761080752088461e-05, + "loss": 4.5373, + "step": 145920 + }, + { + "epoch": 0.02, + "learning_rate": 4.760242157337409e-05, + "loss": 4.5353, + "step": 146432 + }, + { + "epoch": 0.02, + "learning_rate": 4.759403562586357e-05, + "loss": 4.5423, + "step": 146944 + }, + { + "epoch": 0.02, + "learning_rate": 4.758564967835306e-05, + "loss": 4.5374, + "step": 147456 + }, + { + "epoch": 0.02, + "learning_rate": 4.757728010964627e-05, + "loss": 4.5371, + "step": 147968 + }, + { + "epoch": 0.02, + "learning_rate": 4.756889416213575e-05, + "loss": 4.5334, + "step": 148480 + }, + { + "epoch": 0.02, + "learning_rate": 4.756050821462523e-05, + "loss": 4.5196, + "step": 148992 + }, + { + "epoch": 0.02, + "learning_rate": 4.755212226711471e-05, + "loss": 4.5304, + "step": 149504 + }, + { + "epoch": 0.02, + "learning_rate": 4.754373631960419e-05, + "loss": 4.5158, + "step": 150016 + }, + { + "epoch": 0.02, + "learning_rate": 4.75353667508974e-05, + "loss": 4.5221, + "step": 150528 + }, + { + "epoch": 0.02, + "learning_rate": 4.752698080338688e-05, + "loss": 4.5251, + "step": 151040 + }, + { + "epoch": 0.02, + "learning_rate": 4.751859485587636e-05, + "loss": 4.5122, + "step": 151552 + }, + { + "epoch": 0.02, + "learning_rate": 4.751020890836584e-05, + "loss": 4.5178, + "step": 152064 + }, + { + "epoch": 0.02, + "learning_rate": 4.7501839339659046e-05, + "loss": 4.5151, + "step": 152576 + }, + { + "epoch": 0.03, + "eval_loss": 4.481568813323975, + "eval_runtime": 534.6878, + "eval_samples_per_second": 713.671, + "eval_steps_per_second": 22.303, + "step": 152638 + }, + { + "epoch": 0.0, + "learning_rate": 4.7493453392148526e-05, + "loss": 4.5286, + "step": 153088 + }, + { + "epoch": 0.0, + "learning_rate": 4.748506744463801e-05, + "loss": 4.524, + "step": 153600 + }, + { + "epoch": 0.0, + "learning_rate": 4.747668149712749e-05, + "loss": 4.5131, + "step": 154112 + }, + { + "epoch": 0.0, + "learning_rate": 4.746829554961697e-05, + "loss": 4.5121, + "step": 154624 + }, + { + "epoch": 0.0, + "learning_rate": 4.7459909602106446e-05, + "loss": 4.5118, + "step": 155136 + }, + { + "epoch": 0.0, + "learning_rate": 4.7451523654595926e-05, + "loss": 4.4943, + "step": 155648 + }, + { + "epoch": 0.0, + "learning_rate": 4.7443137707085406e-05, + "loss": 4.5092, + "step": 156160 + }, + { + "epoch": 0.0, + "learning_rate": 4.7434751759574886e-05, + "loss": 4.5075, + "step": 156672 + }, + { + "epoch": 0.0, + "learning_rate": 4.7426365812064366e-05, + "loss": 4.4931, + "step": 157184 + }, + { + "epoch": 0.0, + "learning_rate": 4.7417979864553846e-05, + "loss": 4.5115, + "step": 157696 + }, + { + "epoch": 0.0, + "learning_rate": 4.7409593917043326e-05, + "loss": 4.5183, + "step": 158208 + }, + { + "epoch": 0.0, + "learning_rate": 4.7401207969532806e-05, + "loss": 4.5013, + "step": 158720 + }, + { + "epoch": 0.0, + "learning_rate": 4.7392838400826015e-05, + "loss": 4.4945, + "step": 159232 + }, + { + "epoch": 0.0, + "learning_rate": 4.7384452453315495e-05, + "loss": 4.4981, + "step": 159744 + }, + { + "epoch": 0.0, + "learning_rate": 4.7376066505804975e-05, + "loss": 4.5022, + "step": 160256 + }, + { + "epoch": 0.0, + "learning_rate": 4.736768055829446e-05, + "loss": 4.4961, + "step": 160768 + }, + { + "epoch": 0.0, + "learning_rate": 4.735929461078394e-05, + "loss": 4.4938, + "step": 161280 + }, + { + "epoch": 0.0, + "learning_rate": 4.735090866327342e-05, + "loss": 4.4925, + "step": 161792 + }, + { + "epoch": 0.0, + "learning_rate": 4.73425227157629e-05, + "loss": 4.4939, + "step": 162304 + }, + { + "epoch": 0.0, + "learning_rate": 4.733413676825238e-05, + "loss": 4.4959, + "step": 162816 + }, + { + "epoch": 0.0, + "learning_rate": 4.732576719954559e-05, + "loss": 4.496, + "step": 163328 + }, + { + "epoch": 0.0, + "learning_rate": 4.731738125203507e-05, + "loss": 4.4993, + "step": 163840 + }, + { + "epoch": 0.0, + "learning_rate": 4.730899530452455e-05, + "loss": 4.4975, + "step": 164352 + }, + { + "epoch": 0.0, + "learning_rate": 4.730060935701403e-05, + "loss": 4.4898, + "step": 164864 + }, + { + "epoch": 0.0, + "learning_rate": 4.729222340950351e-05, + "loss": 4.4924, + "step": 165376 + }, + { + "epoch": 0.0, + "learning_rate": 4.728385384079672e-05, + "loss": 4.4882, + "step": 165888 + }, + { + "epoch": 0.0, + "learning_rate": 4.72754678932862e-05, + "loss": 4.4922, + "step": 166400 + }, + { + "epoch": 0.0, + "learning_rate": 4.726708194577568e-05, + "loss": 4.4761, + "step": 166912 + }, + { + "epoch": 0.0, + "learning_rate": 4.725869599826516e-05, + "loss": 4.4876, + "step": 167424 + }, + { + "epoch": 0.01, + "learning_rate": 4.7250326429558375e-05, + "loss": 4.4727, + "step": 167936 + }, + { + "epoch": 0.01, + "learning_rate": 4.7241940482047855e-05, + "loss": 4.479, + "step": 168448 + }, + { + "epoch": 0.01, + "learning_rate": 4.7233554534537335e-05, + "loss": 4.4911, + "step": 168960 + }, + { + "epoch": 0.01, + "learning_rate": 4.722516858702681e-05, + "loss": 4.4863, + "step": 169472 + }, + { + "epoch": 0.01, + "learning_rate": 4.721678263951629e-05, + "loss": 4.4757, + "step": 169984 + }, + { + "epoch": 0.01, + "learning_rate": 4.7208413070809504e-05, + "loss": 4.4699, + "step": 170496 + }, + { + "epoch": 0.01, + "learning_rate": 4.7200027123298984e-05, + "loss": 4.4908, + "step": 171008 + }, + { + "epoch": 0.01, + "learning_rate": 4.719164117578846e-05, + "loss": 4.472, + "step": 171520 + }, + { + "epoch": 0.01, + "learning_rate": 4.718325522827794e-05, + "loss": 4.4798, + "step": 172032 + }, + { + "epoch": 0.01, + "learning_rate": 4.717486928076742e-05, + "loss": 4.4762, + "step": 172544 + }, + { + "epoch": 0.01, + "learning_rate": 4.71664833332569e-05, + "loss": 4.4604, + "step": 173056 + }, + { + "epoch": 0.01, + "learning_rate": 4.7158097385746384e-05, + "loss": 4.4584, + "step": 173568 + }, + { + "epoch": 0.01, + "learning_rate": 4.7149711438235864e-05, + "loss": 4.4602, + "step": 174080 + }, + { + "epoch": 0.01, + "learning_rate": 4.714134186952907e-05, + "loss": 4.4739, + "step": 174592 + }, + { + "epoch": 0.01, + "learning_rate": 4.713295592201855e-05, + "loss": 4.4638, + "step": 175104 + }, + { + "epoch": 0.01, + "learning_rate": 4.712456997450803e-05, + "loss": 4.4654, + "step": 175616 + }, + { + "epoch": 0.01, + "learning_rate": 4.711620040580124e-05, + "loss": 4.4579, + "step": 176128 + }, + { + "epoch": 0.01, + "learning_rate": 4.710781445829072e-05, + "loss": 4.4622, + "step": 176640 + }, + { + "epoch": 0.01, + "learning_rate": 4.70994285107802e-05, + "loss": 4.4778, + "step": 177152 + }, + { + "epoch": 0.01, + "learning_rate": 4.709104256326968e-05, + "loss": 4.4569, + "step": 177664 + }, + { + "epoch": 0.01, + "learning_rate": 4.708265661575916e-05, + "loss": 4.4575, + "step": 178176 + }, + { + "epoch": 0.01, + "learning_rate": 4.707427066824864e-05, + "loss": 4.4525, + "step": 178688 + }, + { + "epoch": 0.01, + "learning_rate": 4.706590109954185e-05, + "loss": 4.4724, + "step": 179200 + }, + { + "epoch": 0.01, + "learning_rate": 4.705751515203134e-05, + "loss": 4.4442, + "step": 179712 + }, + { + "epoch": 0.01, + "learning_rate": 4.704912920452082e-05, + "loss": 4.4654, + "step": 180224 + }, + { + "epoch": 0.01, + "learning_rate": 4.70407432570103e-05, + "loss": 4.4453, + "step": 180736 + }, + { + "epoch": 0.01, + "learning_rate": 4.703235730949978e-05, + "loss": 4.4569, + "step": 181248 + }, + { + "epoch": 0.01, + "learning_rate": 4.702397136198926e-05, + "loss": 4.4589, + "step": 181760 + }, + { + "epoch": 0.01, + "learning_rate": 4.701558541447874e-05, + "loss": 4.4463, + "step": 182272 + }, + { + "epoch": 0.01, + "learning_rate": 4.7007215845771946e-05, + "loss": 4.453, + "step": 182784 + }, + { + "epoch": 0.01, + "learning_rate": 4.6998829898261426e-05, + "loss": 4.4456, + "step": 183296 + }, + { + "epoch": 0.01, + "learning_rate": 4.6990443950750906e-05, + "loss": 4.4299, + "step": 183808 + }, + { + "epoch": 0.01, + "learning_rate": 4.6982058003240386e-05, + "loss": 4.4503, + "step": 184320 + }, + { + "epoch": 0.01, + "learning_rate": 4.6973688434533595e-05, + "loss": 4.447, + "step": 184832 + }, + { + "epoch": 0.01, + "learning_rate": 4.6965302487023075e-05, + "loss": 4.4395, + "step": 185344 + }, + { + "epoch": 0.01, + "learning_rate": 4.6956916539512555e-05, + "loss": 4.44, + "step": 185856 + }, + { + "epoch": 0.01, + "learning_rate": 4.6948530592002035e-05, + "loss": 4.4398, + "step": 186368 + }, + { + "epoch": 0.01, + "learning_rate": 4.6940144644491515e-05, + "loss": 4.4225, + "step": 186880 + }, + { + "epoch": 0.01, + "learning_rate": 4.6931758696980995e-05, + "loss": 4.4486, + "step": 187392 + }, + { + "epoch": 0.01, + "learning_rate": 4.6923372749470475e-05, + "loss": 4.4301, + "step": 187904 + }, + { + "epoch": 0.01, + "learning_rate": 4.691500318076369e-05, + "loss": 4.4349, + "step": 188416 + }, + { + "epoch": 0.01, + "learning_rate": 4.690661723325317e-05, + "loss": 4.4451, + "step": 188928 + }, + { + "epoch": 0.01, + "learning_rate": 4.6898231285742644e-05, + "loss": 4.4332, + "step": 189440 + }, + { + "epoch": 0.01, + "learning_rate": 4.6889845338232124e-05, + "loss": 4.4422, + "step": 189952 + }, + { + "epoch": 0.01, + "learning_rate": 4.6881459390721604e-05, + "loss": 4.4515, + "step": 190464 + }, + { + "epoch": 0.01, + "learning_rate": 4.687308982201481e-05, + "loss": 4.4383, + "step": 190976 + }, + { + "epoch": 0.01, + "learning_rate": 4.686470387450429e-05, + "loss": 4.4457, + "step": 191488 + }, + { + "epoch": 0.01, + "learning_rate": 4.685631792699377e-05, + "loss": 4.426, + "step": 192000 + }, + { + "epoch": 0.01, + "learning_rate": 4.684793197948325e-05, + "loss": 4.4218, + "step": 192512 + }, + { + "epoch": 0.01, + "learning_rate": 4.683956241077647e-05, + "loss": 4.4309, + "step": 193024 + }, + { + "epoch": 0.01, + "learning_rate": 4.683117646326595e-05, + "loss": 4.4415, + "step": 193536 + }, + { + "epoch": 0.01, + "learning_rate": 4.682279051575543e-05, + "loss": 4.4291, + "step": 194048 + }, + { + "epoch": 0.01, + "learning_rate": 4.681440456824491e-05, + "loss": 4.4234, + "step": 194560 + }, + { + "epoch": 0.01, + "learning_rate": 4.680601862073439e-05, + "loss": 4.4144, + "step": 195072 + }, + { + "epoch": 0.01, + "learning_rate": 4.679763267322387e-05, + "loss": 4.4172, + "step": 195584 + }, + { + "epoch": 0.01, + "learning_rate": 4.678924672571335e-05, + "loss": 4.4204, + "step": 196096 + }, + { + "epoch": 0.01, + "learning_rate": 4.678086077820283e-05, + "loss": 4.4256, + "step": 196608 + }, + { + "epoch": 0.01, + "learning_rate": 4.677247483069231e-05, + "loss": 4.4337, + "step": 197120 + }, + { + "epoch": 0.01, + "learning_rate": 4.676408888318179e-05, + "loss": 4.4311, + "step": 197632 + }, + { + "epoch": 0.01, + "learning_rate": 4.675570293567127e-05, + "loss": 4.4213, + "step": 198144 + }, + { + "epoch": 0.02, + "learning_rate": 4.674733336696448e-05, + "loss": 4.4167, + "step": 198656 + }, + { + "epoch": 0.02, + "learning_rate": 4.673894741945396e-05, + "loss": 4.4112, + "step": 199168 + }, + { + "epoch": 0.02, + "learning_rate": 4.673056147194344e-05, + "loss": 4.4156, + "step": 199680 + }, + { + "epoch": 0.02, + "learning_rate": 4.6722175524432924e-05, + "loss": 4.4159, + "step": 200192 + }, + { + "epoch": 0.02, + "learning_rate": 4.671380595572613e-05, + "loss": 4.4119, + "step": 200704 + }, + { + "epoch": 0.02, + "learning_rate": 4.670542000821561e-05, + "loss": 4.4094, + "step": 201216 + }, + { + "epoch": 0.02, + "learning_rate": 4.669703406070509e-05, + "loss": 4.4203, + "step": 201728 + }, + { + "epoch": 0.02, + "learning_rate": 4.668864811319457e-05, + "loss": 4.4191, + "step": 202240 + }, + { + "epoch": 0.02, + "learning_rate": 4.668026216568405e-05, + "loss": 4.4166, + "step": 202752 + }, + { + "epoch": 0.02, + "learning_rate": 4.667187621817353e-05, + "loss": 4.4169, + "step": 203264 + }, + { + "epoch": 0.02, + "learning_rate": 4.6663490270663006e-05, + "loss": 4.4195, + "step": 203776 + }, + { + "epoch": 0.02, + "learning_rate": 4.6655104323152486e-05, + "loss": 4.4143, + "step": 204288 + }, + { + "epoch": 0.02, + "learning_rate": 4.66467347544457e-05, + "loss": 4.4072, + "step": 204800 + }, + { + "epoch": 0.02, + "learning_rate": 4.6638348806935175e-05, + "loss": 4.4234, + "step": 205312 + }, + { + "epoch": 0.02, + "learning_rate": 4.662997923822839e-05, + "loss": 4.4043, + "step": 205824 + }, + { + "epoch": 0.02, + "learning_rate": 4.662159329071788e-05, + "loss": 4.4241, + "step": 206336 + }, + { + "epoch": 0.02, + "learning_rate": 4.661320734320736e-05, + "loss": 4.3968, + "step": 206848 + }, + { + "epoch": 0.02, + "learning_rate": 4.660482139569683e-05, + "loss": 4.391, + "step": 207360 + }, + { + "epoch": 0.02, + "learning_rate": 4.659643544818631e-05, + "loss": 4.3983, + "step": 207872 + }, + { + "epoch": 0.02, + "learning_rate": 4.658804950067579e-05, + "loss": 4.4121, + "step": 208384 + }, + { + "epoch": 0.02, + "learning_rate": 4.657966355316527e-05, + "loss": 4.4031, + "step": 208896 + }, + { + "epoch": 0.02, + "learning_rate": 4.657129398445848e-05, + "loss": 4.3874, + "step": 209408 + }, + { + "epoch": 0.02, + "learning_rate": 4.6562924415751695e-05, + "loss": 4.3895, + "step": 209920 + }, + { + "epoch": 0.02, + "learning_rate": 4.6554538468241175e-05, + "loss": 4.3963, + "step": 210432 + }, + { + "epoch": 0.02, + "learning_rate": 4.654615252073065e-05, + "loss": 4.4109, + "step": 210944 + }, + { + "epoch": 0.02, + "learning_rate": 4.653776657322013e-05, + "loss": 4.3982, + "step": 211456 + }, + { + "epoch": 0.02, + "learning_rate": 4.6529397004513344e-05, + "loss": 4.4072, + "step": 211968 + }, + { + "epoch": 0.02, + "learning_rate": 4.652101105700283e-05, + "loss": 4.4018, + "step": 212480 + }, + { + "epoch": 0.02, + "learning_rate": 4.6512625109492304e-05, + "loss": 4.4, + "step": 212992 + }, + { + "epoch": 0.02, + "learning_rate": 4.6504239161981784e-05, + "loss": 4.402, + "step": 213504 + }, + { + "epoch": 0.02, + "learning_rate": 4.6495853214471264e-05, + "loss": 4.3896, + "step": 214016 + }, + { + "epoch": 0.02, + "learning_rate": 4.6487467266960744e-05, + "loss": 4.391, + "step": 214528 + }, + { + "epoch": 0.02, + "learning_rate": 4.647909769825395e-05, + "loss": 4.4013, + "step": 215040 + }, + { + "epoch": 0.02, + "learning_rate": 4.647071175074343e-05, + "loss": 4.3849, + "step": 215552 + }, + { + "epoch": 0.02, + "learning_rate": 4.646232580323291e-05, + "loss": 4.3919, + "step": 216064 + }, + { + "epoch": 0.02, + "learning_rate": 4.645393985572239e-05, + "loss": 4.3944, + "step": 216576 + }, + { + "epoch": 0.02, + "learning_rate": 4.64455702870156e-05, + "loss": 4.3988, + "step": 217088 + }, + { + "epoch": 0.02, + "learning_rate": 4.643718433950508e-05, + "loss": 4.3828, + "step": 217600 + }, + { + "epoch": 0.02, + "learning_rate": 4.642879839199457e-05, + "loss": 4.3812, + "step": 218112 + }, + { + "epoch": 0.02, + "learning_rate": 4.642041244448405e-05, + "loss": 4.4068, + "step": 218624 + }, + { + "epoch": 0.02, + "learning_rate": 4.641202649697353e-05, + "loss": 4.385, + "step": 219136 + }, + { + "epoch": 0.02, + "learning_rate": 4.640364054946301e-05, + "loss": 4.3975, + "step": 219648 + }, + { + "epoch": 0.02, + "learning_rate": 4.639525460195249e-05, + "loss": 4.3889, + "step": 220160 + }, + { + "epoch": 0.02, + "learning_rate": 4.638686865444197e-05, + "loss": 4.3955, + "step": 220672 + }, + { + "epoch": 0.02, + "learning_rate": 4.637848270693145e-05, + "loss": 4.385, + "step": 221184 + }, + { + "epoch": 0.02, + "learning_rate": 4.637009675942093e-05, + "loss": 4.3822, + "step": 221696 + }, + { + "epoch": 0.02, + "learning_rate": 4.636172719071414e-05, + "loss": 4.3916, + "step": 222208 + }, + { + "epoch": 0.02, + "learning_rate": 4.635334124320362e-05, + "loss": 4.384, + "step": 222720 + }, + { + "epoch": 0.02, + "learning_rate": 4.63449552956931e-05, + "loss": 4.396, + "step": 223232 + }, + { + "epoch": 0.02, + "learning_rate": 4.633656934818258e-05, + "loss": 4.3905, + "step": 223744 + }, + { + "epoch": 0.02, + "learning_rate": 4.632818340067206e-05, + "loss": 4.3905, + "step": 224256 + }, + { + "epoch": 0.02, + "learning_rate": 4.631979745316154e-05, + "loss": 4.3852, + "step": 224768 + }, + { + "epoch": 0.02, + "learning_rate": 4.631142788445475e-05, + "loss": 4.3766, + "step": 225280 + }, + { + "epoch": 0.02, + "learning_rate": 4.630304193694423e-05, + "loss": 4.3845, + "step": 225792 + }, + { + "epoch": 0.02, + "learning_rate": 4.629465598943371e-05, + "loss": 4.3738, + "step": 226304 + }, + { + "epoch": 0.02, + "learning_rate": 4.628627004192319e-05, + "loss": 4.378, + "step": 226816 + }, + { + "epoch": 0.02, + "learning_rate": 4.6277884094412666e-05, + "loss": 4.3789, + "step": 227328 + }, + { + "epoch": 0.02, + "learning_rate": 4.6269498146902146e-05, + "loss": 4.3709, + "step": 227840 + }, + { + "epoch": 0.02, + "learning_rate": 4.6261112199391626e-05, + "loss": 4.3752, + "step": 228352 + }, + { + "epoch": 0.02, + "learning_rate": 4.6252742630684835e-05, + "loss": 4.374, + "step": 228864 + }, + { + "epoch": 0.03, + "eval_loss": 4.347803592681885, + "eval_runtime": 537.7927, + "eval_samples_per_second": 709.55, + "eval_steps_per_second": 22.174, + "step": 228957 + }, + { + "epoch": 1.0, + "learning_rate": 4.6244356683174315e-05, + "loss": 4.3823, + "step": 229376 + }, + { + "epoch": 1.0, + "learning_rate": 4.6235970735663795e-05, + "loss": 4.3866, + "step": 229888 + }, + { + "epoch": 1.0, + "learning_rate": 4.6227584788153275e-05, + "loss": 4.3735, + "step": 230400 + }, + { + "epoch": 1.0, + "learning_rate": 4.6219198840642755e-05, + "loss": 4.3717, + "step": 230912 + }, + { + "epoch": 1.0, + "learning_rate": 4.6210812893132235e-05, + "loss": 4.3742, + "step": 231424 + }, + { + "epoch": 1.0, + "learning_rate": 4.6202426945621715e-05, + "loss": 4.3549, + "step": 231936 + }, + { + "epoch": 1.0, + "learning_rate": 4.619405737691493e-05, + "loss": 4.3695, + "step": 232448 + }, + { + "epoch": 1.0, + "learning_rate": 4.618567142940441e-05, + "loss": 4.3724, + "step": 232960 + }, + { + "epoch": 1.0, + "learning_rate": 4.617728548189389e-05, + "loss": 4.3604, + "step": 233472 + }, + { + "epoch": 1.0, + "learning_rate": 4.61689159131871e-05, + "loss": 4.3743, + "step": 233984 + }, + { + "epoch": 1.0, + "learning_rate": 4.616052996567658e-05, + "loss": 4.3822, + "step": 234496 + }, + { + "epoch": 1.0, + "learning_rate": 4.615214401816606e-05, + "loss": 4.362, + "step": 235008 + }, + { + "epoch": 1.0, + "learning_rate": 4.614375807065554e-05, + "loss": 4.3618, + "step": 235520 + }, + { + "epoch": 1.0, + "learning_rate": 4.613537212314502e-05, + "loss": 4.3639, + "step": 236032 + }, + { + "epoch": 1.0, + "learning_rate": 4.61269861756345e-05, + "loss": 4.3686, + "step": 236544 + }, + { + "epoch": 1.0, + "learning_rate": 4.611860022812398e-05, + "loss": 4.3596, + "step": 237056 + }, + { + "epoch": 1.0, + "learning_rate": 4.611021428061346e-05, + "loss": 4.3583, + "step": 237568 + }, + { + "epoch": 1.0, + "learning_rate": 4.610182833310294e-05, + "loss": 4.3593, + "step": 238080 + }, + { + "epoch": 1.0, + "learning_rate": 4.609344238559242e-05, + "loss": 4.3625, + "step": 238592 + }, + { + "epoch": 1.0, + "learning_rate": 4.60850564380819e-05, + "loss": 4.3673, + "step": 239104 + }, + { + "epoch": 1.0, + "learning_rate": 4.607667049057138e-05, + "loss": 4.3629, + "step": 239616 + }, + { + "epoch": 1.0, + "learning_rate": 4.6068300921864595e-05, + "loss": 4.3673, + "step": 240128 + }, + { + "epoch": 1.0, + "learning_rate": 4.6059931353157805e-05, + "loss": 4.3658, + "step": 240640 + }, + { + "epoch": 1.0, + "learning_rate": 4.6051545405647285e-05, + "loss": 4.3606, + "step": 241152 + }, + { + "epoch": 1.0, + "learning_rate": 4.6043175836940494e-05, + "loss": 4.3571, + "step": 241664 + }, + { + "epoch": 1.0, + "learning_rate": 4.6034789889429974e-05, + "loss": 4.3573, + "step": 242176 + }, + { + "epoch": 1.0, + "learning_rate": 4.6026403941919454e-05, + "loss": 4.3649, + "step": 242688 + }, + { + "epoch": 1.0, + "learning_rate": 4.6018017994408933e-05, + "loss": 4.3498, + "step": 243200 + }, + { + "epoch": 1.0, + "learning_rate": 4.6009632046898413e-05, + "loss": 4.3596, + "step": 243712 + }, + { + "epoch": 1.01, + "learning_rate": 4.6001246099387893e-05, + "loss": 4.3461, + "step": 244224 + }, + { + "epoch": 1.01, + "learning_rate": 4.599286015187737e-05, + "loss": 4.3492, + "step": 244736 + }, + { + "epoch": 1.01, + "learning_rate": 4.598447420436685e-05, + "loss": 4.3638, + "step": 245248 + }, + { + "epoch": 1.01, + "learning_rate": 4.597610463566007e-05, + "loss": 4.358, + "step": 245760 + }, + { + "epoch": 1.01, + "learning_rate": 4.596771868814955e-05, + "loss": 4.3514, + "step": 246272 + }, + { + "epoch": 1.01, + "learning_rate": 4.595933274063903e-05, + "loss": 4.3472, + "step": 246784 + }, + { + "epoch": 1.01, + "learning_rate": 4.59509467931285e-05, + "loss": 4.36, + "step": 247296 + }, + { + "epoch": 1.01, + "learning_rate": 4.594256084561798e-05, + "loss": 4.3454, + "step": 247808 + }, + { + "epoch": 1.01, + "learning_rate": 4.59341912769112e-05, + "loss": 4.3565, + "step": 248320 + }, + { + "epoch": 1.01, + "learning_rate": 4.592580532940067e-05, + "loss": 4.3498, + "step": 248832 + }, + { + "epoch": 1.01, + "learning_rate": 4.591741938189015e-05, + "loss": 4.3385, + "step": 249344 + }, + { + "epoch": 1.01, + "learning_rate": 4.590903343437963e-05, + "loss": 4.3375, + "step": 249856 + }, + { + "epoch": 1.01, + "learning_rate": 4.590064748686911e-05, + "loss": 4.3388, + "step": 250368 + }, + { + "epoch": 1.01, + "learning_rate": 4.589226153935859e-05, + "loss": 4.3483, + "step": 250880 + }, + { + "epoch": 1.01, + "learning_rate": 4.588387559184808e-05, + "loss": 4.3403, + "step": 251392 + }, + { + "epoch": 1.01, + "learning_rate": 4.587548964433756e-05, + "loss": 4.3436, + "step": 251904 + }, + { + "epoch": 1.01, + "learning_rate": 4.586712007563077e-05, + "loss": 4.3391, + "step": 252416 + }, + { + "epoch": 1.01, + "learning_rate": 4.585873412812025e-05, + "loss": 4.342, + "step": 252928 + }, + { + "epoch": 1.01, + "learning_rate": 4.585034818060973e-05, + "loss": 4.3544, + "step": 253440 + }, + { + "epoch": 1.01, + "learning_rate": 4.584196223309921e-05, + "loss": 4.3362, + "step": 253952 + }, + { + "epoch": 1.01, + "learning_rate": 4.5833592664392416e-05, + "loss": 4.3371, + "step": 254464 + }, + { + "epoch": 1.01, + "learning_rate": 4.5825206716881896e-05, + "loss": 4.3383, + "step": 254976 + }, + { + "epoch": 1.01, + "learning_rate": 4.5816820769371376e-05, + "loss": 4.3485, + "step": 255488 + }, + { + "epoch": 1.01, + "learning_rate": 4.5808434821860856e-05, + "loss": 4.3278, + "step": 256000 + }, + { + "epoch": 1.01, + "learning_rate": 4.5800065253154065e-05, + "loss": 4.3439, + "step": 256512 + }, + { + "epoch": 1.01, + "learning_rate": 4.5791679305643545e-05, + "loss": 4.3272, + "step": 257024 + }, + { + "epoch": 1.01, + "learning_rate": 4.5783293358133025e-05, + "loss": 4.3352, + "step": 257536 + }, + { + "epoch": 1.01, + "learning_rate": 4.577492378942624e-05, + "loss": 4.3411, + "step": 258048 + }, + { + "epoch": 1.01, + "learning_rate": 4.576653784191572e-05, + "loss": 4.3239, + "step": 258560 + }, + { + "epoch": 1.01, + "learning_rate": 4.575816827320893e-05, + "loss": 4.3414, + "step": 259072 + }, + { + "epoch": 1.01, + "learning_rate": 4.574978232569841e-05, + "loss": 4.3316, + "step": 259584 + }, + { + "epoch": 1.01, + "learning_rate": 4.574139637818789e-05, + "loss": 4.3154, + "step": 260096 + }, + { + "epoch": 1.01, + "learning_rate": 4.573301043067737e-05, + "loss": 4.3317, + "step": 260608 + }, + { + "epoch": 1.01, + "learning_rate": 4.572462448316685e-05, + "loss": 4.331, + "step": 261120 + }, + { + "epoch": 1.01, + "learning_rate": 4.571625491446006e-05, + "loss": 4.3296, + "step": 261632 + }, + { + "epoch": 1.01, + "learning_rate": 4.570786896694954e-05, + "loss": 4.3239, + "step": 262144 + }, + { + "epoch": 1.01, + "learning_rate": 4.569948301943902e-05, + "loss": 4.3265, + "step": 262656 + }, + { + "epoch": 1.01, + "learning_rate": 4.56910970719285e-05, + "loss": 4.3087, + "step": 263168 + }, + { + "epoch": 1.01, + "learning_rate": 4.568271112441798e-05, + "loss": 4.335, + "step": 263680 + }, + { + "epoch": 1.01, + "learning_rate": 4.5674325176907465e-05, + "loss": 4.3171, + "step": 264192 + }, + { + "epoch": 1.01, + "learning_rate": 4.5665955608200674e-05, + "loss": 4.322, + "step": 264704 + }, + { + "epoch": 1.01, + "learning_rate": 4.5657569660690154e-05, + "loss": 4.3314, + "step": 265216 + }, + { + "epoch": 1.01, + "learning_rate": 4.5649183713179634e-05, + "loss": 4.3208, + "step": 265728 + }, + { + "epoch": 1.01, + "learning_rate": 4.5640797765669114e-05, + "loss": 4.3304, + "step": 266240 + }, + { + "epoch": 1.01, + "learning_rate": 4.5632411818158594e-05, + "loss": 4.3356, + "step": 266752 + }, + { + "epoch": 1.01, + "learning_rate": 4.562405862825553e-05, + "loss": 4.3303, + "step": 267264 + }, + { + "epoch": 1.01, + "learning_rate": 4.561567268074501e-05, + "loss": 4.3351, + "step": 267776 + }, + { + "epoch": 1.01, + "learning_rate": 4.560728673323449e-05, + "loss": 4.3141, + "step": 268288 + }, + { + "epoch": 1.01, + "learning_rate": 4.559890078572397e-05, + "loss": 4.3132, + "step": 268800 + }, + { + "epoch": 1.01, + "learning_rate": 4.559051483821345e-05, + "loss": 4.3202, + "step": 269312 + }, + { + "epoch": 1.01, + "learning_rate": 4.558212889070293e-05, + "loss": 4.3333, + "step": 269824 + }, + { + "epoch": 1.01, + "learning_rate": 4.557374294319242e-05, + "loss": 4.3201, + "step": 270336 + }, + { + "epoch": 1.01, + "learning_rate": 4.55653569956819e-05, + "loss": 4.3127, + "step": 270848 + }, + { + "epoch": 1.01, + "learning_rate": 4.555697104817138e-05, + "loss": 4.3037, + "step": 271360 + }, + { + "epoch": 1.01, + "learning_rate": 4.554858510066086e-05, + "loss": 4.3117, + "step": 271872 + }, + { + "epoch": 1.01, + "learning_rate": 4.554019915315033e-05, + "loss": 4.3118, + "step": 272384 + }, + { + "epoch": 1.01, + "learning_rate": 4.553181320563981e-05, + "loss": 4.316, + "step": 272896 + }, + { + "epoch": 1.01, + "learning_rate": 4.552344363693303e-05, + "loss": 4.3228, + "step": 273408 + }, + { + "epoch": 1.01, + "learning_rate": 4.551505768942251e-05, + "loss": 4.3262, + "step": 273920 + }, + { + "epoch": 1.01, + "learning_rate": 4.550668812071572e-05, + "loss": 4.3136, + "step": 274432 + }, + { + "epoch": 1.02, + "learning_rate": 4.54983021732052e-05, + "loss": 4.3137, + "step": 274944 + }, + { + "epoch": 1.02, + "learning_rate": 4.5489932604498406e-05, + "loss": 4.3026, + "step": 275456 + }, + { + "epoch": 1.02, + "learning_rate": 4.5481546656987886e-05, + "loss": 4.3073, + "step": 275968 + }, + { + "epoch": 1.02, + "learning_rate": 4.547316070947737e-05, + "loss": 4.3132, + "step": 276480 + }, + { + "epoch": 1.02, + "learning_rate": 4.546477476196685e-05, + "loss": 4.3064, + "step": 276992 + }, + { + "epoch": 1.02, + "learning_rate": 4.545638881445633e-05, + "loss": 4.3016, + "step": 277504 + }, + { + "epoch": 1.02, + "learning_rate": 4.5448002866945806e-05, + "loss": 4.3165, + "step": 278016 + }, + { + "epoch": 1.02, + "learning_rate": 4.5439616919435286e-05, + "loss": 4.3109, + "step": 278528 + }, + { + "epoch": 1.02, + "learning_rate": 4.5431230971924765e-05, + "loss": 4.3189, + "step": 279040 + }, + { + "epoch": 1.02, + "learning_rate": 4.5422845024414245e-05, + "loss": 4.309, + "step": 279552 + }, + { + "epoch": 1.02, + "learning_rate": 4.5414475455707455e-05, + "loss": 4.3157, + "step": 280064 + }, + { + "epoch": 1.02, + "learning_rate": 4.5406089508196934e-05, + "loss": 4.3115, + "step": 280576 + }, + { + "epoch": 1.02, + "learning_rate": 4.5397703560686414e-05, + "loss": 4.3053, + "step": 281088 + }, + { + "epoch": 1.02, + "learning_rate": 4.5389317613175894e-05, + "loss": 4.3222, + "step": 281600 + }, + { + "epoch": 1.02, + "learning_rate": 4.5380931665665374e-05, + "loss": 4.3018, + "step": 282112 + }, + { + "epoch": 1.02, + "learning_rate": 4.5372545718154854e-05, + "loss": 4.3236, + "step": 282624 + }, + { + "epoch": 1.02, + "learning_rate": 4.536415977064434e-05, + "loss": 4.2935, + "step": 283136 + }, + { + "epoch": 1.02, + "learning_rate": 4.535577382313382e-05, + "loss": 4.2914, + "step": 283648 + }, + { + "epoch": 1.02, + "learning_rate": 4.534740425442703e-05, + "loss": 4.2968, + "step": 284160 + }, + { + "epoch": 1.02, + "learning_rate": 4.533901830691651e-05, + "loss": 4.3108, + "step": 284672 + }, + { + "epoch": 1.02, + "learning_rate": 4.533064873820972e-05, + "loss": 4.3032, + "step": 285184 + }, + { + "epoch": 1.02, + "learning_rate": 4.53222627906992e-05, + "loss": 4.2872, + "step": 285696 + }, + { + "epoch": 1.02, + "learning_rate": 4.531387684318868e-05, + "loss": 4.2888, + "step": 286208 + }, + { + "epoch": 1.02, + "learning_rate": 4.530549089567816e-05, + "loss": 4.2945, + "step": 286720 + }, + { + "epoch": 1.02, + "learning_rate": 4.529710494816764e-05, + "loss": 4.3131, + "step": 287232 + }, + { + "epoch": 1.02, + "learning_rate": 4.528871900065712e-05, + "loss": 4.2975, + "step": 287744 + }, + { + "epoch": 1.02, + "learning_rate": 4.52803330531466e-05, + "loss": 4.3107, + "step": 288256 + }, + { + "epoch": 1.02, + "learning_rate": 4.527194710563608e-05, + "loss": 4.3018, + "step": 288768 + }, + { + "epoch": 1.02, + "learning_rate": 4.5263577536929295e-05, + "loss": 4.304, + "step": 289280 + }, + { + "epoch": 1.02, + "learning_rate": 4.5255191589418775e-05, + "loss": 4.3022, + "step": 289792 + }, + { + "epoch": 1.02, + "learning_rate": 4.5246805641908255e-05, + "loss": 4.2934, + "step": 290304 + }, + { + "epoch": 1.02, + "learning_rate": 4.5238419694397735e-05, + "loss": 4.294, + "step": 290816 + }, + { + "epoch": 1.02, + "learning_rate": 4.5230050125690944e-05, + "loss": 4.3005, + "step": 291328 + }, + { + "epoch": 1.02, + "learning_rate": 4.5221664178180424e-05, + "loss": 4.2898, + "step": 291840 + }, + { + "epoch": 1.02, + "learning_rate": 4.5213278230669904e-05, + "loss": 4.295, + "step": 292352 + }, + { + "epoch": 1.02, + "learning_rate": 4.5204892283159384e-05, + "loss": 4.2997, + "step": 292864 + }, + { + "epoch": 1.02, + "learning_rate": 4.519652271445259e-05, + "loss": 4.2973, + "step": 293376 + }, + { + "epoch": 1.02, + "learning_rate": 4.518813676694207e-05, + "loss": 4.2888, + "step": 293888 + }, + { + "epoch": 1.02, + "learning_rate": 4.517975081943155e-05, + "loss": 4.2867, + "step": 294400 + }, + { + "epoch": 1.02, + "learning_rate": 4.517136487192103e-05, + "loss": 4.3086, + "step": 294912 + }, + { + "epoch": 1.02, + "learning_rate": 4.516299530321425e-05, + "loss": 4.2887, + "step": 295424 + }, + { + "epoch": 1.02, + "learning_rate": 4.515460935570373e-05, + "loss": 4.3022, + "step": 295936 + }, + { + "epoch": 1.02, + "learning_rate": 4.514622340819321e-05, + "loss": 4.2963, + "step": 296448 + }, + { + "epoch": 1.02, + "learning_rate": 4.513785383948642e-05, + "loss": 4.299, + "step": 296960 + }, + { + "epoch": 1.02, + "learning_rate": 4.51294678919759e-05, + "loss": 4.2907, + "step": 297472 + }, + { + "epoch": 1.02, + "learning_rate": 4.512108194446538e-05, + "loss": 4.2921, + "step": 297984 + }, + { + "epoch": 1.02, + "learning_rate": 4.511269599695486e-05, + "loss": 4.2944, + "step": 298496 + }, + { + "epoch": 1.02, + "learning_rate": 4.510431004944434e-05, + "loss": 4.29, + "step": 299008 + }, + { + "epoch": 1.02, + "learning_rate": 4.509592410193382e-05, + "loss": 4.3031, + "step": 299520 + }, + { + "epoch": 1.02, + "learning_rate": 4.508753815442329e-05, + "loss": 4.2931, + "step": 300032 + }, + { + "epoch": 1.02, + "learning_rate": 4.507915220691277e-05, + "loss": 4.2989, + "step": 300544 + }, + { + "epoch": 1.02, + "learning_rate": 4.5070782638205986e-05, + "loss": 4.2968, + "step": 301056 + }, + { + "epoch": 1.02, + "learning_rate": 4.5062396690695466e-05, + "loss": 4.2842, + "step": 301568 + }, + { + "epoch": 1.02, + "learning_rate": 4.5054010743184946e-05, + "loss": 4.2919, + "step": 302080 + }, + { + "epoch": 1.02, + "learning_rate": 4.504564117447816e-05, + "loss": 4.2815, + "step": 302592 + }, + { + "epoch": 1.02, + "learning_rate": 4.503725522696764e-05, + "loss": 4.287, + "step": 303104 + }, + { + "epoch": 1.02, + "learning_rate": 4.5028869279457115e-05, + "loss": 4.2849, + "step": 303616 + }, + { + "epoch": 1.02, + "learning_rate": 4.5020483331946595e-05, + "loss": 4.2834, + "step": 304128 + }, + { + "epoch": 1.02, + "learning_rate": 4.5012097384436075e-05, + "loss": 4.2782, + "step": 304640 + }, + { + "epoch": 1.02, + "learning_rate": 4.500372781572929e-05, + "loss": 4.2862, + "step": 305152 + }, + { + "epoch": 1.03, + "eval_loss": 4.264499664306641, + "eval_runtime": 538.3004, + "eval_samples_per_second": 708.881, + "eval_steps_per_second": 22.153, + "step": 305276 + }, + { + "epoch": 0.0, + "learning_rate": 4.49953582470225e-05, + "loss": 4.2956, + "step": 305664 + }, + { + "epoch": 0.0, + "learning_rate": 4.498697229951198e-05, + "loss": 4.2931, + "step": 306176 + }, + { + "epoch": 0.0, + "learning_rate": 4.497858635200146e-05, + "loss": 4.2842, + "step": 306688 + }, + { + "epoch": 0.0, + "learning_rate": 4.497020040449093e-05, + "loss": 4.2781, + "step": 307200 + }, + { + "epoch": 0.0, + "learning_rate": 4.496181445698042e-05, + "loss": 4.2876, + "step": 307712 + }, + { + "epoch": 0.0, + "learning_rate": 4.49534285094699e-05, + "loss": 4.2654, + "step": 308224 + }, + { + "epoch": 0.0, + "learning_rate": 4.494504256195938e-05, + "loss": 4.2838, + "step": 308736 + }, + { + "epoch": 0.0, + "learning_rate": 4.493665661444886e-05, + "loss": 4.2791, + "step": 309248 + }, + { + "epoch": 0.0, + "learning_rate": 4.492827066693834e-05, + "loss": 4.2733, + "step": 309760 + }, + { + "epoch": 0.0, + "learning_rate": 4.491990109823155e-05, + "loss": 4.2848, + "step": 310272 + }, + { + "epoch": 0.0, + "learning_rate": 4.491151515072103e-05, + "loss": 4.2916, + "step": 310784 + }, + { + "epoch": 0.0, + "learning_rate": 4.490312920321051e-05, + "loss": 4.2721, + "step": 311296 + }, + { + "epoch": 0.0, + "learning_rate": 4.489474325569999e-05, + "loss": 4.2769, + "step": 311808 + }, + { + "epoch": 0.0, + "learning_rate": 4.488635730818947e-05, + "loss": 4.2753, + "step": 312320 + }, + { + "epoch": 0.0, + "learning_rate": 4.487797136067895e-05, + "loss": 4.2826, + "step": 312832 + }, + { + "epoch": 0.0, + "learning_rate": 4.486958541316843e-05, + "loss": 4.2692, + "step": 313344 + }, + { + "epoch": 0.0, + "learning_rate": 4.486119946565791e-05, + "loss": 4.2707, + "step": 313856 + }, + { + "epoch": 0.0, + "learning_rate": 4.485281351814739e-05, + "loss": 4.2708, + "step": 314368 + }, + { + "epoch": 0.0, + "learning_rate": 4.484442757063687e-05, + "loss": 4.2736, + "step": 314880 + }, + { + "epoch": 0.0, + "learning_rate": 4.483604162312635e-05, + "loss": 4.2841, + "step": 315392 + }, + { + "epoch": 0.0, + "learning_rate": 4.482765567561583e-05, + "loss": 4.2755, + "step": 315904 + }, + { + "epoch": 0.0, + "learning_rate": 4.481926972810531e-05, + "loss": 4.2828, + "step": 316416 + }, + { + "epoch": 0.0, + "learning_rate": 4.4810900159398524e-05, + "loss": 4.2833, + "step": 316928 + }, + { + "epoch": 0.0, + "learning_rate": 4.480253059069173e-05, + "loss": 4.2746, + "step": 317440 + }, + { + "epoch": 0.0, + "learning_rate": 4.479414464318121e-05, + "loss": 4.2715, + "step": 317952 + }, + { + "epoch": 0.0, + "learning_rate": 4.478575869567069e-05, + "loss": 4.2717, + "step": 318464 + }, + { + "epoch": 0.0, + "learning_rate": 4.477737274816017e-05, + "loss": 4.2794, + "step": 318976 + }, + { + "epoch": 0.0, + "learning_rate": 4.476898680064965e-05, + "loss": 4.2666, + "step": 319488 + }, + { + "epoch": 0.0, + "learning_rate": 4.4760600853139126e-05, + "loss": 4.2722, + "step": 320000 + }, + { + "epoch": 0.0, + "learning_rate": 4.4752214905628606e-05, + "loss": 4.269, + "step": 320512 + }, + { + "epoch": 0.01, + "learning_rate": 4.4743828958118086e-05, + "loss": 4.2605, + "step": 321024 + }, + { + "epoch": 0.01, + "learning_rate": 4.473544301060757e-05, + "loss": 4.277, + "step": 321536 + }, + { + "epoch": 0.01, + "learning_rate": 4.472705706309705e-05, + "loss": 4.2793, + "step": 322048 + }, + { + "epoch": 0.01, + "learning_rate": 4.471867111558653e-05, + "loss": 4.27, + "step": 322560 + }, + { + "epoch": 0.01, + "learning_rate": 4.471030154687974e-05, + "loss": 4.2643, + "step": 323072 + }, + { + "epoch": 0.01, + "learning_rate": 4.470191559936922e-05, + "loss": 4.2729, + "step": 323584 + }, + { + "epoch": 0.01, + "learning_rate": 4.469354603066243e-05, + "loss": 4.2702, + "step": 324096 + }, + { + "epoch": 0.01, + "learning_rate": 4.468516008315191e-05, + "loss": 4.2702, + "step": 324608 + }, + { + "epoch": 0.01, + "learning_rate": 4.467677413564139e-05, + "loss": 4.2683, + "step": 325120 + }, + { + "epoch": 0.01, + "learning_rate": 4.466838818813087e-05, + "loss": 4.2555, + "step": 325632 + }, + { + "epoch": 0.01, + "learning_rate": 4.466000224062035e-05, + "loss": 4.2551, + "step": 326144 + }, + { + "epoch": 0.01, + "learning_rate": 4.465161629310983e-05, + "loss": 4.2537, + "step": 326656 + }, + { + "epoch": 0.01, + "learning_rate": 4.464323034559931e-05, + "loss": 4.2678, + "step": 327168 + }, + { + "epoch": 0.01, + "learning_rate": 4.4634860776892526e-05, + "loss": 4.2607, + "step": 327680 + }, + { + "epoch": 0.01, + "learning_rate": 4.4626474829382006e-05, + "loss": 4.2611, + "step": 328192 + }, + { + "epoch": 0.01, + "learning_rate": 4.4618088881871486e-05, + "loss": 4.2605, + "step": 328704 + }, + { + "epoch": 0.01, + "learning_rate": 4.4609702934360966e-05, + "loss": 4.2588, + "step": 329216 + }, + { + "epoch": 0.01, + "learning_rate": 4.4601333365654175e-05, + "loss": 4.2739, + "step": 329728 + }, + { + "epoch": 0.01, + "learning_rate": 4.4592947418143655e-05, + "loss": 4.2569, + "step": 330240 + }, + { + "epoch": 0.01, + "learning_rate": 4.4584561470633135e-05, + "loss": 4.2547, + "step": 330752 + }, + { + "epoch": 0.01, + "learning_rate": 4.4576175523122615e-05, + "loss": 4.2614, + "step": 331264 + }, + { + "epoch": 0.01, + "learning_rate": 4.4567789575612095e-05, + "loss": 4.2654, + "step": 331776 + }, + { + "epoch": 0.01, + "learning_rate": 4.4559420006905304e-05, + "loss": 4.2506, + "step": 332288 + }, + { + "epoch": 0.01, + "learning_rate": 4.4551034059394784e-05, + "loss": 4.2596, + "step": 332800 + }, + { + "epoch": 0.01, + "learning_rate": 4.4542648111884264e-05, + "loss": 4.2515, + "step": 333312 + }, + { + "epoch": 0.01, + "learning_rate": 4.4534262164373744e-05, + "loss": 4.257, + "step": 333824 + }, + { + "epoch": 0.01, + "learning_rate": 4.4525876216863224e-05, + "loss": 4.2613, + "step": 334336 + }, + { + "epoch": 0.01, + "learning_rate": 4.4517490269352704e-05, + "loss": 4.2451, + "step": 334848 + }, + { + "epoch": 0.01, + "learning_rate": 4.450910432184219e-05, + "loss": 4.2643, + "step": 335360 + }, + { + "epoch": 0.01, + "learning_rate": 4.4500718374331664e-05, + "loss": 4.2502, + "step": 335872 + }, + { + "epoch": 0.01, + "learning_rate": 4.449234880562488e-05, + "loss": 4.2403, + "step": 336384 + }, + { + "epoch": 0.01, + "learning_rate": 4.448396285811436e-05, + "loss": 4.2505, + "step": 336896 + }, + { + "epoch": 0.01, + "learning_rate": 4.447557691060384e-05, + "loss": 4.2533, + "step": 337408 + }, + { + "epoch": 0.01, + "learning_rate": 4.446719096309331e-05, + "loss": 4.2562, + "step": 337920 + }, + { + "epoch": 0.01, + "learning_rate": 4.445880501558279e-05, + "loss": 4.2465, + "step": 338432 + }, + { + "epoch": 0.01, + "learning_rate": 4.445045182567974e-05, + "loss": 4.2511, + "step": 338944 + }, + { + "epoch": 0.01, + "learning_rate": 4.444206587816922e-05, + "loss": 4.2277, + "step": 339456 + }, + { + "epoch": 0.01, + "learning_rate": 4.44336799306587e-05, + "loss": 4.2594, + "step": 339968 + }, + { + "epoch": 0.01, + "learning_rate": 4.442529398314818e-05, + "loss": 4.2409, + "step": 340480 + }, + { + "epoch": 0.01, + "learning_rate": 4.441690803563766e-05, + "loss": 4.2449, + "step": 340992 + }, + { + "epoch": 0.01, + "learning_rate": 4.440852208812714e-05, + "loss": 4.2577, + "step": 341504 + }, + { + "epoch": 0.01, + "learning_rate": 4.440013614061662e-05, + "loss": 4.246, + "step": 342016 + }, + { + "epoch": 0.01, + "learning_rate": 4.43917501931061e-05, + "loss": 4.2513, + "step": 342528 + }, + { + "epoch": 0.01, + "learning_rate": 4.4383380624399314e-05, + "loss": 4.2582, + "step": 343040 + }, + { + "epoch": 0.01, + "learning_rate": 4.437499467688879e-05, + "loss": 4.2553, + "step": 343552 + }, + { + "epoch": 0.01, + "learning_rate": 4.436660872937827e-05, + "loss": 4.2571, + "step": 344064 + }, + { + "epoch": 0.01, + "learning_rate": 4.4358222781867747e-05, + "loss": 4.2425, + "step": 344576 + }, + { + "epoch": 0.01, + "learning_rate": 4.4349836834357227e-05, + "loss": 4.237, + "step": 345088 + }, + { + "epoch": 0.01, + "learning_rate": 4.4341450886846707e-05, + "loss": 4.2461, + "step": 345600 + }, + { + "epoch": 0.01, + "learning_rate": 4.4333064939336186e-05, + "loss": 4.2568, + "step": 346112 + }, + { + "epoch": 0.01, + "learning_rate": 4.4324695370629396e-05, + "loss": 4.2452, + "step": 346624 + }, + { + "epoch": 0.01, + "learning_rate": 4.431630942311888e-05, + "loss": 4.2408, + "step": 347136 + }, + { + "epoch": 0.01, + "learning_rate": 4.430792347560836e-05, + "loss": 4.2288, + "step": 347648 + }, + { + "epoch": 0.01, + "learning_rate": 4.429953752809784e-05, + "loss": 4.2398, + "step": 348160 + }, + { + "epoch": 0.01, + "learning_rate": 4.429115158058732e-05, + "loss": 4.2354, + "step": 348672 + }, + { + "epoch": 0.01, + "learning_rate": 4.42827656330768e-05, + "loss": 4.2453, + "step": 349184 + }, + { + "epoch": 0.01, + "learning_rate": 4.427437968556628e-05, + "loss": 4.2431, + "step": 349696 + }, + { + "epoch": 0.01, + "learning_rate": 4.426601011685949e-05, + "loss": 4.255, + "step": 350208 + }, + { + "epoch": 0.01, + "learning_rate": 4.425762416934897e-05, + "loss": 4.2385, + "step": 350720 + }, + { + "epoch": 0.02, + "learning_rate": 4.424923822183845e-05, + "loss": 4.2418, + "step": 351232 + }, + { + "epoch": 0.02, + "learning_rate": 4.424085227432793e-05, + "loss": 4.2286, + "step": 351744 + }, + { + "epoch": 0.02, + "learning_rate": 4.423248270562114e-05, + "loss": 4.2357, + "step": 352256 + }, + { + "epoch": 0.02, + "learning_rate": 4.422409675811062e-05, + "loss": 4.241, + "step": 352768 + }, + { + "epoch": 0.02, + "learning_rate": 4.42157108106001e-05, + "loss": 4.2309, + "step": 353280 + }, + { + "epoch": 0.02, + "learning_rate": 4.420732486308958e-05, + "loss": 4.2304, + "step": 353792 + }, + { + "epoch": 0.02, + "learning_rate": 4.419893891557907e-05, + "loss": 4.2455, + "step": 354304 + }, + { + "epoch": 0.02, + "learning_rate": 4.419055296806855e-05, + "loss": 4.234, + "step": 354816 + }, + { + "epoch": 0.02, + "learning_rate": 4.418216702055803e-05, + "loss": 4.2542, + "step": 355328 + }, + { + "epoch": 0.02, + "learning_rate": 4.41737810730475e-05, + "loss": 4.235, + "step": 355840 + }, + { + "epoch": 0.02, + "learning_rate": 4.416539512553698e-05, + "loss": 4.2429, + "step": 356352 + }, + { + "epoch": 0.02, + "learning_rate": 4.415700917802646e-05, + "loss": 4.2359, + "step": 356864 + }, + { + "epoch": 0.02, + "learning_rate": 4.414862323051594e-05, + "loss": 4.2338, + "step": 357376 + }, + { + "epoch": 0.02, + "learning_rate": 4.414023728300542e-05, + "loss": 4.2496, + "step": 357888 + }, + { + "epoch": 0.02, + "learning_rate": 4.4131884093102365e-05, + "loss": 4.2317, + "step": 358400 + }, + { + "epoch": 0.02, + "learning_rate": 4.4123498145591845e-05, + "loss": 4.2502, + "step": 358912 + }, + { + "epoch": 0.02, + "learning_rate": 4.411511219808132e-05, + "loss": 4.2271, + "step": 359424 + }, + { + "epoch": 0.02, + "learning_rate": 4.4106726250570805e-05, + "loss": 4.2229, + "step": 359936 + }, + { + "epoch": 0.02, + "learning_rate": 4.4098340303060284e-05, + "loss": 4.2233, + "step": 360448 + }, + { + "epoch": 0.02, + "learning_rate": 4.4089954355549764e-05, + "loss": 4.238, + "step": 360960 + }, + { + "epoch": 0.02, + "learning_rate": 4.4081584786842974e-05, + "loss": 4.2358, + "step": 361472 + }, + { + "epoch": 0.02, + "learning_rate": 4.4073198839332453e-05, + "loss": 4.2216, + "step": 361984 + }, + { + "epoch": 0.02, + "learning_rate": 4.4064812891821933e-05, + "loss": 4.2157, + "step": 362496 + }, + { + "epoch": 0.02, + "learning_rate": 4.4056426944311413e-05, + "loss": 4.2255, + "step": 363008 + }, + { + "epoch": 0.02, + "learning_rate": 4.404804099680089e-05, + "loss": 4.2439, + "step": 363520 + }, + { + "epoch": 0.02, + "learning_rate": 4.403965504929037e-05, + "loss": 4.2261, + "step": 364032 + }, + { + "epoch": 0.02, + "learning_rate": 4.403128548058358e-05, + "loss": 4.2425, + "step": 364544 + }, + { + "epoch": 0.02, + "learning_rate": 4.402289953307306e-05, + "loss": 4.2318, + "step": 365056 + }, + { + "epoch": 0.02, + "learning_rate": 4.401451358556254e-05, + "loss": 4.2369, + "step": 365568 + }, + { + "epoch": 0.02, + "learning_rate": 4.400612763805202e-05, + "loss": 4.235, + "step": 366080 + }, + { + "epoch": 0.02, + "learning_rate": 4.39977416905415e-05, + "loss": 4.2242, + "step": 366592 + }, + { + "epoch": 0.02, + "learning_rate": 4.398935574303098e-05, + "loss": 4.2241, + "step": 367104 + }, + { + "epoch": 0.02, + "learning_rate": 4.398096979552047e-05, + "loss": 4.2314, + "step": 367616 + }, + { + "epoch": 0.02, + "learning_rate": 4.397258384800995e-05, + "loss": 4.2243, + "step": 368128 + }, + { + "epoch": 0.02, + "learning_rate": 4.396421427930316e-05, + "loss": 4.2272, + "step": 368640 + }, + { + "epoch": 0.02, + "learning_rate": 4.395582833179264e-05, + "loss": 4.2293, + "step": 369152 + }, + { + "epoch": 0.02, + "learning_rate": 4.394744238428212e-05, + "loss": 4.2298, + "step": 369664 + }, + { + "epoch": 0.02, + "learning_rate": 4.39390564367716e-05, + "loss": 4.2211, + "step": 370176 + }, + { + "epoch": 0.02, + "learning_rate": 4.393068686806481e-05, + "loss": 4.2202, + "step": 370688 + }, + { + "epoch": 0.02, + "learning_rate": 4.392230092055429e-05, + "loss": 4.2358, + "step": 371200 + }, + { + "epoch": 0.02, + "learning_rate": 4.391391497304377e-05, + "loss": 4.2203, + "step": 371712 + }, + { + "epoch": 0.02, + "learning_rate": 4.390552902553325e-05, + "loss": 4.2362, + "step": 372224 + }, + { + "epoch": 0.02, + "learning_rate": 4.3897159456826456e-05, + "loss": 4.2296, + "step": 372736 + }, + { + "epoch": 0.02, + "learning_rate": 4.3888773509315936e-05, + "loss": 4.2281, + "step": 373248 + }, + { + "epoch": 0.02, + "learning_rate": 4.388038756180542e-05, + "loss": 4.2267, + "step": 373760 + }, + { + "epoch": 0.02, + "learning_rate": 4.38720016142949e-05, + "loss": 4.2238, + "step": 374272 + }, + { + "epoch": 0.02, + "learning_rate": 4.386363204558811e-05, + "loss": 4.228, + "step": 374784 + }, + { + "epoch": 0.02, + "learning_rate": 4.385524609807759e-05, + "loss": 4.2227, + "step": 375296 + }, + { + "epoch": 0.02, + "learning_rate": 4.384686015056707e-05, + "loss": 4.2375, + "step": 375808 + }, + { + "epoch": 0.02, + "learning_rate": 4.383849058186028e-05, + "loss": 4.2304, + "step": 376320 + }, + { + "epoch": 0.02, + "learning_rate": 4.383010463434976e-05, + "loss": 4.2324, + "step": 376832 + }, + { + "epoch": 0.02, + "learning_rate": 4.382171868683924e-05, + "loss": 4.2303, + "step": 377344 + }, + { + "epoch": 0.02, + "learning_rate": 4.381333273932872e-05, + "loss": 4.2213, + "step": 377856 + }, + { + "epoch": 0.02, + "learning_rate": 4.38049467918182e-05, + "loss": 4.2243, + "step": 378368 + }, + { + "epoch": 0.02, + "learning_rate": 4.379656084430768e-05, + "loss": 4.2164, + "step": 378880 + }, + { + "epoch": 0.02, + "learning_rate": 4.378817489679716e-05, + "loss": 4.2196, + "step": 379392 + }, + { + "epoch": 0.02, + "learning_rate": 4.3779805328090376e-05, + "loss": 4.2229, + "step": 379904 + }, + { + "epoch": 0.02, + "learning_rate": 4.3771419380579856e-05, + "loss": 4.2184, + "step": 380416 + }, + { + "epoch": 0.02, + "learning_rate": 4.3763033433069336e-05, + "loss": 4.21, + "step": 380928 + }, + { + "epoch": 0.02, + "learning_rate": 4.375464748555881e-05, + "loss": 4.2226, + "step": 381440 + }, + { + "epoch": 0.03, + "eval_loss": 4.208243370056152, + "eval_runtime": 539.7013, + "eval_samples_per_second": 707.041, + "eval_steps_per_second": 22.096, + "step": 381595 + }, + { + "epoch": 1.0, + "learning_rate": 4.374626153804829e-05, + "loss": 4.228, + "step": 381952 + }, + { + "epoch": 1.0, + "learning_rate": 4.373787559053777e-05, + "loss": 4.2249, + "step": 382464 + }, + { + "epoch": 1.0, + "learning_rate": 4.372948964302725e-05, + "loss": 4.2225, + "step": 382976 + }, + { + "epoch": 1.0, + "learning_rate": 4.372110369551673e-05, + "loss": 4.2185, + "step": 383488 + }, + { + "epoch": 1.0, + "learning_rate": 4.371273412680994e-05, + "loss": 4.2164, + "step": 384000 + }, + { + "epoch": 1.0, + "learning_rate": 4.370434817929942e-05, + "loss": 4.2056, + "step": 384512 + }, + { + "epoch": 1.0, + "learning_rate": 4.36959622317889e-05, + "loss": 4.2187, + "step": 385024 + }, + { + "epoch": 1.0, + "learning_rate": 4.368757628427838e-05, + "loss": 4.2112, + "step": 385536 + }, + { + "epoch": 1.0, + "learning_rate": 4.367919033676786e-05, + "loss": 4.2165, + "step": 386048 + }, + { + "epoch": 1.0, + "learning_rate": 4.3670820768061074e-05, + "loss": 4.2206, + "step": 386560 + }, + { + "epoch": 1.0, + "learning_rate": 4.3662434820550554e-05, + "loss": 4.2248, + "step": 387072 + }, + { + "epoch": 1.0, + "learning_rate": 4.3654048873040034e-05, + "loss": 4.2093, + "step": 387584 + }, + { + "epoch": 1.0, + "learning_rate": 4.3645662925529514e-05, + "loss": 4.2157, + "step": 388096 + }, + { + "epoch": 1.0, + "learning_rate": 4.3637276978018994e-05, + "loss": 4.2111, + "step": 388608 + }, + { + "epoch": 1.0, + "learning_rate": 4.36289074093122e-05, + "loss": 4.2183, + "step": 389120 + }, + { + "epoch": 1.0, + "learning_rate": 4.362052146180168e-05, + "loss": 4.2072, + "step": 389632 + }, + { + "epoch": 1.0, + "learning_rate": 4.361213551429116e-05, + "loss": 4.2073, + "step": 390144 + }, + { + "epoch": 1.0, + "learning_rate": 4.360374956678064e-05, + "loss": 4.2089, + "step": 390656 + }, + { + "epoch": 1.0, + "learning_rate": 4.359536361927012e-05, + "loss": 4.2143, + "step": 391168 + }, + { + "epoch": 1.0, + "learning_rate": 4.35869776717596e-05, + "loss": 4.2217, + "step": 391680 + }, + { + "epoch": 1.0, + "learning_rate": 4.357859172424908e-05, + "loss": 4.2124, + "step": 392192 + }, + { + "epoch": 1.0, + "learning_rate": 4.357020577673856e-05, + "loss": 4.2237, + "step": 392704 + }, + { + "epoch": 1.0, + "learning_rate": 4.356183620803178e-05, + "loss": 4.2168, + "step": 393216 + }, + { + "epoch": 1.0, + "learning_rate": 4.355345026052126e-05, + "loss": 4.2157, + "step": 393728 + }, + { + "epoch": 1.0, + "learning_rate": 4.354506431301074e-05, + "loss": 4.2085, + "step": 394240 + }, + { + "epoch": 1.0, + "learning_rate": 4.353667836550022e-05, + "loss": 4.2089, + "step": 394752 + }, + { + "epoch": 1.0, + "learning_rate": 4.352830879679343e-05, + "loss": 4.2204, + "step": 395264 + }, + { + "epoch": 1.0, + "learning_rate": 4.351992284928291e-05, + "loss": 4.2045, + "step": 395776 + }, + { + "epoch": 1.0, + "learning_rate": 4.351153690177239e-05, + "loss": 4.2128, + "step": 396288 + }, + { + "epoch": 1.0, + "learning_rate": 4.350315095426187e-05, + "loss": 4.2054, + "step": 396800 + }, + { + "epoch": 1.01, + "learning_rate": 4.3494781385555076e-05, + "loss": 4.1978, + "step": 397312 + }, + { + "epoch": 1.01, + "learning_rate": 4.3486411816848285e-05, + "loss": 4.2202, + "step": 397824 + }, + { + "epoch": 1.01, + "learning_rate": 4.3478025869337765e-05, + "loss": 4.215, + "step": 398336 + }, + { + "epoch": 1.01, + "learning_rate": 4.346963992182725e-05, + "loss": 4.2127, + "step": 398848 + }, + { + "epoch": 1.01, + "learning_rate": 4.346125397431673e-05, + "loss": 4.2003, + "step": 399360 + }, + { + "epoch": 1.01, + "learning_rate": 4.345286802680621e-05, + "loss": 4.2157, + "step": 399872 + }, + { + "epoch": 1.01, + "learning_rate": 4.344449845809942e-05, + "loss": 4.2055, + "step": 400384 + }, + { + "epoch": 1.01, + "learning_rate": 4.343612888939263e-05, + "loss": 4.213, + "step": 400896 + }, + { + "epoch": 1.01, + "learning_rate": 4.342774294188211e-05, + "loss": 4.2074, + "step": 401408 + }, + { + "epoch": 1.01, + "learning_rate": 4.341935699437159e-05, + "loss": 4.1952, + "step": 401920 + }, + { + "epoch": 1.01, + "learning_rate": 4.341097104686107e-05, + "loss": 4.1956, + "step": 402432 + }, + { + "epoch": 1.01, + "learning_rate": 4.340258509935055e-05, + "loss": 4.1914, + "step": 402944 + }, + { + "epoch": 1.01, + "learning_rate": 4.339419915184003e-05, + "loss": 4.2113, + "step": 403456 + }, + { + "epoch": 1.01, + "learning_rate": 4.338581320432951e-05, + "loss": 4.2007, + "step": 403968 + }, + { + "epoch": 1.01, + "learning_rate": 4.337742725681899e-05, + "loss": 4.1962, + "step": 404480 + }, + { + "epoch": 1.01, + "learning_rate": 4.336904130930847e-05, + "loss": 4.2038, + "step": 404992 + }, + { + "epoch": 1.01, + "learning_rate": 4.3360671740601686e-05, + "loss": 4.1994, + "step": 405504 + }, + { + "epoch": 1.01, + "learning_rate": 4.3352285793091166e-05, + "loss": 4.2145, + "step": 406016 + }, + { + "epoch": 1.01, + "learning_rate": 4.334389984558064e-05, + "loss": 4.1954, + "step": 406528 + }, + { + "epoch": 1.01, + "learning_rate": 4.333551389807012e-05, + "loss": 4.1971, + "step": 407040 + }, + { + "epoch": 1.01, + "learning_rate": 4.33271279505596e-05, + "loss": 4.2042, + "step": 407552 + }, + { + "epoch": 1.01, + "learning_rate": 4.331874200304908e-05, + "loss": 4.2038, + "step": 408064 + }, + { + "epoch": 1.01, + "learning_rate": 4.331035605553856e-05, + "loss": 4.1928, + "step": 408576 + }, + { + "epoch": 1.01, + "learning_rate": 4.330198648683177e-05, + "loss": 4.1987, + "step": 409088 + }, + { + "epoch": 1.01, + "learning_rate": 4.329360053932125e-05, + "loss": 4.1947, + "step": 409600 + }, + { + "epoch": 1.01, + "learning_rate": 4.328521459181073e-05, + "loss": 4.1995, + "step": 410112 + }, + { + "epoch": 1.01, + "learning_rate": 4.327682864430021e-05, + "loss": 4.1989, + "step": 410624 + }, + { + "epoch": 1.01, + "learning_rate": 4.326844269678969e-05, + "loss": 4.1926, + "step": 411136 + }, + { + "epoch": 1.01, + "learning_rate": 4.326005674927917e-05, + "loss": 4.2054, + "step": 411648 + }, + { + "epoch": 1.01, + "learning_rate": 4.3251670801768654e-05, + "loss": 4.1886, + "step": 412160 + }, + { + "epoch": 1.01, + "learning_rate": 4.3243284854258134e-05, + "loss": 4.1866, + "step": 412672 + }, + { + "epoch": 1.01, + "learning_rate": 4.323491528555134e-05, + "loss": 4.1904, + "step": 413184 + }, + { + "epoch": 1.01, + "learning_rate": 4.322652933804082e-05, + "loss": 4.1896, + "step": 413696 + }, + { + "epoch": 1.01, + "learning_rate": 4.321815976933403e-05, + "loss": 4.2066, + "step": 414208 + }, + { + "epoch": 1.01, + "learning_rate": 4.320977382182351e-05, + "loss": 4.1871, + "step": 414720 + }, + { + "epoch": 1.01, + "learning_rate": 4.320138787431299e-05, + "loss": 4.1966, + "step": 415232 + }, + { + "epoch": 1.01, + "learning_rate": 4.319300192680247e-05, + "loss": 4.1755, + "step": 415744 + }, + { + "epoch": 1.01, + "learning_rate": 4.318463235809568e-05, + "loss": 4.1966, + "step": 416256 + }, + { + "epoch": 1.01, + "learning_rate": 4.317624641058516e-05, + "loss": 4.1843, + "step": 416768 + }, + { + "epoch": 1.01, + "learning_rate": 4.316786046307464e-05, + "loss": 4.1862, + "step": 417280 + }, + { + "epoch": 1.01, + "learning_rate": 4.315947451556412e-05, + "loss": 4.1988, + "step": 417792 + }, + { + "epoch": 1.01, + "learning_rate": 4.315108856805361e-05, + "loss": 4.1894, + "step": 418304 + }, + { + "epoch": 1.01, + "learning_rate": 4.314270262054309e-05, + "loss": 4.1947, + "step": 418816 + }, + { + "epoch": 1.01, + "learning_rate": 4.313431667303257e-05, + "loss": 4.2015, + "step": 419328 + }, + { + "epoch": 1.01, + "learning_rate": 4.312593072552205e-05, + "loss": 4.196, + "step": 419840 + }, + { + "epoch": 1.01, + "learning_rate": 4.311756115681526e-05, + "loss": 4.2038, + "step": 420352 + }, + { + "epoch": 1.01, + "learning_rate": 4.310917520930474e-05, + "loss": 4.1884, + "step": 420864 + }, + { + "epoch": 1.01, + "learning_rate": 4.310078926179422e-05, + "loss": 4.1801, + "step": 421376 + }, + { + "epoch": 1.01, + "learning_rate": 4.30924033142837e-05, + "loss": 4.1904, + "step": 421888 + }, + { + "epoch": 1.01, + "learning_rate": 4.308401736677318e-05, + "loss": 4.2001, + "step": 422400 + }, + { + "epoch": 1.01, + "learning_rate": 4.3075647798066386e-05, + "loss": 4.1879, + "step": 422912 + }, + { + "epoch": 1.01, + "learning_rate": 4.3067261850555866e-05, + "loss": 4.1871, + "step": 423424 + }, + { + "epoch": 1.01, + "learning_rate": 4.3058875903045346e-05, + "loss": 4.1727, + "step": 423936 + }, + { + "epoch": 1.01, + "learning_rate": 4.3050489955534826e-05, + "loss": 4.186, + "step": 424448 + }, + { + "epoch": 1.01, + "learning_rate": 4.304212038682804e-05, + "loss": 4.1776, + "step": 424960 + }, + { + "epoch": 1.01, + "learning_rate": 4.303373443931752e-05, + "loss": 4.1908, + "step": 425472 + }, + { + "epoch": 1.01, + "learning_rate": 4.3025348491807e-05, + "loss": 4.1916, + "step": 425984 + }, + { + "epoch": 1.01, + "learning_rate": 4.301697892310021e-05, + "loss": 4.1959, + "step": 426496 + }, + { + "epoch": 1.01, + "learning_rate": 4.300859297558969e-05, + "loss": 4.1848, + "step": 427008 + }, + { + "epoch": 1.02, + "learning_rate": 4.300020702807917e-05, + "loss": 4.1884, + "step": 427520 + }, + { + "epoch": 1.02, + "learning_rate": 4.299182108056865e-05, + "loss": 4.1729, + "step": 428032 + }, + { + "epoch": 1.02, + "learning_rate": 4.2983435133058124e-05, + "loss": 4.1807, + "step": 428544 + }, + { + "epoch": 1.02, + "learning_rate": 4.2975049185547604e-05, + "loss": 4.1844, + "step": 429056 + }, + { + "epoch": 1.02, + "learning_rate": 4.2966663238037084e-05, + "loss": 4.1791, + "step": 429568 + }, + { + "epoch": 1.02, + "learning_rate": 4.29582936693303e-05, + "loss": 4.1754, + "step": 430080 + }, + { + "epoch": 1.02, + "learning_rate": 4.294990772181978e-05, + "loss": 4.1926, + "step": 430592 + }, + { + "epoch": 1.02, + "learning_rate": 4.294152177430926e-05, + "loss": 4.1789, + "step": 431104 + }, + { + "epoch": 1.02, + "learning_rate": 4.293313582679874e-05, + "loss": 4.2014, + "step": 431616 + }, + { + "epoch": 1.02, + "learning_rate": 4.292474987928822e-05, + "loss": 4.1746, + "step": 432128 + }, + { + "epoch": 1.02, + "learning_rate": 4.29163639317777e-05, + "loss": 4.1929, + "step": 432640 + }, + { + "epoch": 1.02, + "learning_rate": 4.290797798426718e-05, + "loss": 4.1817, + "step": 433152 + }, + { + "epoch": 1.02, + "learning_rate": 4.289959203675666e-05, + "loss": 4.1801, + "step": 433664 + }, + { + "epoch": 1.02, + "learning_rate": 4.289122246804987e-05, + "loss": 4.1989, + "step": 434176 + }, + { + "epoch": 1.02, + "learning_rate": 4.288283652053935e-05, + "loss": 4.1751, + "step": 434688 + }, + { + "epoch": 1.02, + "learning_rate": 4.287445057302883e-05, + "loss": 4.1956, + "step": 435200 + }, + { + "epoch": 1.02, + "learning_rate": 4.286606462551831e-05, + "loss": 4.173, + "step": 435712 + }, + { + "epoch": 1.02, + "learning_rate": 4.285769505681152e-05, + "loss": 4.1707, + "step": 436224 + }, + { + "epoch": 1.02, + "learning_rate": 4.284932548810473e-05, + "loss": 4.1698, + "step": 436736 + }, + { + "epoch": 1.02, + "learning_rate": 4.284093954059421e-05, + "loss": 4.1827, + "step": 437248 + }, + { + "epoch": 1.02, + "learning_rate": 4.283255359308369e-05, + "loss": 4.1827, + "step": 437760 + }, + { + "epoch": 1.02, + "learning_rate": 4.282416764557317e-05, + "loss": 4.1728, + "step": 438272 + }, + { + "epoch": 1.02, + "learning_rate": 4.281578169806265e-05, + "loss": 4.16, + "step": 438784 + }, + { + "epoch": 1.02, + "learning_rate": 4.280739575055213e-05, + "loss": 4.1725, + "step": 439296 + }, + { + "epoch": 1.02, + "learning_rate": 4.279902618184534e-05, + "loss": 4.1866, + "step": 439808 + }, + { + "epoch": 1.02, + "learning_rate": 4.279064023433482e-05, + "loss": 4.1807, + "step": 440320 + }, + { + "epoch": 1.02, + "learning_rate": 4.27822542868243e-05, + "loss": 4.1848, + "step": 440832 + }, + { + "epoch": 1.02, + "learning_rate": 4.277386833931378e-05, + "loss": 4.1818, + "step": 441344 + }, + { + "epoch": 1.02, + "learning_rate": 4.276548239180326e-05, + "loss": 4.1811, + "step": 441856 + }, + { + "epoch": 1.02, + "learning_rate": 4.275709644429274e-05, + "loss": 4.1817, + "step": 442368 + }, + { + "epoch": 1.02, + "learning_rate": 4.274871049678222e-05, + "loss": 4.1745, + "step": 442880 + }, + { + "epoch": 1.02, + "learning_rate": 4.27403245492717e-05, + "loss": 4.1706, + "step": 443392 + }, + { + "epoch": 1.02, + "learning_rate": 4.273195498056492e-05, + "loss": 4.1788, + "step": 443904 + }, + { + "epoch": 1.02, + "learning_rate": 4.27235690330544e-05, + "loss": 4.172, + "step": 444416 + }, + { + "epoch": 1.02, + "learning_rate": 4.271518308554388e-05, + "loss": 4.1759, + "step": 444928 + }, + { + "epoch": 1.02, + "learning_rate": 4.270679713803336e-05, + "loss": 4.1775, + "step": 445440 + }, + { + "epoch": 1.02, + "learning_rate": 4.269841119052284e-05, + "loss": 4.1771, + "step": 445952 + }, + { + "epoch": 1.02, + "learning_rate": 4.2690041621816046e-05, + "loss": 4.1734, + "step": 446464 + }, + { + "epoch": 1.02, + "learning_rate": 4.2681655674305526e-05, + "loss": 4.1683, + "step": 446976 + }, + { + "epoch": 1.02, + "learning_rate": 4.2673269726795006e-05, + "loss": 4.1839, + "step": 447488 + }, + { + "epoch": 1.02, + "learning_rate": 4.2664883779284486e-05, + "loss": 4.1707, + "step": 448000 + }, + { + "epoch": 1.02, + "learning_rate": 4.2656514210577695e-05, + "loss": 4.1821, + "step": 448512 + }, + { + "epoch": 1.02, + "learning_rate": 4.2648128263067175e-05, + "loss": 4.1843, + "step": 449024 + }, + { + "epoch": 1.02, + "learning_rate": 4.2639742315556655e-05, + "loss": 4.1759, + "step": 449536 + }, + { + "epoch": 1.02, + "learning_rate": 4.2631356368046135e-05, + "loss": 4.1788, + "step": 450048 + }, + { + "epoch": 1.02, + "learning_rate": 4.262298679933935e-05, + "loss": 4.1732, + "step": 450560 + }, + { + "epoch": 1.02, + "learning_rate": 4.261460085182883e-05, + "loss": 4.1744, + "step": 451072 + }, + { + "epoch": 1.02, + "learning_rate": 4.260621490431831e-05, + "loss": 4.1676, + "step": 451584 + }, + { + "epoch": 1.02, + "learning_rate": 4.259784533561152e-05, + "loss": 4.1873, + "step": 452096 + }, + { + "epoch": 1.02, + "learning_rate": 4.2589459388101e-05, + "loss": 4.1825, + "step": 452608 + }, + { + "epoch": 1.02, + "learning_rate": 4.258107344059048e-05, + "loss": 4.1783, + "step": 453120 + }, + { + "epoch": 1.02, + "learning_rate": 4.257268749307996e-05, + "loss": 4.1766, + "step": 453632 + }, + { + "epoch": 1.02, + "learning_rate": 4.256430154556943e-05, + "loss": 4.1733, + "step": 454144 + }, + { + "epoch": 1.02, + "learning_rate": 4.255593197686265e-05, + "loss": 4.1771, + "step": 454656 + }, + { + "epoch": 1.02, + "learning_rate": 4.254754602935213e-05, + "loss": 4.1689, + "step": 455168 + }, + { + "epoch": 1.02, + "learning_rate": 4.253916008184161e-05, + "loss": 4.1644, + "step": 455680 + }, + { + "epoch": 1.02, + "learning_rate": 4.253077413433109e-05, + "loss": 4.1752, + "step": 456192 + }, + { + "epoch": 1.02, + "learning_rate": 4.252238818682057e-05, + "loss": 4.1672, + "step": 456704 + }, + { + "epoch": 1.02, + "learning_rate": 4.251400223931005e-05, + "loss": 4.1592, + "step": 457216 + }, + { + "epoch": 1.02, + "learning_rate": 4.250561629179953e-05, + "loss": 4.1691, + "step": 457728 + }, + { + "epoch": 1.03, + "eval_loss": 4.166858196258545, + "eval_runtime": 544.061, + "eval_samples_per_second": 701.375, + "eval_steps_per_second": 21.918, + "step": 457914 + }, + { + "epoch": 0.0, + "learning_rate": 4.249723034428901e-05, + "loss": 4.1846, + "step": 458240 + }, + { + "epoch": 0.0, + "learning_rate": 4.248884439677849e-05, + "loss": 4.1739, + "step": 458752 + }, + { + "epoch": 0.0, + "learning_rate": 4.248045844926797e-05, + "loss": 4.1704, + "step": 459264 + }, + { + "epoch": 0.0, + "learning_rate": 4.247207250175745e-05, + "loss": 4.1694, + "step": 459776 + }, + { + "epoch": 0.0, + "learning_rate": 4.246368655424693e-05, + "loss": 4.1694, + "step": 460288 + }, + { + "epoch": 0.0, + "learning_rate": 4.245530060673641e-05, + "loss": 4.1586, + "step": 460800 + }, + { + "epoch": 0.0, + "learning_rate": 4.244691465922589e-05, + "loss": 4.1691, + "step": 461312 + }, + { + "epoch": 0.0, + "learning_rate": 4.243852871171537e-05, + "loss": 4.1615, + "step": 461824 + }, + { + "epoch": 0.0, + "learning_rate": 4.243014276420485e-05, + "loss": 4.1649, + "step": 462336 + }, + { + "epoch": 0.0, + "learning_rate": 4.242175681669432e-05, + "loss": 4.1709, + "step": 462848 + }, + { + "epoch": 0.0, + "learning_rate": 4.241337086918381e-05, + "loss": 4.1786, + "step": 463360 + }, + { + "epoch": 0.0, + "learning_rate": 4.240498492167329e-05, + "loss": 4.1558, + "step": 463872 + }, + { + "epoch": 0.0, + "learning_rate": 4.239659897416277e-05, + "loss": 4.1727, + "step": 464384 + }, + { + "epoch": 0.0, + "learning_rate": 4.238821302665225e-05, + "loss": 4.161, + "step": 464896 + }, + { + "epoch": 0.0, + "learning_rate": 4.237982707914173e-05, + "loss": 4.1724, + "step": 465408 + }, + { + "epoch": 0.0, + "learning_rate": 4.237144113163121e-05, + "loss": 4.1556, + "step": 465920 + }, + { + "epoch": 0.0, + "learning_rate": 4.236305518412069e-05, + "loss": 4.1575, + "step": 466432 + }, + { + "epoch": 0.0, + "learning_rate": 4.235466923661017e-05, + "loss": 4.1624, + "step": 466944 + }, + { + "epoch": 0.0, + "learning_rate": 4.234628328909965e-05, + "loss": 4.1621, + "step": 467456 + }, + { + "epoch": 0.0, + "learning_rate": 4.233789734158913e-05, + "loss": 4.1716, + "step": 467968 + }, + { + "epoch": 0.0, + "learning_rate": 4.232952777288234e-05, + "loss": 4.1667, + "step": 468480 + }, + { + "epoch": 0.0, + "learning_rate": 4.232114182537182e-05, + "loss": 4.1745, + "step": 468992 + }, + { + "epoch": 0.0, + "learning_rate": 4.2312772256665026e-05, + "loss": 4.1709, + "step": 469504 + }, + { + "epoch": 0.0, + "learning_rate": 4.2304386309154506e-05, + "loss": 4.1658, + "step": 470016 + }, + { + "epoch": 0.0, + "learning_rate": 4.2296000361643986e-05, + "loss": 4.1597, + "step": 470528 + }, + { + "epoch": 0.0, + "learning_rate": 4.228761441413347e-05, + "loss": 4.1606, + "step": 471040 + }, + { + "epoch": 0.0, + "learning_rate": 4.227922846662295e-05, + "loss": 4.1708, + "step": 471552 + }, + { + "epoch": 0.0, + "learning_rate": 4.227084251911243e-05, + "loss": 4.1587, + "step": 472064 + }, + { + "epoch": 0.0, + "learning_rate": 4.226245657160191e-05, + "loss": 4.1623, + "step": 472576 + }, + { + "epoch": 0.0, + "learning_rate": 4.2254070624091386e-05, + "loss": 4.1606, + "step": 473088 + }, + { + "epoch": 0.01, + "learning_rate": 4.22457010553846e-05, + "loss": 4.151, + "step": 473600 + }, + { + "epoch": 0.01, + "learning_rate": 4.223731510787408e-05, + "loss": 4.1681, + "step": 474112 + }, + { + "epoch": 0.01, + "learning_rate": 4.222892916036356e-05, + "loss": 4.169, + "step": 474624 + }, + { + "epoch": 0.01, + "learning_rate": 4.2220543212853035e-05, + "loss": 4.1655, + "step": 475136 + }, + { + "epoch": 0.01, + "learning_rate": 4.221217364414625e-05, + "loss": 4.1531, + "step": 475648 + }, + { + "epoch": 0.01, + "learning_rate": 4.220378769663573e-05, + "loss": 4.1669, + "step": 476160 + }, + { + "epoch": 0.01, + "learning_rate": 4.219541812792894e-05, + "loss": 4.1599, + "step": 476672 + }, + { + "epoch": 0.01, + "learning_rate": 4.2187032180418426e-05, + "loss": 4.1654, + "step": 477184 + }, + { + "epoch": 0.01, + "learning_rate": 4.2178646232907906e-05, + "loss": 4.1621, + "step": 477696 + }, + { + "epoch": 0.01, + "learning_rate": 4.2170260285397386e-05, + "loss": 4.1507, + "step": 478208 + }, + { + "epoch": 0.01, + "learning_rate": 4.216187433788686e-05, + "loss": 4.1455, + "step": 478720 + }, + { + "epoch": 0.01, + "learning_rate": 4.2153504769180075e-05, + "loss": 4.1445, + "step": 479232 + }, + { + "epoch": 0.01, + "learning_rate": 4.2145118821669555e-05, + "loss": 4.1688, + "step": 479744 + }, + { + "epoch": 0.01, + "learning_rate": 4.2136732874159035e-05, + "loss": 4.1534, + "step": 480256 + }, + { + "epoch": 0.01, + "learning_rate": 4.212834692664851e-05, + "loss": 4.1495, + "step": 480768 + }, + { + "epoch": 0.01, + "learning_rate": 4.2119977357941724e-05, + "loss": 4.16, + "step": 481280 + }, + { + "epoch": 0.01, + "learning_rate": 4.2111591410431204e-05, + "loss": 4.1531, + "step": 481792 + }, + { + "epoch": 0.01, + "learning_rate": 4.210320546292068e-05, + "loss": 4.1649, + "step": 482304 + }, + { + "epoch": 0.01, + "learning_rate": 4.2094819515410164e-05, + "loss": 4.1505, + "step": 482816 + }, + { + "epoch": 0.01, + "learning_rate": 4.2086433567899644e-05, + "loss": 4.1533, + "step": 483328 + }, + { + "epoch": 0.01, + "learning_rate": 4.2078047620389124e-05, + "loss": 4.1522, + "step": 483840 + }, + { + "epoch": 0.01, + "learning_rate": 4.2069661672878604e-05, + "loss": 4.1608, + "step": 484352 + }, + { + "epoch": 0.01, + "learning_rate": 4.2061275725368084e-05, + "loss": 4.1491, + "step": 484864 + }, + { + "epoch": 0.01, + "learning_rate": 4.205290615666129e-05, + "loss": 4.1478, + "step": 485376 + }, + { + "epoch": 0.01, + "learning_rate": 4.204452020915077e-05, + "loss": 4.1505, + "step": 485888 + }, + { + "epoch": 0.01, + "learning_rate": 4.203613426164025e-05, + "loss": 4.1572, + "step": 486400 + }, + { + "epoch": 0.01, + "learning_rate": 4.202774831412973e-05, + "loss": 4.1508, + "step": 486912 + }, + { + "epoch": 0.01, + "learning_rate": 4.201937874542294e-05, + "loss": 4.1459, + "step": 487424 + }, + { + "epoch": 0.01, + "learning_rate": 4.201099279791242e-05, + "loss": 4.1595, + "step": 487936 + }, + { + "epoch": 0.01, + "learning_rate": 4.20026068504019e-05, + "loss": 4.1447, + "step": 488448 + }, + { + "epoch": 0.01, + "learning_rate": 4.199422090289138e-05, + "loss": 4.1414, + "step": 488960 + }, + { + "epoch": 0.01, + "learning_rate": 4.198583495538086e-05, + "loss": 4.1452, + "step": 489472 + }, + { + "epoch": 0.01, + "learning_rate": 4.197744900787035e-05, + "loss": 4.1455, + "step": 489984 + }, + { + "epoch": 0.01, + "learning_rate": 4.196906306035983e-05, + "loss": 4.1621, + "step": 490496 + }, + { + "epoch": 0.01, + "learning_rate": 4.196067711284931e-05, + "loss": 4.1431, + "step": 491008 + }, + { + "epoch": 0.01, + "learning_rate": 4.195230754414252e-05, + "loss": 4.15, + "step": 491520 + }, + { + "epoch": 0.01, + "learning_rate": 4.1943921596632e-05, + "loss": 4.1329, + "step": 492032 + }, + { + "epoch": 0.01, + "learning_rate": 4.193553564912148e-05, + "loss": 4.1494, + "step": 492544 + }, + { + "epoch": 0.01, + "learning_rate": 4.192714970161096e-05, + "loss": 4.1397, + "step": 493056 + }, + { + "epoch": 0.01, + "learning_rate": 4.1918780132904167e-05, + "loss": 4.1425, + "step": 493568 + }, + { + "epoch": 0.01, + "learning_rate": 4.1910394185393647e-05, + "loss": 4.1519, + "step": 494080 + }, + { + "epoch": 0.01, + "learning_rate": 4.1902008237883126e-05, + "loss": 4.1458, + "step": 494592 + }, + { + "epoch": 0.01, + "learning_rate": 4.1893622290372606e-05, + "loss": 4.1517, + "step": 495104 + }, + { + "epoch": 0.01, + "learning_rate": 4.1885252721665816e-05, + "loss": 4.1578, + "step": 495616 + }, + { + "epoch": 0.01, + "learning_rate": 4.18768667741553e-05, + "loss": 4.152, + "step": 496128 + }, + { + "epoch": 0.01, + "learning_rate": 4.186848082664478e-05, + "loss": 4.1572, + "step": 496640 + }, + { + "epoch": 0.01, + "learning_rate": 4.186009487913426e-05, + "loss": 4.1502, + "step": 497152 + }, + { + "epoch": 0.01, + "learning_rate": 4.185172531042747e-05, + "loss": 4.1348, + "step": 497664 + }, + { + "epoch": 0.01, + "learning_rate": 4.184335574172068e-05, + "loss": 4.1455, + "step": 498176 + }, + { + "epoch": 0.01, + "learning_rate": 4.183496979421016e-05, + "loss": 4.152, + "step": 498688 + }, + { + "epoch": 0.01, + "learning_rate": 4.182658384669964e-05, + "loss": 4.1491, + "step": 499200 + }, + { + "epoch": 0.01, + "learning_rate": 4.181819789918912e-05, + "loss": 4.1461, + "step": 499712 + }, + { + "epoch": 0.01, + "learning_rate": 4.18098119516786e-05, + "loss": 4.1255, + "step": 500224 + }, + { + "epoch": 0.01, + "learning_rate": 4.180142600416808e-05, + "loss": 4.1447, + "step": 500736 + }, + { + "epoch": 0.01, + "learning_rate": 4.179304005665756e-05, + "loss": 4.1334, + "step": 501248 + }, + { + "epoch": 0.01, + "learning_rate": 4.178467048795077e-05, + "loss": 4.1484, + "step": 501760 + }, + { + "epoch": 0.01, + "learning_rate": 4.177628454044025e-05, + "loss": 4.1452, + "step": 502272 + }, + { + "epoch": 0.01, + "learning_rate": 4.1767898592929736e-05, + "loss": 4.1539, + "step": 502784 + }, + { + "epoch": 0.01, + "learning_rate": 4.1759512645419216e-05, + "loss": 4.1415, + "step": 503296 + }, + { + "epoch": 0.02, + "learning_rate": 4.1751126697908696e-05, + "loss": 4.144, + "step": 503808 + }, + { + "epoch": 0.02, + "learning_rate": 4.174274075039817e-05, + "loss": 4.1334, + "step": 504320 + }, + { + "epoch": 0.02, + "learning_rate": 4.173435480288765e-05, + "loss": 4.1432, + "step": 504832 + }, + { + "epoch": 0.02, + "learning_rate": 4.172596885537713e-05, + "loss": 4.1374, + "step": 505344 + }, + { + "epoch": 0.02, + "learning_rate": 4.171759928667034e-05, + "loss": 4.1384, + "step": 505856 + }, + { + "epoch": 0.02, + "learning_rate": 4.170921333915982e-05, + "loss": 4.1311, + "step": 506368 + }, + { + "epoch": 0.02, + "learning_rate": 4.17008273916493e-05, + "loss": 4.1461, + "step": 506880 + }, + { + "epoch": 0.02, + "learning_rate": 4.169244144413878e-05, + "loss": 4.1362, + "step": 507392 + }, + { + "epoch": 0.02, + "learning_rate": 4.168407187543199e-05, + "loss": 4.1612, + "step": 507904 + }, + { + "epoch": 0.02, + "learning_rate": 4.1675685927921474e-05, + "loss": 4.1343, + "step": 508416 + }, + { + "epoch": 0.02, + "learning_rate": 4.1667299980410954e-05, + "loss": 4.1512, + "step": 508928 + }, + { + "epoch": 0.02, + "learning_rate": 4.1658914032900434e-05, + "loss": 4.1398, + "step": 509440 + }, + { + "epoch": 0.02, + "learning_rate": 4.1650528085389914e-05, + "loss": 4.1373, + "step": 509952 + }, + { + "epoch": 0.02, + "learning_rate": 4.1642142137879393e-05, + "loss": 4.1536, + "step": 510464 + }, + { + "epoch": 0.02, + "learning_rate": 4.1633756190368873e-05, + "loss": 4.1368, + "step": 510976 + }, + { + "epoch": 0.02, + "learning_rate": 4.1625370242858353e-05, + "loss": 4.1508, + "step": 511488 + }, + { + "epoch": 0.02, + "learning_rate": 4.161700067415156e-05, + "loss": 4.1357, + "step": 512000 + }, + { + "epoch": 0.02, + "learning_rate": 4.160861472664104e-05, + "loss": 4.1256, + "step": 512512 + }, + { + "epoch": 0.02, + "learning_rate": 4.160024515793425e-05, + "loss": 4.1287, + "step": 513024 + }, + { + "epoch": 0.02, + "learning_rate": 4.159185921042373e-05, + "loss": 4.1408, + "step": 513536 + }, + { + "epoch": 0.02, + "learning_rate": 4.158347326291321e-05, + "loss": 4.1409, + "step": 514048 + }, + { + "epoch": 0.02, + "learning_rate": 4.157508731540269e-05, + "loss": 4.1352, + "step": 514560 + }, + { + "epoch": 0.02, + "learning_rate": 4.156670136789217e-05, + "loss": 4.1136, + "step": 515072 + }, + { + "epoch": 0.02, + "learning_rate": 4.155833179918539e-05, + "loss": 4.1319, + "step": 515584 + }, + { + "epoch": 0.02, + "learning_rate": 4.154994585167487e-05, + "loss": 4.1469, + "step": 516096 + }, + { + "epoch": 0.02, + "learning_rate": 4.154155990416435e-05, + "loss": 4.138, + "step": 516608 + }, + { + "epoch": 0.02, + "learning_rate": 4.153317395665383e-05, + "loss": 4.1443, + "step": 517120 + }, + { + "epoch": 0.02, + "learning_rate": 4.152478800914331e-05, + "loss": 4.1397, + "step": 517632 + }, + { + "epoch": 0.02, + "learning_rate": 4.151640206163279e-05, + "loss": 4.1409, + "step": 518144 + }, + { + "epoch": 0.02, + "learning_rate": 4.150801611412227e-05, + "loss": 4.1404, + "step": 518656 + }, + { + "epoch": 0.02, + "learning_rate": 4.149963016661175e-05, + "loss": 4.1358, + "step": 519168 + }, + { + "epoch": 0.02, + "learning_rate": 4.1491260597904956e-05, + "loss": 4.1282, + "step": 519680 + }, + { + "epoch": 0.02, + "learning_rate": 4.1482874650394436e-05, + "loss": 4.1365, + "step": 520192 + }, + { + "epoch": 0.02, + "learning_rate": 4.1474488702883916e-05, + "loss": 4.1326, + "step": 520704 + }, + { + "epoch": 0.02, + "learning_rate": 4.1466102755373396e-05, + "loss": 4.1352, + "step": 521216 + }, + { + "epoch": 0.02, + "learning_rate": 4.145773318666661e-05, + "loss": 4.1363, + "step": 521728 + }, + { + "epoch": 0.02, + "learning_rate": 4.144934723915609e-05, + "loss": 4.1369, + "step": 522240 + }, + { + "epoch": 0.02, + "learning_rate": 4.144096129164557e-05, + "loss": 4.1307, + "step": 522752 + }, + { + "epoch": 0.02, + "learning_rate": 4.143257534413505e-05, + "loss": 4.127, + "step": 523264 + }, + { + "epoch": 0.02, + "learning_rate": 4.142418939662453e-05, + "loss": 4.1418, + "step": 523776 + }, + { + "epoch": 0.02, + "learning_rate": 4.141583620672147e-05, + "loss": 4.128, + "step": 524288 + }, + { + "epoch": 0.02, + "learning_rate": 4.140745025921095e-05, + "loss": 4.1433, + "step": 524800 + }, + { + "epoch": 0.02, + "learning_rate": 4.139906431170043e-05, + "loss": 4.1446, + "step": 525312 + }, + { + "epoch": 0.02, + "learning_rate": 4.139067836418991e-05, + "loss": 4.1356, + "step": 525824 + }, + { + "epoch": 0.02, + "learning_rate": 4.138229241667939e-05, + "loss": 4.1392, + "step": 526336 + }, + { + "epoch": 0.02, + "learning_rate": 4.137390646916887e-05, + "loss": 4.1351, + "step": 526848 + }, + { + "epoch": 0.02, + "learning_rate": 4.136552052165835e-05, + "loss": 4.1333, + "step": 527360 + }, + { + "epoch": 0.02, + "learning_rate": 4.135713457414783e-05, + "loss": 4.1282, + "step": 527872 + }, + { + "epoch": 0.02, + "learning_rate": 4.1348765005441045e-05, + "loss": 4.1455, + "step": 528384 + }, + { + "epoch": 0.02, + "learning_rate": 4.1340379057930525e-05, + "loss": 4.1428, + "step": 528896 + }, + { + "epoch": 0.02, + "learning_rate": 4.1331993110420005e-05, + "loss": 4.1398, + "step": 529408 + }, + { + "epoch": 0.02, + "learning_rate": 4.132360716290948e-05, + "loss": 4.1379, + "step": 529920 + }, + { + "epoch": 0.02, + "learning_rate": 4.1315237594202694e-05, + "loss": 4.1309, + "step": 530432 + }, + { + "epoch": 0.02, + "learning_rate": 4.1306851646692174e-05, + "loss": 4.1342, + "step": 530944 + }, + { + "epoch": 0.02, + "learning_rate": 4.1298482077985383e-05, + "loss": 4.1293, + "step": 531456 + }, + { + "epoch": 0.02, + "learning_rate": 4.129009613047486e-05, + "loss": 4.1221, + "step": 531968 + }, + { + "epoch": 0.02, + "learning_rate": 4.128171018296434e-05, + "loss": 4.1384, + "step": 532480 + }, + { + "epoch": 0.02, + "learning_rate": 4.127332423545382e-05, + "loss": 4.1275, + "step": 532992 + }, + { + "epoch": 0.02, + "learning_rate": 4.12649382879433e-05, + "loss": 4.1203, + "step": 533504 + }, + { + "epoch": 0.02, + "learning_rate": 4.125655234043278e-05, + "loss": 4.1333, + "step": 534016 + }, + { + "epoch": 0.03, + "eval_loss": 4.136108875274658, + "eval_runtime": 553.0927, + "eval_samples_per_second": 689.922, + "eval_steps_per_second": 21.561, + "step": 534233 + }, + { + "epoch": 1.0, + "learning_rate": 4.1248182771726e-05, + "loss": 4.1349, + "step": 534528 + }, + { + "epoch": 1.0, + "learning_rate": 4.123979682421548e-05, + "loss": 4.1327, + "step": 535040 + }, + { + "epoch": 1.0, + "learning_rate": 4.123141087670495e-05, + "loss": 4.1384, + "step": 535552 + }, + { + "epoch": 1.0, + "learning_rate": 4.122302492919443e-05, + "loss": 4.1294, + "step": 536064 + }, + { + "epoch": 1.0, + "learning_rate": 4.121465536048765e-05, + "loss": 4.1278, + "step": 536576 + }, + { + "epoch": 1.0, + "learning_rate": 4.120626941297712e-05, + "loss": 4.125, + "step": 537088 + }, + { + "epoch": 1.0, + "learning_rate": 4.11978834654666e-05, + "loss": 4.1284, + "step": 537600 + }, + { + "epoch": 1.0, + "learning_rate": 4.118949751795608e-05, + "loss": 4.1247, + "step": 538112 + }, + { + "epoch": 1.0, + "learning_rate": 4.118111157044556e-05, + "loss": 4.1257, + "step": 538624 + }, + { + "epoch": 1.0, + "learning_rate": 4.117274200173877e-05, + "loss": 4.1272, + "step": 539136 + }, + { + "epoch": 1.0, + "learning_rate": 4.1164372433031986e-05, + "loss": 4.1409, + "step": 539648 + }, + { + "epoch": 1.0, + "learning_rate": 4.115598648552147e-05, + "loss": 4.1206, + "step": 540160 + }, + { + "epoch": 1.0, + "learning_rate": 4.114760053801095e-05, + "loss": 4.1327, + "step": 540672 + }, + { + "epoch": 1.0, + "learning_rate": 4.1139214590500426e-05, + "loss": 4.1169, + "step": 541184 + }, + { + "epoch": 1.0, + "learning_rate": 4.1130828642989906e-05, + "loss": 4.136, + "step": 541696 + }, + { + "epoch": 1.0, + "learning_rate": 4.1122442695479386e-05, + "loss": 4.1169, + "step": 542208 + }, + { + "epoch": 1.0, + "learning_rate": 4.1114056747968866e-05, + "loss": 4.1207, + "step": 542720 + }, + { + "epoch": 1.0, + "learning_rate": 4.1105670800458346e-05, + "loss": 4.1235, + "step": 543232 + }, + { + "epoch": 1.0, + "learning_rate": 4.1097284852947826e-05, + "loss": 4.1233, + "step": 543744 + }, + { + "epoch": 1.0, + "learning_rate": 4.1088898905437306e-05, + "loss": 4.1333, + "step": 544256 + }, + { + "epoch": 1.0, + "learning_rate": 4.1080512957926786e-05, + "loss": 4.1285, + "step": 544768 + }, + { + "epoch": 1.0, + "learning_rate": 4.1072143389219995e-05, + "loss": 4.14, + "step": 545280 + }, + { + "epoch": 1.0, + "learning_rate": 4.1063757441709475e-05, + "loss": 4.1285, + "step": 545792 + }, + { + "epoch": 1.0, + "learning_rate": 4.1055371494198955e-05, + "loss": 4.1276, + "step": 546304 + }, + { + "epoch": 1.0, + "learning_rate": 4.1046985546688435e-05, + "loss": 4.1252, + "step": 546816 + }, + { + "epoch": 1.0, + "learning_rate": 4.103859959917792e-05, + "loss": 4.1235, + "step": 547328 + }, + { + "epoch": 1.0, + "learning_rate": 4.10302136516674e-05, + "loss": 4.1294, + "step": 547840 + }, + { + "epoch": 1.0, + "learning_rate": 4.102182770415688e-05, + "loss": 4.1226, + "step": 548352 + }, + { + "epoch": 1.0, + "learning_rate": 4.101344175664636e-05, + "loss": 4.127, + "step": 548864 + }, + { + "epoch": 1.0, + "learning_rate": 4.100507218793957e-05, + "loss": 4.1227, + "step": 549376 + }, + { + "epoch": 1.01, + "learning_rate": 4.099668624042905e-05, + "loss": 4.1147, + "step": 549888 + }, + { + "epoch": 1.01, + "learning_rate": 4.098830029291853e-05, + "loss": 4.1281, + "step": 550400 + }, + { + "epoch": 1.01, + "learning_rate": 4.097991434540801e-05, + "loss": 4.1319, + "step": 550912 + }, + { + "epoch": 1.01, + "learning_rate": 4.097154477670122e-05, + "loss": 4.1269, + "step": 551424 + }, + { + "epoch": 1.01, + "learning_rate": 4.09631588291907e-05, + "loss": 4.1164, + "step": 551936 + }, + { + "epoch": 1.01, + "learning_rate": 4.095477288168018e-05, + "loss": 4.1291, + "step": 552448 + }, + { + "epoch": 1.01, + "learning_rate": 4.094640331297339e-05, + "loss": 4.1232, + "step": 552960 + }, + { + "epoch": 1.01, + "learning_rate": 4.0938017365462875e-05, + "loss": 4.1258, + "step": 553472 + }, + { + "epoch": 1.01, + "learning_rate": 4.0929631417952355e-05, + "loss": 4.1248, + "step": 553984 + }, + { + "epoch": 1.01, + "learning_rate": 4.0921245470441835e-05, + "loss": 4.1173, + "step": 554496 + }, + { + "epoch": 1.01, + "learning_rate": 4.091285952293131e-05, + "loss": 4.1104, + "step": 555008 + }, + { + "epoch": 1.01, + "learning_rate": 4.090447357542079e-05, + "loss": 4.1091, + "step": 555520 + }, + { + "epoch": 1.01, + "learning_rate": 4.089608762791027e-05, + "loss": 4.1276, + "step": 556032 + }, + { + "epoch": 1.01, + "learning_rate": 4.088770168039975e-05, + "loss": 4.1187, + "step": 556544 + }, + { + "epoch": 1.01, + "learning_rate": 4.087931573288923e-05, + "loss": 4.1121, + "step": 557056 + }, + { + "epoch": 1.01, + "learning_rate": 4.087094616418244e-05, + "loss": 4.124, + "step": 557568 + }, + { + "epoch": 1.01, + "learning_rate": 4.086256021667192e-05, + "loss": 4.1143, + "step": 558080 + }, + { + "epoch": 1.01, + "learning_rate": 4.08541742691614e-05, + "loss": 4.1282, + "step": 558592 + }, + { + "epoch": 1.01, + "learning_rate": 4.084578832165088e-05, + "loss": 4.113, + "step": 559104 + }, + { + "epoch": 1.01, + "learning_rate": 4.083741875294409e-05, + "loss": 4.1201, + "step": 559616 + }, + { + "epoch": 1.01, + "learning_rate": 4.082903280543357e-05, + "loss": 4.115, + "step": 560128 + }, + { + "epoch": 1.01, + "learning_rate": 4.082064685792305e-05, + "loss": 4.123, + "step": 560640 + }, + { + "epoch": 1.01, + "learning_rate": 4.081226091041253e-05, + "loss": 4.1137, + "step": 561152 + }, + { + "epoch": 1.01, + "learning_rate": 4.080389134170574e-05, + "loss": 4.1092, + "step": 561664 + }, + { + "epoch": 1.01, + "learning_rate": 4.079550539419522e-05, + "loss": 4.1171, + "step": 562176 + }, + { + "epoch": 1.01, + "learning_rate": 4.07871194466847e-05, + "loss": 4.1169, + "step": 562688 + }, + { + "epoch": 1.01, + "learning_rate": 4.077873349917418e-05, + "loss": 4.1153, + "step": 563200 + }, + { + "epoch": 1.01, + "learning_rate": 4.077036393046739e-05, + "loss": 4.1124, + "step": 563712 + }, + { + "epoch": 1.01, + "learning_rate": 4.076197798295687e-05, + "loss": 4.1244, + "step": 564224 + }, + { + "epoch": 1.01, + "learning_rate": 4.075359203544635e-05, + "loss": 4.1018, + "step": 564736 + }, + { + "epoch": 1.01, + "learning_rate": 4.074520608793583e-05, + "loss": 4.1107, + "step": 565248 + }, + { + "epoch": 1.01, + "learning_rate": 4.0736836519229046e-05, + "loss": 4.1065, + "step": 565760 + }, + { + "epoch": 1.01, + "learning_rate": 4.0728450571718526e-05, + "loss": 4.1129, + "step": 566272 + }, + { + "epoch": 1.01, + "learning_rate": 4.0720064624208006e-05, + "loss": 4.1244, + "step": 566784 + }, + { + "epoch": 1.01, + "learning_rate": 4.0711678676697486e-05, + "loss": 4.1073, + "step": 567296 + }, + { + "epoch": 1.01, + "learning_rate": 4.0703309107990695e-05, + "loss": 4.1138, + "step": 567808 + }, + { + "epoch": 1.01, + "learning_rate": 4.0694923160480175e-05, + "loss": 4.0993, + "step": 568320 + }, + { + "epoch": 1.01, + "learning_rate": 4.0686537212969655e-05, + "loss": 4.1149, + "step": 568832 + }, + { + "epoch": 1.01, + "learning_rate": 4.0678151265459135e-05, + "loss": 4.1039, + "step": 569344 + }, + { + "epoch": 1.01, + "learning_rate": 4.0669781696752344e-05, + "loss": 4.1116, + "step": 569856 + }, + { + "epoch": 1.01, + "learning_rate": 4.0661395749241824e-05, + "loss": 4.1157, + "step": 570368 + }, + { + "epoch": 1.01, + "learning_rate": 4.0653009801731304e-05, + "loss": 4.1093, + "step": 570880 + }, + { + "epoch": 1.01, + "learning_rate": 4.0644623854220784e-05, + "loss": 4.1124, + "step": 571392 + }, + { + "epoch": 1.01, + "learning_rate": 4.0636254285514e-05, + "loss": 4.123, + "step": 571904 + }, + { + "epoch": 1.01, + "learning_rate": 4.062788471680721e-05, + "loss": 4.1149, + "step": 572416 + }, + { + "epoch": 1.01, + "learning_rate": 4.061949876929669e-05, + "loss": 4.1203, + "step": 572928 + }, + { + "epoch": 1.01, + "learning_rate": 4.061111282178617e-05, + "loss": 4.1145, + "step": 573440 + }, + { + "epoch": 1.01, + "learning_rate": 4.060272687427565e-05, + "loss": 4.1008, + "step": 573952 + }, + { + "epoch": 1.01, + "learning_rate": 4.059434092676513e-05, + "loss": 4.1097, + "step": 574464 + }, + { + "epoch": 1.01, + "learning_rate": 4.058595497925461e-05, + "loss": 4.1163, + "step": 574976 + }, + { + "epoch": 1.01, + "learning_rate": 4.057756903174409e-05, + "loss": 4.117, + "step": 575488 + }, + { + "epoch": 1.01, + "learning_rate": 4.056918308423357e-05, + "loss": 4.1064, + "step": 576000 + }, + { + "epoch": 1.01, + "learning_rate": 4.056079713672305e-05, + "loss": 4.0955, + "step": 576512 + }, + { + "epoch": 1.01, + "learning_rate": 4.055242756801626e-05, + "loss": 4.1043, + "step": 577024 + }, + { + "epoch": 1.01, + "learning_rate": 4.054404162050574e-05, + "loss": 4.1008, + "step": 577536 + }, + { + "epoch": 1.01, + "learning_rate": 4.053565567299522e-05, + "loss": 4.1136, + "step": 578048 + }, + { + "epoch": 1.01, + "learning_rate": 4.0527286104288434e-05, + "loss": 4.1093, + "step": 578560 + }, + { + "epoch": 1.01, + "learning_rate": 4.0518900156777914e-05, + "loss": 4.1176, + "step": 579072 + }, + { + "epoch": 1.01, + "learning_rate": 4.0510514209267394e-05, + "loss": 4.1109, + "step": 579584 + }, + { + "epoch": 1.02, + "learning_rate": 4.0502128261756874e-05, + "loss": 4.1046, + "step": 580096 + }, + { + "epoch": 1.02, + "learning_rate": 4.0493742314246353e-05, + "loss": 4.1004, + "step": 580608 + }, + { + "epoch": 1.02, + "learning_rate": 4.048537274553956e-05, + "loss": 4.1073, + "step": 581120 + }, + { + "epoch": 1.02, + "learning_rate": 4.047700317683277e-05, + "loss": 4.101, + "step": 581632 + }, + { + "epoch": 1.02, + "learning_rate": 4.046861722932225e-05, + "loss": 4.1048, + "step": 582144 + }, + { + "epoch": 1.02, + "learning_rate": 4.046023128181173e-05, + "loss": 4.0996, + "step": 582656 + }, + { + "epoch": 1.02, + "learning_rate": 4.045184533430121e-05, + "loss": 4.11, + "step": 583168 + }, + { + "epoch": 1.02, + "learning_rate": 4.044345938679069e-05, + "loss": 4.1037, + "step": 583680 + }, + { + "epoch": 1.02, + "learning_rate": 4.043507343928017e-05, + "loss": 4.1267, + "step": 584192 + }, + { + "epoch": 1.02, + "learning_rate": 4.042668749176965e-05, + "loss": 4.0977, + "step": 584704 + }, + { + "epoch": 1.02, + "learning_rate": 4.041830154425914e-05, + "loss": 4.1146, + "step": 585216 + }, + { + "epoch": 1.02, + "learning_rate": 4.040993197555235e-05, + "loss": 4.1094, + "step": 585728 + }, + { + "epoch": 1.02, + "learning_rate": 4.040154602804183e-05, + "loss": 4.1023, + "step": 586240 + }, + { + "epoch": 1.02, + "learning_rate": 4.039316008053131e-05, + "loss": 4.1129, + "step": 586752 + }, + { + "epoch": 1.02, + "learning_rate": 4.0384790511824516e-05, + "loss": 4.1087, + "step": 587264 + }, + { + "epoch": 1.02, + "learning_rate": 4.0376404564313996e-05, + "loss": 4.115, + "step": 587776 + }, + { + "epoch": 1.02, + "learning_rate": 4.0368018616803476e-05, + "loss": 4.1029, + "step": 588288 + }, + { + "epoch": 1.02, + "learning_rate": 4.0359632669292956e-05, + "loss": 4.0911, + "step": 588800 + }, + { + "epoch": 1.02, + "learning_rate": 4.0351246721782436e-05, + "loss": 4.0962, + "step": 589312 + }, + { + "epoch": 1.02, + "learning_rate": 4.034286077427191e-05, + "loss": 4.107, + "step": 589824 + }, + { + "epoch": 1.02, + "learning_rate": 4.0334491205565125e-05, + "loss": 4.1068, + "step": 590336 + }, + { + "epoch": 1.02, + "learning_rate": 4.0326105258054605e-05, + "loss": 4.0974, + "step": 590848 + }, + { + "epoch": 1.02, + "learning_rate": 4.031771931054409e-05, + "loss": 4.0816, + "step": 591360 + }, + { + "epoch": 1.02, + "learning_rate": 4.0309333363033565e-05, + "loss": 4.0985, + "step": 591872 + }, + { + "epoch": 1.02, + "learning_rate": 4.030096379432678e-05, + "loss": 4.113, + "step": 592384 + }, + { + "epoch": 1.02, + "learning_rate": 4.029257784681626e-05, + "loss": 4.1052, + "step": 592896 + }, + { + "epoch": 1.02, + "learning_rate": 4.028419189930574e-05, + "loss": 4.114, + "step": 593408 + }, + { + "epoch": 1.02, + "learning_rate": 4.0275805951795214e-05, + "loss": 4.1063, + "step": 593920 + }, + { + "epoch": 1.02, + "learning_rate": 4.0267420004284694e-05, + "loss": 4.1027, + "step": 594432 + }, + { + "epoch": 1.02, + "learning_rate": 4.0259034056774174e-05, + "loss": 4.1078, + "step": 594944 + }, + { + "epoch": 1.02, + "learning_rate": 4.0250648109263654e-05, + "loss": 4.103, + "step": 595456 + }, + { + "epoch": 1.02, + "learning_rate": 4.0242262161753134e-05, + "loss": 4.0956, + "step": 595968 + }, + { + "epoch": 1.02, + "learning_rate": 4.0233876214242614e-05, + "loss": 4.1013, + "step": 596480 + }, + { + "epoch": 1.02, + "learning_rate": 4.0225490266732094e-05, + "loss": 4.0998, + "step": 596992 + }, + { + "epoch": 1.02, + "learning_rate": 4.0217104319221574e-05, + "loss": 4.1005, + "step": 597504 + }, + { + "epoch": 1.02, + "learning_rate": 4.020873475051479e-05, + "loss": 4.0978, + "step": 598016 + }, + { + "epoch": 1.02, + "learning_rate": 4.020034880300427e-05, + "loss": 4.108, + "step": 598528 + }, + { + "epoch": 1.02, + "learning_rate": 4.019196285549375e-05, + "loss": 4.0988, + "step": 599040 + }, + { + "epoch": 1.02, + "learning_rate": 4.018357690798323e-05, + "loss": 4.0927, + "step": 599552 + }, + { + "epoch": 1.02, + "learning_rate": 4.017520733927644e-05, + "loss": 4.1114, + "step": 600064 + }, + { + "epoch": 1.02, + "learning_rate": 4.016682139176592e-05, + "loss": 4.0922, + "step": 600576 + }, + { + "epoch": 1.02, + "learning_rate": 4.01584354442554e-05, + "loss": 4.1088, + "step": 601088 + }, + { + "epoch": 1.02, + "learning_rate": 4.015004949674488e-05, + "loss": 4.1144, + "step": 601600 + }, + { + "epoch": 1.02, + "learning_rate": 4.014166354923436e-05, + "loss": 4.1048, + "step": 602112 + }, + { + "epoch": 1.02, + "learning_rate": 4.013327760172384e-05, + "loss": 4.1022, + "step": 602624 + }, + { + "epoch": 1.02, + "learning_rate": 4.012490803301705e-05, + "loss": 4.1003, + "step": 603136 + }, + { + "epoch": 1.02, + "learning_rate": 4.011652208550653e-05, + "loss": 4.1003, + "step": 603648 + }, + { + "epoch": 1.02, + "learning_rate": 4.0108136137996014e-05, + "loss": 4.0982, + "step": 604160 + }, + { + "epoch": 1.02, + "learning_rate": 4.0099750190485494e-05, + "loss": 4.1146, + "step": 604672 + }, + { + "epoch": 1.02, + "learning_rate": 4.0091364242974974e-05, + "loss": 4.1079, + "step": 605184 + }, + { + "epoch": 1.02, + "learning_rate": 4.0082978295464454e-05, + "loss": 4.109, + "step": 605696 + }, + { + "epoch": 1.02, + "learning_rate": 4.007459234795393e-05, + "loss": 4.1023, + "step": 606208 + }, + { + "epoch": 1.02, + "learning_rate": 4.006620640044341e-05, + "loss": 4.0992, + "step": 606720 + }, + { + "epoch": 1.02, + "learning_rate": 4.005783683173662e-05, + "loss": 4.104, + "step": 607232 + }, + { + "epoch": 1.02, + "learning_rate": 4.00494508842261e-05, + "loss": 4.0952, + "step": 607744 + }, + { + "epoch": 1.02, + "learning_rate": 4.0041064936715576e-05, + "loss": 4.0951, + "step": 608256 + }, + { + "epoch": 1.02, + "learning_rate": 4.0032678989205056e-05, + "loss": 4.1048, + "step": 608768 + }, + { + "epoch": 1.02, + "learning_rate": 4.0024293041694536e-05, + "loss": 4.0977, + "step": 609280 + }, + { + "epoch": 1.02, + "learning_rate": 4.001592347298775e-05, + "loss": 4.0849, + "step": 609792 + }, + { + "epoch": 1.02, + "learning_rate": 4.000755390428097e-05, + "loss": 4.1034, + "step": 610304 + }, + { + "epoch": 1.03, + "eval_loss": 4.111530303955078, + "eval_runtime": 550.7994, + "eval_samples_per_second": 692.795, + "eval_steps_per_second": 21.65, + "step": 610552 + }, + { + "epoch": 0.0, + "learning_rate": 3.999916795677045e-05, + "loss": 4.1077, + "step": 610816 + }, + { + "epoch": 0.0, + "learning_rate": 3.999078200925993e-05, + "loss": 4.1081, + "step": 611328 + }, + { + "epoch": 0.0, + "learning_rate": 3.99823960617494e-05, + "loss": 4.105, + "step": 611840 + }, + { + "epoch": 0.0, + "learning_rate": 3.997401011423888e-05, + "loss": 4.0971, + "step": 612352 + }, + { + "epoch": 0.0, + "learning_rate": 3.996562416672836e-05, + "loss": 4.0965, + "step": 612864 + }, + { + "epoch": 0.0, + "learning_rate": 3.995723821921784e-05, + "loss": 4.0931, + "step": 613376 + }, + { + "epoch": 0.0, + "learning_rate": 3.994885227170732e-05, + "loss": 4.0959, + "step": 613888 + }, + { + "epoch": 0.0, + "learning_rate": 3.99404663241968e-05, + "loss": 4.0906, + "step": 614400 + }, + { + "epoch": 0.0, + "learning_rate": 3.993208037668628e-05, + "loss": 4.0925, + "step": 614912 + }, + { + "epoch": 0.0, + "learning_rate": 3.992369442917576e-05, + "loss": 4.0989, + "step": 615424 + }, + { + "epoch": 0.0, + "learning_rate": 3.991530848166524e-05, + "loss": 4.107, + "step": 615936 + }, + { + "epoch": 0.0, + "learning_rate": 3.990692253415472e-05, + "loss": 4.0898, + "step": 616448 + }, + { + "epoch": 0.0, + "learning_rate": 3.98985365866442e-05, + "loss": 4.102, + "step": 616960 + }, + { + "epoch": 0.0, + "learning_rate": 3.989015063913368e-05, + "loss": 4.0848, + "step": 617472 + }, + { + "epoch": 0.0, + "learning_rate": 3.988176469162316e-05, + "loss": 4.1082, + "step": 617984 + }, + { + "epoch": 0.0, + "learning_rate": 3.987337874411264e-05, + "loss": 4.0835, + "step": 618496 + }, + { + "epoch": 0.0, + "learning_rate": 3.986499279660212e-05, + "loss": 4.0911, + "step": 619008 + }, + { + "epoch": 0.0, + "learning_rate": 3.98566068490916e-05, + "loss": 4.0933, + "step": 619520 + }, + { + "epoch": 0.0, + "learning_rate": 3.984822090158108e-05, + "loss": 4.094, + "step": 620032 + }, + { + "epoch": 0.0, + "learning_rate": 3.983983495407056e-05, + "loss": 4.0996, + "step": 620544 + }, + { + "epoch": 0.0, + "learning_rate": 3.983146538536377e-05, + "loss": 4.0949, + "step": 621056 + }, + { + "epoch": 0.0, + "learning_rate": 3.982307943785325e-05, + "loss": 4.1102, + "step": 621568 + }, + { + "epoch": 0.0, + "learning_rate": 3.981469349034273e-05, + "loss": 4.0984, + "step": 622080 + }, + { + "epoch": 0.0, + "learning_rate": 3.980630754283221e-05, + "loss": 4.0986, + "step": 622592 + }, + { + "epoch": 0.0, + "learning_rate": 3.979793797412542e-05, + "loss": 4.0909, + "step": 623104 + }, + { + "epoch": 0.0, + "learning_rate": 3.97895520266149e-05, + "loss": 4.0979, + "step": 623616 + }, + { + "epoch": 0.0, + "learning_rate": 3.9781166079104385e-05, + "loss": 4.0966, + "step": 624128 + }, + { + "epoch": 0.0, + "learning_rate": 3.9772780131593865e-05, + "loss": 4.0918, + "step": 624640 + }, + { + "epoch": 0.0, + "learning_rate": 3.9764410562887074e-05, + "loss": 4.0933, + "step": 625152 + }, + { + "epoch": 0.0, + "learning_rate": 3.9756024615376554e-05, + "loss": 4.0932, + "step": 625664 + }, + { + "epoch": 0.01, + "learning_rate": 3.974765504666976e-05, + "loss": 4.0813, + "step": 626176 + }, + { + "epoch": 0.01, + "learning_rate": 3.973926909915924e-05, + "loss": 4.0993, + "step": 626688 + }, + { + "epoch": 0.01, + "learning_rate": 3.973088315164872e-05, + "loss": 4.1011, + "step": 627200 + }, + { + "epoch": 0.01, + "learning_rate": 3.97224972041382e-05, + "loss": 4.0949, + "step": 627712 + }, + { + "epoch": 0.01, + "learning_rate": 3.971411125662768e-05, + "loss": 4.0853, + "step": 628224 + }, + { + "epoch": 0.01, + "learning_rate": 3.970572530911716e-05, + "loss": 4.1002, + "step": 628736 + }, + { + "epoch": 0.01, + "learning_rate": 3.969735574041037e-05, + "loss": 4.0909, + "step": 629248 + }, + { + "epoch": 0.01, + "learning_rate": 3.968896979289985e-05, + "loss": 4.093, + "step": 629760 + }, + { + "epoch": 0.01, + "learning_rate": 3.968058384538934e-05, + "loss": 4.0961, + "step": 630272 + }, + { + "epoch": 0.01, + "learning_rate": 3.967219789787882e-05, + "loss": 4.085, + "step": 630784 + }, + { + "epoch": 0.01, + "learning_rate": 3.966382832917203e-05, + "loss": 4.0837, + "step": 631296 + }, + { + "epoch": 0.01, + "learning_rate": 3.965544238166151e-05, + "loss": 4.0766, + "step": 631808 + }, + { + "epoch": 0.01, + "learning_rate": 3.964705643415099e-05, + "loss": 4.0996, + "step": 632320 + }, + { + "epoch": 0.01, + "learning_rate": 3.963867048664047e-05, + "loss": 4.0885, + "step": 632832 + }, + { + "epoch": 0.01, + "learning_rate": 3.963028453912995e-05, + "loss": 4.0815, + "step": 633344 + }, + { + "epoch": 0.01, + "learning_rate": 3.962189859161943e-05, + "loss": 4.0966, + "step": 633856 + }, + { + "epoch": 0.01, + "learning_rate": 3.9613529022912636e-05, + "loss": 4.0791, + "step": 634368 + }, + { + "epoch": 0.01, + "learning_rate": 3.9605143075402116e-05, + "loss": 4.1, + "step": 634880 + }, + { + "epoch": 0.01, + "learning_rate": 3.9596757127891596e-05, + "loss": 4.0851, + "step": 635392 + }, + { + "epoch": 0.01, + "learning_rate": 3.9588371180381076e-05, + "loss": 4.0908, + "step": 635904 + }, + { + "epoch": 0.01, + "learning_rate": 3.9579985232870556e-05, + "loss": 4.0863, + "step": 636416 + }, + { + "epoch": 0.01, + "learning_rate": 3.9571599285360036e-05, + "loss": 4.0882, + "step": 636928 + }, + { + "epoch": 0.01, + "learning_rate": 3.956321333784952e-05, + "loss": 4.0906, + "step": 637440 + }, + { + "epoch": 0.01, + "learning_rate": 3.9554827390339e-05, + "loss": 4.0753, + "step": 637952 + }, + { + "epoch": 0.01, + "learning_rate": 3.954645782163221e-05, + "loss": 4.0888, + "step": 638464 + }, + { + "epoch": 0.01, + "learning_rate": 3.953807187412169e-05, + "loss": 4.0849, + "step": 638976 + }, + { + "epoch": 0.01, + "learning_rate": 3.952968592661117e-05, + "loss": 4.0858, + "step": 639488 + }, + { + "epoch": 0.01, + "learning_rate": 3.952129997910065e-05, + "loss": 4.0838, + "step": 640000 + }, + { + "epoch": 0.01, + "learning_rate": 3.951293041039386e-05, + "loss": 4.097, + "step": 640512 + }, + { + "epoch": 0.01, + "learning_rate": 3.950454446288334e-05, + "loss": 4.0698, + "step": 641024 + }, + { + "epoch": 0.01, + "learning_rate": 3.949615851537282e-05, + "loss": 4.0828, + "step": 641536 + }, + { + "epoch": 0.01, + "learning_rate": 3.9487772567862294e-05, + "loss": 4.0794, + "step": 642048 + }, + { + "epoch": 0.01, + "learning_rate": 3.947940299915551e-05, + "loss": 4.0829, + "step": 642560 + }, + { + "epoch": 0.01, + "learning_rate": 3.947101705164499e-05, + "loss": 4.0913, + "step": 643072 + }, + { + "epoch": 0.01, + "learning_rate": 3.946263110413447e-05, + "loss": 4.0812, + "step": 643584 + }, + { + "epoch": 0.01, + "learning_rate": 3.945424515662395e-05, + "loss": 4.0895, + "step": 644096 + }, + { + "epoch": 0.01, + "learning_rate": 3.9445891966720895e-05, + "loss": 4.0635, + "step": 644608 + }, + { + "epoch": 0.01, + "learning_rate": 3.9437506019210375e-05, + "loss": 4.0877, + "step": 645120 + }, + { + "epoch": 0.01, + "learning_rate": 3.9429120071699855e-05, + "loss": 4.0733, + "step": 645632 + }, + { + "epoch": 0.01, + "learning_rate": 3.9420734124189335e-05, + "loss": 4.0798, + "step": 646144 + }, + { + "epoch": 0.01, + "learning_rate": 3.9412348176678815e-05, + "loss": 4.0838, + "step": 646656 + }, + { + "epoch": 0.01, + "learning_rate": 3.9403962229168295e-05, + "loss": 4.0843, + "step": 647168 + }, + { + "epoch": 0.01, + "learning_rate": 3.939557628165777e-05, + "loss": 4.0827, + "step": 647680 + }, + { + "epoch": 0.01, + "learning_rate": 3.938719033414725e-05, + "loss": 4.098, + "step": 648192 + }, + { + "epoch": 0.01, + "learning_rate": 3.9378820765440464e-05, + "loss": 4.0833, + "step": 648704 + }, + { + "epoch": 0.01, + "learning_rate": 3.9370434817929943e-05, + "loss": 4.0911, + "step": 649216 + }, + { + "epoch": 0.01, + "learning_rate": 3.9362048870419423e-05, + "loss": 4.0884, + "step": 649728 + }, + { + "epoch": 0.01, + "learning_rate": 3.9353662922908903e-05, + "loss": 4.0738, + "step": 650240 + }, + { + "epoch": 0.01, + "learning_rate": 3.934527697539838e-05, + "loss": 4.0751, + "step": 650752 + }, + { + "epoch": 0.01, + "learning_rate": 3.933689102788786e-05, + "loss": 4.0864, + "step": 651264 + }, + { + "epoch": 0.01, + "learning_rate": 3.932850508037734e-05, + "loss": 4.0919, + "step": 651776 + }, + { + "epoch": 0.01, + "learning_rate": 3.932011913286682e-05, + "loss": 4.0852, + "step": 652288 + }, + { + "epoch": 0.01, + "learning_rate": 3.93117331853563e-05, + "loss": 4.0626, + "step": 652800 + }, + { + "epoch": 0.01, + "learning_rate": 3.930334723784578e-05, + "loss": 4.0739, + "step": 653312 + }, + { + "epoch": 0.01, + "learning_rate": 3.929499404794272e-05, + "loss": 4.0754, + "step": 653824 + }, + { + "epoch": 0.01, + "learning_rate": 3.92866081004322e-05, + "loss": 4.0826, + "step": 654336 + }, + { + "epoch": 0.01, + "learning_rate": 3.927822215292168e-05, + "loss": 4.0834, + "step": 654848 + }, + { + "epoch": 0.01, + "learning_rate": 3.926983620541116e-05, + "loss": 4.0881, + "step": 655360 + }, + { + "epoch": 0.01, + "learning_rate": 3.926145025790065e-05, + "loss": 4.0796, + "step": 655872 + }, + { + "epoch": 0.02, + "learning_rate": 3.925306431039013e-05, + "loss": 4.0778, + "step": 656384 + }, + { + "epoch": 0.02, + "learning_rate": 3.924467836287961e-05, + "loss": 4.0758, + "step": 656896 + }, + { + "epoch": 0.02, + "learning_rate": 3.923629241536909e-05, + "loss": 4.0766, + "step": 657408 + }, + { + "epoch": 0.02, + "learning_rate": 3.9227939225466026e-05, + "loss": 4.067, + "step": 657920 + }, + { + "epoch": 0.02, + "learning_rate": 3.9219553277955506e-05, + "loss": 4.0821, + "step": 658432 + }, + { + "epoch": 0.02, + "learning_rate": 3.9211167330444986e-05, + "loss": 4.0725, + "step": 658944 + }, + { + "epoch": 0.02, + "learning_rate": 3.9202781382934466e-05, + "loss": 4.079, + "step": 659456 + }, + { + "epoch": 0.02, + "learning_rate": 3.9194395435423946e-05, + "loss": 4.0753, + "step": 659968 + }, + { + "epoch": 0.02, + "learning_rate": 3.9186009487913426e-05, + "loss": 4.095, + "step": 660480 + }, + { + "epoch": 0.02, + "learning_rate": 3.9177623540402906e-05, + "loss": 4.0732, + "step": 660992 + }, + { + "epoch": 0.02, + "learning_rate": 3.9169237592892386e-05, + "loss": 4.0869, + "step": 661504 + }, + { + "epoch": 0.02, + "learning_rate": 3.91608680241856e-05, + "loss": 4.0858, + "step": 662016 + }, + { + "epoch": 0.02, + "learning_rate": 3.915249845547881e-05, + "loss": 4.0766, + "step": 662528 + }, + { + "epoch": 0.02, + "learning_rate": 3.914411250796829e-05, + "loss": 4.0821, + "step": 663040 + }, + { + "epoch": 0.02, + "learning_rate": 3.913572656045777e-05, + "loss": 4.0819, + "step": 663552 + }, + { + "epoch": 0.02, + "learning_rate": 3.912734061294725e-05, + "loss": 4.0866, + "step": 664064 + }, + { + "epoch": 0.02, + "learning_rate": 3.911895466543673e-05, + "loss": 4.0739, + "step": 664576 + }, + { + "epoch": 0.02, + "learning_rate": 3.911056871792621e-05, + "loss": 4.0656, + "step": 665088 + }, + { + "epoch": 0.02, + "learning_rate": 3.910218277041569e-05, + "loss": 4.0694, + "step": 665600 + }, + { + "epoch": 0.02, + "learning_rate": 3.909379682290517e-05, + "loss": 4.0774, + "step": 666112 + }, + { + "epoch": 0.02, + "learning_rate": 3.908542725419838e-05, + "loss": 4.0763, + "step": 666624 + }, + { + "epoch": 0.02, + "learning_rate": 3.907704130668786e-05, + "loss": 4.0752, + "step": 667136 + }, + { + "epoch": 0.02, + "learning_rate": 3.906865535917734e-05, + "loss": 4.052, + "step": 667648 + }, + { + "epoch": 0.02, + "learning_rate": 3.906026941166682e-05, + "loss": 4.0745, + "step": 668160 + }, + { + "epoch": 0.02, + "learning_rate": 3.9051899842960035e-05, + "loss": 4.0808, + "step": 668672 + }, + { + "epoch": 0.02, + "learning_rate": 3.9043513895449515e-05, + "loss": 4.0771, + "step": 669184 + }, + { + "epoch": 0.02, + "learning_rate": 3.9035144326742724e-05, + "loss": 4.0858, + "step": 669696 + }, + { + "epoch": 0.02, + "learning_rate": 3.9026758379232204e-05, + "loss": 4.0814, + "step": 670208 + }, + { + "epoch": 0.02, + "learning_rate": 3.9018372431721684e-05, + "loss": 4.0718, + "step": 670720 + }, + { + "epoch": 0.02, + "learning_rate": 3.9009986484211164e-05, + "loss": 4.0806, + "step": 671232 + }, + { + "epoch": 0.02, + "learning_rate": 3.9001600536700644e-05, + "loss": 4.0786, + "step": 671744 + }, + { + "epoch": 0.02, + "learning_rate": 3.899323096799385e-05, + "loss": 4.0665, + "step": 672256 + }, + { + "epoch": 0.02, + "learning_rate": 3.898484502048333e-05, + "loss": 4.0763, + "step": 672768 + }, + { + "epoch": 0.02, + "learning_rate": 3.897645907297281e-05, + "loss": 4.0734, + "step": 673280 + }, + { + "epoch": 0.02, + "learning_rate": 3.896807312546229e-05, + "loss": 4.0742, + "step": 673792 + }, + { + "epoch": 0.02, + "learning_rate": 3.895968717795177e-05, + "loss": 4.0712, + "step": 674304 + }, + { + "epoch": 0.02, + "learning_rate": 3.895130123044125e-05, + "loss": 4.0778, + "step": 674816 + }, + { + "epoch": 0.02, + "learning_rate": 3.894291528293073e-05, + "loss": 4.0759, + "step": 675328 + }, + { + "epoch": 0.02, + "learning_rate": 3.893454571422395e-05, + "loss": 4.0641, + "step": 675840 + }, + { + "epoch": 0.02, + "learning_rate": 3.892615976671343e-05, + "loss": 4.0815, + "step": 676352 + }, + { + "epoch": 0.02, + "learning_rate": 3.891779019800664e-05, + "loss": 4.0668, + "step": 676864 + }, + { + "epoch": 0.02, + "learning_rate": 3.890940425049612e-05, + "loss": 4.0818, + "step": 677376 + }, + { + "epoch": 0.02, + "learning_rate": 3.89010183029856e-05, + "loss": 4.0877, + "step": 677888 + }, + { + "epoch": 0.02, + "learning_rate": 3.889263235547508e-05, + "loss": 4.0753, + "step": 678400 + }, + { + "epoch": 0.02, + "learning_rate": 3.888424640796455e-05, + "loss": 4.0772, + "step": 678912 + }, + { + "epoch": 0.02, + "learning_rate": 3.887586046045403e-05, + "loss": 4.0723, + "step": 679424 + }, + { + "epoch": 0.02, + "learning_rate": 3.886747451294351e-05, + "loss": 4.0741, + "step": 679936 + }, + { + "epoch": 0.02, + "learning_rate": 3.885908856543299e-05, + "loss": 4.0722, + "step": 680448 + }, + { + "epoch": 0.02, + "learning_rate": 3.885070261792248e-05, + "loss": 4.0852, + "step": 680960 + }, + { + "epoch": 0.02, + "learning_rate": 3.884231667041196e-05, + "loss": 4.0864, + "step": 681472 + }, + { + "epoch": 0.02, + "learning_rate": 3.883393072290144e-05, + "loss": 4.0778, + "step": 681984 + }, + { + "epoch": 0.02, + "learning_rate": 3.882554477539092e-05, + "loss": 4.0751, + "step": 682496 + }, + { + "epoch": 0.02, + "learning_rate": 3.88171588278804e-05, + "loss": 4.0782, + "step": 683008 + }, + { + "epoch": 0.02, + "learning_rate": 3.8808789259173606e-05, + "loss": 4.0731, + "step": 683520 + }, + { + "epoch": 0.02, + "learning_rate": 3.8800403311663086e-05, + "loss": 4.0706, + "step": 684032 + }, + { + "epoch": 0.02, + "learning_rate": 3.8792017364152566e-05, + "loss": 4.0675, + "step": 684544 + }, + { + "epoch": 0.02, + "learning_rate": 3.8783647795445775e-05, + "loss": 4.0765, + "step": 685056 + }, + { + "epoch": 0.02, + "learning_rate": 3.8775261847935255e-05, + "loss": 4.0729, + "step": 685568 + }, + { + "epoch": 0.02, + "learning_rate": 3.8766875900424735e-05, + "loss": 4.0609, + "step": 686080 + }, + { + "epoch": 0.02, + "learning_rate": 3.8758489952914215e-05, + "loss": 4.0735, + "step": 686592 + }, + { + "epoch": 0.03, + "eval_loss": 4.0929107666015625, + "eval_runtime": 530.2994, + "eval_samples_per_second": 719.577, + "eval_steps_per_second": 22.487, + "step": 686871 + }, + { + "epoch": 1.0, + "learning_rate": 3.8750104005403695e-05, + "loss": 4.0729, + "step": 687104 + }, + { + "epoch": 1.0, + "learning_rate": 3.874173443669691e-05, + "loss": 4.0814, + "step": 687616 + }, + { + "epoch": 1.0, + "learning_rate": 3.873334848918639e-05, + "loss": 4.0791, + "step": 688128 + }, + { + "epoch": 1.0, + "learning_rate": 3.872496254167587e-05, + "loss": 4.0662, + "step": 688640 + }, + { + "epoch": 1.0, + "learning_rate": 3.871657659416535e-05, + "loss": 4.0735, + "step": 689152 + }, + { + "epoch": 1.0, + "learning_rate": 3.870819064665483e-05, + "loss": 4.0688, + "step": 689664 + }, + { + "epoch": 1.0, + "learning_rate": 3.869980469914431e-05, + "loss": 4.0665, + "step": 690176 + }, + { + "epoch": 1.0, + "learning_rate": 3.869143513043752e-05, + "loss": 4.0667, + "step": 690688 + }, + { + "epoch": 1.0, + "learning_rate": 3.8683049182927e-05, + "loss": 4.0642, + "step": 691200 + }, + { + "epoch": 1.0, + "learning_rate": 3.867466323541648e-05, + "loss": 4.0749, + "step": 691712 + }, + { + "epoch": 1.0, + "learning_rate": 3.866627728790596e-05, + "loss": 4.0776, + "step": 692224 + }, + { + "epoch": 1.0, + "learning_rate": 3.865790771919917e-05, + "loss": 4.065, + "step": 692736 + }, + { + "epoch": 1.0, + "learning_rate": 3.864952177168865e-05, + "loss": 4.0764, + "step": 693248 + }, + { + "epoch": 1.0, + "learning_rate": 3.864113582417813e-05, + "loss": 4.0592, + "step": 693760 + }, + { + "epoch": 1.0, + "learning_rate": 3.863274987666761e-05, + "loss": 4.0767, + "step": 694272 + }, + { + "epoch": 1.0, + "learning_rate": 3.862436392915709e-05, + "loss": 4.0553, + "step": 694784 + }, + { + "epoch": 1.0, + "learning_rate": 3.861597798164657e-05, + "loss": 4.0668, + "step": 695296 + }, + { + "epoch": 1.0, + "learning_rate": 3.860759203413605e-05, + "loss": 4.0695, + "step": 695808 + }, + { + "epoch": 1.0, + "learning_rate": 3.859920608662553e-05, + "loss": 4.0668, + "step": 696320 + }, + { + "epoch": 1.0, + "learning_rate": 3.859083651791874e-05, + "loss": 4.0788, + "step": 696832 + }, + { + "epoch": 1.0, + "learning_rate": 3.858245057040822e-05, + "loss": 4.0628, + "step": 697344 + }, + { + "epoch": 1.0, + "learning_rate": 3.85740646228977e-05, + "loss": 4.0891, + "step": 697856 + }, + { + "epoch": 1.0, + "learning_rate": 3.856567867538718e-05, + "loss": 4.0689, + "step": 698368 + }, + { + "epoch": 1.0, + "learning_rate": 3.855729272787666e-05, + "loss": 4.0708, + "step": 698880 + }, + { + "epoch": 1.0, + "learning_rate": 3.854892315916987e-05, + "loss": 4.0683, + "step": 699392 + }, + { + "epoch": 1.0, + "learning_rate": 3.854053721165935e-05, + "loss": 4.073, + "step": 699904 + }, + { + "epoch": 1.0, + "learning_rate": 3.8532151264148833e-05, + "loss": 4.0699, + "step": 700416 + }, + { + "epoch": 1.0, + "learning_rate": 3.852376531663831e-05, + "loss": 4.0635, + "step": 700928 + }, + { + "epoch": 1.0, + "learning_rate": 3.851539574793152e-05, + "loss": 4.0682, + "step": 701440 + }, + { + "epoch": 1.0, + "learning_rate": 3.850702617922474e-05, + "loss": 4.0707, + "step": 701952 + }, + { + "epoch": 1.01, + "learning_rate": 3.849864023171421e-05, + "loss": 4.0497, + "step": 702464 + }, + { + "epoch": 1.01, + "learning_rate": 3.849025428420369e-05, + "loss": 4.0745, + "step": 702976 + }, + { + "epoch": 1.01, + "learning_rate": 3.848186833669317e-05, + "loss": 4.0749, + "step": 703488 + }, + { + "epoch": 1.01, + "learning_rate": 3.847348238918265e-05, + "loss": 4.0692, + "step": 704000 + }, + { + "epoch": 1.01, + "learning_rate": 3.846509644167213e-05, + "loss": 4.0594, + "step": 704512 + }, + { + "epoch": 1.01, + "learning_rate": 3.845672687296534e-05, + "loss": 4.0734, + "step": 705024 + }, + { + "epoch": 1.01, + "learning_rate": 3.844834092545482e-05, + "loss": 4.0666, + "step": 705536 + }, + { + "epoch": 1.01, + "learning_rate": 3.84399549779443e-05, + "loss": 4.0676, + "step": 706048 + }, + { + "epoch": 1.01, + "learning_rate": 3.843156903043379e-05, + "loss": 4.0703, + "step": 706560 + }, + { + "epoch": 1.01, + "learning_rate": 3.842318308292327e-05, + "loss": 4.0563, + "step": 707072 + }, + { + "epoch": 1.01, + "learning_rate": 3.841479713541275e-05, + "loss": 4.0597, + "step": 707584 + }, + { + "epoch": 1.01, + "learning_rate": 3.840641118790223e-05, + "loss": 4.0494, + "step": 708096 + }, + { + "epoch": 1.01, + "learning_rate": 3.839802524039171e-05, + "loss": 4.0711, + "step": 708608 + }, + { + "epoch": 1.01, + "learning_rate": 3.8389655671684916e-05, + "loss": 4.0626, + "step": 709120 + }, + { + "epoch": 1.01, + "learning_rate": 3.8381269724174396e-05, + "loss": 4.0558, + "step": 709632 + }, + { + "epoch": 1.01, + "learning_rate": 3.8372900155467605e-05, + "loss": 4.0697, + "step": 710144 + }, + { + "epoch": 1.01, + "learning_rate": 3.8364514207957085e-05, + "loss": 4.0544, + "step": 710656 + }, + { + "epoch": 1.01, + "learning_rate": 3.8356128260446565e-05, + "loss": 4.0715, + "step": 711168 + }, + { + "epoch": 1.01, + "learning_rate": 3.8347742312936045e-05, + "loss": 4.0658, + "step": 711680 + }, + { + "epoch": 1.01, + "learning_rate": 3.8339356365425525e-05, + "loss": 4.0635, + "step": 712192 + }, + { + "epoch": 1.01, + "learning_rate": 3.833098679671874e-05, + "loss": 4.0621, + "step": 712704 + }, + { + "epoch": 1.01, + "learning_rate": 3.832260084920822e-05, + "loss": 4.0584, + "step": 713216 + }, + { + "epoch": 1.01, + "learning_rate": 3.83142149016977e-05, + "loss": 4.067, + "step": 713728 + }, + { + "epoch": 1.01, + "learning_rate": 3.830582895418718e-05, + "loss": 4.0481, + "step": 714240 + }, + { + "epoch": 1.01, + "learning_rate": 3.829744300667666e-05, + "loss": 4.0659, + "step": 714752 + }, + { + "epoch": 1.01, + "learning_rate": 3.828907343796987e-05, + "loss": 4.0573, + "step": 715264 + }, + { + "epoch": 1.01, + "learning_rate": 3.828068749045935e-05, + "loss": 4.0615, + "step": 715776 + }, + { + "epoch": 1.01, + "learning_rate": 3.827230154294883e-05, + "loss": 4.0551, + "step": 716288 + }, + { + "epoch": 1.01, + "learning_rate": 3.826391559543831e-05, + "loss": 4.0714, + "step": 716800 + }, + { + "epoch": 1.01, + "learning_rate": 3.825552964792779e-05, + "loss": 4.052, + "step": 717312 + }, + { + "epoch": 1.01, + "learning_rate": 3.824714370041727e-05, + "loss": 4.0532, + "step": 717824 + }, + { + "epoch": 1.01, + "learning_rate": 3.823875775290675e-05, + "loss": 4.0593, + "step": 718336 + }, + { + "epoch": 1.01, + "learning_rate": 3.823037180539622e-05, + "loss": 4.0532, + "step": 718848 + }, + { + "epoch": 1.01, + "learning_rate": 3.822200223668944e-05, + "loss": 4.0623, + "step": 719360 + }, + { + "epoch": 1.01, + "learning_rate": 3.8213616289178925e-05, + "loss": 4.0602, + "step": 719872 + }, + { + "epoch": 1.01, + "learning_rate": 3.82052303416684e-05, + "loss": 4.0638, + "step": 720384 + }, + { + "epoch": 1.01, + "learning_rate": 3.819684439415788e-05, + "loss": 4.0398, + "step": 720896 + }, + { + "epoch": 1.01, + "learning_rate": 3.818845844664736e-05, + "loss": 4.0617, + "step": 721408 + }, + { + "epoch": 1.01, + "learning_rate": 3.8180088877940574e-05, + "loss": 4.0509, + "step": 721920 + }, + { + "epoch": 1.01, + "learning_rate": 3.817170293043005e-05, + "loss": 4.0528, + "step": 722432 + }, + { + "epoch": 1.01, + "learning_rate": 3.816331698291953e-05, + "loss": 4.0578, + "step": 722944 + }, + { + "epoch": 1.01, + "learning_rate": 3.815493103540901e-05, + "loss": 4.0575, + "step": 723456 + }, + { + "epoch": 1.01, + "learning_rate": 3.814654508789849e-05, + "loss": 4.0568, + "step": 723968 + }, + { + "epoch": 1.01, + "learning_rate": 3.813815914038797e-05, + "loss": 4.0734, + "step": 724480 + }, + { + "epoch": 1.01, + "learning_rate": 3.8129789571681176e-05, + "loss": 4.0599, + "step": 724992 + }, + { + "epoch": 1.01, + "learning_rate": 3.812140362417066e-05, + "loss": 4.0661, + "step": 725504 + }, + { + "epoch": 1.01, + "learning_rate": 3.811301767666014e-05, + "loss": 4.0621, + "step": 726016 + }, + { + "epoch": 1.01, + "learning_rate": 3.810463172914962e-05, + "loss": 4.0523, + "step": 726528 + }, + { + "epoch": 1.01, + "learning_rate": 3.80962457816391e-05, + "loss": 4.0512, + "step": 727040 + }, + { + "epoch": 1.01, + "learning_rate": 3.808785983412858e-05, + "loss": 4.0598, + "step": 727552 + }, + { + "epoch": 1.01, + "learning_rate": 3.807947388661806e-05, + "loss": 4.0627, + "step": 728064 + }, + { + "epoch": 1.01, + "learning_rate": 3.807108793910754e-05, + "loss": 4.0663, + "step": 728576 + }, + { + "epoch": 1.01, + "learning_rate": 3.806271837040075e-05, + "loss": 4.0378, + "step": 729088 + }, + { + "epoch": 1.01, + "learning_rate": 3.805433242289023e-05, + "loss": 4.0455, + "step": 729600 + }, + { + "epoch": 1.01, + "learning_rate": 3.804594647537971e-05, + "loss": 4.0498, + "step": 730112 + }, + { + "epoch": 1.01, + "learning_rate": 3.803756052786919e-05, + "loss": 4.058, + "step": 730624 + }, + { + "epoch": 1.01, + "learning_rate": 3.80291909591624e-05, + "loss": 4.0561, + "step": 731136 + }, + { + "epoch": 1.01, + "learning_rate": 3.802082139045561e-05, + "loss": 4.0666, + "step": 731648 + }, + { + "epoch": 1.01, + "learning_rate": 3.8012435442945097e-05, + "loss": 4.0567, + "step": 732160 + }, + { + "epoch": 1.02, + "learning_rate": 3.8004049495434577e-05, + "loss": 4.0541, + "step": 732672 + }, + { + "epoch": 1.02, + "learning_rate": 3.7995663547924056e-05, + "loss": 4.0505, + "step": 733184 + }, + { + "epoch": 1.02, + "learning_rate": 3.7987277600413536e-05, + "loss": 4.0521, + "step": 733696 + }, + { + "epoch": 1.02, + "learning_rate": 3.7978908031706746e-05, + "loss": 4.0428, + "step": 734208 + }, + { + "epoch": 1.02, + "learning_rate": 3.7970522084196225e-05, + "loss": 4.0548, + "step": 734720 + }, + { + "epoch": 1.02, + "learning_rate": 3.7962136136685705e-05, + "loss": 4.0512, + "step": 735232 + }, + { + "epoch": 1.02, + "learning_rate": 3.7953750189175185e-05, + "loss": 4.051, + "step": 735744 + }, + { + "epoch": 1.02, + "learning_rate": 3.7945364241664665e-05, + "loss": 4.0535, + "step": 736256 + }, + { + "epoch": 1.02, + "learning_rate": 3.7936978294154145e-05, + "loss": 4.067, + "step": 736768 + }, + { + "epoch": 1.02, + "learning_rate": 3.7928608725447354e-05, + "loss": 4.0504, + "step": 737280 + }, + { + "epoch": 1.02, + "learning_rate": 3.7920222777936834e-05, + "loss": 4.0623, + "step": 737792 + }, + { + "epoch": 1.02, + "learning_rate": 3.7911836830426314e-05, + "loss": 4.0597, + "step": 738304 + }, + { + "epoch": 1.02, + "learning_rate": 3.7903450882915794e-05, + "loss": 4.0553, + "step": 738816 + }, + { + "epoch": 1.02, + "learning_rate": 3.789506493540528e-05, + "loss": 4.0511, + "step": 739328 + }, + { + "epoch": 1.02, + "learning_rate": 3.788669536669849e-05, + "loss": 4.0586, + "step": 739840 + }, + { + "epoch": 1.02, + "learning_rate": 3.787830941918797e-05, + "loss": 4.0624, + "step": 740352 + }, + { + "epoch": 1.02, + "learning_rate": 3.786992347167745e-05, + "loss": 4.0479, + "step": 740864 + }, + { + "epoch": 1.02, + "learning_rate": 3.786153752416693e-05, + "loss": 4.0462, + "step": 741376 + }, + { + "epoch": 1.02, + "learning_rate": 3.785315157665641e-05, + "loss": 4.0424, + "step": 741888 + }, + { + "epoch": 1.02, + "learning_rate": 3.784476562914588e-05, + "loss": 4.0488, + "step": 742400 + }, + { + "epoch": 1.02, + "learning_rate": 3.78363960604391e-05, + "loss": 4.055, + "step": 742912 + }, + { + "epoch": 1.02, + "learning_rate": 3.782801011292858e-05, + "loss": 4.0507, + "step": 743424 + }, + { + "epoch": 1.02, + "learning_rate": 3.781962416541805e-05, + "loss": 4.0338, + "step": 743936 + }, + { + "epoch": 1.02, + "learning_rate": 3.781123821790753e-05, + "loss": 4.0456, + "step": 744448 + }, + { + "epoch": 1.02, + "learning_rate": 3.780286864920075e-05, + "loss": 4.055, + "step": 744960 + }, + { + "epoch": 1.02, + "learning_rate": 3.7794482701690235e-05, + "loss": 4.0543, + "step": 745472 + }, + { + "epoch": 1.02, + "learning_rate": 3.778609675417971e-05, + "loss": 4.0651, + "step": 745984 + }, + { + "epoch": 1.02, + "learning_rate": 3.777771080666919e-05, + "loss": 4.0556, + "step": 746496 + }, + { + "epoch": 1.02, + "learning_rate": 3.776932485915867e-05, + "loss": 4.0488, + "step": 747008 + }, + { + "epoch": 1.02, + "learning_rate": 3.776093891164815e-05, + "loss": 4.0533, + "step": 747520 + }, + { + "epoch": 1.02, + "learning_rate": 3.775255296413763e-05, + "loss": 4.0562, + "step": 748032 + }, + { + "epoch": 1.02, + "learning_rate": 3.774416701662711e-05, + "loss": 4.047, + "step": 748544 + }, + { + "epoch": 1.02, + "learning_rate": 3.773578106911659e-05, + "loss": 4.0484, + "step": 749056 + }, + { + "epoch": 1.02, + "learning_rate": 3.77274115004098e-05, + "loss": 4.0526, + "step": 749568 + }, + { + "epoch": 1.02, + "learning_rate": 3.771902555289928e-05, + "loss": 4.0431, + "step": 750080 + }, + { + "epoch": 1.02, + "learning_rate": 3.7710655984192486e-05, + "loss": 4.0523, + "step": 750592 + }, + { + "epoch": 1.02, + "learning_rate": 3.770227003668197e-05, + "loss": 4.0519, + "step": 751104 + }, + { + "epoch": 1.02, + "learning_rate": 3.769388408917145e-05, + "loss": 4.0554, + "step": 751616 + }, + { + "epoch": 1.02, + "learning_rate": 3.768549814166093e-05, + "loss": 4.0409, + "step": 752128 + }, + { + "epoch": 1.02, + "learning_rate": 3.767711219415041e-05, + "loss": 4.055, + "step": 752640 + }, + { + "epoch": 1.02, + "learning_rate": 3.766874262544362e-05, + "loss": 4.045, + "step": 753152 + }, + { + "epoch": 1.02, + "learning_rate": 3.76603566779331e-05, + "loss": 4.0601, + "step": 753664 + }, + { + "epoch": 1.02, + "learning_rate": 3.765197073042258e-05, + "loss": 4.0651, + "step": 754176 + }, + { + "epoch": 1.02, + "learning_rate": 3.764358478291206e-05, + "loss": 4.0503, + "step": 754688 + }, + { + "epoch": 1.02, + "learning_rate": 3.763519883540154e-05, + "loss": 4.0515, + "step": 755200 + }, + { + "epoch": 1.02, + "learning_rate": 3.762681288789102e-05, + "loss": 4.0541, + "step": 755712 + }, + { + "epoch": 1.02, + "learning_rate": 3.76184269403805e-05, + "loss": 4.0483, + "step": 756224 + }, + { + "epoch": 1.02, + "learning_rate": 3.761004099286998e-05, + "loss": 4.048, + "step": 756736 + }, + { + "epoch": 1.02, + "learning_rate": 3.760167142416319e-05, + "loss": 4.062, + "step": 757248 + }, + { + "epoch": 1.02, + "learning_rate": 3.759328547665267e-05, + "loss": 4.0574, + "step": 757760 + }, + { + "epoch": 1.02, + "learning_rate": 3.758489952914215e-05, + "loss": 4.0567, + "step": 758272 + }, + { + "epoch": 1.02, + "learning_rate": 3.757651358163164e-05, + "loss": 4.0515, + "step": 758784 + }, + { + "epoch": 1.02, + "learning_rate": 3.756812763412112e-05, + "loss": 4.0573, + "step": 759296 + }, + { + "epoch": 1.02, + "learning_rate": 3.75597416866106e-05, + "loss": 4.0468, + "step": 759808 + }, + { + "epoch": 1.02, + "learning_rate": 3.7551388496707535e-05, + "loss": 4.0453, + "step": 760320 + }, + { + "epoch": 1.02, + "learning_rate": 3.7543002549197015e-05, + "loss": 4.0458, + "step": 760832 + }, + { + "epoch": 1.02, + "learning_rate": 3.7534616601686495e-05, + "loss": 4.0533, + "step": 761344 + }, + { + "epoch": 1.02, + "learning_rate": 3.7526230654175975e-05, + "loss": 4.0475, + "step": 761856 + }, + { + "epoch": 1.02, + "learning_rate": 3.7517844706665455e-05, + "loss": 4.0319, + "step": 762368 + }, + { + "epoch": 1.02, + "learning_rate": 3.7509458759154935e-05, + "loss": 4.051, + "step": 762880 + }, + { + "epoch": 1.03, + "eval_loss": 4.076761722564697, + "eval_runtime": 530.228, + "eval_samples_per_second": 719.673, + "eval_steps_per_second": 22.49, + "step": 763190 + }, + { + "epoch": 0.0, + "learning_rate": 3.7501072811644415e-05, + "loss": 4.0675, + "step": 763392 + }, + { + "epoch": 0.0, + "learning_rate": 3.749268686413389e-05, + "loss": 4.0537, + "step": 763904 + }, + { + "epoch": 0.0, + "learning_rate": 3.7484300916623375e-05, + "loss": 4.0572, + "step": 764416 + }, + { + "epoch": 0.0, + "learning_rate": 3.7475914969112855e-05, + "loss": 4.0398, + "step": 764928 + }, + { + "epoch": 0.0, + "learning_rate": 3.7467529021602335e-05, + "loss": 4.0542, + "step": 765440 + }, + { + "epoch": 0.0, + "learning_rate": 3.7459143074091815e-05, + "loss": 4.046, + "step": 765952 + }, + { + "epoch": 0.0, + "learning_rate": 3.7450757126581294e-05, + "loss": 4.043, + "step": 766464 + }, + { + "epoch": 0.0, + "learning_rate": 3.7442371179070774e-05, + "loss": 4.042, + "step": 766976 + }, + { + "epoch": 0.0, + "learning_rate": 3.7433985231560254e-05, + "loss": 4.0412, + "step": 767488 + }, + { + "epoch": 0.0, + "learning_rate": 3.7425599284049734e-05, + "loss": 4.0534, + "step": 768000 + }, + { + "epoch": 0.0, + "learning_rate": 3.7417213336539214e-05, + "loss": 4.0516, + "step": 768512 + }, + { + "epoch": 0.0, + "learning_rate": 3.7408827389028694e-05, + "loss": 4.0447, + "step": 769024 + }, + { + "epoch": 0.0, + "learning_rate": 3.7400441441518174e-05, + "loss": 4.0531, + "step": 769536 + }, + { + "epoch": 0.0, + "learning_rate": 3.7392055494007654e-05, + "loss": 4.0347, + "step": 770048 + }, + { + "epoch": 0.0, + "learning_rate": 3.738366954649713e-05, + "loss": 4.0555, + "step": 770560 + }, + { + "epoch": 0.0, + "learning_rate": 3.737528359898661e-05, + "loss": 4.0346, + "step": 771072 + }, + { + "epoch": 0.0, + "learning_rate": 3.736689765147609e-05, + "loss": 4.0424, + "step": 771584 + }, + { + "epoch": 0.0, + "learning_rate": 3.7358511703965574e-05, + "loss": 4.0462, + "step": 772096 + }, + { + "epoch": 0.0, + "learning_rate": 3.7350125756455054e-05, + "loss": 4.0468, + "step": 772608 + }, + { + "epoch": 0.0, + "learning_rate": 3.7341739808944534e-05, + "loss": 4.0515, + "step": 773120 + }, + { + "epoch": 0.0, + "learning_rate": 3.733337024023774e-05, + "loss": 4.0377, + "step": 773632 + }, + { + "epoch": 0.0, + "learning_rate": 3.732498429272722e-05, + "loss": 4.0647, + "step": 774144 + }, + { + "epoch": 0.0, + "learning_rate": 3.73165983452167e-05, + "loss": 4.0451, + "step": 774656 + }, + { + "epoch": 0.0, + "learning_rate": 3.730821239770618e-05, + "loss": 4.0515, + "step": 775168 + }, + { + "epoch": 0.0, + "learning_rate": 3.729982645019566e-05, + "loss": 4.0436, + "step": 775680 + }, + { + "epoch": 0.0, + "learning_rate": 3.729145688148887e-05, + "loss": 4.0533, + "step": 776192 + }, + { + "epoch": 0.0, + "learning_rate": 3.728307093397835e-05, + "loss": 4.0466, + "step": 776704 + }, + { + "epoch": 0.0, + "learning_rate": 3.727468498646783e-05, + "loss": 4.042, + "step": 777216 + }, + { + "epoch": 0.0, + "learning_rate": 3.726629903895731e-05, + "loss": 4.0487, + "step": 777728 + }, + { + "epoch": 0.0, + "learning_rate": 3.725792947025053e-05, + "loss": 4.0462, + "step": 778240 + }, + { + "epoch": 0.01, + "learning_rate": 3.724955990154374e-05, + "loss": 4.0295, + "step": 778752 + }, + { + "epoch": 0.01, + "learning_rate": 3.724117395403322e-05, + "loss": 4.0501, + "step": 779264 + }, + { + "epoch": 0.01, + "learning_rate": 3.72327880065227e-05, + "loss": 4.0496, + "step": 779776 + }, + { + "epoch": 0.01, + "learning_rate": 3.722440205901218e-05, + "loss": 4.046, + "step": 780288 + }, + { + "epoch": 0.01, + "learning_rate": 3.7216016111501657e-05, + "loss": 4.0363, + "step": 780800 + }, + { + "epoch": 0.01, + "learning_rate": 3.7207630163991137e-05, + "loss": 4.0568, + "step": 781312 + }, + { + "epoch": 0.01, + "learning_rate": 3.7199260595284346e-05, + "loss": 4.0408, + "step": 781824 + }, + { + "epoch": 0.01, + "learning_rate": 3.7190874647773826e-05, + "loss": 4.0501, + "step": 782336 + }, + { + "epoch": 0.01, + "learning_rate": 3.7182488700263306e-05, + "loss": 4.0458, + "step": 782848 + }, + { + "epoch": 0.01, + "learning_rate": 3.7174102752752785e-05, + "loss": 4.0337, + "step": 783360 + }, + { + "epoch": 0.01, + "learning_rate": 3.7165716805242265e-05, + "loss": 4.039, + "step": 783872 + }, + { + "epoch": 0.01, + "learning_rate": 3.7157330857731745e-05, + "loss": 4.0261, + "step": 784384 + }, + { + "epoch": 0.01, + "learning_rate": 3.7148944910221225e-05, + "loss": 4.047, + "step": 784896 + }, + { + "epoch": 0.01, + "learning_rate": 3.7140558962710705e-05, + "loss": 4.042, + "step": 785408 + }, + { + "epoch": 0.01, + "learning_rate": 3.713218939400392e-05, + "loss": 4.0324, + "step": 785920 + }, + { + "epoch": 0.01, + "learning_rate": 3.71238034464934e-05, + "loss": 4.0492, + "step": 786432 + }, + { + "epoch": 0.01, + "learning_rate": 3.711541749898288e-05, + "loss": 4.0342, + "step": 786944 + }, + { + "epoch": 0.01, + "learning_rate": 3.710703155147236e-05, + "loss": 4.0487, + "step": 787456 + }, + { + "epoch": 0.01, + "learning_rate": 3.70986783615693e-05, + "loss": 4.0432, + "step": 787968 + }, + { + "epoch": 0.01, + "learning_rate": 3.709029241405878e-05, + "loss": 4.0384, + "step": 788480 + }, + { + "epoch": 0.01, + "learning_rate": 3.708190646654826e-05, + "loss": 4.0417, + "step": 788992 + }, + { + "epoch": 0.01, + "learning_rate": 3.707352051903774e-05, + "loss": 4.0359, + "step": 789504 + }, + { + "epoch": 0.01, + "learning_rate": 3.706513457152722e-05, + "loss": 4.0423, + "step": 790016 + }, + { + "epoch": 0.01, + "learning_rate": 3.70567486240167e-05, + "loss": 4.0284, + "step": 790528 + }, + { + "epoch": 0.01, + "learning_rate": 3.7048379055309915e-05, + "loss": 4.0422, + "step": 791040 + }, + { + "epoch": 0.01, + "learning_rate": 3.7039993107799395e-05, + "loss": 4.0382, + "step": 791552 + }, + { + "epoch": 0.01, + "learning_rate": 3.7031607160288875e-05, + "loss": 4.0389, + "step": 792064 + }, + { + "epoch": 0.01, + "learning_rate": 3.7023221212778355e-05, + "loss": 4.0345, + "step": 792576 + }, + { + "epoch": 0.01, + "learning_rate": 3.7014835265267835e-05, + "loss": 4.0466, + "step": 793088 + }, + { + "epoch": 0.01, + "learning_rate": 3.7006449317757315e-05, + "loss": 4.0307, + "step": 793600 + }, + { + "epoch": 0.01, + "learning_rate": 3.6998063370246795e-05, + "loss": 4.0304, + "step": 794112 + }, + { + "epoch": 0.01, + "learning_rate": 3.698967742273627e-05, + "loss": 4.036, + "step": 794624 + }, + { + "epoch": 0.01, + "learning_rate": 3.6981307854029484e-05, + "loss": 4.03, + "step": 795136 + }, + { + "epoch": 0.01, + "learning_rate": 3.6972921906518964e-05, + "loss": 4.0409, + "step": 795648 + }, + { + "epoch": 0.01, + "learning_rate": 3.696453595900844e-05, + "loss": 4.038, + "step": 796160 + }, + { + "epoch": 0.01, + "learning_rate": 3.695615001149792e-05, + "loss": 4.0406, + "step": 796672 + }, + { + "epoch": 0.01, + "learning_rate": 3.694778044279113e-05, + "loss": 4.0195, + "step": 797184 + }, + { + "epoch": 0.01, + "learning_rate": 3.693939449528061e-05, + "loss": 4.0403, + "step": 797696 + }, + { + "epoch": 0.01, + "learning_rate": 3.693100854777009e-05, + "loss": 4.0281, + "step": 798208 + }, + { + "epoch": 0.01, + "learning_rate": 3.692263897906331e-05, + "loss": 4.0313, + "step": 798720 + }, + { + "epoch": 0.01, + "learning_rate": 3.691425303155279e-05, + "loss": 4.0334, + "step": 799232 + }, + { + "epoch": 0.01, + "learning_rate": 3.690586708404226e-05, + "loss": 4.0394, + "step": 799744 + }, + { + "epoch": 0.01, + "learning_rate": 3.689748113653174e-05, + "loss": 4.0315, + "step": 800256 + }, + { + "epoch": 0.01, + "learning_rate": 3.688909518902122e-05, + "loss": 4.0515, + "step": 800768 + }, + { + "epoch": 0.01, + "learning_rate": 3.68807092415107e-05, + "loss": 4.0407, + "step": 801280 + }, + { + "epoch": 0.01, + "learning_rate": 3.687232329400018e-05, + "loss": 4.042, + "step": 801792 + }, + { + "epoch": 0.01, + "learning_rate": 3.686395372529339e-05, + "loss": 4.0412, + "step": 802304 + }, + { + "epoch": 0.01, + "learning_rate": 3.685556777778287e-05, + "loss": 4.0294, + "step": 802816 + }, + { + "epoch": 0.01, + "learning_rate": 3.684718183027235e-05, + "loss": 4.0322, + "step": 803328 + }, + { + "epoch": 0.01, + "learning_rate": 3.683879588276184e-05, + "loss": 4.0386, + "step": 803840 + }, + { + "epoch": 0.01, + "learning_rate": 3.6830426314055046e-05, + "loss": 4.04, + "step": 804352 + }, + { + "epoch": 0.01, + "learning_rate": 3.6822040366544526e-05, + "loss": 4.0448, + "step": 804864 + }, + { + "epoch": 0.01, + "learning_rate": 3.6813654419034006e-05, + "loss": 4.0186, + "step": 805376 + }, + { + "epoch": 0.01, + "learning_rate": 3.6805268471523486e-05, + "loss": 4.0234, + "step": 805888 + }, + { + "epoch": 0.01, + "learning_rate": 3.6796898902816695e-05, + "loss": 4.025, + "step": 806400 + }, + { + "epoch": 0.01, + "learning_rate": 3.678852933410991e-05, + "loss": 4.0431, + "step": 806912 + }, + { + "epoch": 0.01, + "learning_rate": 3.6780143386599384e-05, + "loss": 4.0342, + "step": 807424 + }, + { + "epoch": 0.01, + "learning_rate": 3.6771757439088864e-05, + "loss": 4.0438, + "step": 807936 + }, + { + "epoch": 0.01, + "learning_rate": 3.6763371491578344e-05, + "loss": 4.0335, + "step": 808448 + }, + { + "epoch": 0.01, + "learning_rate": 3.6754985544067824e-05, + "loss": 4.0363, + "step": 808960 + }, + { + "epoch": 0.02, + "learning_rate": 3.6746599596557304e-05, + "loss": 4.0241, + "step": 809472 + }, + { + "epoch": 0.02, + "learning_rate": 3.673821364904679e-05, + "loss": 4.0325, + "step": 809984 + }, + { + "epoch": 0.02, + "learning_rate": 3.672982770153627e-05, + "loss": 4.0247, + "step": 810496 + }, + { + "epoch": 0.02, + "learning_rate": 3.672144175402575e-05, + "loss": 4.0319, + "step": 811008 + }, + { + "epoch": 0.02, + "learning_rate": 3.671305580651523e-05, + "loss": 4.0289, + "step": 811520 + }, + { + "epoch": 0.02, + "learning_rate": 3.670468623780844e-05, + "loss": 4.032, + "step": 812032 + }, + { + "epoch": 0.02, + "learning_rate": 3.669630029029792e-05, + "loss": 4.0344, + "step": 812544 + }, + { + "epoch": 0.02, + "learning_rate": 3.66879143427874e-05, + "loss": 4.0437, + "step": 813056 + }, + { + "epoch": 0.02, + "learning_rate": 3.667952839527688e-05, + "loss": 4.0357, + "step": 813568 + }, + { + "epoch": 0.02, + "learning_rate": 3.667114244776636e-05, + "loss": 4.041, + "step": 814080 + }, + { + "epoch": 0.02, + "learning_rate": 3.666277287905957e-05, + "loss": 4.0363, + "step": 814592 + }, + { + "epoch": 0.02, + "learning_rate": 3.665438693154905e-05, + "loss": 4.0358, + "step": 815104 + }, + { + "epoch": 0.02, + "learning_rate": 3.664600098403853e-05, + "loss": 4.0338, + "step": 815616 + }, + { + "epoch": 0.02, + "learning_rate": 3.663761503652801e-05, + "loss": 4.0406, + "step": 816128 + }, + { + "epoch": 0.02, + "learning_rate": 3.662922908901749e-05, + "loss": 4.0398, + "step": 816640 + }, + { + "epoch": 0.02, + "learning_rate": 3.6620843141506975e-05, + "loss": 4.028, + "step": 817152 + }, + { + "epoch": 0.02, + "learning_rate": 3.6612457193996455e-05, + "loss": 4.0253, + "step": 817664 + }, + { + "epoch": 0.02, + "learning_rate": 3.660407124648593e-05, + "loss": 4.02, + "step": 818176 + }, + { + "epoch": 0.02, + "learning_rate": 3.6595701677779144e-05, + "loss": 4.0312, + "step": 818688 + }, + { + "epoch": 0.02, + "learning_rate": 3.658733210907235e-05, + "loss": 4.0326, + "step": 819200 + }, + { + "epoch": 0.02, + "learning_rate": 3.657894616156183e-05, + "loss": 4.036, + "step": 819712 + }, + { + "epoch": 0.02, + "learning_rate": 3.657056021405131e-05, + "loss": 4.0111, + "step": 820224 + }, + { + "epoch": 0.02, + "learning_rate": 3.656217426654079e-05, + "loss": 4.0176, + "step": 820736 + }, + { + "epoch": 0.02, + "learning_rate": 3.655378831903027e-05, + "loss": 4.0376, + "step": 821248 + }, + { + "epoch": 0.02, + "learning_rate": 3.6545402371519746e-05, + "loss": 4.0349, + "step": 821760 + }, + { + "epoch": 0.02, + "learning_rate": 3.6537016424009226e-05, + "loss": 4.0446, + "step": 822272 + }, + { + "epoch": 0.02, + "learning_rate": 3.652863047649871e-05, + "loss": 4.0377, + "step": 822784 + }, + { + "epoch": 0.02, + "learning_rate": 3.652024452898819e-05, + "loss": 4.0256, + "step": 823296 + }, + { + "epoch": 0.02, + "learning_rate": 3.65118749602814e-05, + "loss": 4.0358, + "step": 823808 + }, + { + "epoch": 0.02, + "learning_rate": 3.650348901277088e-05, + "loss": 4.0331, + "step": 824320 + }, + { + "epoch": 0.02, + "learning_rate": 3.649510306526036e-05, + "loss": 4.0261, + "step": 824832 + }, + { + "epoch": 0.02, + "learning_rate": 3.648673349655357e-05, + "loss": 4.0274, + "step": 825344 + }, + { + "epoch": 0.02, + "learning_rate": 3.647834754904305e-05, + "loss": 4.0327, + "step": 825856 + }, + { + "epoch": 0.02, + "learning_rate": 3.646996160153253e-05, + "loss": 4.0283, + "step": 826368 + }, + { + "epoch": 0.02, + "learning_rate": 3.646157565402201e-05, + "loss": 4.0307, + "step": 826880 + }, + { + "epoch": 0.02, + "learning_rate": 3.645318970651149e-05, + "loss": 4.0295, + "step": 827392 + }, + { + "epoch": 0.02, + "learning_rate": 3.644480375900097e-05, + "loss": 4.0383, + "step": 827904 + }, + { + "epoch": 0.02, + "learning_rate": 3.643641781149045e-05, + "loss": 4.0237, + "step": 828416 + }, + { + "epoch": 0.02, + "learning_rate": 3.642803186397993e-05, + "loss": 4.0297, + "step": 828928 + }, + { + "epoch": 0.02, + "learning_rate": 3.641966229527315e-05, + "loss": 4.029, + "step": 829440 + }, + { + "epoch": 0.02, + "learning_rate": 3.641127634776263e-05, + "loss": 4.0395, + "step": 829952 + }, + { + "epoch": 0.02, + "learning_rate": 3.6402890400252107e-05, + "loss": 4.0411, + "step": 830464 + }, + { + "epoch": 0.02, + "learning_rate": 3.6394504452741587e-05, + "loss": 4.0323, + "step": 830976 + }, + { + "epoch": 0.02, + "learning_rate": 3.6386134884034796e-05, + "loss": 4.0308, + "step": 831488 + }, + { + "epoch": 0.02, + "learning_rate": 3.6377748936524276e-05, + "loss": 4.0362, + "step": 832000 + }, + { + "epoch": 0.02, + "learning_rate": 3.6369362989013756e-05, + "loss": 4.0287, + "step": 832512 + }, + { + "epoch": 0.02, + "learning_rate": 3.6360977041503236e-05, + "loss": 4.0275, + "step": 833024 + }, + { + "epoch": 0.02, + "learning_rate": 3.6352607472796445e-05, + "loss": 4.0413, + "step": 833536 + }, + { + "epoch": 0.02, + "learning_rate": 3.6344221525285925e-05, + "loss": 4.038, + "step": 834048 + }, + { + "epoch": 0.02, + "learning_rate": 3.6335835577775405e-05, + "loss": 4.0401, + "step": 834560 + }, + { + "epoch": 0.02, + "learning_rate": 3.6327449630264884e-05, + "loss": 4.0334, + "step": 835072 + }, + { + "epoch": 0.02, + "learning_rate": 3.63190800615581e-05, + "loss": 4.0392, + "step": 835584 + }, + { + "epoch": 0.02, + "learning_rate": 3.631069411404758e-05, + "loss": 4.0231, + "step": 836096 + }, + { + "epoch": 0.02, + "learning_rate": 3.630230816653706e-05, + "loss": 4.0271, + "step": 836608 + }, + { + "epoch": 0.02, + "learning_rate": 3.629393859783027e-05, + "loss": 4.0277, + "step": 837120 + }, + { + "epoch": 0.02, + "learning_rate": 3.628555265031975e-05, + "loss": 4.0321, + "step": 837632 + }, + { + "epoch": 0.02, + "learning_rate": 3.627716670280923e-05, + "loss": 4.0281, + "step": 838144 + }, + { + "epoch": 0.02, + "learning_rate": 3.626878075529871e-05, + "loss": 4.0155, + "step": 838656 + }, + { + "epoch": 0.02, + "learning_rate": 3.626041118659192e-05, + "loss": 4.0334, + "step": 839168 + }, + { + "epoch": 0.03, + "eval_loss": 4.0648016929626465, + "eval_runtime": 543.7091, + "eval_samples_per_second": 701.829, + "eval_steps_per_second": 21.933, + "step": 839509 + }, + { + "epoch": 1.0, + "learning_rate": 3.62520252390814e-05, + "loss": 4.0312, + "step": 839680 + }, + { + "epoch": 1.0, + "learning_rate": 3.624363929157088e-05, + "loss": 4.0365, + "step": 840192 + }, + { + "epoch": 1.0, + "learning_rate": 3.623525334406036e-05, + "loss": 4.0358, + "step": 840704 + }, + { + "epoch": 1.0, + "learning_rate": 3.622686739654984e-05, + "loss": 4.0243, + "step": 841216 + }, + { + "epoch": 1.0, + "learning_rate": 3.621848144903932e-05, + "loss": 4.0331, + "step": 841728 + }, + { + "epoch": 1.0, + "learning_rate": 3.62100955015288e-05, + "loss": 4.0277, + "step": 842240 + }, + { + "epoch": 1.0, + "learning_rate": 3.6201709554018285e-05, + "loss": 4.0211, + "step": 842752 + }, + { + "epoch": 1.0, + "learning_rate": 3.619332360650776e-05, + "loss": 4.0242, + "step": 843264 + }, + { + "epoch": 1.0, + "learning_rate": 3.618493765899724e-05, + "loss": 4.0274, + "step": 843776 + }, + { + "epoch": 1.0, + "learning_rate": 3.617655171148672e-05, + "loss": 4.0309, + "step": 844288 + }, + { + "epoch": 1.0, + "learning_rate": 3.6168182142779934e-05, + "loss": 4.0287, + "step": 844800 + }, + { + "epoch": 1.0, + "learning_rate": 3.615979619526941e-05, + "loss": 4.026, + "step": 845312 + }, + { + "epoch": 1.0, + "learning_rate": 3.615141024775889e-05, + "loss": 4.0333, + "step": 845824 + }, + { + "epoch": 1.0, + "learning_rate": 3.614302430024837e-05, + "loss": 4.0215, + "step": 846336 + }, + { + "epoch": 1.0, + "learning_rate": 3.613463835273785e-05, + "loss": 4.0306, + "step": 846848 + }, + { + "epoch": 1.0, + "learning_rate": 3.612625240522733e-05, + "loss": 4.0168, + "step": 847360 + }, + { + "epoch": 1.0, + "learning_rate": 3.611786645771681e-05, + "loss": 4.0212, + "step": 847872 + }, + { + "epoch": 1.0, + "learning_rate": 3.610948051020629e-05, + "loss": 4.0288, + "step": 848384 + }, + { + "epoch": 1.0, + "learning_rate": 3.61011109414995e-05, + "loss": 4.0255, + "step": 848896 + }, + { + "epoch": 1.0, + "learning_rate": 3.609272499398898e-05, + "loss": 4.0305, + "step": 849408 + }, + { + "epoch": 1.0, + "learning_rate": 3.608433904647846e-05, + "loss": 4.0218, + "step": 849920 + }, + { + "epoch": 1.0, + "learning_rate": 3.607595309896794e-05, + "loss": 4.0468, + "step": 850432 + }, + { + "epoch": 1.0, + "learning_rate": 3.606758353026115e-05, + "loss": 4.0251, + "step": 850944 + }, + { + "epoch": 1.0, + "learning_rate": 3.605919758275063e-05, + "loss": 4.0289, + "step": 851456 + }, + { + "epoch": 1.0, + "learning_rate": 3.605081163524011e-05, + "loss": 4.0273, + "step": 851968 + }, + { + "epoch": 1.0, + "learning_rate": 3.604242568772959e-05, + "loss": 4.0365, + "step": 852480 + }, + { + "epoch": 1.0, + "learning_rate": 3.60340561190228e-05, + "loss": 4.0261, + "step": 852992 + }, + { + "epoch": 1.0, + "learning_rate": 3.602567017151228e-05, + "loss": 4.02, + "step": 853504 + }, + { + "epoch": 1.0, + "learning_rate": 3.601728422400176e-05, + "loss": 4.0281, + "step": 854016 + }, + { + "epoch": 1.0, + "learning_rate": 3.600889827649124e-05, + "loss": 4.0303, + "step": 854528 + }, + { + "epoch": 1.01, + "learning_rate": 3.600051232898072e-05, + "loss": 4.0121, + "step": 855040 + }, + { + "epoch": 1.01, + "learning_rate": 3.599212638147021e-05, + "loss": 4.0261, + "step": 855552 + }, + { + "epoch": 1.01, + "learning_rate": 3.598374043395969e-05, + "loss": 4.0337, + "step": 856064 + }, + { + "epoch": 1.01, + "learning_rate": 3.597535448644917e-05, + "loss": 4.0294, + "step": 856576 + }, + { + "epoch": 1.01, + "learning_rate": 3.596696853893865e-05, + "loss": 4.0168, + "step": 857088 + }, + { + "epoch": 1.01, + "learning_rate": 3.5958598970231856e-05, + "loss": 4.0315, + "step": 857600 + }, + { + "epoch": 1.01, + "learning_rate": 3.5950229401525065e-05, + "loss": 4.026, + "step": 858112 + }, + { + "epoch": 1.01, + "learning_rate": 3.5941843454014545e-05, + "loss": 4.028, + "step": 858624 + }, + { + "epoch": 1.01, + "learning_rate": 3.5933457506504025e-05, + "loss": 4.0254, + "step": 859136 + }, + { + "epoch": 1.01, + "learning_rate": 3.5925071558993505e-05, + "loss": 4.0182, + "step": 859648 + }, + { + "epoch": 1.01, + "learning_rate": 3.5916685611482985e-05, + "loss": 4.021, + "step": 860160 + }, + { + "epoch": 1.01, + "learning_rate": 3.5908316042776194e-05, + "loss": 4.0081, + "step": 860672 + }, + { + "epoch": 1.01, + "learning_rate": 3.5899930095265674e-05, + "loss": 4.0255, + "step": 861184 + }, + { + "epoch": 1.01, + "learning_rate": 3.5891544147755154e-05, + "loss": 4.0251, + "step": 861696 + }, + { + "epoch": 1.01, + "learning_rate": 3.588315820024464e-05, + "loss": 4.0165, + "step": 862208 + }, + { + "epoch": 1.01, + "learning_rate": 3.587477225273412e-05, + "loss": 4.0257, + "step": 862720 + }, + { + "epoch": 1.01, + "learning_rate": 3.5866386305223594e-05, + "loss": 4.0154, + "step": 863232 + }, + { + "epoch": 1.01, + "learning_rate": 3.5858000357713074e-05, + "loss": 4.0332, + "step": 863744 + }, + { + "epoch": 1.01, + "learning_rate": 3.5849614410202554e-05, + "loss": 4.0263, + "step": 864256 + }, + { + "epoch": 1.01, + "learning_rate": 3.5841228462692034e-05, + "loss": 4.013, + "step": 864768 + }, + { + "epoch": 1.01, + "learning_rate": 3.583285889398524e-05, + "loss": 4.0212, + "step": 865280 + }, + { + "epoch": 1.01, + "learning_rate": 3.582447294647472e-05, + "loss": 4.0172, + "step": 865792 + }, + { + "epoch": 1.01, + "learning_rate": 3.58160869989642e-05, + "loss": 4.0292, + "step": 866304 + }, + { + "epoch": 1.01, + "learning_rate": 3.580770105145368e-05, + "loss": 4.0068, + "step": 866816 + }, + { + "epoch": 1.01, + "learning_rate": 3.579933148274689e-05, + "loss": 4.0269, + "step": 867328 + }, + { + "epoch": 1.01, + "learning_rate": 3.579094553523638e-05, + "loss": 4.0183, + "step": 867840 + }, + { + "epoch": 1.01, + "learning_rate": 3.578255958772586e-05, + "loss": 4.0211, + "step": 868352 + }, + { + "epoch": 1.01, + "learning_rate": 3.577417364021534e-05, + "loss": 4.0186, + "step": 868864 + }, + { + "epoch": 1.01, + "learning_rate": 3.576580407150855e-05, + "loss": 4.0288, + "step": 869376 + }, + { + "epoch": 1.01, + "learning_rate": 3.575741812399803e-05, + "loss": 4.0078, + "step": 869888 + }, + { + "epoch": 1.01, + "learning_rate": 3.574903217648751e-05, + "loss": 4.0127, + "step": 870400 + }, + { + "epoch": 1.01, + "learning_rate": 3.574064622897699e-05, + "loss": 4.0182, + "step": 870912 + }, + { + "epoch": 1.01, + "learning_rate": 3.573226028146647e-05, + "loss": 4.0124, + "step": 871424 + }, + { + "epoch": 1.01, + "learning_rate": 3.5723890712759676e-05, + "loss": 4.0213, + "step": 871936 + }, + { + "epoch": 1.01, + "learning_rate": 3.5715504765249156e-05, + "loss": 4.02, + "step": 872448 + }, + { + "epoch": 1.01, + "learning_rate": 3.5707118817738636e-05, + "loss": 4.0263, + "step": 872960 + }, + { + "epoch": 1.01, + "learning_rate": 3.5698732870228116e-05, + "loss": 3.9986, + "step": 873472 + }, + { + "epoch": 1.01, + "learning_rate": 3.569036330152133e-05, + "loss": 4.0198, + "step": 873984 + }, + { + "epoch": 1.01, + "learning_rate": 3.568197735401081e-05, + "loss": 4.0147, + "step": 874496 + }, + { + "epoch": 1.01, + "learning_rate": 3.567359140650029e-05, + "loss": 4.0115, + "step": 875008 + }, + { + "epoch": 1.01, + "learning_rate": 3.566520545898977e-05, + "loss": 4.0159, + "step": 875520 + }, + { + "epoch": 1.01, + "learning_rate": 3.565683589028298e-05, + "loss": 4.0236, + "step": 876032 + }, + { + "epoch": 1.01, + "learning_rate": 3.564844994277246e-05, + "loss": 4.0092, + "step": 876544 + }, + { + "epoch": 1.01, + "learning_rate": 3.564006399526194e-05, + "loss": 4.0338, + "step": 877056 + }, + { + "epoch": 1.01, + "learning_rate": 3.563167804775142e-05, + "loss": 4.0217, + "step": 877568 + }, + { + "epoch": 1.01, + "learning_rate": 3.562330847904463e-05, + "loss": 4.0254, + "step": 878080 + }, + { + "epoch": 1.01, + "learning_rate": 3.561492253153411e-05, + "loss": 4.0229, + "step": 878592 + }, + { + "epoch": 1.01, + "learning_rate": 3.560653658402359e-05, + "loss": 4.0127, + "step": 879104 + }, + { + "epoch": 1.01, + "learning_rate": 3.559815063651307e-05, + "loss": 4.0133, + "step": 879616 + }, + { + "epoch": 1.01, + "learning_rate": 3.558976468900255e-05, + "loss": 4.0213, + "step": 880128 + }, + { + "epoch": 1.01, + "learning_rate": 3.5581411499099495e-05, + "loss": 4.0247, + "step": 880640 + }, + { + "epoch": 1.01, + "learning_rate": 3.5573025551588975e-05, + "loss": 4.022, + "step": 881152 + }, + { + "epoch": 1.01, + "learning_rate": 3.5564639604078455e-05, + "loss": 4.002, + "step": 881664 + }, + { + "epoch": 1.01, + "learning_rate": 3.5556253656567935e-05, + "loss": 4.0076, + "step": 882176 + }, + { + "epoch": 1.01, + "learning_rate": 3.5547867709057415e-05, + "loss": 4.0037, + "step": 882688 + }, + { + "epoch": 1.01, + "learning_rate": 3.5539498140350624e-05, + "loss": 4.0251, + "step": 883200 + }, + { + "epoch": 1.01, + "learning_rate": 3.5531112192840104e-05, + "loss": 4.0132, + "step": 883712 + }, + { + "epoch": 1.01, + "learning_rate": 3.5522726245329584e-05, + "loss": 4.0263, + "step": 884224 + }, + { + "epoch": 1.01, + "learning_rate": 3.5514340297819064e-05, + "loss": 4.0177, + "step": 884736 + }, + { + "epoch": 1.01, + "learning_rate": 3.5505954350308544e-05, + "loss": 4.0143, + "step": 885248 + }, + { + "epoch": 1.02, + "learning_rate": 3.549758478160175e-05, + "loss": 4.0087, + "step": 885760 + }, + { + "epoch": 1.02, + "learning_rate": 3.548919883409124e-05, + "loss": 4.0128, + "step": 886272 + }, + { + "epoch": 1.02, + "learning_rate": 3.548081288658072e-05, + "loss": 4.011, + "step": 886784 + }, + { + "epoch": 1.02, + "learning_rate": 3.54724269390702e-05, + "loss": 4.011, + "step": 887296 + }, + { + "epoch": 1.02, + "learning_rate": 3.546404099155968e-05, + "loss": 4.0089, + "step": 887808 + }, + { + "epoch": 1.02, + "learning_rate": 3.545567142285289e-05, + "loss": 4.017, + "step": 888320 + }, + { + "epoch": 1.02, + "learning_rate": 3.54473018541461e-05, + "loss": 4.0169, + "step": 888832 + }, + { + "epoch": 1.02, + "learning_rate": 3.543891590663558e-05, + "loss": 4.023, + "step": 889344 + }, + { + "epoch": 1.02, + "learning_rate": 3.543052995912506e-05, + "loss": 4.0198, + "step": 889856 + }, + { + "epoch": 1.02, + "learning_rate": 3.542214401161454e-05, + "loss": 4.0206, + "step": 890368 + }, + { + "epoch": 1.02, + "learning_rate": 3.541375806410402e-05, + "loss": 4.0171, + "step": 890880 + }, + { + "epoch": 1.02, + "learning_rate": 3.54053721165935e-05, + "loss": 4.0213, + "step": 891392 + }, + { + "epoch": 1.02, + "learning_rate": 3.539698616908298e-05, + "loss": 4.015, + "step": 891904 + }, + { + "epoch": 1.02, + "learning_rate": 3.538860022157246e-05, + "loss": 4.0222, + "step": 892416 + }, + { + "epoch": 1.02, + "learning_rate": 3.538021427406194e-05, + "loss": 4.0164, + "step": 892928 + }, + { + "epoch": 1.02, + "learning_rate": 3.5371828326551424e-05, + "loss": 4.0168, + "step": 893440 + }, + { + "epoch": 1.02, + "learning_rate": 3.5363442379040904e-05, + "loss": 4.0097, + "step": 893952 + }, + { + "epoch": 1.02, + "learning_rate": 3.535505643153038e-05, + "loss": 3.9967, + "step": 894464 + }, + { + "epoch": 1.02, + "learning_rate": 3.534668686282359e-05, + "loss": 4.0141, + "step": 894976 + }, + { + "epoch": 1.02, + "learning_rate": 3.533833367292053e-05, + "loss": 4.0179, + "step": 895488 + }, + { + "epoch": 1.02, + "learning_rate": 3.532994772541001e-05, + "loss": 4.0139, + "step": 896000 + }, + { + "epoch": 1.02, + "learning_rate": 3.532156177789949e-05, + "loss": 3.9941, + "step": 896512 + }, + { + "epoch": 1.02, + "learning_rate": 3.531317583038897e-05, + "loss": 4.0011, + "step": 897024 + }, + { + "epoch": 1.02, + "learning_rate": 3.530478988287845e-05, + "loss": 4.0189, + "step": 897536 + }, + { + "epoch": 1.02, + "learning_rate": 3.529640393536793e-05, + "loss": 4.0193, + "step": 898048 + }, + { + "epoch": 1.02, + "learning_rate": 3.528801798785741e-05, + "loss": 4.0247, + "step": 898560 + }, + { + "epoch": 1.02, + "learning_rate": 3.527963204034689e-05, + "loss": 4.0197, + "step": 899072 + }, + { + "epoch": 1.02, + "learning_rate": 3.527124609283637e-05, + "loss": 4.0136, + "step": 899584 + }, + { + "epoch": 1.02, + "learning_rate": 3.526286014532585e-05, + "loss": 4.0152, + "step": 900096 + }, + { + "epoch": 1.02, + "learning_rate": 3.525447419781533e-05, + "loss": 4.0155, + "step": 900608 + }, + { + "epoch": 1.02, + "learning_rate": 3.524608825030481e-05, + "loss": 4.0052, + "step": 901120 + }, + { + "epoch": 1.02, + "learning_rate": 3.523770230279429e-05, + "loss": 4.0076, + "step": 901632 + }, + { + "epoch": 1.02, + "learning_rate": 3.52293327340875e-05, + "loss": 4.0177, + "step": 902144 + }, + { + "epoch": 1.02, + "learning_rate": 3.522094678657698e-05, + "loss": 4.0047, + "step": 902656 + }, + { + "epoch": 1.02, + "learning_rate": 3.521256083906646e-05, + "loss": 4.015, + "step": 903168 + }, + { + "epoch": 1.02, + "learning_rate": 3.520417489155594e-05, + "loss": 4.0093, + "step": 903680 + }, + { + "epoch": 1.02, + "learning_rate": 3.5195821701652885e-05, + "loss": 4.0213, + "step": 904192 + }, + { + "epoch": 1.02, + "learning_rate": 3.5187435754142365e-05, + "loss": 4.0043, + "step": 904704 + }, + { + "epoch": 1.02, + "learning_rate": 3.5179049806631844e-05, + "loss": 4.0155, + "step": 905216 + }, + { + "epoch": 1.02, + "learning_rate": 3.5170663859121324e-05, + "loss": 4.0098, + "step": 905728 + }, + { + "epoch": 1.02, + "learning_rate": 3.5162277911610804e-05, + "loss": 4.0217, + "step": 906240 + }, + { + "epoch": 1.02, + "learning_rate": 3.5153891964100284e-05, + "loss": 4.0223, + "step": 906752 + }, + { + "epoch": 1.02, + "learning_rate": 3.5145506016589764e-05, + "loss": 4.0169, + "step": 907264 + }, + { + "epoch": 1.02, + "learning_rate": 3.5137120069079244e-05, + "loss": 4.0103, + "step": 907776 + }, + { + "epoch": 1.02, + "learning_rate": 3.512875050037245e-05, + "loss": 4.0147, + "step": 908288 + }, + { + "epoch": 1.02, + "learning_rate": 3.512036455286193e-05, + "loss": 4.0124, + "step": 908800 + }, + { + "epoch": 1.02, + "learning_rate": 3.511197860535141e-05, + "loss": 4.0104, + "step": 909312 + }, + { + "epoch": 1.02, + "learning_rate": 3.510359265784089e-05, + "loss": 4.0232, + "step": 909824 + }, + { + "epoch": 1.02, + "learning_rate": 3.509520671033037e-05, + "loss": 4.0185, + "step": 910336 + }, + { + "epoch": 1.02, + "learning_rate": 3.508682076281985e-05, + "loss": 4.0221, + "step": 910848 + }, + { + "epoch": 1.02, + "learning_rate": 3.507843481530933e-05, + "loss": 4.0186, + "step": 911360 + }, + { + "epoch": 1.02, + "learning_rate": 3.507004886779881e-05, + "loss": 4.0183, + "step": 911872 + }, + { + "epoch": 1.02, + "learning_rate": 3.506167929909203e-05, + "loss": 4.0063, + "step": 912384 + }, + { + "epoch": 1.02, + "learning_rate": 3.505329335158151e-05, + "loss": 4.0095, + "step": 912896 + }, + { + "epoch": 1.02, + "learning_rate": 3.504490740407099e-05, + "loss": 4.0104, + "step": 913408 + }, + { + "epoch": 1.02, + "learning_rate": 3.503652145656047e-05, + "loss": 4.0159, + "step": 913920 + }, + { + "epoch": 1.02, + "learning_rate": 3.502815188785368e-05, + "loss": 4.0134, + "step": 914432 + }, + { + "epoch": 1.02, + "learning_rate": 3.501976594034316e-05, + "loss": 3.9974, + "step": 914944 + }, + { + "epoch": 1.02, + "learning_rate": 3.501137999283264e-05, + "loss": 4.0153, + "step": 915456 + }, + { + "epoch": 1.03, + "eval_loss": 4.053233623504639, + "eval_runtime": 542.9356, + "eval_samples_per_second": 702.829, + "eval_steps_per_second": 21.964, + "step": 915828 + }, + { + "epoch": 0.0, + "learning_rate": 3.500299404532212e-05, + "loss": 4.029, + "step": 915968 + }, + { + "epoch": 0.0, + "learning_rate": 3.49946080978116e-05, + "loss": 4.0177, + "step": 916480 + }, + { + "epoch": 0.0, + "learning_rate": 3.498622215030108e-05, + "loss": 4.0173, + "step": 916992 + }, + { + "epoch": 0.0, + "learning_rate": 3.497783620279056e-05, + "loss": 4.0077, + "step": 917504 + }, + { + "epoch": 0.0, + "learning_rate": 3.496945025528003e-05, + "loss": 4.0158, + "step": 918016 + }, + { + "epoch": 0.0, + "learning_rate": 3.496106430776952e-05, + "loss": 4.0119, + "step": 918528 + }, + { + "epoch": 0.0, + "learning_rate": 3.4952678360259e-05, + "loss": 4.0036, + "step": 919040 + }, + { + "epoch": 0.0, + "learning_rate": 3.494429241274848e-05, + "loss": 4.0081, + "step": 919552 + }, + { + "epoch": 0.0, + "learning_rate": 3.493590646523796e-05, + "loss": 4.0078, + "step": 920064 + }, + { + "epoch": 0.0, + "learning_rate": 3.492752051772744e-05, + "loss": 4.0113, + "step": 920576 + }, + { + "epoch": 0.0, + "learning_rate": 3.491913457021692e-05, + "loss": 4.0146, + "step": 921088 + }, + { + "epoch": 0.0, + "learning_rate": 3.49107486227064e-05, + "loss": 4.0091, + "step": 921600 + }, + { + "epoch": 0.0, + "learning_rate": 3.490236267519588e-05, + "loss": 4.0156, + "step": 922112 + }, + { + "epoch": 0.0, + "learning_rate": 3.489397672768536e-05, + "loss": 4.0042, + "step": 922624 + }, + { + "epoch": 0.0, + "learning_rate": 3.488559078017484e-05, + "loss": 4.0129, + "step": 923136 + }, + { + "epoch": 0.0, + "learning_rate": 3.4877221211468046e-05, + "loss": 4.0011, + "step": 923648 + }, + { + "epoch": 0.0, + "learning_rate": 3.4868835263957526e-05, + "loss": 4.0085, + "step": 924160 + }, + { + "epoch": 0.0, + "learning_rate": 3.4860449316447006e-05, + "loss": 4.0063, + "step": 924672 + }, + { + "epoch": 0.0, + "learning_rate": 3.4852063368936486e-05, + "loss": 4.0073, + "step": 925184 + }, + { + "epoch": 0.0, + "learning_rate": 3.4843677421425966e-05, + "loss": 4.0187, + "step": 925696 + }, + { + "epoch": 0.0, + "learning_rate": 3.4835291473915446e-05, + "loss": 4.0032, + "step": 926208 + }, + { + "epoch": 0.0, + "learning_rate": 3.4826905526404926e-05, + "loss": 4.0243, + "step": 926720 + }, + { + "epoch": 0.0, + "learning_rate": 3.4818519578894406e-05, + "loss": 4.0101, + "step": 927232 + }, + { + "epoch": 0.0, + "learning_rate": 3.4810133631383886e-05, + "loss": 4.0149, + "step": 927744 + }, + { + "epoch": 0.0, + "learning_rate": 3.4801747683873366e-05, + "loss": 4.0057, + "step": 928256 + }, + { + "epoch": 0.0, + "learning_rate": 3.4793378115166575e-05, + "loss": 4.0189, + "step": 928768 + }, + { + "epoch": 0.0, + "learning_rate": 3.4784992167656055e-05, + "loss": 4.0128, + "step": 929280 + }, + { + "epoch": 0.0, + "learning_rate": 3.4776606220145535e-05, + "loss": 4.0046, + "step": 929792 + }, + { + "epoch": 0.0, + "learning_rate": 3.4768220272635015e-05, + "loss": 4.0075, + "step": 930304 + }, + { + "epoch": 0.0, + "learning_rate": 3.4759834325124495e-05, + "loss": 4.0149, + "step": 930816 + }, + { + "epoch": 0.01, + "learning_rate": 3.4751448377613975e-05, + "loss": 3.9985, + "step": 931328 + }, + { + "epoch": 0.01, + "learning_rate": 3.4743062430103455e-05, + "loss": 4.0067, + "step": 931840 + }, + { + "epoch": 0.01, + "learning_rate": 3.4734676482592935e-05, + "loss": 4.0176, + "step": 932352 + }, + { + "epoch": 0.01, + "learning_rate": 3.472630691388615e-05, + "loss": 4.0131, + "step": 932864 + }, + { + "epoch": 0.01, + "learning_rate": 3.471792096637563e-05, + "loss": 3.9995, + "step": 933376 + }, + { + "epoch": 0.01, + "learning_rate": 3.470955139766884e-05, + "loss": 4.0098, + "step": 933888 + }, + { + "epoch": 0.01, + "learning_rate": 3.470118182896205e-05, + "loss": 4.0142, + "step": 934400 + }, + { + "epoch": 0.01, + "learning_rate": 3.469279588145153e-05, + "loss": 4.012, + "step": 934912 + }, + { + "epoch": 0.01, + "learning_rate": 3.468440993394101e-05, + "loss": 4.0053, + "step": 935424 + }, + { + "epoch": 0.01, + "learning_rate": 3.467602398643049e-05, + "loss": 4.0071, + "step": 935936 + }, + { + "epoch": 0.01, + "learning_rate": 3.466763803891997e-05, + "loss": 4.0037, + "step": 936448 + }, + { + "epoch": 0.01, + "learning_rate": 3.465925209140945e-05, + "loss": 3.9884, + "step": 936960 + }, + { + "epoch": 0.01, + "learning_rate": 3.465086614389893e-05, + "loss": 4.0129, + "step": 937472 + }, + { + "epoch": 0.01, + "learning_rate": 3.464248019638841e-05, + "loss": 4.0064, + "step": 937984 + }, + { + "epoch": 0.01, + "learning_rate": 3.463409424887789e-05, + "loss": 3.9977, + "step": 938496 + }, + { + "epoch": 0.01, + "learning_rate": 3.462570830136737e-05, + "loss": 4.0105, + "step": 939008 + }, + { + "epoch": 0.01, + "learning_rate": 3.4617338732660584e-05, + "loss": 3.9998, + "step": 939520 + }, + { + "epoch": 0.01, + "learning_rate": 3.4608952785150064e-05, + "loss": 4.0157, + "step": 940032 + }, + { + "epoch": 0.01, + "learning_rate": 3.4600566837639544e-05, + "loss": 4.01, + "step": 940544 + }, + { + "epoch": 0.01, + "learning_rate": 3.4592180890129024e-05, + "loss": 3.9946, + "step": 941056 + }, + { + "epoch": 0.01, + "learning_rate": 3.458381132142223e-05, + "loss": 4.0077, + "step": 941568 + }, + { + "epoch": 0.01, + "learning_rate": 3.457542537391171e-05, + "loss": 4.0029, + "step": 942080 + }, + { + "epoch": 0.01, + "learning_rate": 3.456703942640119e-05, + "loss": 4.0123, + "step": 942592 + }, + { + "epoch": 0.01, + "learning_rate": 3.455865347889067e-05, + "loss": 3.9894, + "step": 943104 + }, + { + "epoch": 0.01, + "learning_rate": 3.455026753138015e-05, + "loss": 4.0109, + "step": 943616 + }, + { + "epoch": 0.01, + "learning_rate": 3.454188158386963e-05, + "loss": 3.9955, + "step": 944128 + }, + { + "epoch": 0.01, + "learning_rate": 3.4533495636359106e-05, + "loss": 4.0055, + "step": 944640 + }, + { + "epoch": 0.01, + "learning_rate": 3.4525109688848586e-05, + "loss": 4.003, + "step": 945152 + }, + { + "epoch": 0.01, + "learning_rate": 3.451675649894554e-05, + "loss": 4.0112, + "step": 945664 + }, + { + "epoch": 0.01, + "learning_rate": 3.450837055143502e-05, + "loss": 3.9959, + "step": 946176 + }, + { + "epoch": 0.01, + "learning_rate": 3.44999846039245e-05, + "loss": 3.9921, + "step": 946688 + }, + { + "epoch": 0.01, + "learning_rate": 3.449159865641398e-05, + "loss": 4.0019, + "step": 947200 + }, + { + "epoch": 0.01, + "learning_rate": 3.448321270890346e-05, + "loss": 3.9975, + "step": 947712 + }, + { + "epoch": 0.01, + "learning_rate": 3.447482676139293e-05, + "loss": 4.0031, + "step": 948224 + }, + { + "epoch": 0.01, + "learning_rate": 3.446644081388241e-05, + "loss": 4.0078, + "step": 948736 + }, + { + "epoch": 0.01, + "learning_rate": 3.445805486637189e-05, + "loss": 4.0048, + "step": 949248 + }, + { + "epoch": 0.01, + "learning_rate": 3.444966891886137e-05, + "loss": 3.9879, + "step": 949760 + }, + { + "epoch": 0.01, + "learning_rate": 3.444128297135085e-05, + "loss": 4.0004, + "step": 950272 + }, + { + "epoch": 0.01, + "learning_rate": 3.443289702384033e-05, + "loss": 3.9994, + "step": 950784 + }, + { + "epoch": 0.01, + "learning_rate": 3.442451107632981e-05, + "loss": 3.9923, + "step": 951296 + }, + { + "epoch": 0.01, + "learning_rate": 3.4416141507623026e-05, + "loss": 3.9989, + "step": 951808 + }, + { + "epoch": 0.01, + "learning_rate": 3.4407755560112506e-05, + "loss": 4.0075, + "step": 952320 + }, + { + "epoch": 0.01, + "learning_rate": 3.4399385991405715e-05, + "loss": 3.9971, + "step": 952832 + }, + { + "epoch": 0.01, + "learning_rate": 3.4391000043895195e-05, + "loss": 4.0087, + "step": 953344 + }, + { + "epoch": 0.01, + "learning_rate": 3.4382614096384675e-05, + "loss": 4.012, + "step": 953856 + }, + { + "epoch": 0.01, + "learning_rate": 3.4374228148874155e-05, + "loss": 4.01, + "step": 954368 + }, + { + "epoch": 0.01, + "learning_rate": 3.4365842201363635e-05, + "loss": 4.0102, + "step": 954880 + }, + { + "epoch": 0.01, + "learning_rate": 3.4357472632656844e-05, + "loss": 3.9943, + "step": 955392 + }, + { + "epoch": 0.01, + "learning_rate": 3.4349086685146324e-05, + "loss": 3.9957, + "step": 955904 + }, + { + "epoch": 0.01, + "learning_rate": 3.4340700737635804e-05, + "loss": 4.0053, + "step": 956416 + }, + { + "epoch": 0.01, + "learning_rate": 3.433233116892901e-05, + "loss": 4.0084, + "step": 956928 + }, + { + "epoch": 0.01, + "learning_rate": 3.432394522141849e-05, + "loss": 4.0019, + "step": 957440 + }, + { + "epoch": 0.01, + "learning_rate": 3.431555927390798e-05, + "loss": 3.9828, + "step": 957952 + }, + { + "epoch": 0.01, + "learning_rate": 3.430717332639746e-05, + "loss": 3.9971, + "step": 958464 + }, + { + "epoch": 0.01, + "learning_rate": 3.429878737888694e-05, + "loss": 3.9882, + "step": 958976 + }, + { + "epoch": 0.01, + "learning_rate": 3.429040143137642e-05, + "loss": 4.0069, + "step": 959488 + }, + { + "epoch": 0.01, + "learning_rate": 3.42820154838659e-05, + "loss": 3.9991, + "step": 960000 + }, + { + "epoch": 0.01, + "learning_rate": 3.427362953635538e-05, + "loss": 4.0119, + "step": 960512 + }, + { + "epoch": 0.01, + "learning_rate": 3.426524358884486e-05, + "loss": 4.0007, + "step": 961024 + }, + { + "epoch": 0.01, + "learning_rate": 3.425687402013807e-05, + "loss": 3.9987, + "step": 961536 + }, + { + "epoch": 0.02, + "learning_rate": 3.424848807262755e-05, + "loss": 3.9917, + "step": 962048 + }, + { + "epoch": 0.02, + "learning_rate": 3.424010212511703e-05, + "loss": 3.9975, + "step": 962560 + }, + { + "epoch": 0.02, + "learning_rate": 3.423171617760651e-05, + "loss": 3.9973, + "step": 963072 + }, + { + "epoch": 0.02, + "learning_rate": 3.422333023009599e-05, + "loss": 3.9951, + "step": 963584 + }, + { + "epoch": 0.02, + "learning_rate": 3.421494428258547e-05, + "loss": 3.9903, + "step": 964096 + }, + { + "epoch": 0.02, + "learning_rate": 3.420655833507495e-05, + "loss": 3.9988, + "step": 964608 + }, + { + "epoch": 0.02, + "learning_rate": 3.419817238756443e-05, + "loss": 4.0004, + "step": 965120 + }, + { + "epoch": 0.02, + "learning_rate": 3.418978644005391e-05, + "loss": 4.0089, + "step": 965632 + }, + { + "epoch": 0.02, + "learning_rate": 3.4181416871347124e-05, + "loss": 4.0041, + "step": 966144 + }, + { + "epoch": 0.02, + "learning_rate": 3.41730309238366e-05, + "loss": 4.005, + "step": 966656 + }, + { + "epoch": 0.02, + "learning_rate": 3.416464497632608e-05, + "loss": 4.002, + "step": 967168 + }, + { + "epoch": 0.02, + "learning_rate": 3.415625902881556e-05, + "loss": 4.0013, + "step": 967680 + }, + { + "epoch": 0.02, + "learning_rate": 3.4147889460108767e-05, + "loss": 3.9997, + "step": 968192 + }, + { + "epoch": 0.02, + "learning_rate": 3.4139503512598247e-05, + "loss": 4.0088, + "step": 968704 + }, + { + "epoch": 0.02, + "learning_rate": 3.413113394389146e-05, + "loss": 3.9988, + "step": 969216 + }, + { + "epoch": 0.02, + "learning_rate": 3.412274799638094e-05, + "loss": 4.0036, + "step": 969728 + }, + { + "epoch": 0.02, + "learning_rate": 3.4114362048870416e-05, + "loss": 3.9896, + "step": 970240 + }, + { + "epoch": 0.02, + "learning_rate": 3.4105976101359895e-05, + "loss": 3.9844, + "step": 970752 + }, + { + "epoch": 0.02, + "learning_rate": 3.409760653265311e-05, + "loss": 3.9963, + "step": 971264 + }, + { + "epoch": 0.02, + "learning_rate": 3.40892205851426e-05, + "loss": 4.0036, + "step": 971776 + }, + { + "epoch": 0.02, + "learning_rate": 3.408083463763207e-05, + "loss": 3.9988, + "step": 972288 + }, + { + "epoch": 0.02, + "learning_rate": 3.407244869012155e-05, + "loss": 3.9783, + "step": 972800 + }, + { + "epoch": 0.02, + "learning_rate": 3.406406274261103e-05, + "loss": 3.9825, + "step": 973312 + }, + { + "epoch": 0.02, + "learning_rate": 3.405567679510051e-05, + "loss": 4.0063, + "step": 973824 + }, + { + "epoch": 0.02, + "learning_rate": 3.404729084758999e-05, + "loss": 4.0052, + "step": 974336 + }, + { + "epoch": 0.02, + "learning_rate": 3.403890490007947e-05, + "loss": 4.006, + "step": 974848 + }, + { + "epoch": 0.02, + "learning_rate": 3.403053533137268e-05, + "loss": 4.0033, + "step": 975360 + }, + { + "epoch": 0.02, + "learning_rate": 3.402214938386216e-05, + "loss": 4.0017, + "step": 975872 + }, + { + "epoch": 0.02, + "learning_rate": 3.401376343635164e-05, + "loss": 4.0008, + "step": 976384 + }, + { + "epoch": 0.02, + "learning_rate": 3.400537748884112e-05, + "loss": 3.9972, + "step": 976896 + }, + { + "epoch": 0.02, + "learning_rate": 3.39969915413306e-05, + "loss": 3.9888, + "step": 977408 + }, + { + "epoch": 0.02, + "learning_rate": 3.398860559382008e-05, + "loss": 3.9924, + "step": 977920 + }, + { + "epoch": 0.02, + "learning_rate": 3.398021964630957e-05, + "loss": 4.0033, + "step": 978432 + }, + { + "epoch": 0.02, + "learning_rate": 3.397183369879905e-05, + "loss": 3.9886, + "step": 978944 + }, + { + "epoch": 0.02, + "learning_rate": 3.3963447751288527e-05, + "loss": 3.9968, + "step": 979456 + }, + { + "epoch": 0.02, + "learning_rate": 3.3955078182581736e-05, + "loss": 3.9972, + "step": 979968 + }, + { + "epoch": 0.02, + "learning_rate": 3.3946692235071216e-05, + "loss": 4.0023, + "step": 980480 + }, + { + "epoch": 0.02, + "learning_rate": 3.3938306287560696e-05, + "loss": 3.9901, + "step": 980992 + }, + { + "epoch": 0.02, + "learning_rate": 3.3929920340050176e-05, + "loss": 4.0001, + "step": 981504 + }, + { + "epoch": 0.02, + "learning_rate": 3.3921550771343385e-05, + "loss": 3.9983, + "step": 982016 + }, + { + "epoch": 0.02, + "learning_rate": 3.3913164823832865e-05, + "loss": 3.9991, + "step": 982528 + }, + { + "epoch": 0.02, + "learning_rate": 3.3904795255126074e-05, + "loss": 4.0073, + "step": 983040 + }, + { + "epoch": 0.02, + "learning_rate": 3.3896409307615554e-05, + "loss": 4.0057, + "step": 983552 + }, + { + "epoch": 0.02, + "learning_rate": 3.3888023360105034e-05, + "loss": 3.9958, + "step": 984064 + }, + { + "epoch": 0.02, + "learning_rate": 3.387963741259452e-05, + "loss": 3.9961, + "step": 984576 + }, + { + "epoch": 0.02, + "learning_rate": 3.3871251465084e-05, + "loss": 3.9983, + "step": 985088 + }, + { + "epoch": 0.02, + "learning_rate": 3.386286551757348e-05, + "loss": 3.9897, + "step": 985600 + }, + { + "epoch": 0.02, + "learning_rate": 3.3854479570062953e-05, + "loss": 4.0081, + "step": 986112 + }, + { + "epoch": 0.02, + "learning_rate": 3.384609362255243e-05, + "loss": 4.004, + "step": 986624 + }, + { + "epoch": 0.02, + "learning_rate": 3.383772405384565e-05, + "loss": 4.0084, + "step": 987136 + }, + { + "epoch": 0.02, + "learning_rate": 3.382933810633513e-05, + "loss": 4.001, + "step": 987648 + }, + { + "epoch": 0.02, + "learning_rate": 3.38209521588246e-05, + "loss": 4.0046, + "step": 988160 + }, + { + "epoch": 0.02, + "learning_rate": 3.381256621131408e-05, + "loss": 3.9889, + "step": 988672 + }, + { + "epoch": 0.02, + "learning_rate": 3.38041966426073e-05, + "loss": 3.9953, + "step": 989184 + }, + { + "epoch": 0.02, + "learning_rate": 3.379582707390051e-05, + "loss": 3.9931, + "step": 989696 + }, + { + "epoch": 0.02, + "learning_rate": 3.378744112638999e-05, + "loss": 4.0029, + "step": 990208 + }, + { + "epoch": 0.02, + "learning_rate": 3.3779055178879474e-05, + "loss": 3.9997, + "step": 990720 + }, + { + "epoch": 0.02, + "learning_rate": 3.3770669231368954e-05, + "loss": 3.981, + "step": 991232 + }, + { + "epoch": 0.02, + "learning_rate": 3.376228328385843e-05, + "loss": 3.9936, + "step": 991744 + }, + { + "epoch": 0.03, + "eval_loss": 4.043963432312012, + "eval_runtime": 541.3915, + "eval_samples_per_second": 704.834, + "eval_steps_per_second": 22.027, + "step": 992147 + }, + { + "epoch": 0.0, + "learning_rate": 3.375389733634791e-05, + "loss": 4.0165, + "step": 992256 + }, + { + "epoch": 0.0, + "learning_rate": 3.374551138883739e-05, + "loss": 4.0004, + "step": 992768 + }, + { + "epoch": 0.0, + "learning_rate": 3.373712544132687e-05, + "loss": 4.005, + "step": 993280 + }, + { + "epoch": 0.0, + "learning_rate": 3.372873949381635e-05, + "loss": 3.9927, + "step": 993792 + }, + { + "epoch": 0.0, + "learning_rate": 3.372035354630583e-05, + "loss": 4.0009, + "step": 994304 + }, + { + "epoch": 0.0, + "learning_rate": 3.371196759879531e-05, + "loss": 4.0007, + "step": 994816 + }, + { + "epoch": 0.0, + "learning_rate": 3.370358165128479e-05, + "loss": 3.9811, + "step": 995328 + }, + { + "epoch": 0.0, + "learning_rate": 3.369519570377427e-05, + "loss": 3.9962, + "step": 995840 + }, + { + "epoch": 0.0, + "learning_rate": 3.368680975626375e-05, + "loss": 3.9914, + "step": 996352 + }, + { + "epoch": 0.0, + "learning_rate": 3.367842380875323e-05, + "loss": 3.9893, + "step": 996864 + }, + { + "epoch": 0.0, + "learning_rate": 3.367003786124271e-05, + "loss": 3.9997, + "step": 997376 + }, + { + "epoch": 0.0, + "learning_rate": 3.3661651913732187e-05, + "loss": 3.9986, + "step": 997888 + }, + { + "epoch": 0.0, + "learning_rate": 3.3653265966221667e-05, + "loss": 3.9913, + "step": 998400 + }, + { + "epoch": 0.0, + "learning_rate": 3.3644880018711147e-05, + "loss": 3.9933, + "step": 998912 + }, + { + "epoch": 0.0, + "learning_rate": 3.3636494071200626e-05, + "loss": 3.9971, + "step": 999424 + }, + { + "epoch": 0.0, + "learning_rate": 3.3628108123690106e-05, + "loss": 3.9917, + "step": 999936 + }, + { + "epoch": 0.0, + "learning_rate": 3.3619722176179586e-05, + "loss": 3.9861, + "step": 1000448 + }, + { + "epoch": 0.0, + "learning_rate": 3.3611336228669066e-05, + "loss": 3.9892, + "step": 1000960 + }, + { + "epoch": 0.0, + "learning_rate": 3.3602950281158546e-05, + "loss": 3.9918, + "step": 1001472 + }, + { + "epoch": 0.0, + "learning_rate": 3.3594580712451755e-05, + "loss": 4.0001, + "step": 1001984 + }, + { + "epoch": 0.0, + "learning_rate": 3.3586194764941235e-05, + "loss": 3.9926, + "step": 1002496 + }, + { + "epoch": 0.0, + "learning_rate": 3.3577808817430715e-05, + "loss": 4.0057, + "step": 1003008 + }, + { + "epoch": 0.0, + "learning_rate": 3.3569422869920195e-05, + "loss": 3.9924, + "step": 1003520 + }, + { + "epoch": 0.0, + "learning_rate": 3.3561036922409675e-05, + "loss": 4.0021, + "step": 1004032 + }, + { + "epoch": 0.0, + "learning_rate": 3.3552650974899155e-05, + "loss": 3.9928, + "step": 1004544 + }, + { + "epoch": 0.0, + "learning_rate": 3.3544265027388635e-05, + "loss": 4.001, + "step": 1005056 + }, + { + "epoch": 0.0, + "learning_rate": 3.353587907987812e-05, + "loss": 3.9986, + "step": 1005568 + }, + { + "epoch": 0.0, + "learning_rate": 3.352752588997506e-05, + "loss": 3.9913, + "step": 1006080 + }, + { + "epoch": 0.0, + "learning_rate": 3.351913994246454e-05, + "loss": 3.9923, + "step": 1006592 + }, + { + "epoch": 0.0, + "learning_rate": 3.351075399495402e-05, + "loss": 3.9993, + "step": 1007104 + }, + { + "epoch": 0.01, + "learning_rate": 3.35023680474435e-05, + "loss": 3.9826, + "step": 1007616 + }, + { + "epoch": 0.01, + "learning_rate": 3.349398209993298e-05, + "loss": 3.9928, + "step": 1008128 + }, + { + "epoch": 0.01, + "learning_rate": 3.348559615242246e-05, + "loss": 4.0013, + "step": 1008640 + }, + { + "epoch": 0.01, + "learning_rate": 3.347721020491194e-05, + "loss": 4.0001, + "step": 1009152 + }, + { + "epoch": 0.01, + "learning_rate": 3.346882425740142e-05, + "loss": 3.982, + "step": 1009664 + }, + { + "epoch": 0.01, + "learning_rate": 3.346045468869463e-05, + "loss": 3.9949, + "step": 1010176 + }, + { + "epoch": 0.01, + "learning_rate": 3.3452085119987845e-05, + "loss": 4.0005, + "step": 1010688 + }, + { + "epoch": 0.01, + "learning_rate": 3.3443699172477325e-05, + "loss": 3.9958, + "step": 1011200 + }, + { + "epoch": 0.01, + "learning_rate": 3.3435313224966805e-05, + "loss": 3.99, + "step": 1011712 + }, + { + "epoch": 0.01, + "learning_rate": 3.3426927277456285e-05, + "loss": 3.9932, + "step": 1012224 + }, + { + "epoch": 0.01, + "learning_rate": 3.3418541329945765e-05, + "loss": 3.9911, + "step": 1012736 + }, + { + "epoch": 0.01, + "learning_rate": 3.3410155382435245e-05, + "loss": 3.9724, + "step": 1013248 + }, + { + "epoch": 0.01, + "learning_rate": 3.3401769434924724e-05, + "loss": 3.9971, + "step": 1013760 + }, + { + "epoch": 0.01, + "learning_rate": 3.3393399866217934e-05, + "loss": 3.9895, + "step": 1014272 + }, + { + "epoch": 0.01, + "learning_rate": 3.3385013918707414e-05, + "loss": 3.9851, + "step": 1014784 + }, + { + "epoch": 0.01, + "learning_rate": 3.3376627971196893e-05, + "loss": 3.993, + "step": 1015296 + }, + { + "epoch": 0.01, + "learning_rate": 3.3368242023686373e-05, + "loss": 3.9845, + "step": 1015808 + }, + { + "epoch": 0.01, + "learning_rate": 3.3359856076175853e-05, + "loss": 4.0021, + "step": 1016320 + }, + { + "epoch": 0.01, + "learning_rate": 3.335148650746906e-05, + "loss": 3.9944, + "step": 1016832 + }, + { + "epoch": 0.01, + "learning_rate": 3.334310055995854e-05, + "loss": 3.9818, + "step": 1017344 + }, + { + "epoch": 0.01, + "learning_rate": 3.333471461244803e-05, + "loss": 3.9935, + "step": 1017856 + }, + { + "epoch": 0.01, + "learning_rate": 3.33263286649375e-05, + "loss": 3.9858, + "step": 1018368 + }, + { + "epoch": 0.01, + "learning_rate": 3.331795909623072e-05, + "loss": 3.996, + "step": 1018880 + }, + { + "epoch": 0.01, + "learning_rate": 3.33095731487202e-05, + "loss": 3.9736, + "step": 1019392 + }, + { + "epoch": 0.01, + "learning_rate": 3.330118720120968e-05, + "loss": 3.9959, + "step": 1019904 + }, + { + "epoch": 0.01, + "learning_rate": 3.329280125369915e-05, + "loss": 3.9785, + "step": 1020416 + }, + { + "epoch": 0.01, + "learning_rate": 3.328441530618863e-05, + "loss": 3.9905, + "step": 1020928 + }, + { + "epoch": 0.01, + "learning_rate": 3.327602935867811e-05, + "loss": 3.9925, + "step": 1021440 + }, + { + "epoch": 0.01, + "learning_rate": 3.326764341116759e-05, + "loss": 3.9913, + "step": 1021952 + }, + { + "epoch": 0.01, + "learning_rate": 3.325925746365707e-05, + "loss": 3.9862, + "step": 1022464 + }, + { + "epoch": 0.01, + "learning_rate": 3.325088789495028e-05, + "loss": 3.9771, + "step": 1022976 + }, + { + "epoch": 0.01, + "learning_rate": 3.3242518326243496e-05, + "loss": 3.9859, + "step": 1023488 + }, + { + "epoch": 0.01, + "learning_rate": 3.3234132378732976e-05, + "loss": 3.9843, + "step": 1024000 + }, + { + "epoch": 0.01, + "learning_rate": 3.3225746431222456e-05, + "loss": 3.9848, + "step": 1024512 + }, + { + "epoch": 0.01, + "learning_rate": 3.3217360483711936e-05, + "loss": 3.9927, + "step": 1025024 + }, + { + "epoch": 0.01, + "learning_rate": 3.3208974536201416e-05, + "loss": 3.9863, + "step": 1025536 + }, + { + "epoch": 0.01, + "learning_rate": 3.3200588588690896e-05, + "loss": 3.9774, + "step": 1026048 + }, + { + "epoch": 0.01, + "learning_rate": 3.3192202641180376e-05, + "loss": 3.9817, + "step": 1026560 + }, + { + "epoch": 0.01, + "learning_rate": 3.3183816693669856e-05, + "loss": 3.9894, + "step": 1027072 + }, + { + "epoch": 0.01, + "learning_rate": 3.3175447124963065e-05, + "loss": 3.9753, + "step": 1027584 + }, + { + "epoch": 0.01, + "learning_rate": 3.3167077556256274e-05, + "loss": 3.984, + "step": 1028096 + }, + { + "epoch": 0.01, + "learning_rate": 3.3158691608745754e-05, + "loss": 3.9964, + "step": 1028608 + }, + { + "epoch": 0.01, + "learning_rate": 3.3150305661235234e-05, + "loss": 3.9809, + "step": 1029120 + }, + { + "epoch": 0.01, + "learning_rate": 3.314191971372472e-05, + "loss": 3.9925, + "step": 1029632 + }, + { + "epoch": 0.01, + "learning_rate": 3.31335337662142e-05, + "loss": 3.999, + "step": 1030144 + }, + { + "epoch": 0.01, + "learning_rate": 3.312514781870368e-05, + "loss": 3.9968, + "step": 1030656 + }, + { + "epoch": 0.01, + "learning_rate": 3.311676187119316e-05, + "loss": 3.9958, + "step": 1031168 + }, + { + "epoch": 0.01, + "learning_rate": 3.310839230248637e-05, + "loss": 3.978, + "step": 1031680 + }, + { + "epoch": 0.01, + "learning_rate": 3.310000635497585e-05, + "loss": 3.979, + "step": 1032192 + }, + { + "epoch": 0.01, + "learning_rate": 3.309162040746533e-05, + "loss": 3.9875, + "step": 1032704 + }, + { + "epoch": 0.01, + "learning_rate": 3.308325083875854e-05, + "loss": 3.9944, + "step": 1033216 + }, + { + "epoch": 0.01, + "learning_rate": 3.307486489124802e-05, + "loss": 3.9894, + "step": 1033728 + }, + { + "epoch": 0.01, + "learning_rate": 3.30664789437375e-05, + "loss": 3.9715, + "step": 1034240 + }, + { + "epoch": 0.01, + "learning_rate": 3.305809299622698e-05, + "loss": 3.9839, + "step": 1034752 + }, + { + "epoch": 0.01, + "learning_rate": 3.304970704871646e-05, + "loss": 3.9759, + "step": 1035264 + }, + { + "epoch": 0.01, + "learning_rate": 3.304133748000967e-05, + "loss": 3.9859, + "step": 1035776 + }, + { + "epoch": 0.01, + "learning_rate": 3.3032951532499154e-05, + "loss": 3.9855, + "step": 1036288 + }, + { + "epoch": 0.01, + "learning_rate": 3.3024565584988634e-05, + "loss": 3.9992, + "step": 1036800 + }, + { + "epoch": 0.01, + "learning_rate": 3.3016179637478114e-05, + "loss": 3.9858, + "step": 1037312 + }, + { + "epoch": 0.01, + "learning_rate": 3.3007793689967594e-05, + "loss": 3.9861, + "step": 1037824 + }, + { + "epoch": 0.02, + "learning_rate": 3.2999407742457074e-05, + "loss": 3.977, + "step": 1038336 + }, + { + "epoch": 0.02, + "learning_rate": 3.2991021794946554e-05, + "loss": 3.9788, + "step": 1038848 + }, + { + "epoch": 0.02, + "learning_rate": 3.2982635847436034e-05, + "loss": 3.9813, + "step": 1039360 + }, + { + "epoch": 0.02, + "learning_rate": 3.2974249899925514e-05, + "loss": 3.9872, + "step": 1039872 + }, + { + "epoch": 0.02, + "learning_rate": 3.296586395241499e-05, + "loss": 3.9727, + "step": 1040384 + }, + { + "epoch": 0.02, + "learning_rate": 3.295747800490447e-05, + "loss": 3.9851, + "step": 1040896 + }, + { + "epoch": 0.02, + "learning_rate": 3.294910843619768e-05, + "loss": 3.9865, + "step": 1041408 + }, + { + "epoch": 0.02, + "learning_rate": 3.294072248868716e-05, + "loss": 3.9909, + "step": 1041920 + }, + { + "epoch": 0.02, + "learning_rate": 3.2932336541176636e-05, + "loss": 3.9937, + "step": 1042432 + }, + { + "epoch": 0.02, + "learning_rate": 3.292395059366612e-05, + "loss": 3.9909, + "step": 1042944 + }, + { + "epoch": 0.02, + "learning_rate": 3.29155646461556e-05, + "loss": 3.9888, + "step": 1043456 + }, + { + "epoch": 0.02, + "learning_rate": 3.290717869864508e-05, + "loss": 3.9843, + "step": 1043968 + }, + { + "epoch": 0.02, + "learning_rate": 3.289879275113456e-05, + "loss": 3.9859, + "step": 1044480 + }, + { + "epoch": 0.02, + "learning_rate": 3.289040680362404e-05, + "loss": 4.0008, + "step": 1044992 + }, + { + "epoch": 0.02, + "learning_rate": 3.288203723491725e-05, + "loss": 3.9807, + "step": 1045504 + }, + { + "epoch": 0.02, + "learning_rate": 3.287365128740673e-05, + "loss": 3.9911, + "step": 1046016 + }, + { + "epoch": 0.02, + "learning_rate": 3.286526533989621e-05, + "loss": 3.9754, + "step": 1046528 + }, + { + "epoch": 0.02, + "learning_rate": 3.285687939238569e-05, + "loss": 3.9681, + "step": 1047040 + }, + { + "epoch": 0.02, + "learning_rate": 3.28485098236789e-05, + "loss": 3.9842, + "step": 1047552 + }, + { + "epoch": 0.02, + "learning_rate": 3.284012387616838e-05, + "loss": 3.9905, + "step": 1048064 + }, + { + "epoch": 0.02, + "learning_rate": 3.283173792865786e-05, + "loss": 3.9849, + "step": 1048576 + }, + { + "epoch": 0.02, + "learning_rate": 3.282335198114734e-05, + "loss": 3.9646, + "step": 1049088 + }, + { + "epoch": 0.02, + "learning_rate": 3.2814982412440556e-05, + "loss": 3.9722, + "step": 1049600 + }, + { + "epoch": 0.02, + "learning_rate": 3.2806596464930036e-05, + "loss": 3.9855, + "step": 1050112 + }, + { + "epoch": 0.02, + "learning_rate": 3.2798210517419516e-05, + "loss": 3.9914, + "step": 1050624 + }, + { + "epoch": 0.02, + "learning_rate": 3.2789824569908996e-05, + "loss": 3.9911, + "step": 1051136 + }, + { + "epoch": 0.02, + "learning_rate": 3.2781455001202205e-05, + "loss": 3.9858, + "step": 1051648 + }, + { + "epoch": 0.02, + "learning_rate": 3.2773085432495415e-05, + "loss": 3.9894, + "step": 1052160 + }, + { + "epoch": 0.02, + "learning_rate": 3.2764699484984894e-05, + "loss": 3.9855, + "step": 1052672 + }, + { + "epoch": 0.02, + "learning_rate": 3.2756313537474374e-05, + "loss": 3.9804, + "step": 1053184 + }, + { + "epoch": 0.02, + "learning_rate": 3.2747927589963854e-05, + "loss": 3.9806, + "step": 1053696 + }, + { + "epoch": 0.02, + "learning_rate": 3.2739541642453334e-05, + "loss": 3.9774, + "step": 1054208 + }, + { + "epoch": 0.02, + "learning_rate": 3.2731155694942814e-05, + "loss": 3.9877, + "step": 1054720 + }, + { + "epoch": 0.02, + "learning_rate": 3.2722769747432294e-05, + "loss": 3.9731, + "step": 1055232 + }, + { + "epoch": 0.02, + "learning_rate": 3.2714383799921774e-05, + "loss": 3.9875, + "step": 1055744 + }, + { + "epoch": 0.02, + "learning_rate": 3.270601423121499e-05, + "loss": 3.9812, + "step": 1056256 + }, + { + "epoch": 0.02, + "learning_rate": 3.269762828370447e-05, + "loss": 3.9932, + "step": 1056768 + }, + { + "epoch": 0.02, + "learning_rate": 3.268925871499768e-05, + "loss": 3.9715, + "step": 1057280 + }, + { + "epoch": 0.02, + "learning_rate": 3.268087276748716e-05, + "loss": 3.9851, + "step": 1057792 + }, + { + "epoch": 0.02, + "learning_rate": 3.267248681997664e-05, + "loss": 3.9836, + "step": 1058304 + }, + { + "epoch": 0.02, + "learning_rate": 3.266410087246612e-05, + "loss": 3.9877, + "step": 1058816 + }, + { + "epoch": 0.02, + "learning_rate": 3.26557149249556e-05, + "loss": 3.9935, + "step": 1059328 + }, + { + "epoch": 0.02, + "learning_rate": 3.264732897744508e-05, + "loss": 3.9929, + "step": 1059840 + }, + { + "epoch": 0.02, + "learning_rate": 3.263895940873829e-05, + "loss": 3.9774, + "step": 1060352 + }, + { + "epoch": 0.02, + "learning_rate": 3.263057346122777e-05, + "loss": 3.9879, + "step": 1060864 + }, + { + "epoch": 0.02, + "learning_rate": 3.262218751371725e-05, + "loss": 3.9814, + "step": 1061376 + }, + { + "epoch": 0.02, + "learning_rate": 3.261380156620673e-05, + "loss": 3.9762, + "step": 1061888 + }, + { + "epoch": 0.02, + "learning_rate": 3.260541561869621e-05, + "loss": 3.9972, + "step": 1062400 + }, + { + "epoch": 0.02, + "learning_rate": 3.2597029671185695e-05, + "loss": 3.9894, + "step": 1062912 + }, + { + "epoch": 0.02, + "learning_rate": 3.2588660102478904e-05, + "loss": 3.998, + "step": 1063424 + }, + { + "epoch": 0.02, + "learning_rate": 3.2580274154968384e-05, + "loss": 3.9848, + "step": 1063936 + }, + { + "epoch": 0.02, + "learning_rate": 3.2571888207457864e-05, + "loss": 3.9905, + "step": 1064448 + }, + { + "epoch": 0.02, + "learning_rate": 3.2563502259947344e-05, + "loss": 3.9719, + "step": 1064960 + }, + { + "epoch": 0.02, + "learning_rate": 3.2555116312436823e-05, + "loss": 3.985, + "step": 1065472 + }, + { + "epoch": 0.02, + "learning_rate": 3.25467303649263e-05, + "loss": 3.9782, + "step": 1065984 + }, + { + "epoch": 0.02, + "learning_rate": 3.2538344417415777e-05, + "loss": 3.9887, + "step": 1066496 + }, + { + "epoch": 0.02, + "learning_rate": 3.2529958469905257e-05, + "loss": 3.9835, + "step": 1067008 + }, + { + "epoch": 0.02, + "learning_rate": 3.2521588901198466e-05, + "loss": 3.9675, + "step": 1067520 + }, + { + "epoch": 0.02, + "learning_rate": 3.2513202953687946e-05, + "loss": 3.9834, + "step": 1068032 + }, + { + "epoch": 0.03, + "eval_loss": 4.036366939544678, + "eval_runtime": 556.0365, + "eval_samples_per_second": 686.27, + "eval_steps_per_second": 21.446, + "step": 1068466 + }, + { + "epoch": 1.0, + "learning_rate": 3.250481700617743e-05, + "loss": 3.9811, + "step": 1068544 + }, + { + "epoch": 1.0, + "learning_rate": 3.249644743747065e-05, + "loss": 3.9922, + "step": 1069056 + }, + { + "epoch": 1.0, + "learning_rate": 3.248806148996012e-05, + "loss": 3.9875, + "step": 1069568 + }, + { + "epoch": 1.0, + "learning_rate": 3.24796755424496e-05, + "loss": 3.9817, + "step": 1070080 + }, + { + "epoch": 1.0, + "learning_rate": 3.247128959493908e-05, + "loss": 3.9875, + "step": 1070592 + }, + { + "epoch": 1.0, + "learning_rate": 3.246290364742856e-05, + "loss": 3.986, + "step": 1071104 + }, + { + "epoch": 1.0, + "learning_rate": 3.245451769991804e-05, + "loss": 3.9695, + "step": 1071616 + }, + { + "epoch": 1.0, + "learning_rate": 3.244614813121125e-05, + "loss": 3.9806, + "step": 1072128 + }, + { + "epoch": 1.0, + "learning_rate": 3.243776218370073e-05, + "loss": 3.98, + "step": 1072640 + }, + { + "epoch": 1.0, + "learning_rate": 3.242937623619021e-05, + "loss": 3.9763, + "step": 1073152 + }, + { + "epoch": 1.0, + "learning_rate": 3.242099028867969e-05, + "loss": 3.9858, + "step": 1073664 + }, + { + "epoch": 1.0, + "learning_rate": 3.241260434116917e-05, + "loss": 3.9886, + "step": 1074176 + }, + { + "epoch": 1.0, + "learning_rate": 3.2404234772462386e-05, + "loss": 3.9755, + "step": 1074688 + }, + { + "epoch": 1.0, + "learning_rate": 3.2395848824951866e-05, + "loss": 3.9841, + "step": 1075200 + }, + { + "epoch": 1.0, + "learning_rate": 3.2387462877441346e-05, + "loss": 3.9803, + "step": 1075712 + }, + { + "epoch": 1.0, + "learning_rate": 3.2379076929930826e-05, + "loss": 3.9771, + "step": 1076224 + }, + { + "epoch": 1.0, + "learning_rate": 3.2370690982420306e-05, + "loss": 3.9764, + "step": 1076736 + }, + { + "epoch": 1.0, + "learning_rate": 3.2362305034909786e-05, + "loss": 3.9797, + "step": 1077248 + }, + { + "epoch": 1.0, + "learning_rate": 3.2353935466202995e-05, + "loss": 3.9821, + "step": 1077760 + }, + { + "epoch": 1.0, + "learning_rate": 3.2345549518692475e-05, + "loss": 3.9813, + "step": 1078272 + }, + { + "epoch": 1.0, + "learning_rate": 3.2337163571181955e-05, + "loss": 3.9828, + "step": 1078784 + }, + { + "epoch": 1.0, + "learning_rate": 3.2328777623671435e-05, + "loss": 3.9873, + "step": 1079296 + }, + { + "epoch": 1.0, + "learning_rate": 3.2320391676160915e-05, + "loss": 3.9801, + "step": 1079808 + }, + { + "epoch": 1.0, + "learning_rate": 3.2312005728650395e-05, + "loss": 3.9874, + "step": 1080320 + }, + { + "epoch": 1.0, + "learning_rate": 3.2303619781139875e-05, + "loss": 3.9817, + "step": 1080832 + }, + { + "epoch": 1.0, + "learning_rate": 3.2295233833629355e-05, + "loss": 3.991, + "step": 1081344 + }, + { + "epoch": 1.0, + "learning_rate": 3.228686426492257e-05, + "loss": 3.977, + "step": 1081856 + }, + { + "epoch": 1.0, + "learning_rate": 3.227849469621578e-05, + "loss": 3.9842, + "step": 1082368 + }, + { + "epoch": 1.0, + "learning_rate": 3.227010874870526e-05, + "loss": 3.9746, + "step": 1082880 + }, + { + "epoch": 1.0, + "learning_rate": 3.226172280119474e-05, + "loss": 3.9883, + "step": 1083392 + }, + { + "epoch": 1.01, + "learning_rate": 3.225333685368422e-05, + "loss": 3.9685, + "step": 1083904 + }, + { + "epoch": 1.01, + "learning_rate": 3.22449509061737e-05, + "loss": 3.9758, + "step": 1084416 + }, + { + "epoch": 1.01, + "learning_rate": 3.223656495866318e-05, + "loss": 3.9927, + "step": 1084928 + }, + { + "epoch": 1.01, + "learning_rate": 3.222817901115265e-05, + "loss": 3.9835, + "step": 1085440 + }, + { + "epoch": 1.01, + "learning_rate": 3.221979306364213e-05, + "loss": 3.969, + "step": 1085952 + }, + { + "epoch": 1.01, + "learning_rate": 3.221142349493535e-05, + "loss": 3.9823, + "step": 1086464 + }, + { + "epoch": 1.01, + "learning_rate": 3.220305392622856e-05, + "loss": 3.9857, + "step": 1086976 + }, + { + "epoch": 1.01, + "learning_rate": 3.219466797871804e-05, + "loss": 3.9813, + "step": 1087488 + }, + { + "epoch": 1.01, + "learning_rate": 3.2186282031207524e-05, + "loss": 3.9777, + "step": 1088000 + }, + { + "epoch": 1.01, + "learning_rate": 3.2177896083697004e-05, + "loss": 3.9807, + "step": 1088512 + }, + { + "epoch": 1.01, + "learning_rate": 3.2169510136186484e-05, + "loss": 3.9748, + "step": 1089024 + }, + { + "epoch": 1.01, + "learning_rate": 3.216112418867596e-05, + "loss": 3.963, + "step": 1089536 + }, + { + "epoch": 1.01, + "learning_rate": 3.215273824116544e-05, + "loss": 3.9794, + "step": 1090048 + }, + { + "epoch": 1.01, + "learning_rate": 3.214435229365492e-05, + "loss": 3.9789, + "step": 1090560 + }, + { + "epoch": 1.01, + "learning_rate": 3.213599910375186e-05, + "loss": 3.9749, + "step": 1091072 + }, + { + "epoch": 1.01, + "learning_rate": 3.212761315624134e-05, + "loss": 3.9762, + "step": 1091584 + }, + { + "epoch": 1.01, + "learning_rate": 3.211922720873082e-05, + "loss": 3.9718, + "step": 1092096 + }, + { + "epoch": 1.01, + "learning_rate": 3.21108412612203e-05, + "loss": 3.9864, + "step": 1092608 + }, + { + "epoch": 1.01, + "learning_rate": 3.2102455313709775e-05, + "loss": 3.9839, + "step": 1093120 + }, + { + "epoch": 1.01, + "learning_rate": 3.209406936619926e-05, + "loss": 3.9675, + "step": 1093632 + }, + { + "epoch": 1.01, + "learning_rate": 3.208568341868874e-05, + "loss": 3.9799, + "step": 1094144 + }, + { + "epoch": 1.01, + "learning_rate": 3.207729747117822e-05, + "loss": 3.9713, + "step": 1094656 + }, + { + "epoch": 1.01, + "learning_rate": 3.206892790247143e-05, + "loss": 3.986, + "step": 1095168 + }, + { + "epoch": 1.01, + "learning_rate": 3.206054195496091e-05, + "loss": 3.9644, + "step": 1095680 + }, + { + "epoch": 1.01, + "learning_rate": 3.205215600745039e-05, + "loss": 3.9831, + "step": 1096192 + }, + { + "epoch": 1.01, + "learning_rate": 3.20437864387436e-05, + "loss": 3.9634, + "step": 1096704 + }, + { + "epoch": 1.01, + "learning_rate": 3.203540049123308e-05, + "loss": 3.9734, + "step": 1097216 + }, + { + "epoch": 1.01, + "learning_rate": 3.202701454372256e-05, + "loss": 3.9848, + "step": 1097728 + }, + { + "epoch": 1.01, + "learning_rate": 3.201862859621204e-05, + "loss": 3.9753, + "step": 1098240 + }, + { + "epoch": 1.01, + "learning_rate": 3.201025902750525e-05, + "loss": 3.9743, + "step": 1098752 + }, + { + "epoch": 1.01, + "learning_rate": 3.200187307999473e-05, + "loss": 3.9629, + "step": 1099264 + }, + { + "epoch": 1.01, + "learning_rate": 3.1993487132484216e-05, + "loss": 3.9711, + "step": 1099776 + }, + { + "epoch": 1.01, + "learning_rate": 3.1985101184973696e-05, + "loss": 3.977, + "step": 1100288 + }, + { + "epoch": 1.01, + "learning_rate": 3.1976715237463175e-05, + "loss": 3.9688, + "step": 1100800 + }, + { + "epoch": 1.01, + "learning_rate": 3.1968329289952655e-05, + "loss": 3.979, + "step": 1101312 + }, + { + "epoch": 1.01, + "learning_rate": 3.1959943342442135e-05, + "loss": 3.9744, + "step": 1101824 + }, + { + "epoch": 1.01, + "learning_rate": 3.1951557394931615e-05, + "loss": 3.9646, + "step": 1102336 + }, + { + "epoch": 1.01, + "learning_rate": 3.1943171447421095e-05, + "loss": 3.9662, + "step": 1102848 + }, + { + "epoch": 1.01, + "learning_rate": 3.1934785499910575e-05, + "loss": 3.9804, + "step": 1103360 + }, + { + "epoch": 1.01, + "learning_rate": 3.1926399552400055e-05, + "loss": 3.9595, + "step": 1103872 + }, + { + "epoch": 1.01, + "learning_rate": 3.1918013604889535e-05, + "loss": 3.9715, + "step": 1104384 + }, + { + "epoch": 1.01, + "learning_rate": 3.1909644036182744e-05, + "loss": 3.9887, + "step": 1104896 + }, + { + "epoch": 1.01, + "learning_rate": 3.1901258088672224e-05, + "loss": 3.9631, + "step": 1105408 + }, + { + "epoch": 1.01, + "learning_rate": 3.189288851996543e-05, + "loss": 3.9827, + "step": 1105920 + }, + { + "epoch": 1.01, + "learning_rate": 3.188450257245491e-05, + "loss": 3.988, + "step": 1106432 + }, + { + "epoch": 1.01, + "learning_rate": 3.187611662494439e-05, + "loss": 3.9806, + "step": 1106944 + }, + { + "epoch": 1.01, + "learning_rate": 3.186773067743388e-05, + "loss": 3.9808, + "step": 1107456 + }, + { + "epoch": 1.01, + "learning_rate": 3.185934472992336e-05, + "loss": 3.9662, + "step": 1107968 + }, + { + "epoch": 1.01, + "learning_rate": 3.185095878241284e-05, + "loss": 3.9703, + "step": 1108480 + }, + { + "epoch": 1.01, + "learning_rate": 3.184257283490232e-05, + "loss": 3.9729, + "step": 1108992 + }, + { + "epoch": 1.01, + "learning_rate": 3.183420326619553e-05, + "loss": 3.9814, + "step": 1109504 + }, + { + "epoch": 1.01, + "learning_rate": 3.182583369748874e-05, + "loss": 3.9749, + "step": 1110016 + }, + { + "epoch": 1.01, + "learning_rate": 3.181744774997822e-05, + "loss": 3.9592, + "step": 1110528 + }, + { + "epoch": 1.01, + "learning_rate": 3.18090618024677e-05, + "loss": 3.9683, + "step": 1111040 + }, + { + "epoch": 1.01, + "learning_rate": 3.180067585495718e-05, + "loss": 3.9645, + "step": 1111552 + }, + { + "epoch": 1.01, + "learning_rate": 3.179228990744666e-05, + "loss": 3.9693, + "step": 1112064 + }, + { + "epoch": 1.01, + "learning_rate": 3.178390395993614e-05, + "loss": 3.9767, + "step": 1112576 + }, + { + "epoch": 1.01, + "learning_rate": 3.177551801242562e-05, + "loss": 3.9818, + "step": 1113088 + }, + { + "epoch": 1.01, + "learning_rate": 3.17671320649151e-05, + "loss": 3.9781, + "step": 1113600 + }, + { + "epoch": 1.01, + "learning_rate": 3.175874611740458e-05, + "loss": 3.971, + "step": 1114112 + }, + { + "epoch": 1.02, + "learning_rate": 3.175036016989406e-05, + "loss": 3.9629, + "step": 1114624 + }, + { + "epoch": 1.02, + "learning_rate": 3.174197422238354e-05, + "loss": 3.9633, + "step": 1115136 + }, + { + "epoch": 1.02, + "learning_rate": 3.173358827487302e-05, + "loss": 3.9703, + "step": 1115648 + }, + { + "epoch": 1.02, + "learning_rate": 3.172523508496996e-05, + "loss": 3.9739, + "step": 1116160 + }, + { + "epoch": 1.02, + "learning_rate": 3.1716849137459436e-05, + "loss": 3.9611, + "step": 1116672 + }, + { + "epoch": 1.02, + "learning_rate": 3.1708463189948916e-05, + "loss": 3.9673, + "step": 1117184 + }, + { + "epoch": 1.02, + "learning_rate": 3.170009362124213e-05, + "loss": 3.9754, + "step": 1117696 + }, + { + "epoch": 1.02, + "learning_rate": 3.169170767373161e-05, + "loss": 3.9786, + "step": 1118208 + }, + { + "epoch": 1.02, + "learning_rate": 3.1683321726221085e-05, + "loss": 3.977, + "step": 1118720 + }, + { + "epoch": 1.02, + "learning_rate": 3.167493577871057e-05, + "loss": 3.9778, + "step": 1119232 + }, + { + "epoch": 1.02, + "learning_rate": 3.166654983120005e-05, + "loss": 3.9796, + "step": 1119744 + }, + { + "epoch": 1.02, + "learning_rate": 3.165816388368953e-05, + "loss": 3.9697, + "step": 1120256 + }, + { + "epoch": 1.02, + "learning_rate": 3.164977793617901e-05, + "loss": 3.9703, + "step": 1120768 + }, + { + "epoch": 1.02, + "learning_rate": 3.164139198866849e-05, + "loss": 3.9885, + "step": 1121280 + }, + { + "epoch": 1.02, + "learning_rate": 3.16330224199617e-05, + "loss": 3.967, + "step": 1121792 + }, + { + "epoch": 1.02, + "learning_rate": 3.162463647245118e-05, + "loss": 3.9795, + "step": 1122304 + }, + { + "epoch": 1.02, + "learning_rate": 3.161625052494066e-05, + "loss": 3.9631, + "step": 1122816 + }, + { + "epoch": 1.02, + "learning_rate": 3.160786457743014e-05, + "loss": 3.958, + "step": 1123328 + }, + { + "epoch": 1.02, + "learning_rate": 3.159949500872335e-05, + "loss": 3.9718, + "step": 1123840 + }, + { + "epoch": 1.02, + "learning_rate": 3.159110906121283e-05, + "loss": 3.9735, + "step": 1124352 + }, + { + "epoch": 1.02, + "learning_rate": 3.158272311370231e-05, + "loss": 3.9714, + "step": 1124864 + }, + { + "epoch": 1.02, + "learning_rate": 3.157433716619179e-05, + "loss": 3.9503, + "step": 1125376 + }, + { + "epoch": 1.02, + "learning_rate": 3.156595121868127e-05, + "loss": 3.9635, + "step": 1125888 + }, + { + "epoch": 1.02, + "learning_rate": 3.1557565271170756e-05, + "loss": 3.9731, + "step": 1126400 + }, + { + "epoch": 1.02, + "learning_rate": 3.1549195702463965e-05, + "loss": 3.9775, + "step": 1126912 + }, + { + "epoch": 1.02, + "learning_rate": 3.1540809754953445e-05, + "loss": 3.9743, + "step": 1127424 + }, + { + "epoch": 1.02, + "learning_rate": 3.1532423807442925e-05, + "loss": 3.9736, + "step": 1127936 + }, + { + "epoch": 1.02, + "learning_rate": 3.1524037859932405e-05, + "loss": 3.975, + "step": 1128448 + }, + { + "epoch": 1.02, + "learning_rate": 3.1515651912421885e-05, + "loss": 3.975, + "step": 1128960 + }, + { + "epoch": 1.02, + "learning_rate": 3.1507282343715094e-05, + "loss": 3.9704, + "step": 1129472 + }, + { + "epoch": 1.02, + "learning_rate": 3.1498896396204574e-05, + "loss": 3.9698, + "step": 1129984 + }, + { + "epoch": 1.02, + "learning_rate": 3.1490510448694054e-05, + "loss": 3.9615, + "step": 1130496 + }, + { + "epoch": 1.02, + "learning_rate": 3.1482124501183534e-05, + "loss": 3.9759, + "step": 1131008 + }, + { + "epoch": 1.02, + "learning_rate": 3.1473738553673014e-05, + "loss": 3.9585, + "step": 1131520 + }, + { + "epoch": 1.02, + "learning_rate": 3.1465352606162494e-05, + "loss": 3.9769, + "step": 1132032 + }, + { + "epoch": 1.02, + "learning_rate": 3.1456966658651974e-05, + "loss": 3.9666, + "step": 1132544 + }, + { + "epoch": 1.02, + "learning_rate": 3.144859708994519e-05, + "loss": 3.9776, + "step": 1133056 + }, + { + "epoch": 1.02, + "learning_rate": 3.144021114243467e-05, + "loss": 3.9582, + "step": 1133568 + }, + { + "epoch": 1.02, + "learning_rate": 3.143182519492415e-05, + "loss": 3.9758, + "step": 1134080 + }, + { + "epoch": 1.02, + "learning_rate": 3.142343924741362e-05, + "loss": 3.9703, + "step": 1134592 + }, + { + "epoch": 1.02, + "learning_rate": 3.141506967870684e-05, + "loss": 3.9733, + "step": 1135104 + }, + { + "epoch": 1.02, + "learning_rate": 3.140668373119632e-05, + "loss": 3.9803, + "step": 1135616 + }, + { + "epoch": 1.02, + "learning_rate": 3.13982977836858e-05, + "loss": 3.9763, + "step": 1136128 + }, + { + "epoch": 1.02, + "learning_rate": 3.138991183617527e-05, + "loss": 3.9671, + "step": 1136640 + }, + { + "epoch": 1.02, + "learning_rate": 3.138152588866475e-05, + "loss": 3.9793, + "step": 1137152 + }, + { + "epoch": 1.02, + "learning_rate": 3.137315631995797e-05, + "loss": 3.9679, + "step": 1137664 + }, + { + "epoch": 1.02, + "learning_rate": 3.136477037244745e-05, + "loss": 3.9628, + "step": 1138176 + }, + { + "epoch": 1.02, + "learning_rate": 3.135638442493693e-05, + "loss": 3.9842, + "step": 1138688 + }, + { + "epoch": 1.02, + "learning_rate": 3.134799847742641e-05, + "loss": 3.9767, + "step": 1139200 + }, + { + "epoch": 1.02, + "learning_rate": 3.133961252991589e-05, + "loss": 3.9835, + "step": 1139712 + }, + { + "epoch": 1.02, + "learning_rate": 3.133122658240537e-05, + "loss": 3.9735, + "step": 1140224 + }, + { + "epoch": 1.02, + "learning_rate": 3.132284063489485e-05, + "loss": 3.9786, + "step": 1140736 + }, + { + "epoch": 1.02, + "learning_rate": 3.131445468738433e-05, + "loss": 3.9631, + "step": 1141248 + }, + { + "epoch": 1.02, + "learning_rate": 3.1306085118677536e-05, + "loss": 3.9693, + "step": 1141760 + }, + { + "epoch": 1.02, + "learning_rate": 3.1297715549970745e-05, + "loss": 3.966, + "step": 1142272 + }, + { + "epoch": 1.02, + "learning_rate": 3.1289329602460225e-05, + "loss": 3.9731, + "step": 1142784 + }, + { + "epoch": 1.02, + "learning_rate": 3.1280943654949705e-05, + "loss": 3.971, + "step": 1143296 + }, + { + "epoch": 1.02, + "learning_rate": 3.1272557707439185e-05, + "loss": 3.9608, + "step": 1143808 + }, + { + "epoch": 1.02, + "learning_rate": 3.1264171759928665e-05, + "loss": 3.9681, + "step": 1144320 + }, + { + "epoch": 1.03, + "eval_loss": 4.029379844665527, + "eval_runtime": 541.4534, + "eval_samples_per_second": 704.753, + "eval_steps_per_second": 22.024, + "step": 1144785 + }, + { + "epoch": 0.0, + "learning_rate": 3.1255785812418145e-05, + "loss": 3.9663, + "step": 1144832 + }, + { + "epoch": 0.0, + "learning_rate": 3.1247399864907625e-05, + "loss": 3.9707, + "step": 1145344 + }, + { + "epoch": 0.0, + "learning_rate": 3.123901391739711e-05, + "loss": 3.9657, + "step": 1145856 + }, + { + "epoch": 0.0, + "learning_rate": 3.123062796988659e-05, + "loss": 3.9805, + "step": 1146368 + }, + { + "epoch": 0.0, + "learning_rate": 3.122224202237607e-05, + "loss": 3.9738, + "step": 1146880 + }, + { + "epoch": 0.0, + "learning_rate": 3.121385607486555e-05, + "loss": 3.9799, + "step": 1147392 + }, + { + "epoch": 0.0, + "learning_rate": 3.120547012735503e-05, + "loss": 3.9704, + "step": 1147904 + }, + { + "epoch": 0.0, + "learning_rate": 3.119708417984451e-05, + "loss": 3.9679, + "step": 1148416 + }, + { + "epoch": 0.0, + "learning_rate": 3.1188698232333985e-05, + "loss": 3.9592, + "step": 1148928 + }, + { + "epoch": 0.0, + "learning_rate": 3.1180312284823465e-05, + "loss": 3.9729, + "step": 1149440 + }, + { + "epoch": 0.0, + "learning_rate": 3.1171926337312945e-05, + "loss": 3.9716, + "step": 1149952 + }, + { + "epoch": 0.0, + "learning_rate": 3.1163540389802425e-05, + "loss": 3.9709, + "step": 1150464 + }, + { + "epoch": 0.0, + "learning_rate": 3.1155154442291904e-05, + "loss": 3.978, + "step": 1150976 + }, + { + "epoch": 0.0, + "learning_rate": 3.1146768494781384e-05, + "loss": 3.9607, + "step": 1151488 + }, + { + "epoch": 0.0, + "learning_rate": 3.1138382547270864e-05, + "loss": 3.9641, + "step": 1152000 + }, + { + "epoch": 0.0, + "learning_rate": 3.1129996599760344e-05, + "loss": 3.9562, + "step": 1152512 + }, + { + "epoch": 0.0, + "learning_rate": 3.1121610652249824e-05, + "loss": 3.9645, + "step": 1153024 + }, + { + "epoch": 0.0, + "learning_rate": 3.111322470473931e-05, + "loss": 3.9708, + "step": 1153536 + }, + { + "epoch": 0.0, + "learning_rate": 3.110483875722879e-05, + "loss": 3.9688, + "step": 1154048 + }, + { + "epoch": 0.0, + "learning_rate": 3.109645280971827e-05, + "loss": 3.9676, + "step": 1154560 + }, + { + "epoch": 0.0, + "learning_rate": 3.108808324101148e-05, + "loss": 3.9812, + "step": 1155072 + }, + { + "epoch": 0.0, + "learning_rate": 3.107969729350096e-05, + "loss": 3.9714, + "step": 1155584 + }, + { + "epoch": 0.0, + "learning_rate": 3.107131134599044e-05, + "loss": 3.9745, + "step": 1156096 + }, + { + "epoch": 0.0, + "learning_rate": 3.106292539847992e-05, + "loss": 3.9699, + "step": 1156608 + }, + { + "epoch": 0.0, + "learning_rate": 3.105455582977313e-05, + "loss": 3.9829, + "step": 1157120 + }, + { + "epoch": 0.0, + "learning_rate": 3.104616988226261e-05, + "loss": 3.9656, + "step": 1157632 + }, + { + "epoch": 0.0, + "learning_rate": 3.103778393475209e-05, + "loss": 3.9669, + "step": 1158144 + }, + { + "epoch": 0.0, + "learning_rate": 3.102939798724157e-05, + "loss": 3.9616, + "step": 1158656 + }, + { + "epoch": 0.0, + "learning_rate": 3.102101203973105e-05, + "loss": 3.9651, + "step": 1159168 + }, + { + "epoch": 0.0, + "learning_rate": 3.101262609222053e-05, + "loss": 3.9627, + "step": 1159680 + }, + { + "epoch": 0.01, + "learning_rate": 3.1004256523513745e-05, + "loss": 3.9685, + "step": 1160192 + }, + { + "epoch": 0.01, + "learning_rate": 3.0995870576003225e-05, + "loss": 3.9706, + "step": 1160704 + }, + { + "epoch": 0.01, + "learning_rate": 3.09874846284927e-05, + "loss": 3.9743, + "step": 1161216 + }, + { + "epoch": 0.01, + "learning_rate": 3.097909868098218e-05, + "loss": 3.9748, + "step": 1161728 + }, + { + "epoch": 0.01, + "learning_rate": 3.0970729112275394e-05, + "loss": 3.9666, + "step": 1162240 + }, + { + "epoch": 0.01, + "learning_rate": 3.09623595435686e-05, + "loss": 3.9779, + "step": 1162752 + }, + { + "epoch": 0.01, + "learning_rate": 3.095397359605808e-05, + "loss": 3.972, + "step": 1163264 + }, + { + "epoch": 0.01, + "learning_rate": 3.094558764854756e-05, + "loss": 3.9587, + "step": 1163776 + }, + { + "epoch": 0.01, + "learning_rate": 3.093720170103704e-05, + "loss": 3.9648, + "step": 1164288 + }, + { + "epoch": 0.01, + "learning_rate": 3.092883213233025e-05, + "loss": 3.9504, + "step": 1164800 + }, + { + "epoch": 0.01, + "learning_rate": 3.092044618481973e-05, + "loss": 3.963, + "step": 1165312 + }, + { + "epoch": 0.01, + "learning_rate": 3.091206023730921e-05, + "loss": 3.9615, + "step": 1165824 + }, + { + "epoch": 0.01, + "learning_rate": 3.09036742897987e-05, + "loss": 3.982, + "step": 1166336 + }, + { + "epoch": 0.01, + "learning_rate": 3.089528834228817e-05, + "loss": 3.9574, + "step": 1166848 + }, + { + "epoch": 0.01, + "learning_rate": 3.088690239477765e-05, + "loss": 3.9622, + "step": 1167360 + }, + { + "epoch": 0.01, + "learning_rate": 3.087853282607087e-05, + "loss": 3.9637, + "step": 1167872 + }, + { + "epoch": 0.01, + "learning_rate": 3.087014687856035e-05, + "loss": 3.9672, + "step": 1168384 + }, + { + "epoch": 0.01, + "learning_rate": 3.0861777309853556e-05, + "loss": 3.9655, + "step": 1168896 + }, + { + "epoch": 0.01, + "learning_rate": 3.0853391362343036e-05, + "loss": 3.9459, + "step": 1169408 + }, + { + "epoch": 0.01, + "learning_rate": 3.0845005414832516e-05, + "loss": 3.963, + "step": 1169920 + }, + { + "epoch": 0.01, + "learning_rate": 3.0836619467321996e-05, + "loss": 3.9808, + "step": 1170432 + }, + { + "epoch": 0.01, + "learning_rate": 3.082823351981147e-05, + "loss": 3.9707, + "step": 1170944 + }, + { + "epoch": 0.01, + "learning_rate": 3.081984757230095e-05, + "loss": 3.9637, + "step": 1171456 + }, + { + "epoch": 0.01, + "learning_rate": 3.0811478003594165e-05, + "loss": 3.9554, + "step": 1171968 + }, + { + "epoch": 0.01, + "learning_rate": 3.0803092056083645e-05, + "loss": 3.9704, + "step": 1172480 + }, + { + "epoch": 0.01, + "learning_rate": 3.0794706108573125e-05, + "loss": 3.952, + "step": 1172992 + }, + { + "epoch": 0.01, + "learning_rate": 3.0786320161062605e-05, + "loss": 3.9782, + "step": 1173504 + }, + { + "epoch": 0.01, + "learning_rate": 3.0777934213552085e-05, + "loss": 3.9469, + "step": 1174016 + }, + { + "epoch": 0.01, + "learning_rate": 3.0769548266041565e-05, + "loss": 3.982, + "step": 1174528 + }, + { + "epoch": 0.01, + "learning_rate": 3.0761178697334774e-05, + "loss": 3.9697, + "step": 1175040 + }, + { + "epoch": 0.01, + "learning_rate": 3.0752792749824254e-05, + "loss": 3.9477, + "step": 1175552 + }, + { + "epoch": 0.01, + "learning_rate": 3.0744406802313734e-05, + "loss": 3.9597, + "step": 1176064 + }, + { + "epoch": 0.01, + "learning_rate": 3.0736020854803214e-05, + "loss": 3.9724, + "step": 1176576 + }, + { + "epoch": 0.01, + "learning_rate": 3.0727634907292694e-05, + "loss": 3.9408, + "step": 1177088 + }, + { + "epoch": 0.01, + "learning_rate": 3.0719248959782174e-05, + "loss": 3.9575, + "step": 1177600 + }, + { + "epoch": 0.01, + "learning_rate": 3.071087939107539e-05, + "loss": 3.978, + "step": 1178112 + }, + { + "epoch": 0.01, + "learning_rate": 3.070249344356487e-05, + "loss": 3.9638, + "step": 1178624 + }, + { + "epoch": 0.01, + "learning_rate": 3.069410749605435e-05, + "loss": 3.9578, + "step": 1179136 + }, + { + "epoch": 0.01, + "learning_rate": 3.068572154854383e-05, + "loss": 3.9438, + "step": 1179648 + }, + { + "epoch": 0.01, + "learning_rate": 3.067733560103331e-05, + "loss": 3.9599, + "step": 1180160 + }, + { + "epoch": 0.01, + "learning_rate": 3.066894965352279e-05, + "loss": 3.9555, + "step": 1180672 + }, + { + "epoch": 0.01, + "learning_rate": 3.066056370601227e-05, + "loss": 3.9723, + "step": 1181184 + }, + { + "epoch": 0.01, + "learning_rate": 3.065217775850175e-05, + "loss": 3.9556, + "step": 1181696 + }, + { + "epoch": 0.01, + "learning_rate": 3.064379181099123e-05, + "loss": 3.9671, + "step": 1182208 + }, + { + "epoch": 0.01, + "learning_rate": 3.063540586348071e-05, + "loss": 3.969, + "step": 1182720 + }, + { + "epoch": 0.01, + "learning_rate": 3.062701991597018e-05, + "loss": 3.9685, + "step": 1183232 + }, + { + "epoch": 0.01, + "learning_rate": 3.061863396845966e-05, + "loss": 3.9602, + "step": 1183744 + }, + { + "epoch": 0.01, + "learning_rate": 3.061026439975288e-05, + "loss": 3.962, + "step": 1184256 + }, + { + "epoch": 0.01, + "learning_rate": 3.060187845224236e-05, + "loss": 3.9659, + "step": 1184768 + }, + { + "epoch": 0.01, + "learning_rate": 3.059349250473184e-05, + "loss": 3.9665, + "step": 1185280 + }, + { + "epoch": 0.01, + "learning_rate": 3.058510655722132e-05, + "loss": 3.9583, + "step": 1185792 + }, + { + "epoch": 0.01, + "learning_rate": 3.0576736988514534e-05, + "loss": 3.9628, + "step": 1186304 + }, + { + "epoch": 0.01, + "learning_rate": 3.056836741980774e-05, + "loss": 3.9622, + "step": 1186816 + }, + { + "epoch": 0.01, + "learning_rate": 3.055998147229722e-05, + "loss": 3.9659, + "step": 1187328 + }, + { + "epoch": 0.01, + "learning_rate": 3.05515955247867e-05, + "loss": 3.9473, + "step": 1187840 + }, + { + "epoch": 0.01, + "learning_rate": 3.054320957727618e-05, + "loss": 3.9622, + "step": 1188352 + }, + { + "epoch": 0.01, + "learning_rate": 3.0534823629765656e-05, + "loss": 3.9607, + "step": 1188864 + }, + { + "epoch": 0.01, + "learning_rate": 3.0526437682255136e-05, + "loss": 3.9629, + "step": 1189376 + }, + { + "epoch": 0.01, + "learning_rate": 3.0518051734744616e-05, + "loss": 3.9617, + "step": 1189888 + }, + { + "epoch": 0.01, + "learning_rate": 3.05096657872341e-05, + "loss": 3.9595, + "step": 1190400 + }, + { + "epoch": 0.02, + "learning_rate": 3.050129621852731e-05, + "loss": 3.955, + "step": 1190912 + }, + { + "epoch": 0.02, + "learning_rate": 3.049291027101679e-05, + "loss": 3.9519, + "step": 1191424 + }, + { + "epoch": 0.02, + "learning_rate": 3.0484540702310004e-05, + "loss": 3.9558, + "step": 1191936 + }, + { + "epoch": 0.02, + "learning_rate": 3.047615475479948e-05, + "loss": 3.9707, + "step": 1192448 + }, + { + "epoch": 0.02, + "learning_rate": 3.046776880728896e-05, + "loss": 3.9579, + "step": 1192960 + }, + { + "epoch": 0.02, + "learning_rate": 3.045938285977844e-05, + "loss": 3.9682, + "step": 1193472 + }, + { + "epoch": 0.02, + "learning_rate": 3.045099691226792e-05, + "loss": 3.9572, + "step": 1193984 + }, + { + "epoch": 0.02, + "learning_rate": 3.04426109647574e-05, + "loss": 3.9606, + "step": 1194496 + }, + { + "epoch": 0.02, + "learning_rate": 3.043422501724688e-05, + "loss": 3.9538, + "step": 1195008 + }, + { + "epoch": 0.02, + "learning_rate": 3.042585544854009e-05, + "loss": 3.9706, + "step": 1195520 + }, + { + "epoch": 0.02, + "learning_rate": 3.0417485879833302e-05, + "loss": 3.9643, + "step": 1196032 + }, + { + "epoch": 0.02, + "learning_rate": 3.0409099932322782e-05, + "loss": 3.9652, + "step": 1196544 + }, + { + "epoch": 0.02, + "learning_rate": 3.0400713984812262e-05, + "loss": 3.9606, + "step": 1197056 + }, + { + "epoch": 0.02, + "learning_rate": 3.0392328037301742e-05, + "loss": 3.9522, + "step": 1197568 + }, + { + "epoch": 0.02, + "learning_rate": 3.0383942089791222e-05, + "loss": 3.973, + "step": 1198080 + }, + { + "epoch": 0.02, + "learning_rate": 3.0375556142280702e-05, + "loss": 3.9496, + "step": 1198592 + }, + { + "epoch": 0.02, + "learning_rate": 3.0367186573573915e-05, + "loss": 3.955, + "step": 1199104 + }, + { + "epoch": 0.02, + "learning_rate": 3.0358800626063395e-05, + "loss": 3.9627, + "step": 1199616 + }, + { + "epoch": 0.02, + "learning_rate": 3.0350414678552875e-05, + "loss": 3.9546, + "step": 1200128 + }, + { + "epoch": 0.02, + "learning_rate": 3.0342028731042355e-05, + "loss": 3.9547, + "step": 1200640 + }, + { + "epoch": 0.02, + "learning_rate": 3.0333659162335564e-05, + "loss": 3.9574, + "step": 1201152 + }, + { + "epoch": 0.02, + "learning_rate": 3.0325273214825044e-05, + "loss": 3.947, + "step": 1201664 + }, + { + "epoch": 0.02, + "learning_rate": 3.0316887267314527e-05, + "loss": 3.9617, + "step": 1202176 + }, + { + "epoch": 0.02, + "learning_rate": 3.0308501319804007e-05, + "loss": 3.9587, + "step": 1202688 + }, + { + "epoch": 0.02, + "learning_rate": 3.0300115372293487e-05, + "loss": 3.9598, + "step": 1203200 + }, + { + "epoch": 0.02, + "learning_rate": 3.0291729424782967e-05, + "loss": 3.9646, + "step": 1203712 + }, + { + "epoch": 0.02, + "learning_rate": 3.0283343477272447e-05, + "loss": 3.9752, + "step": 1204224 + }, + { + "epoch": 0.02, + "learning_rate": 3.0274957529761927e-05, + "loss": 3.9624, + "step": 1204736 + }, + { + "epoch": 0.02, + "learning_rate": 3.0266571582251407e-05, + "loss": 3.9555, + "step": 1205248 + }, + { + "epoch": 0.02, + "learning_rate": 3.0258185634740887e-05, + "loss": 3.9549, + "step": 1205760 + }, + { + "epoch": 0.02, + "learning_rate": 3.0249799687230367e-05, + "loss": 3.9577, + "step": 1206272 + }, + { + "epoch": 0.02, + "learning_rate": 3.0241413739719843e-05, + "loss": 3.959, + "step": 1206784 + }, + { + "epoch": 0.02, + "learning_rate": 3.023304417101306e-05, + "loss": 3.9601, + "step": 1207296 + }, + { + "epoch": 0.02, + "learning_rate": 3.022465822350254e-05, + "loss": 3.9525, + "step": 1207808 + }, + { + "epoch": 0.02, + "learning_rate": 3.021627227599202e-05, + "loss": 3.9654, + "step": 1208320 + }, + { + "epoch": 0.02, + "learning_rate": 3.0207886328481495e-05, + "loss": 3.9563, + "step": 1208832 + }, + { + "epoch": 0.02, + "learning_rate": 3.0199500380970975e-05, + "loss": 3.9634, + "step": 1209344 + }, + { + "epoch": 0.02, + "learning_rate": 3.019113081226419e-05, + "loss": 3.9536, + "step": 1209856 + }, + { + "epoch": 0.02, + "learning_rate": 3.0182744864753664e-05, + "loss": 3.9565, + "step": 1210368 + }, + { + "epoch": 0.02, + "learning_rate": 3.0174358917243144e-05, + "loss": 3.9525, + "step": 1210880 + }, + { + "epoch": 0.02, + "learning_rate": 3.0165972969732624e-05, + "loss": 3.9693, + "step": 1211392 + }, + { + "epoch": 0.02, + "learning_rate": 3.015760340102584e-05, + "loss": 3.9678, + "step": 1211904 + }, + { + "epoch": 0.02, + "learning_rate": 3.0149217453515317e-05, + "loss": 3.9662, + "step": 1212416 + }, + { + "epoch": 0.02, + "learning_rate": 3.0140831506004797e-05, + "loss": 3.9534, + "step": 1212928 + }, + { + "epoch": 0.02, + "learning_rate": 3.0132445558494277e-05, + "loss": 3.9591, + "step": 1213440 + }, + { + "epoch": 0.02, + "learning_rate": 3.0124059610983757e-05, + "loss": 3.9623, + "step": 1213952 + }, + { + "epoch": 0.02, + "learning_rate": 3.0115690042276966e-05, + "loss": 3.9626, + "step": 1214464 + }, + { + "epoch": 0.02, + "learning_rate": 3.010732047357018e-05, + "loss": 3.9619, + "step": 1214976 + }, + { + "epoch": 0.02, + "learning_rate": 3.0098934526059665e-05, + "loss": 3.9674, + "step": 1215488 + }, + { + "epoch": 0.02, + "learning_rate": 3.0090548578549138e-05, + "loss": 3.9618, + "step": 1216000 + }, + { + "epoch": 0.02, + "learning_rate": 3.0082162631038618e-05, + "loss": 3.9716, + "step": 1216512 + }, + { + "epoch": 0.02, + "learning_rate": 3.0073776683528098e-05, + "loss": 3.96, + "step": 1217024 + }, + { + "epoch": 0.02, + "learning_rate": 3.0065390736017578e-05, + "loss": 3.9672, + "step": 1217536 + }, + { + "epoch": 0.02, + "learning_rate": 3.0057004788507058e-05, + "loss": 3.9611, + "step": 1218048 + }, + { + "epoch": 0.02, + "learning_rate": 3.004861884099654e-05, + "loss": 3.9523, + "step": 1218560 + }, + { + "epoch": 0.02, + "learning_rate": 3.004024927228975e-05, + "loss": 3.9625, + "step": 1219072 + }, + { + "epoch": 0.02, + "learning_rate": 3.003186332477923e-05, + "loss": 3.9478, + "step": 1219584 + }, + { + "epoch": 0.02, + "learning_rate": 3.002347737726871e-05, + "loss": 3.9576, + "step": 1220096 + }, + { + "epoch": 0.02, + "learning_rate": 3.001510780856192e-05, + "loss": 3.9586, + "step": 1220608 + }, + { + "epoch": 0.03, + "eval_loss": 4.022876262664795, + "eval_runtime": 301.5546, + "eval_samples_per_second": 1265.412, + "eval_steps_per_second": 39.545, + "step": 1221105 + }, + { + "epoch": 1.0, + "learning_rate": 3.0006721861051403e-05, + "loss": 3.9559, + "step": 1221120 + }, + { + "epoch": 1.0, + "learning_rate": 2.9998335913540883e-05, + "loss": 3.957, + "step": 1221632 + }, + { + "epoch": 1.0, + "learning_rate": 2.9989949966030363e-05, + "loss": 3.9541, + "step": 1222144 + }, + { + "epoch": 1.0, + "learning_rate": 2.9981564018519843e-05, + "loss": 3.9644, + "step": 1222656 + }, + { + "epoch": 1.0, + "learning_rate": 2.9973178071009323e-05, + "loss": 3.9636, + "step": 1223168 + }, + { + "epoch": 1.0, + "learning_rate": 2.9964792123498803e-05, + "loss": 3.9706, + "step": 1223680 + }, + { + "epoch": 1.0, + "learning_rate": 2.9956406175988283e-05, + "loss": 3.9567, + "step": 1224192 + }, + { + "epoch": 1.0, + "learning_rate": 2.994803660728149e-05, + "loss": 3.9519, + "step": 1224704 + }, + { + "epoch": 1.0, + "learning_rate": 2.9939667038574704e-05, + "loss": 3.9496, + "step": 1225216 + }, + { + "epoch": 1.0, + "learning_rate": 2.9931281091064184e-05, + "loss": 3.9623, + "step": 1225728 + }, + { + "epoch": 1.0, + "learning_rate": 2.9922895143553664e-05, + "loss": 3.9577, + "step": 1226240 + }, + { + "epoch": 1.0, + "learning_rate": 2.9914509196043144e-05, + "loss": 3.9608, + "step": 1226752 + }, + { + "epoch": 1.0, + "learning_rate": 2.9906123248532624e-05, + "loss": 3.9627, + "step": 1227264 + }, + { + "epoch": 1.0, + "learning_rate": 2.9897753679825836e-05, + "loss": 3.9498, + "step": 1227776 + }, + { + "epoch": 1.0, + "learning_rate": 2.9889367732315316e-05, + "loss": 3.954, + "step": 1228288 + }, + { + "epoch": 1.0, + "learning_rate": 2.9880981784804796e-05, + "loss": 3.9498, + "step": 1228800 + }, + { + "epoch": 1.0, + "learning_rate": 2.9872595837294276e-05, + "loss": 3.9484, + "step": 1229312 + }, + { + "epoch": 1.0, + "learning_rate": 2.9864209889783756e-05, + "loss": 3.9582, + "step": 1229824 + }, + { + "epoch": 1.0, + "learning_rate": 2.9855823942273236e-05, + "loss": 3.9557, + "step": 1230336 + }, + { + "epoch": 1.0, + "learning_rate": 2.9847437994762716e-05, + "loss": 3.9573, + "step": 1230848 + }, + { + "epoch": 1.0, + "learning_rate": 2.983906842605593e-05, + "loss": 3.9731, + "step": 1231360 + }, + { + "epoch": 1.0, + "learning_rate": 2.983068247854541e-05, + "loss": 3.9552, + "step": 1231872 + }, + { + "epoch": 1.0, + "learning_rate": 2.982229653103489e-05, + "loss": 3.9633, + "step": 1232384 + }, + { + "epoch": 1.0, + "learning_rate": 2.981391058352437e-05, + "loss": 3.9574, + "step": 1232896 + }, + { + "epoch": 1.0, + "learning_rate": 2.9805541014817578e-05, + "loss": 3.9707, + "step": 1233408 + }, + { + "epoch": 1.0, + "learning_rate": 2.9797155067307058e-05, + "loss": 3.9541, + "step": 1233920 + }, + { + "epoch": 1.0, + "learning_rate": 2.978878549860027e-05, + "loss": 3.9564, + "step": 1234432 + }, + { + "epoch": 1.0, + "learning_rate": 2.978039955108975e-05, + "loss": 3.9491, + "step": 1234944 + }, + { + "epoch": 1.0, + "learning_rate": 2.977201360357923e-05, + "loss": 3.956, + "step": 1235456 + }, + { + "epoch": 1.0, + "learning_rate": 2.976362765606871e-05, + "loss": 3.9504, + "step": 1235968 + }, + { + "epoch": 1.01, + "learning_rate": 2.975524170855819e-05, + "loss": 3.9555, + "step": 1236480 + }, + { + "epoch": 1.01, + "learning_rate": 2.974685576104767e-05, + "loss": 3.9626, + "step": 1236992 + }, + { + "epoch": 1.01, + "learning_rate": 2.973846981353715e-05, + "loss": 3.9604, + "step": 1237504 + }, + { + "epoch": 1.01, + "learning_rate": 2.9730083866026626e-05, + "loss": 3.9621, + "step": 1238016 + }, + { + "epoch": 1.01, + "learning_rate": 2.9721697918516106e-05, + "loss": 3.96, + "step": 1238528 + }, + { + "epoch": 1.01, + "learning_rate": 2.9713328349809322e-05, + "loss": 3.9588, + "step": 1239040 + }, + { + "epoch": 1.01, + "learning_rate": 2.9704942402298795e-05, + "loss": 3.9629, + "step": 1239552 + }, + { + "epoch": 1.01, + "learning_rate": 2.969657283359201e-05, + "loss": 3.9456, + "step": 1240064 + }, + { + "epoch": 1.01, + "learning_rate": 2.968818688608149e-05, + "loss": 3.9555, + "step": 1240576 + }, + { + "epoch": 1.01, + "learning_rate": 2.9679800938570975e-05, + "loss": 3.9386, + "step": 1241088 + }, + { + "epoch": 1.01, + "learning_rate": 2.9671414991060448e-05, + "loss": 3.9518, + "step": 1241600 + }, + { + "epoch": 1.01, + "learning_rate": 2.9663029043549928e-05, + "loss": 3.9473, + "step": 1242112 + }, + { + "epoch": 1.01, + "learning_rate": 2.9654643096039408e-05, + "loss": 3.9661, + "step": 1242624 + }, + { + "epoch": 1.01, + "learning_rate": 2.9646257148528888e-05, + "loss": 3.9517, + "step": 1243136 + }, + { + "epoch": 1.01, + "learning_rate": 2.9637871201018368e-05, + "loss": 3.948, + "step": 1243648 + }, + { + "epoch": 1.01, + "learning_rate": 2.962948525350785e-05, + "loss": 3.9532, + "step": 1244160 + }, + { + "epoch": 1.01, + "learning_rate": 2.962109930599733e-05, + "loss": 3.956, + "step": 1244672 + }, + { + "epoch": 1.01, + "learning_rate": 2.961271335848681e-05, + "loss": 3.9504, + "step": 1245184 + }, + { + "epoch": 1.01, + "learning_rate": 2.960432741097629e-05, + "loss": 3.939, + "step": 1245696 + }, + { + "epoch": 1.01, + "learning_rate": 2.95959578422695e-05, + "loss": 3.9473, + "step": 1246208 + }, + { + "epoch": 1.01, + "learning_rate": 2.958757189475898e-05, + "loss": 3.9682, + "step": 1246720 + }, + { + "epoch": 1.01, + "learning_rate": 2.957918594724846e-05, + "loss": 3.9645, + "step": 1247232 + }, + { + "epoch": 1.01, + "learning_rate": 2.9570816378541672e-05, + "loss": 3.9505, + "step": 1247744 + }, + { + "epoch": 1.01, + "learning_rate": 2.9562430431031152e-05, + "loss": 3.9495, + "step": 1248256 + }, + { + "epoch": 1.01, + "learning_rate": 2.9554044483520632e-05, + "loss": 3.9594, + "step": 1248768 + }, + { + "epoch": 1.01, + "learning_rate": 2.9545658536010112e-05, + "loss": 3.939, + "step": 1249280 + }, + { + "epoch": 1.01, + "learning_rate": 2.953728896730332e-05, + "loss": 3.9668, + "step": 1249792 + }, + { + "epoch": 1.01, + "learning_rate": 2.9528903019792805e-05, + "loss": 3.9368, + "step": 1250304 + }, + { + "epoch": 1.01, + "learning_rate": 2.9520517072282285e-05, + "loss": 3.9656, + "step": 1250816 + }, + { + "epoch": 1.01, + "learning_rate": 2.9512131124771764e-05, + "loss": 3.9622, + "step": 1251328 + }, + { + "epoch": 1.01, + "learning_rate": 2.9503745177261244e-05, + "loss": 3.9382, + "step": 1251840 + }, + { + "epoch": 1.01, + "learning_rate": 2.9495359229750724e-05, + "loss": 3.9474, + "step": 1252352 + }, + { + "epoch": 1.01, + "learning_rate": 2.9486973282240204e-05, + "loss": 3.9596, + "step": 1252864 + }, + { + "epoch": 1.01, + "learning_rate": 2.9478587334729684e-05, + "loss": 3.9292, + "step": 1253376 + }, + { + "epoch": 1.01, + "learning_rate": 2.947020138721916e-05, + "loss": 3.9442, + "step": 1253888 + }, + { + "epoch": 1.01, + "learning_rate": 2.946181543970864e-05, + "loss": 3.9644, + "step": 1254400 + }, + { + "epoch": 1.01, + "learning_rate": 2.945342949219812e-05, + "loss": 3.9551, + "step": 1254912 + }, + { + "epoch": 1.01, + "learning_rate": 2.9445059923491337e-05, + "loss": 3.9494, + "step": 1255424 + }, + { + "epoch": 1.01, + "learning_rate": 2.943667397598081e-05, + "loss": 3.9353, + "step": 1255936 + }, + { + "epoch": 1.01, + "learning_rate": 2.942828802847029e-05, + "loss": 3.9486, + "step": 1256448 + }, + { + "epoch": 1.01, + "learning_rate": 2.941990208095977e-05, + "loss": 3.9417, + "step": 1256960 + }, + { + "epoch": 1.01, + "learning_rate": 2.9411532512252986e-05, + "loss": 3.9652, + "step": 1257472 + }, + { + "epoch": 1.01, + "learning_rate": 2.9403162943546198e-05, + "loss": 3.943, + "step": 1257984 + }, + { + "epoch": 1.01, + "learning_rate": 2.9394776996035678e-05, + "loss": 3.9563, + "step": 1258496 + }, + { + "epoch": 1.01, + "learning_rate": 2.9386391048525158e-05, + "loss": 3.9554, + "step": 1259008 + }, + { + "epoch": 1.01, + "learning_rate": 2.9378005101014635e-05, + "loss": 3.9583, + "step": 1259520 + }, + { + "epoch": 1.01, + "learning_rate": 2.9369635532307847e-05, + "loss": 3.9496, + "step": 1260032 + }, + { + "epoch": 1.01, + "learning_rate": 2.936124958479733e-05, + "loss": 3.9509, + "step": 1260544 + }, + { + "epoch": 1.01, + "learning_rate": 2.935286363728681e-05, + "loss": 3.9573, + "step": 1261056 + }, + { + "epoch": 1.01, + "learning_rate": 2.9344477689776284e-05, + "loss": 3.9543, + "step": 1261568 + }, + { + "epoch": 1.01, + "learning_rate": 2.9336091742265763e-05, + "loss": 3.9489, + "step": 1262080 + }, + { + "epoch": 1.01, + "learning_rate": 2.932772217355898e-05, + "loss": 3.9479, + "step": 1262592 + }, + { + "epoch": 1.01, + "learning_rate": 2.931933622604846e-05, + "loss": 3.9553, + "step": 1263104 + }, + { + "epoch": 1.01, + "learning_rate": 2.9310950278537936e-05, + "loss": 3.9547, + "step": 1263616 + }, + { + "epoch": 1.01, + "learning_rate": 2.9302564331027416e-05, + "loss": 3.9384, + "step": 1264128 + }, + { + "epoch": 1.01, + "learning_rate": 2.9294178383516896e-05, + "loss": 3.9471, + "step": 1264640 + }, + { + "epoch": 1.01, + "learning_rate": 2.9285792436006376e-05, + "loss": 3.9486, + "step": 1265152 + }, + { + "epoch": 1.01, + "learning_rate": 2.9277406488495856e-05, + "loss": 3.9514, + "step": 1265664 + }, + { + "epoch": 1.01, + "learning_rate": 2.9269020540985336e-05, + "loss": 3.9494, + "step": 1266176 + }, + { + "epoch": 1.01, + "learning_rate": 2.9260634593474816e-05, + "loss": 3.9498, + "step": 1266688 + }, + { + "epoch": 1.02, + "learning_rate": 2.9252281403571757e-05, + "loss": 3.943, + "step": 1267200 + }, + { + "epoch": 1.02, + "learning_rate": 2.9243895456061237e-05, + "loss": 3.9385, + "step": 1267712 + }, + { + "epoch": 1.02, + "learning_rate": 2.9235509508550717e-05, + "loss": 3.9457, + "step": 1268224 + }, + { + "epoch": 1.02, + "learning_rate": 2.9227123561040197e-05, + "loss": 3.9584, + "step": 1268736 + }, + { + "epoch": 1.02, + "learning_rate": 2.9218737613529677e-05, + "loss": 3.9474, + "step": 1269248 + }, + { + "epoch": 1.02, + "learning_rate": 2.921035166601916e-05, + "loss": 3.9593, + "step": 1269760 + }, + { + "epoch": 1.02, + "learning_rate": 2.920196571850864e-05, + "loss": 3.9496, + "step": 1270272 + }, + { + "epoch": 1.02, + "learning_rate": 2.919357977099812e-05, + "loss": 3.9441, + "step": 1270784 + }, + { + "epoch": 1.02, + "learning_rate": 2.918521020229133e-05, + "loss": 3.9446, + "step": 1271296 + }, + { + "epoch": 1.02, + "learning_rate": 2.917682425478081e-05, + "loss": 3.9586, + "step": 1271808 + }, + { + "epoch": 1.02, + "learning_rate": 2.916843830727029e-05, + "loss": 3.951, + "step": 1272320 + }, + { + "epoch": 1.02, + "learning_rate": 2.916005235975977e-05, + "loss": 3.9581, + "step": 1272832 + }, + { + "epoch": 1.02, + "learning_rate": 2.9151682791052982e-05, + "loss": 3.9457, + "step": 1273344 + }, + { + "epoch": 1.02, + "learning_rate": 2.9143296843542462e-05, + "loss": 3.9413, + "step": 1273856 + }, + { + "epoch": 1.02, + "learning_rate": 2.913491089603194e-05, + "loss": 3.9646, + "step": 1274368 + }, + { + "epoch": 1.02, + "learning_rate": 2.912652494852142e-05, + "loss": 3.9376, + "step": 1274880 + }, + { + "epoch": 1.02, + "learning_rate": 2.911815537981463e-05, + "loss": 3.9465, + "step": 1275392 + }, + { + "epoch": 1.02, + "learning_rate": 2.9109769432304114e-05, + "loss": 3.9482, + "step": 1275904 + }, + { + "epoch": 1.02, + "learning_rate": 2.9101383484793594e-05, + "loss": 3.9474, + "step": 1276416 + }, + { + "epoch": 1.02, + "learning_rate": 2.9092997537283074e-05, + "loss": 3.9426, + "step": 1276928 + }, + { + "epoch": 1.02, + "learning_rate": 2.9084611589772554e-05, + "loss": 3.9436, + "step": 1277440 + }, + { + "epoch": 1.02, + "learning_rate": 2.9076225642262034e-05, + "loss": 3.9433, + "step": 1277952 + }, + { + "epoch": 1.02, + "learning_rate": 2.9067839694751514e-05, + "loss": 3.9493, + "step": 1278464 + }, + { + "epoch": 1.02, + "learning_rate": 2.9059453747240994e-05, + "loss": 3.9497, + "step": 1278976 + }, + { + "epoch": 1.02, + "learning_rate": 2.9051084178534206e-05, + "loss": 3.9459, + "step": 1279488 + }, + { + "epoch": 1.02, + "learning_rate": 2.9042698231023686e-05, + "loss": 3.9562, + "step": 1280000 + }, + { + "epoch": 1.02, + "learning_rate": 2.9034312283513166e-05, + "loss": 3.9638, + "step": 1280512 + }, + { + "epoch": 1.02, + "learning_rate": 2.9025926336002646e-05, + "loss": 3.9517, + "step": 1281024 + }, + { + "epoch": 1.02, + "learning_rate": 2.9017556767295855e-05, + "loss": 3.9464, + "step": 1281536 + }, + { + "epoch": 1.02, + "learning_rate": 2.9009187198589068e-05, + "loss": 3.943, + "step": 1282048 + }, + { + "epoch": 1.02, + "learning_rate": 2.9000801251078548e-05, + "loss": 3.9452, + "step": 1282560 + }, + { + "epoch": 1.02, + "learning_rate": 2.8992415303568028e-05, + "loss": 3.9494, + "step": 1283072 + }, + { + "epoch": 1.02, + "learning_rate": 2.8984029356057508e-05, + "loss": 3.9499, + "step": 1283584 + }, + { + "epoch": 1.02, + "learning_rate": 2.8975643408546988e-05, + "loss": 3.9436, + "step": 1284096 + }, + { + "epoch": 1.02, + "learning_rate": 2.8967257461036468e-05, + "loss": 3.9525, + "step": 1284608 + }, + { + "epoch": 1.02, + "learning_rate": 2.8958871513525944e-05, + "loss": 3.9474, + "step": 1285120 + }, + { + "epoch": 1.02, + "learning_rate": 2.8950485566015424e-05, + "loss": 3.9498, + "step": 1285632 + }, + { + "epoch": 1.02, + "learning_rate": 2.894211599730864e-05, + "loss": 3.9463, + "step": 1286144 + }, + { + "epoch": 1.02, + "learning_rate": 2.893373004979812e-05, + "loss": 3.9454, + "step": 1286656 + }, + { + "epoch": 1.02, + "learning_rate": 2.892536048109133e-05, + "loss": 3.9425, + "step": 1287168 + }, + { + "epoch": 1.02, + "learning_rate": 2.891697453358081e-05, + "loss": 3.9574, + "step": 1287680 + }, + { + "epoch": 1.02, + "learning_rate": 2.890858858607029e-05, + "loss": 3.958, + "step": 1288192 + }, + { + "epoch": 1.02, + "learning_rate": 2.89002190173635e-05, + "loss": 3.9568, + "step": 1288704 + }, + { + "epoch": 1.02, + "learning_rate": 2.889183306985298e-05, + "loss": 3.9404, + "step": 1289216 + }, + { + "epoch": 1.02, + "learning_rate": 2.888344712234246e-05, + "loss": 3.9515, + "step": 1289728 + }, + { + "epoch": 1.02, + "learning_rate": 2.887506117483194e-05, + "loss": 3.9489, + "step": 1290240 + }, + { + "epoch": 1.02, + "learning_rate": 2.8866675227321414e-05, + "loss": 3.9531, + "step": 1290752 + }, + { + "epoch": 1.02, + "learning_rate": 2.8858289279810898e-05, + "loss": 3.9468, + "step": 1291264 + }, + { + "epoch": 1.02, + "learning_rate": 2.8849903332300378e-05, + "loss": 3.961, + "step": 1291776 + }, + { + "epoch": 1.02, + "learning_rate": 2.8841533763593594e-05, + "loss": 3.9468, + "step": 1292288 + }, + { + "epoch": 1.02, + "learning_rate": 2.8833147816083067e-05, + "loss": 3.9659, + "step": 1292800 + }, + { + "epoch": 1.02, + "learning_rate": 2.8824761868572547e-05, + "loss": 3.9446, + "step": 1293312 + }, + { + "epoch": 1.02, + "learning_rate": 2.8816375921062027e-05, + "loss": 3.959, + "step": 1293824 + }, + { + "epoch": 1.02, + "learning_rate": 2.8807989973551507e-05, + "loss": 3.947, + "step": 1294336 + }, + { + "epoch": 1.02, + "learning_rate": 2.879960402604099e-05, + "loss": 3.9425, + "step": 1294848 + }, + { + "epoch": 1.02, + "learning_rate": 2.879121807853047e-05, + "loss": 3.952, + "step": 1295360 + }, + { + "epoch": 1.02, + "learning_rate": 2.878283213101995e-05, + "loss": 3.9394, + "step": 1295872 + }, + { + "epoch": 1.02, + "learning_rate": 2.877444618350943e-05, + "loss": 3.9463, + "step": 1296384 + }, + { + "epoch": 1.02, + "learning_rate": 2.876606023599891e-05, + "loss": 3.9497, + "step": 1296896 + }, + { + "epoch": 1.02, + "learning_rate": 2.875767428848839e-05, + "loss": 3.9442, + "step": 1297408 + }, + { + "epoch": 1.03, + "eval_loss": 4.017317771911621, + "eval_runtime": 297.868, + "eval_samples_per_second": 1281.074, + "eval_steps_per_second": 40.035, + "step": 1297425 + }, + { + "epoch": 0.0, + "learning_rate": 2.874928834097787e-05, + "loss": 3.9466, + "step": 1297920 + }, + { + "epoch": 0.0, + "learning_rate": 2.874090239346735e-05, + "loss": 3.9452, + "step": 1298432 + }, + { + "epoch": 0.0, + "learning_rate": 2.873251644595683e-05, + "loss": 3.9533, + "step": 1298944 + }, + { + "epoch": 0.0, + "learning_rate": 2.8724130498446306e-05, + "loss": 3.9506, + "step": 1299456 + }, + { + "epoch": 0.0, + "learning_rate": 2.8715744550935786e-05, + "loss": 3.965, + "step": 1299968 + }, + { + "epoch": 0.0, + "learning_rate": 2.8707358603425266e-05, + "loss": 3.9436, + "step": 1300480 + }, + { + "epoch": 0.0, + "learning_rate": 2.8698972655914746e-05, + "loss": 3.9437, + "step": 1300992 + }, + { + "epoch": 0.0, + "learning_rate": 2.8690586708404226e-05, + "loss": 3.9389, + "step": 1301504 + }, + { + "epoch": 0.0, + "learning_rate": 2.8682200760893706e-05, + "loss": 3.9511, + "step": 1302016 + }, + { + "epoch": 0.0, + "learning_rate": 2.8673814813383186e-05, + "loss": 3.9487, + "step": 1302528 + }, + { + "epoch": 0.0, + "learning_rate": 2.866542886587267e-05, + "loss": 3.9499, + "step": 1303040 + }, + { + "epoch": 0.0, + "learning_rate": 2.865704291836215e-05, + "loss": 3.9518, + "step": 1303552 + }, + { + "epoch": 0.0, + "learning_rate": 2.864867334965536e-05, + "loss": 3.9361, + "step": 1304064 + }, + { + "epoch": 0.0, + "learning_rate": 2.8640287402144838e-05, + "loss": 3.9495, + "step": 1304576 + }, + { + "epoch": 0.0, + "learning_rate": 2.8631901454634318e-05, + "loss": 3.9382, + "step": 1305088 + }, + { + "epoch": 0.0, + "learning_rate": 2.8623515507123798e-05, + "loss": 3.9394, + "step": 1305600 + }, + { + "epoch": 0.0, + "learning_rate": 2.8615129559613278e-05, + "loss": 3.943, + "step": 1306112 + }, + { + "epoch": 0.0, + "learning_rate": 2.860674361210276e-05, + "loss": 3.9443, + "step": 1306624 + }, + { + "epoch": 0.0, + "learning_rate": 2.859835766459224e-05, + "loss": 3.949, + "step": 1307136 + }, + { + "epoch": 0.0, + "learning_rate": 2.858997171708172e-05, + "loss": 3.9666, + "step": 1307648 + }, + { + "epoch": 0.0, + "learning_rate": 2.8581585769571195e-05, + "loss": 3.9447, + "step": 1308160 + }, + { + "epoch": 0.0, + "learning_rate": 2.8573199822060675e-05, + "loss": 3.9509, + "step": 1308672 + }, + { + "epoch": 0.0, + "learning_rate": 2.8564813874550154e-05, + "loss": 3.9484, + "step": 1309184 + }, + { + "epoch": 0.0, + "learning_rate": 2.8556444305843367e-05, + "loss": 3.9536, + "step": 1309696 + }, + { + "epoch": 0.0, + "learning_rate": 2.8548058358332847e-05, + "loss": 3.9446, + "step": 1310208 + }, + { + "epoch": 0.0, + "learning_rate": 2.8539672410822327e-05, + "loss": 3.9489, + "step": 1310720 + }, + { + "epoch": 0.0, + "learning_rate": 2.8531286463311807e-05, + "loss": 3.9381, + "step": 1311232 + }, + { + "epoch": 0.0, + "learning_rate": 2.8522900515801287e-05, + "loss": 3.9471, + "step": 1311744 + }, + { + "epoch": 0.0, + "learning_rate": 2.85145309470945e-05, + "loss": 3.9377, + "step": 1312256 + }, + { + "epoch": 0.01, + "learning_rate": 2.850614499958398e-05, + "loss": 3.9462, + "step": 1312768 + }, + { + "epoch": 0.01, + "learning_rate": 2.849775905207346e-05, + "loss": 3.9516, + "step": 1313280 + }, + { + "epoch": 0.01, + "learning_rate": 2.848937310456294e-05, + "loss": 3.9489, + "step": 1313792 + }, + { + "epoch": 0.01, + "learning_rate": 2.848098715705242e-05, + "loss": 3.9551, + "step": 1314304 + }, + { + "epoch": 0.01, + "learning_rate": 2.8472617588345628e-05, + "loss": 3.9455, + "step": 1314816 + }, + { + "epoch": 0.01, + "learning_rate": 2.846424801963884e-05, + "loss": 3.9516, + "step": 1315328 + }, + { + "epoch": 0.01, + "learning_rate": 2.845586207212832e-05, + "loss": 3.9491, + "step": 1315840 + }, + { + "epoch": 0.01, + "learning_rate": 2.84474761246178e-05, + "loss": 3.9417, + "step": 1316352 + }, + { + "epoch": 0.01, + "learning_rate": 2.843909017710728e-05, + "loss": 3.9416, + "step": 1316864 + }, + { + "epoch": 0.01, + "learning_rate": 2.843070422959676e-05, + "loss": 3.9285, + "step": 1317376 + }, + { + "epoch": 0.01, + "learning_rate": 2.842231828208624e-05, + "loss": 3.9424, + "step": 1317888 + }, + { + "epoch": 0.01, + "learning_rate": 2.841393233457572e-05, + "loss": 3.9394, + "step": 1318400 + }, + { + "epoch": 0.01, + "learning_rate": 2.84055463870652e-05, + "loss": 3.9536, + "step": 1318912 + }, + { + "epoch": 0.01, + "learning_rate": 2.839716043955468e-05, + "loss": 3.9389, + "step": 1319424 + }, + { + "epoch": 0.01, + "learning_rate": 2.8388774492044164e-05, + "loss": 3.9413, + "step": 1319936 + }, + { + "epoch": 0.01, + "learning_rate": 2.8380404923337373e-05, + "loss": 3.9429, + "step": 1320448 + }, + { + "epoch": 0.01, + "learning_rate": 2.8372018975826853e-05, + "loss": 3.9429, + "step": 1320960 + }, + { + "epoch": 0.01, + "learning_rate": 2.8363633028316333e-05, + "loss": 3.9395, + "step": 1321472 + }, + { + "epoch": 0.01, + "learning_rate": 2.8355247080805813e-05, + "loss": 3.9308, + "step": 1321984 + }, + { + "epoch": 0.01, + "learning_rate": 2.8346861133295293e-05, + "loss": 3.9332, + "step": 1322496 + }, + { + "epoch": 0.01, + "learning_rate": 2.8338475185784773e-05, + "loss": 3.9576, + "step": 1323008 + }, + { + "epoch": 0.01, + "learning_rate": 2.8330089238274256e-05, + "loss": 3.9536, + "step": 1323520 + }, + { + "epoch": 0.01, + "learning_rate": 2.832170329076373e-05, + "loss": 3.9372, + "step": 1324032 + }, + { + "epoch": 0.01, + "learning_rate": 2.831331734325321e-05, + "loss": 3.9407, + "step": 1324544 + }, + { + "epoch": 0.01, + "learning_rate": 2.8304947774546425e-05, + "loss": 3.949, + "step": 1325056 + }, + { + "epoch": 0.01, + "learning_rate": 2.8296561827035905e-05, + "loss": 3.9309, + "step": 1325568 + }, + { + "epoch": 0.01, + "learning_rate": 2.8288192258329117e-05, + "loss": 3.9524, + "step": 1326080 + }, + { + "epoch": 0.01, + "learning_rate": 2.8279806310818597e-05, + "loss": 3.9262, + "step": 1326592 + }, + { + "epoch": 0.01, + "learning_rate": 2.8271420363308077e-05, + "loss": 3.9587, + "step": 1327104 + }, + { + "epoch": 0.01, + "learning_rate": 2.8263034415797557e-05, + "loss": 3.9477, + "step": 1327616 + }, + { + "epoch": 0.01, + "learning_rate": 2.825464846828703e-05, + "loss": 3.935, + "step": 1328128 + }, + { + "epoch": 0.01, + "learning_rate": 2.8246278899580246e-05, + "loss": 3.9328, + "step": 1328640 + }, + { + "epoch": 0.01, + "learning_rate": 2.8237892952069726e-05, + "loss": 3.9501, + "step": 1329152 + }, + { + "epoch": 0.01, + "learning_rate": 2.8229507004559203e-05, + "loss": 3.9206, + "step": 1329664 + }, + { + "epoch": 0.01, + "learning_rate": 2.8221121057048683e-05, + "loss": 3.9352, + "step": 1330176 + }, + { + "epoch": 0.01, + "learning_rate": 2.8212735109538163e-05, + "loss": 3.9511, + "step": 1330688 + }, + { + "epoch": 0.01, + "learning_rate": 2.8204349162027643e-05, + "loss": 3.9469, + "step": 1331200 + }, + { + "epoch": 0.01, + "learning_rate": 2.8195963214517123e-05, + "loss": 3.9386, + "step": 1331712 + }, + { + "epoch": 0.01, + "learning_rate": 2.8187577267006603e-05, + "loss": 3.9251, + "step": 1332224 + }, + { + "epoch": 0.01, + "learning_rate": 2.8179191319496086e-05, + "loss": 3.9364, + "step": 1332736 + }, + { + "epoch": 0.01, + "learning_rate": 2.8170821750789295e-05, + "loss": 3.935, + "step": 1333248 + }, + { + "epoch": 0.01, + "learning_rate": 2.8162452182082504e-05, + "loss": 3.9486, + "step": 1333760 + }, + { + "epoch": 0.01, + "learning_rate": 2.8154066234571984e-05, + "loss": 3.9374, + "step": 1334272 + }, + { + "epoch": 0.01, + "learning_rate": 2.8145680287061464e-05, + "loss": 3.9438, + "step": 1334784 + }, + { + "epoch": 0.01, + "learning_rate": 2.8137294339550947e-05, + "loss": 3.9465, + "step": 1335296 + }, + { + "epoch": 0.01, + "learning_rate": 2.8128908392040427e-05, + "loss": 3.952, + "step": 1335808 + }, + { + "epoch": 0.01, + "learning_rate": 2.8120538823333636e-05, + "loss": 3.9352, + "step": 1336320 + }, + { + "epoch": 0.01, + "learning_rate": 2.8112152875823116e-05, + "loss": 3.9415, + "step": 1336832 + }, + { + "epoch": 0.01, + "learning_rate": 2.8103766928312596e-05, + "loss": 3.9449, + "step": 1337344 + }, + { + "epoch": 0.01, + "learning_rate": 2.8095380980802076e-05, + "loss": 3.9441, + "step": 1337856 + }, + { + "epoch": 0.01, + "learning_rate": 2.8086995033291556e-05, + "loss": 3.9412, + "step": 1338368 + }, + { + "epoch": 0.01, + "learning_rate": 2.807860908578104e-05, + "loss": 3.9374, + "step": 1338880 + }, + { + "epoch": 0.01, + "learning_rate": 2.807022313827052e-05, + "loss": 3.9447, + "step": 1339392 + }, + { + "epoch": 0.01, + "learning_rate": 2.806183719076e-05, + "loss": 3.9425, + "step": 1339904 + }, + { + "epoch": 0.01, + "learning_rate": 2.805346762205321e-05, + "loss": 3.9305, + "step": 1340416 + }, + { + "epoch": 0.01, + "learning_rate": 2.804508167454269e-05, + "loss": 3.9343, + "step": 1340928 + }, + { + "epoch": 0.01, + "learning_rate": 2.803669572703217e-05, + "loss": 3.9402, + "step": 1341440 + }, + { + "epoch": 0.01, + "learning_rate": 2.802832615832538e-05, + "loss": 3.9439, + "step": 1341952 + }, + { + "epoch": 0.01, + "learning_rate": 2.801994021081486e-05, + "loss": 3.9402, + "step": 1342464 + }, + { + "epoch": 0.01, + "learning_rate": 2.801155426330434e-05, + "loss": 3.937, + "step": 1342976 + }, + { + "epoch": 0.02, + "learning_rate": 2.800316831579382e-05, + "loss": 3.9341, + "step": 1343488 + }, + { + "epoch": 0.02, + "learning_rate": 2.79947823682833e-05, + "loss": 3.9304, + "step": 1344000 + }, + { + "epoch": 0.02, + "learning_rate": 2.798639642077278e-05, + "loss": 3.932, + "step": 1344512 + }, + { + "epoch": 0.02, + "learning_rate": 2.797801047326226e-05, + "loss": 3.9452, + "step": 1345024 + }, + { + "epoch": 0.02, + "learning_rate": 2.7969640904555473e-05, + "loss": 3.9397, + "step": 1345536 + }, + { + "epoch": 0.02, + "learning_rate": 2.7961271335848682e-05, + "loss": 3.9471, + "step": 1346048 + }, + { + "epoch": 0.02, + "learning_rate": 2.7952885388338162e-05, + "loss": 3.9452, + "step": 1346560 + }, + { + "epoch": 0.02, + "learning_rate": 2.7944499440827642e-05, + "loss": 3.9348, + "step": 1347072 + }, + { + "epoch": 0.02, + "learning_rate": 2.7936113493317122e-05, + "loss": 3.9367, + "step": 1347584 + }, + { + "epoch": 0.02, + "learning_rate": 2.7927727545806602e-05, + "loss": 3.9455, + "step": 1348096 + }, + { + "epoch": 0.02, + "learning_rate": 2.7919341598296085e-05, + "loss": 3.947, + "step": 1348608 + }, + { + "epoch": 0.02, + "learning_rate": 2.7910955650785565e-05, + "loss": 3.9473, + "step": 1349120 + }, + { + "epoch": 0.02, + "learning_rate": 2.790256970327504e-05, + "loss": 3.9377, + "step": 1349632 + }, + { + "epoch": 0.02, + "learning_rate": 2.789418375576452e-05, + "loss": 3.9317, + "step": 1350144 + }, + { + "epoch": 0.02, + "learning_rate": 2.7885797808254e-05, + "loss": 3.9528, + "step": 1350656 + }, + { + "epoch": 0.02, + "learning_rate": 2.7877428239547214e-05, + "loss": 3.9273, + "step": 1351168 + }, + { + "epoch": 0.02, + "learning_rate": 2.786904229203669e-05, + "loss": 3.9384, + "step": 1351680 + }, + { + "epoch": 0.02, + "learning_rate": 2.786065634452617e-05, + "loss": 3.9376, + "step": 1352192 + }, + { + "epoch": 0.02, + "learning_rate": 2.785227039701565e-05, + "loss": 3.9384, + "step": 1352704 + }, + { + "epoch": 0.02, + "learning_rate": 2.784388444950513e-05, + "loss": 3.9285, + "step": 1353216 + }, + { + "epoch": 0.02, + "learning_rate": 2.783549850199461e-05, + "loss": 3.9377, + "step": 1353728 + }, + { + "epoch": 0.02, + "learning_rate": 2.782711255448409e-05, + "loss": 3.9296, + "step": 1354240 + }, + { + "epoch": 0.02, + "learning_rate": 2.781872660697357e-05, + "loss": 3.9399, + "step": 1354752 + }, + { + "epoch": 0.02, + "learning_rate": 2.781034065946305e-05, + "loss": 3.9382, + "step": 1355264 + }, + { + "epoch": 0.02, + "learning_rate": 2.7801987469559992e-05, + "loss": 3.9332, + "step": 1355776 + }, + { + "epoch": 0.02, + "learning_rate": 2.7793601522049472e-05, + "loss": 3.9496, + "step": 1356288 + }, + { + "epoch": 0.02, + "learning_rate": 2.7785215574538952e-05, + "loss": 3.9526, + "step": 1356800 + }, + { + "epoch": 0.02, + "learning_rate": 2.7776829627028432e-05, + "loss": 3.9416, + "step": 1357312 + }, + { + "epoch": 0.02, + "learning_rate": 2.7768443679517912e-05, + "loss": 3.9372, + "step": 1357824 + }, + { + "epoch": 0.02, + "learning_rate": 2.7760057732007395e-05, + "loss": 3.9336, + "step": 1358336 + }, + { + "epoch": 0.02, + "learning_rate": 2.7751671784496875e-05, + "loss": 3.9356, + "step": 1358848 + }, + { + "epoch": 0.02, + "learning_rate": 2.7743285836986355e-05, + "loss": 3.9365, + "step": 1359360 + }, + { + "epoch": 0.02, + "learning_rate": 2.7734916268279564e-05, + "loss": 3.9456, + "step": 1359872 + }, + { + "epoch": 0.02, + "learning_rate": 2.7726530320769044e-05, + "loss": 3.9293, + "step": 1360384 + }, + { + "epoch": 0.02, + "learning_rate": 2.7718144373258524e-05, + "loss": 3.9418, + "step": 1360896 + }, + { + "epoch": 0.02, + "learning_rate": 2.7709774804551737e-05, + "loss": 3.9385, + "step": 1361408 + }, + { + "epoch": 0.02, + "learning_rate": 2.7701388857041217e-05, + "loss": 3.9434, + "step": 1361920 + }, + { + "epoch": 0.02, + "learning_rate": 2.7693002909530697e-05, + "loss": 3.9309, + "step": 1362432 + }, + { + "epoch": 0.02, + "learning_rate": 2.7684616962020177e-05, + "loss": 3.9368, + "step": 1362944 + }, + { + "epoch": 0.02, + "learning_rate": 2.7676231014509657e-05, + "loss": 3.9386, + "step": 1363456 + }, + { + "epoch": 0.02, + "learning_rate": 2.7667845066999137e-05, + "loss": 3.9434, + "step": 1363968 + }, + { + "epoch": 0.02, + "learning_rate": 2.7659459119488617e-05, + "loss": 3.9474, + "step": 1364480 + }, + { + "epoch": 0.02, + "learning_rate": 2.7651073171978097e-05, + "loss": 3.9494, + "step": 1364992 + }, + { + "epoch": 0.02, + "learning_rate": 2.764270360327131e-05, + "loss": 3.9299, + "step": 1365504 + }, + { + "epoch": 0.02, + "learning_rate": 2.763431765576079e-05, + "loss": 3.9407, + "step": 1366016 + }, + { + "epoch": 0.02, + "learning_rate": 2.762593170825027e-05, + "loss": 3.9418, + "step": 1366528 + }, + { + "epoch": 0.02, + "learning_rate": 2.7617562139543478e-05, + "loss": 3.9393, + "step": 1367040 + }, + { + "epoch": 0.02, + "learning_rate": 2.7609176192032958e-05, + "loss": 3.939, + "step": 1367552 + }, + { + "epoch": 0.02, + "learning_rate": 2.760079024452244e-05, + "loss": 3.9487, + "step": 1368064 + }, + { + "epoch": 0.02, + "learning_rate": 2.759240429701192e-05, + "loss": 3.937, + "step": 1368576 + }, + { + "epoch": 0.02, + "learning_rate": 2.75840183495014e-05, + "loss": 3.9561, + "step": 1369088 + }, + { + "epoch": 0.02, + "learning_rate": 2.7575632401990874e-05, + "loss": 3.9327, + "step": 1369600 + }, + { + "epoch": 0.02, + "learning_rate": 2.7567246454480354e-05, + "loss": 3.9511, + "step": 1370112 + }, + { + "epoch": 0.02, + "learning_rate": 2.7558860506969834e-05, + "loss": 3.9384, + "step": 1370624 + }, + { + "epoch": 0.02, + "learning_rate": 2.755049093826305e-05, + "loss": 3.9322, + "step": 1371136 + }, + { + "epoch": 0.02, + "learning_rate": 2.7542104990752527e-05, + "loss": 3.9407, + "step": 1371648 + }, + { + "epoch": 0.02, + "learning_rate": 2.7533719043242007e-05, + "loss": 3.933, + "step": 1372160 + }, + { + "epoch": 0.02, + "learning_rate": 2.7525333095731487e-05, + "loss": 3.9341, + "step": 1372672 + }, + { + "epoch": 0.02, + "learning_rate": 2.7516963527024696e-05, + "loss": 3.9392, + "step": 1373184 + }, + { + "epoch": 0.02, + "learning_rate": 2.750857757951418e-05, + "loss": 3.9351, + "step": 1373696 + }, + { + "epoch": 0.03, + "eval_loss": 4.012390613555908, + "eval_runtime": 306.2869, + "eval_samples_per_second": 1245.861, + "eval_steps_per_second": 38.934, + "step": 1373745 + }, + { + "epoch": 1.0, + "learning_rate": 2.7500208010807395e-05, + "loss": 3.9382, + "step": 1374208 + }, + { + "epoch": 1.0, + "learning_rate": 2.7491822063296875e-05, + "loss": 3.9335, + "step": 1374720 + }, + { + "epoch": 1.0, + "learning_rate": 2.7483436115786348e-05, + "loss": 3.9476, + "step": 1375232 + }, + { + "epoch": 1.0, + "learning_rate": 2.7475050168275828e-05, + "loss": 3.9396, + "step": 1375744 + }, + { + "epoch": 1.0, + "learning_rate": 2.7466664220765308e-05, + "loss": 3.9528, + "step": 1376256 + }, + { + "epoch": 1.0, + "learning_rate": 2.745829465205852e-05, + "loss": 3.9357, + "step": 1376768 + }, + { + "epoch": 1.0, + "learning_rate": 2.7449908704548e-05, + "loss": 3.9373, + "step": 1377280 + }, + { + "epoch": 1.0, + "learning_rate": 2.744152275703748e-05, + "loss": 3.9262, + "step": 1377792 + }, + { + "epoch": 1.0, + "learning_rate": 2.743313680952696e-05, + "loss": 3.9452, + "step": 1378304 + }, + { + "epoch": 1.0, + "learning_rate": 2.742475086201644e-05, + "loss": 3.9373, + "step": 1378816 + }, + { + "epoch": 1.0, + "learning_rate": 2.741636491450592e-05, + "loss": 3.9344, + "step": 1379328 + }, + { + "epoch": 1.0, + "learning_rate": 2.7407995345799133e-05, + "loss": 3.9497, + "step": 1379840 + }, + { + "epoch": 1.0, + "learning_rate": 2.739962577709235e-05, + "loss": 3.9256, + "step": 1380352 + }, + { + "epoch": 1.0, + "learning_rate": 2.7391239829581822e-05, + "loss": 3.9423, + "step": 1380864 + }, + { + "epoch": 1.0, + "learning_rate": 2.7382853882071302e-05, + "loss": 3.9259, + "step": 1381376 + }, + { + "epoch": 1.0, + "learning_rate": 2.7374467934560782e-05, + "loss": 3.9315, + "step": 1381888 + }, + { + "epoch": 1.0, + "learning_rate": 2.736608198705026e-05, + "loss": 3.9305, + "step": 1382400 + }, + { + "epoch": 1.0, + "learning_rate": 2.735769603953974e-05, + "loss": 3.9364, + "step": 1382912 + }, + { + "epoch": 1.0, + "learning_rate": 2.7349310092029225e-05, + "loss": 3.9369, + "step": 1383424 + }, + { + "epoch": 1.0, + "learning_rate": 2.7340924144518705e-05, + "loss": 3.9566, + "step": 1383936 + }, + { + "epoch": 1.0, + "learning_rate": 2.7332538197008185e-05, + "loss": 3.932, + "step": 1384448 + }, + { + "epoch": 1.0, + "learning_rate": 2.7324168628301394e-05, + "loss": 3.9397, + "step": 1384960 + }, + { + "epoch": 1.0, + "learning_rate": 2.7315782680790874e-05, + "loss": 3.9401, + "step": 1385472 + }, + { + "epoch": 1.0, + "learning_rate": 2.7307396733280354e-05, + "loss": 3.946, + "step": 1385984 + }, + { + "epoch": 1.0, + "learning_rate": 2.7299010785769834e-05, + "loss": 3.9336, + "step": 1386496 + }, + { + "epoch": 1.0, + "learning_rate": 2.7290624838259314e-05, + "loss": 3.9425, + "step": 1387008 + }, + { + "epoch": 1.0, + "learning_rate": 2.7282238890748797e-05, + "loss": 3.9268, + "step": 1387520 + }, + { + "epoch": 1.0, + "learning_rate": 2.7273869322042006e-05, + "loss": 3.9377, + "step": 1388032 + }, + { + "epoch": 1.0, + "learning_rate": 2.7265483374531486e-05, + "loss": 3.9282, + "step": 1388544 + }, + { + "epoch": 1.01, + "learning_rate": 2.7257097427020966e-05, + "loss": 3.939, + "step": 1389056 + }, + { + "epoch": 1.01, + "learning_rate": 2.7248711479510446e-05, + "loss": 3.9382, + "step": 1389568 + }, + { + "epoch": 1.01, + "learning_rate": 2.7240325531999926e-05, + "loss": 3.9398, + "step": 1390080 + }, + { + "epoch": 1.01, + "learning_rate": 2.7231939584489406e-05, + "loss": 3.9491, + "step": 1390592 + }, + { + "epoch": 1.01, + "learning_rate": 2.7223553636978883e-05, + "loss": 3.9331, + "step": 1391104 + }, + { + "epoch": 1.01, + "learning_rate": 2.72151840682721e-05, + "loss": 3.9439, + "step": 1391616 + }, + { + "epoch": 1.01, + "learning_rate": 2.720679812076158e-05, + "loss": 3.9409, + "step": 1392128 + }, + { + "epoch": 1.01, + "learning_rate": 2.7198428552054788e-05, + "loss": 3.9326, + "step": 1392640 + }, + { + "epoch": 1.01, + "learning_rate": 2.7190042604544267e-05, + "loss": 3.93, + "step": 1393152 + }, + { + "epoch": 1.01, + "learning_rate": 2.718165665703375e-05, + "loss": 3.923, + "step": 1393664 + }, + { + "epoch": 1.01, + "learning_rate": 2.717327070952323e-05, + "loss": 3.9293, + "step": 1394176 + }, + { + "epoch": 1.01, + "learning_rate": 2.716488476201271e-05, + "loss": 3.931, + "step": 1394688 + }, + { + "epoch": 1.01, + "learning_rate": 2.7156498814502184e-05, + "loss": 3.9422, + "step": 1395200 + }, + { + "epoch": 1.01, + "learning_rate": 2.7148112866991664e-05, + "loss": 3.9345, + "step": 1395712 + }, + { + "epoch": 1.01, + "learning_rate": 2.7139726919481144e-05, + "loss": 3.9287, + "step": 1396224 + }, + { + "epoch": 1.01, + "learning_rate": 2.7131340971970627e-05, + "loss": 3.9387, + "step": 1396736 + }, + { + "epoch": 1.01, + "learning_rate": 2.7122955024460107e-05, + "loss": 3.9314, + "step": 1397248 + }, + { + "epoch": 1.01, + "learning_rate": 2.7114569076949587e-05, + "loss": 3.9312, + "step": 1397760 + }, + { + "epoch": 1.01, + "learning_rate": 2.7106183129439067e-05, + "loss": 3.9252, + "step": 1398272 + }, + { + "epoch": 1.01, + "learning_rate": 2.7097813560732276e-05, + "loss": 3.9214, + "step": 1398784 + }, + { + "epoch": 1.01, + "learning_rate": 2.708944399202549e-05, + "loss": 3.9494, + "step": 1399296 + }, + { + "epoch": 1.01, + "learning_rate": 2.708105804451497e-05, + "loss": 3.9427, + "step": 1399808 + }, + { + "epoch": 1.01, + "learning_rate": 2.707267209700445e-05, + "loss": 3.9309, + "step": 1400320 + }, + { + "epoch": 1.01, + "learning_rate": 2.706428614949393e-05, + "loss": 3.9319, + "step": 1400832 + }, + { + "epoch": 1.01, + "learning_rate": 2.705590020198341e-05, + "loss": 3.9393, + "step": 1401344 + }, + { + "epoch": 1.01, + "learning_rate": 2.704751425447289e-05, + "loss": 3.9202, + "step": 1401856 + }, + { + "epoch": 1.01, + "learning_rate": 2.7039144685766097e-05, + "loss": 3.9419, + "step": 1402368 + }, + { + "epoch": 1.01, + "learning_rate": 2.703075873825558e-05, + "loss": 3.9187, + "step": 1402880 + }, + { + "epoch": 1.01, + "learning_rate": 2.702237279074506e-05, + "loss": 3.9474, + "step": 1403392 + }, + { + "epoch": 1.01, + "learning_rate": 2.701398684323454e-05, + "loss": 3.943, + "step": 1403904 + }, + { + "epoch": 1.01, + "learning_rate": 2.700560089572402e-05, + "loss": 3.9261, + "step": 1404416 + }, + { + "epoch": 1.01, + "learning_rate": 2.69972149482135e-05, + "loss": 3.9231, + "step": 1404928 + }, + { + "epoch": 1.01, + "learning_rate": 2.698882900070298e-05, + "loss": 3.9429, + "step": 1405440 + }, + { + "epoch": 1.01, + "learning_rate": 2.698044305319246e-05, + "loss": 3.9099, + "step": 1405952 + }, + { + "epoch": 1.01, + "learning_rate": 2.6972073484485673e-05, + "loss": 3.926, + "step": 1406464 + }, + { + "epoch": 1.01, + "learning_rate": 2.6963687536975153e-05, + "loss": 3.942, + "step": 1406976 + }, + { + "epoch": 1.01, + "learning_rate": 2.6955301589464633e-05, + "loss": 3.9385, + "step": 1407488 + }, + { + "epoch": 1.01, + "learning_rate": 2.6946915641954113e-05, + "loss": 3.9283, + "step": 1408000 + }, + { + "epoch": 1.01, + "learning_rate": 2.6938546073247322e-05, + "loss": 3.9174, + "step": 1408512 + }, + { + "epoch": 1.01, + "learning_rate": 2.6930160125736802e-05, + "loss": 3.9261, + "step": 1409024 + }, + { + "epoch": 1.01, + "learning_rate": 2.6921774178226282e-05, + "loss": 3.9251, + "step": 1409536 + }, + { + "epoch": 1.01, + "learning_rate": 2.6913388230715765e-05, + "loss": 3.9367, + "step": 1410048 + }, + { + "epoch": 1.01, + "learning_rate": 2.6905018662008974e-05, + "loss": 3.9255, + "step": 1410560 + }, + { + "epoch": 1.01, + "learning_rate": 2.6896649093302183e-05, + "loss": 3.9384, + "step": 1411072 + }, + { + "epoch": 1.01, + "learning_rate": 2.6888263145791663e-05, + "loss": 3.9394, + "step": 1411584 + }, + { + "epoch": 1.01, + "learning_rate": 2.6879877198281143e-05, + "loss": 3.9399, + "step": 1412096 + }, + { + "epoch": 1.01, + "learning_rate": 2.6871491250770627e-05, + "loss": 3.9289, + "step": 1412608 + }, + { + "epoch": 1.01, + "learning_rate": 2.6863105303260107e-05, + "loss": 3.9322, + "step": 1413120 + }, + { + "epoch": 1.01, + "learning_rate": 2.6854719355749587e-05, + "loss": 3.9382, + "step": 1413632 + }, + { + "epoch": 1.01, + "learning_rate": 2.6846333408239067e-05, + "loss": 3.9326, + "step": 1414144 + }, + { + "epoch": 1.01, + "learning_rate": 2.683794746072854e-05, + "loss": 3.9314, + "step": 1414656 + }, + { + "epoch": 1.01, + "learning_rate": 2.682956151321802e-05, + "loss": 3.9315, + "step": 1415168 + }, + { + "epoch": 1.01, + "learning_rate": 2.6821175565707503e-05, + "loss": 3.9333, + "step": 1415680 + }, + { + "epoch": 1.01, + "learning_rate": 2.681280599700072e-05, + "loss": 3.935, + "step": 1416192 + }, + { + "epoch": 1.01, + "learning_rate": 2.6804420049490192e-05, + "loss": 3.919, + "step": 1416704 + }, + { + "epoch": 1.01, + "learning_rate": 2.6796034101979672e-05, + "loss": 3.9281, + "step": 1417216 + }, + { + "epoch": 1.01, + "learning_rate": 2.6787648154469152e-05, + "loss": 3.9286, + "step": 1417728 + }, + { + "epoch": 1.01, + "learning_rate": 2.6779278585762368e-05, + "loss": 3.9367, + "step": 1418240 + }, + { + "epoch": 1.01, + "learning_rate": 2.6770892638251844e-05, + "loss": 3.9338, + "step": 1418752 + }, + { + "epoch": 1.01, + "learning_rate": 2.676252306954506e-05, + "loss": 3.9296, + "step": 1419264 + }, + { + "epoch": 1.02, + "learning_rate": 2.675413712203454e-05, + "loss": 3.925, + "step": 1419776 + }, + { + "epoch": 1.02, + "learning_rate": 2.6745751174524013e-05, + "loss": 3.9193, + "step": 1420288 + }, + { + "epoch": 1.02, + "learning_rate": 2.6737365227013493e-05, + "loss": 3.9237, + "step": 1420800 + }, + { + "epoch": 1.02, + "learning_rate": 2.6728979279502973e-05, + "loss": 3.9323, + "step": 1421312 + }, + { + "epoch": 1.02, + "learning_rate": 2.6720593331992457e-05, + "loss": 3.9363, + "step": 1421824 + }, + { + "epoch": 1.02, + "learning_rate": 2.6712223763285666e-05, + "loss": 3.9392, + "step": 1422336 + }, + { + "epoch": 1.02, + "learning_rate": 2.6703837815775146e-05, + "loss": 3.9314, + "step": 1422848 + }, + { + "epoch": 1.02, + "learning_rate": 2.6695451868264626e-05, + "loss": 3.9278, + "step": 1423360 + }, + { + "epoch": 1.02, + "learning_rate": 2.6687065920754106e-05, + "loss": 3.9263, + "step": 1423872 + }, + { + "epoch": 1.02, + "learning_rate": 2.6678679973243586e-05, + "loss": 3.9315, + "step": 1424384 + }, + { + "epoch": 1.02, + "learning_rate": 2.6670310404536798e-05, + "loss": 3.9414, + "step": 1424896 + }, + { + "epoch": 1.02, + "learning_rate": 2.6661924457026278e-05, + "loss": 3.9372, + "step": 1425408 + }, + { + "epoch": 1.02, + "learning_rate": 2.6653554888319487e-05, + "loss": 3.9281, + "step": 1425920 + }, + { + "epoch": 1.02, + "learning_rate": 2.6645168940808967e-05, + "loss": 3.9241, + "step": 1426432 + }, + { + "epoch": 1.02, + "learning_rate": 2.6636782993298447e-05, + "loss": 3.9415, + "step": 1426944 + }, + { + "epoch": 1.02, + "learning_rate": 2.6628397045787927e-05, + "loss": 3.92, + "step": 1427456 + }, + { + "epoch": 1.02, + "learning_rate": 2.6620011098277407e-05, + "loss": 3.9315, + "step": 1427968 + }, + { + "epoch": 1.02, + "learning_rate": 2.661162515076689e-05, + "loss": 3.9268, + "step": 1428480 + }, + { + "epoch": 1.02, + "learning_rate": 2.660323920325637e-05, + "loss": 3.9298, + "step": 1428992 + }, + { + "epoch": 1.02, + "learning_rate": 2.659485325574585e-05, + "loss": 3.9206, + "step": 1429504 + }, + { + "epoch": 1.02, + "learning_rate": 2.658648368703906e-05, + "loss": 3.9292, + "step": 1430016 + }, + { + "epoch": 1.02, + "learning_rate": 2.657809773952854e-05, + "loss": 3.9193, + "step": 1430528 + }, + { + "epoch": 1.02, + "learning_rate": 2.656971179201802e-05, + "loss": 3.9283, + "step": 1431040 + }, + { + "epoch": 1.02, + "learning_rate": 2.65613258445075e-05, + "loss": 3.9334, + "step": 1431552 + }, + { + "epoch": 1.02, + "learning_rate": 2.6552939896996983e-05, + "loss": 3.9197, + "step": 1432064 + }, + { + "epoch": 1.02, + "learning_rate": 2.6544553949486463e-05, + "loss": 3.9427, + "step": 1432576 + }, + { + "epoch": 1.02, + "learning_rate": 2.6536168001975942e-05, + "loss": 3.9441, + "step": 1433088 + }, + { + "epoch": 1.02, + "learning_rate": 2.6527782054465422e-05, + "loss": 3.9358, + "step": 1433600 + }, + { + "epoch": 1.02, + "learning_rate": 2.651941248575863e-05, + "loss": 3.9271, + "step": 1434112 + }, + { + "epoch": 1.02, + "learning_rate": 2.651102653824811e-05, + "loss": 3.923, + "step": 1434624 + }, + { + "epoch": 1.02, + "learning_rate": 2.650264059073759e-05, + "loss": 3.9277, + "step": 1435136 + }, + { + "epoch": 1.02, + "learning_rate": 2.6494254643227075e-05, + "loss": 3.9244, + "step": 1435648 + }, + { + "epoch": 1.02, + "learning_rate": 2.6485885074520284e-05, + "loss": 3.939, + "step": 1436160 + }, + { + "epoch": 1.02, + "learning_rate": 2.6477499127009764e-05, + "loss": 3.919, + "step": 1436672 + }, + { + "epoch": 1.02, + "learning_rate": 2.6469113179499244e-05, + "loss": 3.9334, + "step": 1437184 + }, + { + "epoch": 1.02, + "learning_rate": 2.6460727231988724e-05, + "loss": 3.9353, + "step": 1437696 + }, + { + "epoch": 1.02, + "learning_rate": 2.6452341284478204e-05, + "loss": 3.9306, + "step": 1438208 + }, + { + "epoch": 1.02, + "learning_rate": 2.6443971715771416e-05, + "loss": 3.9225, + "step": 1438720 + }, + { + "epoch": 1.02, + "learning_rate": 2.6435602147064625e-05, + "loss": 3.9288, + "step": 1439232 + }, + { + "epoch": 1.02, + "learning_rate": 2.6427232578357834e-05, + "loss": 3.9289, + "step": 1439744 + }, + { + "epoch": 1.02, + "learning_rate": 2.6418846630847314e-05, + "loss": 3.9337, + "step": 1440256 + }, + { + "epoch": 1.02, + "learning_rate": 2.6410460683336798e-05, + "loss": 3.9349, + "step": 1440768 + }, + { + "epoch": 1.02, + "learning_rate": 2.6402074735826278e-05, + "loss": 3.941, + "step": 1441280 + }, + { + "epoch": 1.02, + "learning_rate": 2.6393688788315758e-05, + "loss": 3.9211, + "step": 1441792 + }, + { + "epoch": 1.02, + "learning_rate": 2.6385302840805238e-05, + "loss": 3.9331, + "step": 1442304 + }, + { + "epoch": 1.02, + "learning_rate": 2.6376916893294718e-05, + "loss": 3.9375, + "step": 1442816 + }, + { + "epoch": 1.02, + "learning_rate": 2.6368530945784197e-05, + "loss": 3.9287, + "step": 1443328 + }, + { + "epoch": 1.02, + "learning_rate": 2.6360144998273677e-05, + "loss": 3.9292, + "step": 1443840 + }, + { + "epoch": 1.02, + "learning_rate": 2.6351759050763154e-05, + "loss": 3.938, + "step": 1444352 + }, + { + "epoch": 1.02, + "learning_rate": 2.6343373103252634e-05, + "loss": 3.9323, + "step": 1444864 + }, + { + "epoch": 1.02, + "learning_rate": 2.6334987155742114e-05, + "loss": 3.9455, + "step": 1445376 + }, + { + "epoch": 1.02, + "learning_rate": 2.6326617587035323e-05, + "loss": 3.9272, + "step": 1445888 + }, + { + "epoch": 1.02, + "learning_rate": 2.6318231639524803e-05, + "loss": 3.9395, + "step": 1446400 + }, + { + "epoch": 1.02, + "learning_rate": 2.6309845692014283e-05, + "loss": 3.9323, + "step": 1446912 + }, + { + "epoch": 1.02, + "learning_rate": 2.6301459744503766e-05, + "loss": 3.9272, + "step": 1447424 + }, + { + "epoch": 1.02, + "learning_rate": 2.6293073796993246e-05, + "loss": 3.9294, + "step": 1447936 + }, + { + "epoch": 1.02, + "learning_rate": 2.6284687849482726e-05, + "loss": 3.9253, + "step": 1448448 + }, + { + "epoch": 1.02, + "learning_rate": 2.6276301901972206e-05, + "loss": 3.925, + "step": 1448960 + }, + { + "epoch": 1.02, + "learning_rate": 2.6267932333265415e-05, + "loss": 3.9274, + "step": 1449472 + }, + { + "epoch": 1.02, + "learning_rate": 2.6259546385754895e-05, + "loss": 3.9238, + "step": 1449984 + }, + { + "epoch": 1.03, + "eval_loss": 4.008549690246582, + "eval_runtime": 292.6994, + "eval_samples_per_second": 1303.696, + "eval_steps_per_second": 40.741, + "step": 1450065 + }, + { + "epoch": 0.0, + "learning_rate": 2.6251160438244375e-05, + "loss": 3.9315, + "step": 1450496 + }, + { + "epoch": 0.0, + "learning_rate": 2.624277449073386e-05, + "loss": 3.9247, + "step": 1451008 + }, + { + "epoch": 0.0, + "learning_rate": 2.623438854322334e-05, + "loss": 3.9391, + "step": 1451520 + }, + { + "epoch": 0.0, + "learning_rate": 2.622600259571282e-05, + "loss": 3.9274, + "step": 1452032 + }, + { + "epoch": 0.0, + "learning_rate": 2.62176166482023e-05, + "loss": 3.9496, + "step": 1452544 + }, + { + "epoch": 0.0, + "learning_rate": 2.6209230700691778e-05, + "loss": 3.9229, + "step": 1453056 + }, + { + "epoch": 0.0, + "learning_rate": 2.6200844753181258e-05, + "loss": 3.9303, + "step": 1453568 + }, + { + "epoch": 0.0, + "learning_rate": 2.6192458805670738e-05, + "loss": 3.9178, + "step": 1454080 + }, + { + "epoch": 0.0, + "learning_rate": 2.6184072858160215e-05, + "loss": 3.9355, + "step": 1454592 + }, + { + "epoch": 0.0, + "learning_rate": 2.6175686910649695e-05, + "loss": 3.9311, + "step": 1455104 + }, + { + "epoch": 0.0, + "learning_rate": 2.6167300963139175e-05, + "loss": 3.925, + "step": 1455616 + }, + { + "epoch": 0.0, + "learning_rate": 2.615893139443239e-05, + "loss": 3.9389, + "step": 1456128 + }, + { + "epoch": 0.0, + "learning_rate": 2.6150545446921864e-05, + "loss": 3.9193, + "step": 1456640 + }, + { + "epoch": 0.0, + "learning_rate": 2.6142159499411344e-05, + "loss": 3.9346, + "step": 1457152 + }, + { + "epoch": 0.0, + "learning_rate": 2.6133773551900824e-05, + "loss": 3.9199, + "step": 1457664 + }, + { + "epoch": 0.0, + "learning_rate": 2.6125387604390307e-05, + "loss": 3.9205, + "step": 1458176 + }, + { + "epoch": 0.0, + "learning_rate": 2.6117001656879787e-05, + "loss": 3.921, + "step": 1458688 + }, + { + "epoch": 0.0, + "learning_rate": 2.6108615709369267e-05, + "loss": 3.93, + "step": 1459200 + }, + { + "epoch": 0.0, + "learning_rate": 2.6100229761858747e-05, + "loss": 3.9307, + "step": 1459712 + }, + { + "epoch": 0.0, + "learning_rate": 2.6091843814348227e-05, + "loss": 3.946, + "step": 1460224 + }, + { + "epoch": 0.0, + "learning_rate": 2.6083457866837707e-05, + "loss": 3.9268, + "step": 1460736 + }, + { + "epoch": 0.0, + "learning_rate": 2.6075071919327187e-05, + "loss": 3.9339, + "step": 1461248 + }, + { + "epoch": 0.0, + "learning_rate": 2.6066685971816667e-05, + "loss": 3.9312, + "step": 1461760 + }, + { + "epoch": 0.0, + "learning_rate": 2.6058300024306147e-05, + "loss": 3.9369, + "step": 1462272 + }, + { + "epoch": 0.0, + "learning_rate": 2.604991407679563e-05, + "loss": 3.9281, + "step": 1462784 + }, + { + "epoch": 0.0, + "learning_rate": 2.6041528129285103e-05, + "loss": 3.9312, + "step": 1463296 + }, + { + "epoch": 0.0, + "learning_rate": 2.603315856057832e-05, + "loss": 3.9256, + "step": 1463808 + }, + { + "epoch": 0.0, + "learning_rate": 2.60247726130678e-05, + "loss": 3.9245, + "step": 1464320 + }, + { + "epoch": 0.0, + "learning_rate": 2.601638666555728e-05, + "loss": 3.9183, + "step": 1464832 + }, + { + "epoch": 0.01, + "learning_rate": 2.6008000718046755e-05, + "loss": 3.9303, + "step": 1465344 + }, + { + "epoch": 0.01, + "learning_rate": 2.599963114933997e-05, + "loss": 3.9322, + "step": 1465856 + }, + { + "epoch": 0.01, + "learning_rate": 2.599124520182945e-05, + "loss": 3.9293, + "step": 1466368 + }, + { + "epoch": 0.01, + "learning_rate": 2.5982859254318925e-05, + "loss": 3.9428, + "step": 1466880 + }, + { + "epoch": 0.01, + "learning_rate": 2.5974473306808404e-05, + "loss": 3.9267, + "step": 1467392 + }, + { + "epoch": 0.01, + "learning_rate": 2.596610373810162e-05, + "loss": 3.9308, + "step": 1467904 + }, + { + "epoch": 0.01, + "learning_rate": 2.59577177905911e-05, + "loss": 3.9307, + "step": 1468416 + }, + { + "epoch": 0.01, + "learning_rate": 2.5949348221884313e-05, + "loss": 3.9296, + "step": 1468928 + }, + { + "epoch": 0.01, + "learning_rate": 2.5940962274373793e-05, + "loss": 3.9188, + "step": 1469440 + }, + { + "epoch": 0.01, + "learning_rate": 2.5932576326863273e-05, + "loss": 3.9131, + "step": 1469952 + }, + { + "epoch": 0.01, + "learning_rate": 2.5924190379352753e-05, + "loss": 3.9208, + "step": 1470464 + }, + { + "epoch": 0.01, + "learning_rate": 2.591580443184223e-05, + "loss": 3.926, + "step": 1470976 + }, + { + "epoch": 0.01, + "learning_rate": 2.590741848433171e-05, + "loss": 3.9325, + "step": 1471488 + }, + { + "epoch": 0.01, + "learning_rate": 2.589903253682119e-05, + "loss": 3.9242, + "step": 1472000 + }, + { + "epoch": 0.01, + "learning_rate": 2.589064658931067e-05, + "loss": 3.9187, + "step": 1472512 + }, + { + "epoch": 0.01, + "learning_rate": 2.588226064180015e-05, + "loss": 3.9328, + "step": 1473024 + }, + { + "epoch": 0.01, + "learning_rate": 2.587387469428963e-05, + "loss": 3.9243, + "step": 1473536 + }, + { + "epoch": 0.01, + "learning_rate": 2.5865505125582838e-05, + "loss": 3.9226, + "step": 1474048 + }, + { + "epoch": 0.01, + "learning_rate": 2.585711917807232e-05, + "loss": 3.9181, + "step": 1474560 + }, + { + "epoch": 0.01, + "learning_rate": 2.584874960936553e-05, + "loss": 3.9065, + "step": 1475072 + }, + { + "epoch": 0.01, + "learning_rate": 2.584036366185501e-05, + "loss": 3.9455, + "step": 1475584 + }, + { + "epoch": 0.01, + "learning_rate": 2.583197771434449e-05, + "loss": 3.9326, + "step": 1476096 + }, + { + "epoch": 0.01, + "learning_rate": 2.582359176683397e-05, + "loss": 3.924, + "step": 1476608 + }, + { + "epoch": 0.01, + "learning_rate": 2.581520581932345e-05, + "loss": 3.9257, + "step": 1477120 + }, + { + "epoch": 0.01, + "learning_rate": 2.580681987181293e-05, + "loss": 3.9258, + "step": 1477632 + }, + { + "epoch": 0.01, + "learning_rate": 2.5798433924302414e-05, + "loss": 3.9107, + "step": 1478144 + }, + { + "epoch": 0.01, + "learning_rate": 2.5790047976791894e-05, + "loss": 3.9324, + "step": 1478656 + }, + { + "epoch": 0.01, + "learning_rate": 2.5781662029281374e-05, + "loss": 3.9149, + "step": 1479168 + }, + { + "epoch": 0.01, + "learning_rate": 2.5773276081770854e-05, + "loss": 3.9306, + "step": 1479680 + }, + { + "epoch": 0.01, + "learning_rate": 2.5764890134260333e-05, + "loss": 3.9388, + "step": 1480192 + }, + { + "epoch": 0.01, + "learning_rate": 2.5756504186749813e-05, + "loss": 3.9179, + "step": 1480704 + }, + { + "epoch": 0.01, + "learning_rate": 2.5748134618043023e-05, + "loss": 3.9119, + "step": 1481216 + }, + { + "epoch": 0.01, + "learning_rate": 2.5739748670532502e-05, + "loss": 3.9374, + "step": 1481728 + }, + { + "epoch": 0.01, + "learning_rate": 2.5731362723021986e-05, + "loss": 3.9023, + "step": 1482240 + }, + { + "epoch": 0.01, + "learning_rate": 2.5722993154315195e-05, + "loss": 3.9168, + "step": 1482752 + }, + { + "epoch": 0.01, + "learning_rate": 2.5714607206804675e-05, + "loss": 3.9346, + "step": 1483264 + }, + { + "epoch": 0.01, + "learning_rate": 2.5706221259294155e-05, + "loss": 3.9303, + "step": 1483776 + }, + { + "epoch": 0.01, + "learning_rate": 2.5697835311783635e-05, + "loss": 3.9224, + "step": 1484288 + }, + { + "epoch": 0.01, + "learning_rate": 2.568944936427311e-05, + "loss": 3.9122, + "step": 1484800 + }, + { + "epoch": 0.01, + "learning_rate": 2.568106341676259e-05, + "loss": 3.9118, + "step": 1485312 + }, + { + "epoch": 0.01, + "learning_rate": 2.567267746925207e-05, + "loss": 3.9163, + "step": 1485824 + }, + { + "epoch": 0.01, + "learning_rate": 2.566429152174155e-05, + "loss": 3.9339, + "step": 1486336 + }, + { + "epoch": 0.01, + "learning_rate": 2.5655938331838496e-05, + "loss": 3.9154, + "step": 1486848 + }, + { + "epoch": 0.01, + "learning_rate": 2.5647552384327976e-05, + "loss": 3.929, + "step": 1487360 + }, + { + "epoch": 0.01, + "learning_rate": 2.5639166436817456e-05, + "loss": 3.9294, + "step": 1487872 + }, + { + "epoch": 0.01, + "learning_rate": 2.563078048930694e-05, + "loss": 3.9344, + "step": 1488384 + }, + { + "epoch": 0.01, + "learning_rate": 2.5622394541796413e-05, + "loss": 3.9209, + "step": 1488896 + }, + { + "epoch": 0.01, + "learning_rate": 2.5614008594285893e-05, + "loss": 3.9202, + "step": 1489408 + }, + { + "epoch": 0.01, + "learning_rate": 2.5605622646775373e-05, + "loss": 3.9326, + "step": 1489920 + }, + { + "epoch": 0.01, + "learning_rate": 2.5597236699264853e-05, + "loss": 3.9203, + "step": 1490432 + }, + { + "epoch": 0.01, + "learning_rate": 2.5588867130558065e-05, + "loss": 3.9279, + "step": 1490944 + }, + { + "epoch": 0.01, + "learning_rate": 2.5580481183047545e-05, + "loss": 3.9231, + "step": 1491456 + }, + { + "epoch": 0.01, + "learning_rate": 2.5572095235537025e-05, + "loss": 3.9227, + "step": 1491968 + }, + { + "epoch": 0.01, + "learning_rate": 2.5563709288026505e-05, + "loss": 3.9289, + "step": 1492480 + }, + { + "epoch": 0.01, + "learning_rate": 2.5555323340515985e-05, + "loss": 3.9079, + "step": 1492992 + }, + { + "epoch": 0.01, + "learning_rate": 2.5546953771809194e-05, + "loss": 3.9225, + "step": 1493504 + }, + { + "epoch": 0.01, + "learning_rate": 2.5538567824298677e-05, + "loss": 3.9236, + "step": 1494016 + }, + { + "epoch": 0.01, + "learning_rate": 2.5530181876788157e-05, + "loss": 3.9277, + "step": 1494528 + }, + { + "epoch": 0.01, + "learning_rate": 2.5521795929277637e-05, + "loss": 3.9217, + "step": 1495040 + }, + { + "epoch": 0.01, + "learning_rate": 2.5513426360570846e-05, + "loss": 3.9202, + "step": 1495552 + }, + { + "epoch": 0.02, + "learning_rate": 2.5505056791864055e-05, + "loss": 3.9155, + "step": 1496064 + }, + { + "epoch": 0.02, + "learning_rate": 2.549667084435354e-05, + "loss": 3.914, + "step": 1496576 + }, + { + "epoch": 0.02, + "learning_rate": 2.548828489684302e-05, + "loss": 3.9144, + "step": 1497088 + }, + { + "epoch": 0.02, + "learning_rate": 2.54798989493325e-05, + "loss": 3.924, + "step": 1497600 + }, + { + "epoch": 0.02, + "learning_rate": 2.547151300182198e-05, + "loss": 3.9339, + "step": 1498112 + }, + { + "epoch": 0.02, + "learning_rate": 2.5463143433115188e-05, + "loss": 3.9305, + "step": 1498624 + }, + { + "epoch": 0.02, + "learning_rate": 2.5454757485604668e-05, + "loss": 3.9251, + "step": 1499136 + }, + { + "epoch": 0.02, + "learning_rate": 2.5446371538094148e-05, + "loss": 3.9147, + "step": 1499648 + }, + { + "epoch": 0.02, + "learning_rate": 2.543798559058363e-05, + "loss": 3.9198, + "step": 1500160 + }, + { + "epoch": 0.02, + "learning_rate": 2.542959964307311e-05, + "loss": 3.9231, + "step": 1500672 + }, + { + "epoch": 0.02, + "learning_rate": 2.542121369556259e-05, + "loss": 3.9297, + "step": 1501184 + }, + { + "epoch": 0.02, + "learning_rate": 2.541282774805207e-05, + "loss": 3.9293, + "step": 1501696 + }, + { + "epoch": 0.02, + "learning_rate": 2.540444180054155e-05, + "loss": 3.9224, + "step": 1502208 + }, + { + "epoch": 0.02, + "learning_rate": 2.539605585303103e-05, + "loss": 3.9147, + "step": 1502720 + }, + { + "epoch": 0.02, + "learning_rate": 2.538768628432424e-05, + "loss": 3.9362, + "step": 1503232 + }, + { + "epoch": 0.02, + "learning_rate": 2.5379300336813723e-05, + "loss": 3.9111, + "step": 1503744 + }, + { + "epoch": 0.02, + "learning_rate": 2.5370914389303203e-05, + "loss": 3.9242, + "step": 1504256 + }, + { + "epoch": 0.02, + "learning_rate": 2.5362528441792683e-05, + "loss": 3.9143, + "step": 1504768 + }, + { + "epoch": 0.02, + "learning_rate": 2.5354158873085892e-05, + "loss": 3.9269, + "step": 1505280 + }, + { + "epoch": 0.02, + "learning_rate": 2.5345772925575372e-05, + "loss": 3.9134, + "step": 1505792 + }, + { + "epoch": 0.02, + "learning_rate": 2.5337403356868585e-05, + "loss": 3.9212, + "step": 1506304 + }, + { + "epoch": 0.02, + "learning_rate": 2.5329033788161794e-05, + "loss": 3.9165, + "step": 1506816 + }, + { + "epoch": 0.02, + "learning_rate": 2.5320647840651274e-05, + "loss": 3.9139, + "step": 1507328 + }, + { + "epoch": 0.02, + "learning_rate": 2.5312261893140754e-05, + "loss": 3.9267, + "step": 1507840 + }, + { + "epoch": 0.02, + "learning_rate": 2.5303875945630234e-05, + "loss": 3.9147, + "step": 1508352 + }, + { + "epoch": 0.02, + "learning_rate": 2.5295489998119714e-05, + "loss": 3.9332, + "step": 1508864 + }, + { + "epoch": 0.02, + "learning_rate": 2.5287104050609193e-05, + "loss": 3.9367, + "step": 1509376 + }, + { + "epoch": 0.02, + "learning_rate": 2.5278718103098677e-05, + "loss": 3.9292, + "step": 1509888 + }, + { + "epoch": 0.02, + "learning_rate": 2.5270332155588157e-05, + "loss": 3.9181, + "step": 1510400 + }, + { + "epoch": 0.02, + "learning_rate": 2.5261946208077637e-05, + "loss": 3.9165, + "step": 1510912 + }, + { + "epoch": 0.02, + "learning_rate": 2.5253560260567117e-05, + "loss": 3.919, + "step": 1511424 + }, + { + "epoch": 0.02, + "learning_rate": 2.5245174313056597e-05, + "loss": 3.9206, + "step": 1511936 + }, + { + "epoch": 0.02, + "learning_rate": 2.523678836554607e-05, + "loss": 3.9306, + "step": 1512448 + }, + { + "epoch": 0.02, + "learning_rate": 2.5228402418035553e-05, + "loss": 3.9092, + "step": 1512960 + }, + { + "epoch": 0.02, + "learning_rate": 2.5220016470525033e-05, + "loss": 3.9257, + "step": 1513472 + }, + { + "epoch": 0.02, + "learning_rate": 2.521164690181825e-05, + "loss": 3.9242, + "step": 1513984 + }, + { + "epoch": 0.02, + "learning_rate": 2.5203260954307722e-05, + "loss": 3.9235, + "step": 1514496 + }, + { + "epoch": 0.02, + "learning_rate": 2.5194891385600938e-05, + "loss": 3.9176, + "step": 1515008 + }, + { + "epoch": 0.02, + "learning_rate": 2.5186505438090418e-05, + "loss": 3.922, + "step": 1515520 + }, + { + "epoch": 0.02, + "learning_rate": 2.5178119490579895e-05, + "loss": 3.92, + "step": 1516032 + }, + { + "epoch": 0.02, + "learning_rate": 2.5169733543069375e-05, + "loss": 3.9285, + "step": 1516544 + }, + { + "epoch": 0.02, + "learning_rate": 2.5161347595558854e-05, + "loss": 3.9271, + "step": 1517056 + }, + { + "epoch": 0.02, + "learning_rate": 2.515297802685207e-05, + "loss": 3.9327, + "step": 1517568 + }, + { + "epoch": 0.02, + "learning_rate": 2.5144592079341544e-05, + "loss": 3.915, + "step": 1518080 + }, + { + "epoch": 0.02, + "learning_rate": 2.5136206131831023e-05, + "loss": 3.9274, + "step": 1518592 + }, + { + "epoch": 0.02, + "learning_rate": 2.5127820184320507e-05, + "loss": 3.9276, + "step": 1519104 + }, + { + "epoch": 0.02, + "learning_rate": 2.5119434236809987e-05, + "loss": 3.9202, + "step": 1519616 + }, + { + "epoch": 0.02, + "learning_rate": 2.5111064668103196e-05, + "loss": 3.9224, + "step": 1520128 + }, + { + "epoch": 0.02, + "learning_rate": 2.5102678720592676e-05, + "loss": 3.933, + "step": 1520640 + }, + { + "epoch": 0.02, + "learning_rate": 2.5094309151885892e-05, + "loss": 3.9215, + "step": 1521152 + }, + { + "epoch": 0.02, + "learning_rate": 2.5085923204375368e-05, + "loss": 3.9374, + "step": 1521664 + }, + { + "epoch": 0.02, + "learning_rate": 2.5077537256864848e-05, + "loss": 3.9176, + "step": 1522176 + }, + { + "epoch": 0.02, + "learning_rate": 2.5069151309354328e-05, + "loss": 3.9306, + "step": 1522688 + }, + { + "epoch": 0.02, + "learning_rate": 2.5060765361843808e-05, + "loss": 3.9236, + "step": 1523200 + }, + { + "epoch": 0.02, + "learning_rate": 2.5052379414333288e-05, + "loss": 3.9204, + "step": 1523712 + }, + { + "epoch": 0.02, + "learning_rate": 2.5043993466822768e-05, + "loss": 3.9179, + "step": 1524224 + }, + { + "epoch": 0.02, + "learning_rate": 2.5035623898115977e-05, + "loss": 3.9225, + "step": 1524736 + }, + { + "epoch": 0.02, + "learning_rate": 2.502723795060546e-05, + "loss": 3.9122, + "step": 1525248 + }, + { + "epoch": 0.02, + "learning_rate": 2.501885200309494e-05, + "loss": 3.9189, + "step": 1525760 + }, + { + "epoch": 0.02, + "learning_rate": 2.501046605558442e-05, + "loss": 3.9209, + "step": 1526272 + }, + { + "epoch": 0.03, + "eval_loss": 4.005068302154541, + "eval_runtime": 302.1078, + "eval_samples_per_second": 1263.095, + "eval_steps_per_second": 39.473, + "step": 1526385 + }, + { + "epoch": 1.0, + "learning_rate": 2.50020801080739e-05, + "loss": 3.9256, + "step": 1526784 + }, + { + "epoch": 1.0, + "learning_rate": 2.499369416056338e-05, + "loss": 3.9172, + "step": 1527296 + }, + { + "epoch": 1.0, + "learning_rate": 2.498530821305286e-05, + "loss": 3.928, + "step": 1527808 + }, + { + "epoch": 1.0, + "learning_rate": 2.497692226554234e-05, + "loss": 3.9212, + "step": 1528320 + }, + { + "epoch": 1.0, + "learning_rate": 2.4968536318031817e-05, + "loss": 3.9394, + "step": 1528832 + }, + { + "epoch": 1.0, + "learning_rate": 2.49601503705213e-05, + "loss": 3.9203, + "step": 1529344 + }, + { + "epoch": 1.0, + "learning_rate": 2.4951780801814513e-05, + "loss": 3.9227, + "step": 1529856 + }, + { + "epoch": 1.0, + "learning_rate": 2.4943411233107722e-05, + "loss": 3.9079, + "step": 1530368 + }, + { + "epoch": 1.0, + "learning_rate": 2.49350252855972e-05, + "loss": 3.9234, + "step": 1530880 + }, + { + "epoch": 1.0, + "learning_rate": 2.492663933808668e-05, + "loss": 3.9236, + "step": 1531392 + }, + { + "epoch": 1.0, + "learning_rate": 2.491825339057616e-05, + "loss": 3.9209, + "step": 1531904 + }, + { + "epoch": 1.0, + "learning_rate": 2.490986744306564e-05, + "loss": 3.9255, + "step": 1532416 + }, + { + "epoch": 1.0, + "learning_rate": 2.4901497874358854e-05, + "loss": 3.9155, + "step": 1532928 + }, + { + "epoch": 1.0, + "learning_rate": 2.4893111926848334e-05, + "loss": 3.9254, + "step": 1533440 + }, + { + "epoch": 1.0, + "learning_rate": 2.4884742358141543e-05, + "loss": 3.9111, + "step": 1533952 + }, + { + "epoch": 1.0, + "learning_rate": 2.4876356410631023e-05, + "loss": 3.9131, + "step": 1534464 + }, + { + "epoch": 1.0, + "learning_rate": 2.4867970463120503e-05, + "loss": 3.9141, + "step": 1534976 + }, + { + "epoch": 1.0, + "learning_rate": 2.4859584515609986e-05, + "loss": 3.9229, + "step": 1535488 + }, + { + "epoch": 1.0, + "learning_rate": 2.4851198568099463e-05, + "loss": 3.92, + "step": 1536000 + }, + { + "epoch": 1.0, + "learning_rate": 2.4842812620588943e-05, + "loss": 3.9367, + "step": 1536512 + }, + { + "epoch": 1.0, + "learning_rate": 2.4834426673078423e-05, + "loss": 3.9212, + "step": 1537024 + }, + { + "epoch": 1.0, + "learning_rate": 2.4826040725567903e-05, + "loss": 3.9294, + "step": 1537536 + }, + { + "epoch": 1.0, + "learning_rate": 2.4817654778057383e-05, + "loss": 3.9202, + "step": 1538048 + }, + { + "epoch": 1.0, + "learning_rate": 2.4809268830546863e-05, + "loss": 3.926, + "step": 1538560 + }, + { + "epoch": 1.0, + "learning_rate": 2.4800882883036343e-05, + "loss": 3.9184, + "step": 1539072 + }, + { + "epoch": 1.0, + "learning_rate": 2.4792496935525823e-05, + "loss": 3.9279, + "step": 1539584 + }, + { + "epoch": 1.0, + "learning_rate": 2.4784110988015303e-05, + "loss": 3.9134, + "step": 1540096 + }, + { + "epoch": 1.0, + "learning_rate": 2.4775725040504783e-05, + "loss": 3.918, + "step": 1540608 + }, + { + "epoch": 1.0, + "learning_rate": 2.4767339092994262e-05, + "loss": 3.9135, + "step": 1541120 + }, + { + "epoch": 1.0, + "learning_rate": 2.475896952428747e-05, + "loss": 3.9208, + "step": 1541632 + }, + { + "epoch": 1.01, + "learning_rate": 2.4750583576776955e-05, + "loss": 3.9265, + "step": 1542144 + }, + { + "epoch": 1.01, + "learning_rate": 2.4742197629266435e-05, + "loss": 3.9219, + "step": 1542656 + }, + { + "epoch": 1.01, + "learning_rate": 2.4733811681755915e-05, + "loss": 3.9341, + "step": 1543168 + }, + { + "epoch": 1.01, + "learning_rate": 2.4725442113049124e-05, + "loss": 3.9187, + "step": 1543680 + }, + { + "epoch": 1.01, + "learning_rate": 2.4717072544342336e-05, + "loss": 3.9218, + "step": 1544192 + }, + { + "epoch": 1.01, + "learning_rate": 2.4708686596831816e-05, + "loss": 3.9247, + "step": 1544704 + }, + { + "epoch": 1.01, + "learning_rate": 2.470031702812503e-05, + "loss": 3.9242, + "step": 1545216 + }, + { + "epoch": 1.01, + "learning_rate": 2.469193108061451e-05, + "loss": 3.9089, + "step": 1545728 + }, + { + "epoch": 1.01, + "learning_rate": 2.468354513310399e-05, + "loss": 3.9076, + "step": 1546240 + }, + { + "epoch": 1.01, + "learning_rate": 2.467515918559347e-05, + "loss": 3.9101, + "step": 1546752 + }, + { + "epoch": 1.01, + "learning_rate": 2.4666773238082945e-05, + "loss": 3.9192, + "step": 1547264 + }, + { + "epoch": 1.01, + "learning_rate": 2.4658387290572425e-05, + "loss": 3.9285, + "step": 1547776 + }, + { + "epoch": 1.01, + "learning_rate": 2.465000134306191e-05, + "loss": 3.9183, + "step": 1548288 + }, + { + "epoch": 1.01, + "learning_rate": 2.464161539555139e-05, + "loss": 3.9097, + "step": 1548800 + }, + { + "epoch": 1.01, + "learning_rate": 2.463322944804087e-05, + "loss": 3.9223, + "step": 1549312 + }, + { + "epoch": 1.01, + "learning_rate": 2.462484350053035e-05, + "loss": 3.9174, + "step": 1549824 + }, + { + "epoch": 1.01, + "learning_rate": 2.4616457553019825e-05, + "loss": 3.917, + "step": 1550336 + }, + { + "epoch": 1.01, + "learning_rate": 2.4608071605509305e-05, + "loss": 3.9109, + "step": 1550848 + }, + { + "epoch": 1.01, + "learning_rate": 2.4599685657998785e-05, + "loss": 3.8949, + "step": 1551360 + }, + { + "epoch": 1.01, + "learning_rate": 2.4591299710488265e-05, + "loss": 3.9393, + "step": 1551872 + }, + { + "epoch": 1.01, + "learning_rate": 2.4582930141781477e-05, + "loss": 3.9231, + "step": 1552384 + }, + { + "epoch": 1.01, + "learning_rate": 2.4574544194270957e-05, + "loss": 3.9201, + "step": 1552896 + }, + { + "epoch": 1.01, + "learning_rate": 2.4566158246760437e-05, + "loss": 3.9189, + "step": 1553408 + }, + { + "epoch": 1.01, + "learning_rate": 2.4557772299249917e-05, + "loss": 3.9178, + "step": 1553920 + }, + { + "epoch": 1.01, + "learning_rate": 2.4549402730543126e-05, + "loss": 3.9085, + "step": 1554432 + }, + { + "epoch": 1.01, + "learning_rate": 2.454101678303261e-05, + "loss": 3.92, + "step": 1554944 + }, + { + "epoch": 1.01, + "learning_rate": 2.453263083552209e-05, + "loss": 3.9117, + "step": 1555456 + }, + { + "epoch": 1.01, + "learning_rate": 2.452424488801157e-05, + "loss": 3.9193, + "step": 1555968 + }, + { + "epoch": 1.01, + "learning_rate": 2.451585894050105e-05, + "loss": 3.9352, + "step": 1556480 + }, + { + "epoch": 1.01, + "learning_rate": 2.450748937179426e-05, + "loss": 3.9107, + "step": 1556992 + }, + { + "epoch": 1.01, + "learning_rate": 2.449910342428374e-05, + "loss": 3.9045, + "step": 1557504 + }, + { + "epoch": 1.01, + "learning_rate": 2.449071747677322e-05, + "loss": 3.9277, + "step": 1558016 + }, + { + "epoch": 1.01, + "learning_rate": 2.4482331529262702e-05, + "loss": 3.8991, + "step": 1558528 + }, + { + "epoch": 1.01, + "learning_rate": 2.4473945581752182e-05, + "loss": 3.9096, + "step": 1559040 + }, + { + "epoch": 1.01, + "learning_rate": 2.446555963424166e-05, + "loss": 3.9269, + "step": 1559552 + }, + { + "epoch": 1.01, + "learning_rate": 2.445719006553487e-05, + "loss": 3.9199, + "step": 1560064 + }, + { + "epoch": 1.01, + "learning_rate": 2.444880411802435e-05, + "loss": 3.918, + "step": 1560576 + }, + { + "epoch": 1.01, + "learning_rate": 2.4440418170513827e-05, + "loss": 3.9071, + "step": 1561088 + }, + { + "epoch": 1.01, + "learning_rate": 2.443203222300331e-05, + "loss": 3.8984, + "step": 1561600 + }, + { + "epoch": 1.01, + "learning_rate": 2.442364627549279e-05, + "loss": 3.911, + "step": 1562112 + }, + { + "epoch": 1.01, + "learning_rate": 2.441526032798227e-05, + "loss": 3.926, + "step": 1562624 + }, + { + "epoch": 1.01, + "learning_rate": 2.440687438047175e-05, + "loss": 3.9114, + "step": 1563136 + }, + { + "epoch": 1.01, + "learning_rate": 2.439850481176496e-05, + "loss": 3.9204, + "step": 1563648 + }, + { + "epoch": 1.01, + "learning_rate": 2.439011886425444e-05, + "loss": 3.9217, + "step": 1564160 + }, + { + "epoch": 1.01, + "learning_rate": 2.4381749295547656e-05, + "loss": 3.9307, + "step": 1564672 + }, + { + "epoch": 1.01, + "learning_rate": 2.4373363348037132e-05, + "loss": 3.9127, + "step": 1565184 + }, + { + "epoch": 1.01, + "learning_rate": 2.4364977400526612e-05, + "loss": 3.912, + "step": 1565696 + }, + { + "epoch": 1.01, + "learning_rate": 2.4356591453016092e-05, + "loss": 3.9248, + "step": 1566208 + }, + { + "epoch": 1.01, + "learning_rate": 2.4348205505505572e-05, + "loss": 3.9079, + "step": 1566720 + }, + { + "epoch": 1.01, + "learning_rate": 2.4339819557995052e-05, + "loss": 3.9263, + "step": 1567232 + }, + { + "epoch": 1.01, + "learning_rate": 2.4331433610484532e-05, + "loss": 3.9154, + "step": 1567744 + }, + { + "epoch": 1.01, + "learning_rate": 2.4323047662974012e-05, + "loss": 3.9141, + "step": 1568256 + }, + { + "epoch": 1.01, + "learning_rate": 2.4314678094267224e-05, + "loss": 3.9225, + "step": 1568768 + }, + { + "epoch": 1.01, + "learning_rate": 2.4306292146756704e-05, + "loss": 3.9034, + "step": 1569280 + }, + { + "epoch": 1.01, + "learning_rate": 2.4297906199246184e-05, + "loss": 3.9128, + "step": 1569792 + }, + { + "epoch": 1.01, + "learning_rate": 2.428952025173566e-05, + "loss": 3.9131, + "step": 1570304 + }, + { + "epoch": 1.01, + "learning_rate": 2.428113430422514e-05, + "loss": 3.9213, + "step": 1570816 + }, + { + "epoch": 1.01, + "learning_rate": 2.4272748356714624e-05, + "loss": 3.9128, + "step": 1571328 + }, + { + "epoch": 1.01, + "learning_rate": 2.4264378788007837e-05, + "loss": 3.9167, + "step": 1571840 + }, + { + "epoch": 1.02, + "learning_rate": 2.4255992840497313e-05, + "loss": 3.9046, + "step": 1572352 + }, + { + "epoch": 1.02, + "learning_rate": 2.4247606892986793e-05, + "loss": 3.911, + "step": 1572864 + }, + { + "epoch": 1.02, + "learning_rate": 2.4239220945476273e-05, + "loss": 3.906, + "step": 1573376 + }, + { + "epoch": 1.02, + "learning_rate": 2.4230851376769486e-05, + "loss": 3.9147, + "step": 1573888 + }, + { + "epoch": 1.02, + "learning_rate": 2.4222465429258966e-05, + "loss": 3.927, + "step": 1574400 + }, + { + "epoch": 1.02, + "learning_rate": 2.4214079481748445e-05, + "loss": 3.9217, + "step": 1574912 + }, + { + "epoch": 1.02, + "learning_rate": 2.4205693534237925e-05, + "loss": 3.9194, + "step": 1575424 + }, + { + "epoch": 1.02, + "learning_rate": 2.4197307586727405e-05, + "loss": 3.9091, + "step": 1575936 + }, + { + "epoch": 1.02, + "learning_rate": 2.4188921639216885e-05, + "loss": 3.9134, + "step": 1576448 + }, + { + "epoch": 1.02, + "learning_rate": 2.4180535691706365e-05, + "loss": 3.9142, + "step": 1576960 + }, + { + "epoch": 1.02, + "learning_rate": 2.4172149744195842e-05, + "loss": 3.9247, + "step": 1577472 + }, + { + "epoch": 1.02, + "learning_rate": 2.4163763796685325e-05, + "loss": 3.92, + "step": 1577984 + }, + { + "epoch": 1.02, + "learning_rate": 2.4155394227978538e-05, + "loss": 3.9154, + "step": 1578496 + }, + { + "epoch": 1.02, + "learning_rate": 2.4147008280468018e-05, + "loss": 3.9081, + "step": 1579008 + }, + { + "epoch": 1.02, + "learning_rate": 2.4138622332957494e-05, + "loss": 3.9311, + "step": 1579520 + }, + { + "epoch": 1.02, + "learning_rate": 2.4130252764250707e-05, + "loss": 3.903, + "step": 1580032 + }, + { + "epoch": 1.02, + "learning_rate": 2.4121866816740187e-05, + "loss": 3.9186, + "step": 1580544 + }, + { + "epoch": 1.02, + "learning_rate": 2.4113480869229667e-05, + "loss": 3.9078, + "step": 1581056 + }, + { + "epoch": 1.02, + "learning_rate": 2.4105094921719147e-05, + "loss": 3.914, + "step": 1581568 + }, + { + "epoch": 1.02, + "learning_rate": 2.4096708974208627e-05, + "loss": 3.9057, + "step": 1582080 + }, + { + "epoch": 1.02, + "learning_rate": 2.4088323026698106e-05, + "loss": 3.9179, + "step": 1582592 + }, + { + "epoch": 1.02, + "learning_rate": 2.4079937079187586e-05, + "loss": 3.9074, + "step": 1583104 + }, + { + "epoch": 1.02, + "learning_rate": 2.4071551131677066e-05, + "loss": 3.91, + "step": 1583616 + }, + { + "epoch": 1.02, + "learning_rate": 2.406318156297028e-05, + "loss": 3.9135, + "step": 1584128 + }, + { + "epoch": 1.02, + "learning_rate": 2.405479561545976e-05, + "loss": 3.9115, + "step": 1584640 + }, + { + "epoch": 1.02, + "learning_rate": 2.404640966794924e-05, + "loss": 3.9251, + "step": 1585152 + }, + { + "epoch": 1.02, + "learning_rate": 2.4038040099242448e-05, + "loss": 3.9297, + "step": 1585664 + }, + { + "epoch": 1.02, + "learning_rate": 2.4029654151731928e-05, + "loss": 3.9165, + "step": 1586176 + }, + { + "epoch": 1.02, + "learning_rate": 2.4021268204221408e-05, + "loss": 3.9131, + "step": 1586688 + }, + { + "epoch": 1.02, + "learning_rate": 2.4012882256710888e-05, + "loss": 3.9115, + "step": 1587200 + }, + { + "epoch": 1.02, + "learning_rate": 2.400449630920037e-05, + "loss": 3.9116, + "step": 1587712 + }, + { + "epoch": 1.02, + "learning_rate": 2.3996110361689848e-05, + "loss": 3.9117, + "step": 1588224 + }, + { + "epoch": 1.02, + "learning_rate": 2.3987724414179328e-05, + "loss": 3.9237, + "step": 1588736 + }, + { + "epoch": 1.02, + "learning_rate": 2.3979338466668808e-05, + "loss": 3.8998, + "step": 1589248 + }, + { + "epoch": 1.02, + "learning_rate": 2.397096889796202e-05, + "loss": 3.919, + "step": 1589760 + }, + { + "epoch": 1.02, + "learning_rate": 2.3962582950451497e-05, + "loss": 3.9117, + "step": 1590272 + }, + { + "epoch": 1.02, + "learning_rate": 2.395419700294098e-05, + "loss": 3.9206, + "step": 1590784 + }, + { + "epoch": 1.02, + "learning_rate": 2.394581105543046e-05, + "loss": 3.9104, + "step": 1591296 + }, + { + "epoch": 1.02, + "learning_rate": 2.393744148672367e-05, + "loss": 3.9167, + "step": 1591808 + }, + { + "epoch": 1.02, + "learning_rate": 2.392905553921315e-05, + "loss": 3.9077, + "step": 1592320 + }, + { + "epoch": 1.02, + "learning_rate": 2.392066959170263e-05, + "loss": 3.9233, + "step": 1592832 + }, + { + "epoch": 1.02, + "learning_rate": 2.391228364419211e-05, + "loss": 3.9173, + "step": 1593344 + }, + { + "epoch": 1.02, + "learning_rate": 2.390391407548532e-05, + "loss": 3.9255, + "step": 1593856 + }, + { + "epoch": 1.02, + "learning_rate": 2.38955281279748e-05, + "loss": 3.9046, + "step": 1594368 + }, + { + "epoch": 1.02, + "learning_rate": 2.388714218046428e-05, + "loss": 3.9189, + "step": 1594880 + }, + { + "epoch": 1.02, + "learning_rate": 2.387875623295376e-05, + "loss": 3.922, + "step": 1595392 + }, + { + "epoch": 1.02, + "learning_rate": 2.387037028544324e-05, + "loss": 3.9129, + "step": 1595904 + }, + { + "epoch": 1.02, + "learning_rate": 2.3862017095540183e-05, + "loss": 3.9139, + "step": 1596416 + }, + { + "epoch": 1.02, + "learning_rate": 2.3853631148029666e-05, + "loss": 3.9245, + "step": 1596928 + }, + { + "epoch": 1.02, + "learning_rate": 2.3845245200519143e-05, + "loss": 3.9168, + "step": 1597440 + }, + { + "epoch": 1.02, + "learning_rate": 2.3836859253008623e-05, + "loss": 3.9341, + "step": 1597952 + }, + { + "epoch": 1.02, + "learning_rate": 2.3828473305498103e-05, + "loss": 3.9085, + "step": 1598464 + }, + { + "epoch": 1.02, + "learning_rate": 2.3820103736791315e-05, + "loss": 3.9222, + "step": 1598976 + }, + { + "epoch": 1.02, + "learning_rate": 2.3811717789280795e-05, + "loss": 3.9152, + "step": 1599488 + }, + { + "epoch": 1.02, + "learning_rate": 2.3803331841770275e-05, + "loss": 3.9132, + "step": 1600000 + }, + { + "epoch": 1.02, + "learning_rate": 2.3794945894259755e-05, + "loss": 3.9156, + "step": 1600512 + }, + { + "epoch": 1.02, + "learning_rate": 2.3786576325552967e-05, + "loss": 3.909, + "step": 1601024 + }, + { + "epoch": 1.02, + "learning_rate": 2.3778206756846177e-05, + "loss": 3.9099, + "step": 1601536 + }, + { + "epoch": 1.02, + "learning_rate": 2.3769820809335657e-05, + "loss": 3.9072, + "step": 1602048 + }, + { + "epoch": 1.02, + "learning_rate": 2.3761434861825136e-05, + "loss": 3.9142, + "step": 1602560 + }, + { + "epoch": 1.03, + "eval_loss": 4.002368927001953, + "eval_runtime": 300.4913, + "eval_samples_per_second": 1269.891, + "eval_steps_per_second": 39.685, + "step": 1602705 + }, + { + "epoch": 0.0, + "learning_rate": 2.3753048914314616e-05, + "loss": 3.9174, + "step": 1603072 + }, + { + "epoch": 0.0, + "learning_rate": 2.3744662966804096e-05, + "loss": 3.9128, + "step": 1603584 + }, + { + "epoch": 0.0, + "learning_rate": 2.3736277019293576e-05, + "loss": 3.9202, + "step": 1604096 + }, + { + "epoch": 0.0, + "learning_rate": 2.3727891071783056e-05, + "loss": 3.9174, + "step": 1604608 + }, + { + "epoch": 0.0, + "learning_rate": 2.3719505124272536e-05, + "loss": 3.9283, + "step": 1605120 + }, + { + "epoch": 0.0, + "learning_rate": 2.3711119176762016e-05, + "loss": 3.9147, + "step": 1605632 + }, + { + "epoch": 0.0, + "learning_rate": 2.3702733229251496e-05, + "loss": 3.9114, + "step": 1606144 + }, + { + "epoch": 0.0, + "learning_rate": 2.3694347281740976e-05, + "loss": 3.9055, + "step": 1606656 + }, + { + "epoch": 0.0, + "learning_rate": 2.3685961334230456e-05, + "loss": 3.9145, + "step": 1607168 + }, + { + "epoch": 0.0, + "learning_rate": 2.3677575386719936e-05, + "loss": 3.9151, + "step": 1607680 + }, + { + "epoch": 0.0, + "learning_rate": 2.3669189439209416e-05, + "loss": 3.9172, + "step": 1608192 + }, + { + "epoch": 0.0, + "learning_rate": 2.3660803491698896e-05, + "loss": 3.9153, + "step": 1608704 + }, + { + "epoch": 0.0, + "learning_rate": 2.3652433922992105e-05, + "loss": 3.9102, + "step": 1609216 + }, + { + "epoch": 0.0, + "learning_rate": 2.364404797548159e-05, + "loss": 3.9141, + "step": 1609728 + }, + { + "epoch": 0.0, + "learning_rate": 2.363566202797107e-05, + "loss": 3.9075, + "step": 1610240 + }, + { + "epoch": 0.0, + "learning_rate": 2.362727608046055e-05, + "loss": 3.902, + "step": 1610752 + }, + { + "epoch": 0.0, + "learning_rate": 2.3618890132950028e-05, + "loss": 3.9095, + "step": 1611264 + }, + { + "epoch": 0.0, + "learning_rate": 2.3610504185439505e-05, + "loss": 3.9133, + "step": 1611776 + }, + { + "epoch": 0.0, + "learning_rate": 2.3602118237928985e-05, + "loss": 3.9135, + "step": 1612288 + }, + { + "epoch": 0.0, + "learning_rate": 2.3593732290418465e-05, + "loss": 3.9295, + "step": 1612800 + }, + { + "epoch": 0.0, + "learning_rate": 2.3585346342907945e-05, + "loss": 3.9153, + "step": 1613312 + }, + { + "epoch": 0.0, + "learning_rate": 2.3576976774201157e-05, + "loss": 3.9219, + "step": 1613824 + }, + { + "epoch": 0.0, + "learning_rate": 2.3568590826690637e-05, + "loss": 3.909, + "step": 1614336 + }, + { + "epoch": 0.0, + "learning_rate": 2.3560204879180117e-05, + "loss": 3.9188, + "step": 1614848 + }, + { + "epoch": 0.0, + "learning_rate": 2.3551818931669597e-05, + "loss": 3.9109, + "step": 1615360 + }, + { + "epoch": 0.0, + "learning_rate": 2.3543449362962806e-05, + "loss": 3.9202, + "step": 1615872 + }, + { + "epoch": 0.0, + "learning_rate": 2.353506341545229e-05, + "loss": 3.9102, + "step": 1616384 + }, + { + "epoch": 0.0, + "learning_rate": 2.352667746794177e-05, + "loss": 3.9073, + "step": 1616896 + }, + { + "epoch": 0.0, + "learning_rate": 2.351829152043125e-05, + "loss": 3.9071, + "step": 1617408 + }, + { + "epoch": 0.0, + "learning_rate": 2.350990557292073e-05, + "loss": 3.9133, + "step": 1617920 + }, + { + "epoch": 0.01, + "learning_rate": 2.350151962541021e-05, + "loss": 3.9155, + "step": 1618432 + }, + { + "epoch": 0.01, + "learning_rate": 2.3493133677899686e-05, + "loss": 3.9172, + "step": 1618944 + }, + { + "epoch": 0.01, + "learning_rate": 2.3484747730389166e-05, + "loss": 3.9233, + "step": 1619456 + }, + { + "epoch": 0.01, + "learning_rate": 2.347636178287865e-05, + "loss": 3.9133, + "step": 1619968 + }, + { + "epoch": 0.01, + "learning_rate": 2.346799221417186e-05, + "loss": 3.9112, + "step": 1620480 + }, + { + "epoch": 0.01, + "learning_rate": 2.345962264546507e-05, + "loss": 3.9252, + "step": 1620992 + }, + { + "epoch": 0.01, + "learning_rate": 2.345123669795455e-05, + "loss": 3.9171, + "step": 1621504 + }, + { + "epoch": 0.01, + "learning_rate": 2.344285075044403e-05, + "loss": 3.8997, + "step": 1622016 + }, + { + "epoch": 0.01, + "learning_rate": 2.3434481181737243e-05, + "loss": 3.9032, + "step": 1622528 + }, + { + "epoch": 0.01, + "learning_rate": 2.3426095234226723e-05, + "loss": 3.9018, + "step": 1623040 + }, + { + "epoch": 0.01, + "learning_rate": 2.3417709286716203e-05, + "loss": 3.906, + "step": 1623552 + }, + { + "epoch": 0.01, + "learning_rate": 2.3409323339205683e-05, + "loss": 3.9223, + "step": 1624064 + }, + { + "epoch": 0.01, + "learning_rate": 2.340093739169516e-05, + "loss": 3.909, + "step": 1624576 + }, + { + "epoch": 0.01, + "learning_rate": 2.339255144418464e-05, + "loss": 3.9052, + "step": 1625088 + }, + { + "epoch": 0.01, + "learning_rate": 2.338416549667412e-05, + "loss": 3.9155, + "step": 1625600 + }, + { + "epoch": 0.01, + "learning_rate": 2.33757795491636e-05, + "loss": 3.907, + "step": 1626112 + }, + { + "epoch": 0.01, + "learning_rate": 2.3367409980456812e-05, + "loss": 3.9124, + "step": 1626624 + }, + { + "epoch": 0.01, + "learning_rate": 2.3359024032946292e-05, + "loss": 3.9028, + "step": 1627136 + }, + { + "epoch": 0.01, + "learning_rate": 2.3350638085435772e-05, + "loss": 3.8851, + "step": 1627648 + }, + { + "epoch": 0.01, + "learning_rate": 2.3342252137925252e-05, + "loss": 3.9358, + "step": 1628160 + }, + { + "epoch": 0.01, + "learning_rate": 2.3333866190414732e-05, + "loss": 3.9141, + "step": 1628672 + }, + { + "epoch": 0.01, + "learning_rate": 2.3325496621707944e-05, + "loss": 3.9181, + "step": 1629184 + }, + { + "epoch": 0.01, + "learning_rate": 2.3317110674197424e-05, + "loss": 3.9074, + "step": 1629696 + }, + { + "epoch": 0.01, + "learning_rate": 2.3308724726686904e-05, + "loss": 3.912, + "step": 1630208 + }, + { + "epoch": 0.01, + "learning_rate": 2.3300338779176384e-05, + "loss": 3.9037, + "step": 1630720 + }, + { + "epoch": 0.01, + "learning_rate": 2.3291952831665864e-05, + "loss": 3.9086, + "step": 1631232 + }, + { + "epoch": 0.01, + "learning_rate": 2.3283583262959073e-05, + "loss": 3.9069, + "step": 1631744 + }, + { + "epoch": 0.01, + "learning_rate": 2.3275197315448553e-05, + "loss": 3.9137, + "step": 1632256 + }, + { + "epoch": 0.01, + "learning_rate": 2.3266811367938036e-05, + "loss": 3.9265, + "step": 1632768 + }, + { + "epoch": 0.01, + "learning_rate": 2.3258425420427516e-05, + "loss": 3.9034, + "step": 1633280 + }, + { + "epoch": 0.01, + "learning_rate": 2.3250039472916993e-05, + "loss": 3.8979, + "step": 1633792 + }, + { + "epoch": 0.01, + "learning_rate": 2.3241653525406473e-05, + "loss": 3.9208, + "step": 1634304 + }, + { + "epoch": 0.01, + "learning_rate": 2.3233267577895953e-05, + "loss": 3.8916, + "step": 1634816 + }, + { + "epoch": 0.01, + "learning_rate": 2.3224881630385433e-05, + "loss": 3.8988, + "step": 1635328 + }, + { + "epoch": 0.01, + "learning_rate": 2.3216512061678645e-05, + "loss": 3.9175, + "step": 1635840 + }, + { + "epoch": 0.01, + "learning_rate": 2.3208126114168125e-05, + "loss": 3.9201, + "step": 1636352 + }, + { + "epoch": 0.01, + "learning_rate": 2.3199740166657605e-05, + "loss": 3.9057, + "step": 1636864 + }, + { + "epoch": 0.01, + "learning_rate": 2.3191354219147085e-05, + "loss": 3.9046, + "step": 1637376 + }, + { + "epoch": 0.01, + "learning_rate": 2.3182968271636565e-05, + "loss": 3.889, + "step": 1637888 + }, + { + "epoch": 0.01, + "learning_rate": 2.3174582324126045e-05, + "loss": 3.9055, + "step": 1638400 + }, + { + "epoch": 0.01, + "learning_rate": 2.316619637661552e-05, + "loss": 3.9169, + "step": 1638912 + }, + { + "epoch": 0.01, + "learning_rate": 2.3157826807908738e-05, + "loss": 3.9024, + "step": 1639424 + }, + { + "epoch": 0.01, + "learning_rate": 2.3149440860398218e-05, + "loss": 3.9113, + "step": 1639936 + }, + { + "epoch": 0.01, + "learning_rate": 2.3141071291691427e-05, + "loss": 3.9144, + "step": 1640448 + }, + { + "epoch": 0.01, + "learning_rate": 2.3132685344180907e-05, + "loss": 3.9251, + "step": 1640960 + }, + { + "epoch": 0.01, + "learning_rate": 2.3124299396670387e-05, + "loss": 3.9049, + "step": 1641472 + }, + { + "epoch": 0.01, + "learning_rate": 2.3115913449159866e-05, + "loss": 3.9058, + "step": 1641984 + }, + { + "epoch": 0.01, + "learning_rate": 2.3107527501649346e-05, + "loss": 3.9148, + "step": 1642496 + }, + { + "epoch": 0.01, + "learning_rate": 2.3099141554138826e-05, + "loss": 3.901, + "step": 1643008 + }, + { + "epoch": 0.01, + "learning_rate": 2.3090755606628306e-05, + "loss": 3.9153, + "step": 1643520 + }, + { + "epoch": 0.01, + "learning_rate": 2.3082369659117786e-05, + "loss": 3.9084, + "step": 1644032 + }, + { + "epoch": 0.01, + "learning_rate": 2.3074000090410995e-05, + "loss": 3.9082, + "step": 1644544 + }, + { + "epoch": 0.01, + "learning_rate": 2.3065614142900475e-05, + "loss": 3.9165, + "step": 1645056 + }, + { + "epoch": 0.01, + "learning_rate": 2.305722819538996e-05, + "loss": 3.8969, + "step": 1645568 + }, + { + "epoch": 0.01, + "learning_rate": 2.304884224787944e-05, + "loss": 3.9013, + "step": 1646080 + }, + { + "epoch": 0.01, + "learning_rate": 2.3040472679172648e-05, + "loss": 3.9092, + "step": 1646592 + }, + { + "epoch": 0.01, + "learning_rate": 2.303210311046586e-05, + "loss": 3.9158, + "step": 1647104 + }, + { + "epoch": 0.01, + "learning_rate": 2.302371716295534e-05, + "loss": 3.9022, + "step": 1647616 + }, + { + "epoch": 0.01, + "learning_rate": 2.3015347594248553e-05, + "loss": 3.911, + "step": 1648128 + }, + { + "epoch": 0.02, + "learning_rate": 2.3006961646738033e-05, + "loss": 3.8977, + "step": 1648640 + }, + { + "epoch": 0.02, + "learning_rate": 2.2998575699227513e-05, + "loss": 3.9019, + "step": 1649152 + }, + { + "epoch": 0.02, + "learning_rate": 2.2990189751716993e-05, + "loss": 3.8989, + "step": 1649664 + }, + { + "epoch": 0.02, + "learning_rate": 2.298180380420647e-05, + "loss": 3.91, + "step": 1650176 + }, + { + "epoch": 0.02, + "learning_rate": 2.297341785669595e-05, + "loss": 3.9187, + "step": 1650688 + }, + { + "epoch": 0.02, + "learning_rate": 2.296503190918543e-05, + "loss": 3.9118, + "step": 1651200 + }, + { + "epoch": 0.02, + "learning_rate": 2.2956645961674912e-05, + "loss": 3.9134, + "step": 1651712 + }, + { + "epoch": 0.02, + "learning_rate": 2.2948260014164392e-05, + "loss": 3.9005, + "step": 1652224 + }, + { + "epoch": 0.02, + "learning_rate": 2.2939874066653872e-05, + "loss": 3.9041, + "step": 1652736 + }, + { + "epoch": 0.02, + "learning_rate": 2.2931488119143352e-05, + "loss": 3.9058, + "step": 1653248 + }, + { + "epoch": 0.02, + "learning_rate": 2.292311855043656e-05, + "loss": 3.9158, + "step": 1653760 + }, + { + "epoch": 0.02, + "learning_rate": 2.291473260292604e-05, + "loss": 3.9102, + "step": 1654272 + }, + { + "epoch": 0.02, + "learning_rate": 2.290634665541552e-05, + "loss": 3.9085, + "step": 1654784 + }, + { + "epoch": 0.02, + "learning_rate": 2.2897960707905e-05, + "loss": 3.9064, + "step": 1655296 + }, + { + "epoch": 0.02, + "learning_rate": 2.288957476039448e-05, + "loss": 3.9195, + "step": 1655808 + }, + { + "epoch": 0.02, + "learning_rate": 2.2881205191687694e-05, + "loss": 3.8951, + "step": 1656320 + }, + { + "epoch": 0.02, + "learning_rate": 2.2872819244177174e-05, + "loss": 3.9108, + "step": 1656832 + }, + { + "epoch": 0.02, + "learning_rate": 2.286443329666665e-05, + "loss": 3.8974, + "step": 1657344 + }, + { + "epoch": 0.02, + "learning_rate": 2.285604734915613e-05, + "loss": 3.9124, + "step": 1657856 + }, + { + "epoch": 0.02, + "learning_rate": 2.2847661401645613e-05, + "loss": 3.8998, + "step": 1658368 + }, + { + "epoch": 0.02, + "learning_rate": 2.2839275454135093e-05, + "loss": 3.9088, + "step": 1658880 + }, + { + "epoch": 0.02, + "learning_rate": 2.2830905885428303e-05, + "loss": 3.8987, + "step": 1659392 + }, + { + "epoch": 0.02, + "learning_rate": 2.2822519937917782e-05, + "loss": 3.9055, + "step": 1659904 + }, + { + "epoch": 0.02, + "learning_rate": 2.2814133990407262e-05, + "loss": 3.9061, + "step": 1660416 + }, + { + "epoch": 0.02, + "learning_rate": 2.2805748042896742e-05, + "loss": 3.9036, + "step": 1660928 + }, + { + "epoch": 0.02, + "learning_rate": 2.2797378474189955e-05, + "loss": 3.9152, + "step": 1661440 + }, + { + "epoch": 0.02, + "learning_rate": 2.2788992526679435e-05, + "loss": 3.9226, + "step": 1661952 + }, + { + "epoch": 0.02, + "learning_rate": 2.2780606579168915e-05, + "loss": 3.9104, + "step": 1662464 + }, + { + "epoch": 0.02, + "learning_rate": 2.2772220631658395e-05, + "loss": 3.9096, + "step": 1662976 + }, + { + "epoch": 0.02, + "learning_rate": 2.2763834684147875e-05, + "loss": 3.8996, + "step": 1663488 + }, + { + "epoch": 0.02, + "learning_rate": 2.2755448736637355e-05, + "loss": 3.9026, + "step": 1664000 + }, + { + "epoch": 0.02, + "learning_rate": 2.2747079167930567e-05, + "loss": 3.9078, + "step": 1664512 + }, + { + "epoch": 0.02, + "learning_rate": 2.2738693220420047e-05, + "loss": 3.9189, + "step": 1665024 + }, + { + "epoch": 0.02, + "learning_rate": 2.2730307272909527e-05, + "loss": 3.8904, + "step": 1665536 + }, + { + "epoch": 0.02, + "learning_rate": 2.2721921325399004e-05, + "loss": 3.9094, + "step": 1666048 + }, + { + "epoch": 0.02, + "learning_rate": 2.2713535377888484e-05, + "loss": 3.9084, + "step": 1666560 + }, + { + "epoch": 0.02, + "learning_rate": 2.2705149430377964e-05, + "loss": 3.9092, + "step": 1667072 + }, + { + "epoch": 0.02, + "learning_rate": 2.2696779861671176e-05, + "loss": 3.9036, + "step": 1667584 + }, + { + "epoch": 0.02, + "learning_rate": 2.2688393914160656e-05, + "loss": 3.9067, + "step": 1668096 + }, + { + "epoch": 0.02, + "learning_rate": 2.2680007966650136e-05, + "loss": 3.9057, + "step": 1668608 + }, + { + "epoch": 0.02, + "learning_rate": 2.2671622019139616e-05, + "loss": 3.9135, + "step": 1669120 + }, + { + "epoch": 0.02, + "learning_rate": 2.2663236071629096e-05, + "loss": 3.9134, + "step": 1669632 + }, + { + "epoch": 0.02, + "learning_rate": 2.2654850124118576e-05, + "loss": 3.9175, + "step": 1670144 + }, + { + "epoch": 0.02, + "learning_rate": 2.2646464176608056e-05, + "loss": 3.8973, + "step": 1670656 + }, + { + "epoch": 0.02, + "learning_rate": 2.2638094607901268e-05, + "loss": 3.9125, + "step": 1671168 + }, + { + "epoch": 0.02, + "learning_rate": 2.2629708660390748e-05, + "loss": 3.9142, + "step": 1671680 + }, + { + "epoch": 0.02, + "learning_rate": 2.2621322712880228e-05, + "loss": 3.9067, + "step": 1672192 + }, + { + "epoch": 0.02, + "learning_rate": 2.2612936765369708e-05, + "loss": 3.91, + "step": 1672704 + }, + { + "epoch": 0.02, + "learning_rate": 2.2604567196662917e-05, + "loss": 3.9155, + "step": 1673216 + }, + { + "epoch": 0.02, + "learning_rate": 2.2596181249152397e-05, + "loss": 3.9085, + "step": 1673728 + }, + { + "epoch": 0.02, + "learning_rate": 2.2587795301641877e-05, + "loss": 3.9289, + "step": 1674240 + }, + { + "epoch": 0.02, + "learning_rate": 2.257940935413136e-05, + "loss": 3.9056, + "step": 1674752 + }, + { + "epoch": 0.02, + "learning_rate": 2.2571023406620837e-05, + "loss": 3.9108, + "step": 1675264 + }, + { + "epoch": 0.02, + "learning_rate": 2.2562637459110317e-05, + "loss": 3.9056, + "step": 1675776 + }, + { + "epoch": 0.02, + "learning_rate": 2.2554251511599797e-05, + "loss": 3.9098, + "step": 1676288 + }, + { + "epoch": 0.02, + "learning_rate": 2.2545865564089277e-05, + "loss": 3.9082, + "step": 1676800 + }, + { + "epoch": 0.02, + "learning_rate": 2.2537512374186222e-05, + "loss": 3.9031, + "step": 1677312 + }, + { + "epoch": 0.02, + "learning_rate": 2.2529126426675702e-05, + "loss": 3.9071, + "step": 1677824 + }, + { + "epoch": 0.02, + "learning_rate": 2.2520740479165182e-05, + "loss": 3.8955, + "step": 1678336 + }, + { + "epoch": 0.02, + "learning_rate": 2.251235453165466e-05, + "loss": 3.9116, + "step": 1678848 + }, + { + "epoch": 0.03, + "eval_loss": 3.999638557434082, + "eval_runtime": 410.5382, + "eval_samples_per_second": 929.49, + "eval_steps_per_second": 29.047, + "step": 1679025 + }, + { + "epoch": 1.0, + "learning_rate": 2.250396858414414e-05, + "loss": 3.9139, + "step": 1679360 + }, + { + "epoch": 1.0, + "learning_rate": 2.2495582636633618e-05, + "loss": 3.9031, + "step": 1679872 + }, + { + "epoch": 1.0, + "learning_rate": 2.2487196689123098e-05, + "loss": 3.9116, + "step": 1680384 + }, + { + "epoch": 1.0, + "learning_rate": 2.2478810741612578e-05, + "loss": 3.9114, + "step": 1680896 + }, + { + "epoch": 1.0, + "learning_rate": 2.247042479410206e-05, + "loss": 3.9222, + "step": 1681408 + }, + { + "epoch": 1.0, + "learning_rate": 2.246203884659154e-05, + "loss": 3.9043, + "step": 1681920 + }, + { + "epoch": 1.0, + "learning_rate": 2.245366927788475e-05, + "loss": 3.9097, + "step": 1682432 + }, + { + "epoch": 1.0, + "learning_rate": 2.244528333037423e-05, + "loss": 3.8935, + "step": 1682944 + }, + { + "epoch": 1.0, + "learning_rate": 2.243689738286371e-05, + "loss": 3.9086, + "step": 1683456 + }, + { + "epoch": 1.0, + "learning_rate": 2.242851143535319e-05, + "loss": 3.9119, + "step": 1683968 + }, + { + "epoch": 1.0, + "learning_rate": 2.242012548784267e-05, + "loss": 3.9084, + "step": 1684480 + }, + { + "epoch": 1.0, + "learning_rate": 2.241173954033215e-05, + "loss": 3.9087, + "step": 1684992 + }, + { + "epoch": 1.0, + "learning_rate": 2.2403369971625363e-05, + "loss": 3.9037, + "step": 1685504 + }, + { + "epoch": 1.0, + "learning_rate": 2.2395000402918572e-05, + "loss": 3.9095, + "step": 1686016 + }, + { + "epoch": 1.0, + "learning_rate": 2.2386614455408052e-05, + "loss": 3.9014, + "step": 1686528 + }, + { + "epoch": 1.0, + "learning_rate": 2.2378228507897532e-05, + "loss": 3.8957, + "step": 1687040 + }, + { + "epoch": 1.0, + "learning_rate": 2.2369842560387015e-05, + "loss": 3.901, + "step": 1687552 + }, + { + "epoch": 1.0, + "learning_rate": 2.2361456612876492e-05, + "loss": 3.906, + "step": 1688064 + }, + { + "epoch": 1.0, + "learning_rate": 2.2353070665365972e-05, + "loss": 3.9101, + "step": 1688576 + }, + { + "epoch": 1.0, + "learning_rate": 2.234468471785545e-05, + "loss": 3.9167, + "step": 1689088 + }, + { + "epoch": 1.0, + "learning_rate": 2.233629877034493e-05, + "loss": 3.9112, + "step": 1689600 + }, + { + "epoch": 1.0, + "learning_rate": 2.232791282283441e-05, + "loss": 3.9162, + "step": 1690112 + }, + { + "epoch": 1.0, + "learning_rate": 2.2319543254127624e-05, + "loss": 3.9023, + "step": 1690624 + }, + { + "epoch": 1.0, + "learning_rate": 2.2311157306617104e-05, + "loss": 3.9112, + "step": 1691136 + }, + { + "epoch": 1.0, + "learning_rate": 2.2302771359106584e-05, + "loss": 3.9074, + "step": 1691648 + }, + { + "epoch": 1.0, + "learning_rate": 2.2294385411596064e-05, + "loss": 3.9139, + "step": 1692160 + }, + { + "epoch": 1.0, + "learning_rate": 2.2285999464085544e-05, + "loss": 3.9026, + "step": 1692672 + }, + { + "epoch": 1.0, + "learning_rate": 2.227761351657502e-05, + "loss": 3.9025, + "step": 1693184 + }, + { + "epoch": 1.0, + "learning_rate": 2.22692275690645e-05, + "loss": 3.9008, + "step": 1693696 + }, + { + "epoch": 1.0, + "learning_rate": 2.2260841621553984e-05, + "loss": 3.9054, + "step": 1694208 + }, + { + "epoch": 1.01, + "learning_rate": 2.2252472052847196e-05, + "loss": 3.9054, + "step": 1694720 + }, + { + "epoch": 1.01, + "learning_rate": 2.2244086105336673e-05, + "loss": 3.9134, + "step": 1695232 + }, + { + "epoch": 1.01, + "learning_rate": 2.2235700157826153e-05, + "loss": 3.9139, + "step": 1695744 + }, + { + "epoch": 1.01, + "learning_rate": 2.2227314210315633e-05, + "loss": 3.9088, + "step": 1696256 + }, + { + "epoch": 1.01, + "learning_rate": 2.2218944641608845e-05, + "loss": 3.911, + "step": 1696768 + }, + { + "epoch": 1.01, + "learning_rate": 2.2210575072902058e-05, + "loss": 3.911, + "step": 1697280 + }, + { + "epoch": 1.01, + "learning_rate": 2.2202189125391538e-05, + "loss": 3.9095, + "step": 1697792 + }, + { + "epoch": 1.01, + "learning_rate": 2.2193803177881018e-05, + "loss": 3.8986, + "step": 1698304 + }, + { + "epoch": 1.01, + "learning_rate": 2.2185417230370494e-05, + "loss": 3.8957, + "step": 1698816 + }, + { + "epoch": 1.01, + "learning_rate": 2.2177031282859974e-05, + "loss": 3.8958, + "step": 1699328 + }, + { + "epoch": 1.01, + "learning_rate": 2.2168645335349454e-05, + "loss": 3.8996, + "step": 1699840 + }, + { + "epoch": 1.01, + "learning_rate": 2.2160259387838937e-05, + "loss": 3.9127, + "step": 1700352 + }, + { + "epoch": 1.01, + "learning_rate": 2.2151873440328417e-05, + "loss": 3.9043, + "step": 1700864 + }, + { + "epoch": 1.01, + "learning_rate": 2.2143487492817897e-05, + "loss": 3.9013, + "step": 1701376 + }, + { + "epoch": 1.01, + "learning_rate": 2.2135101545307377e-05, + "loss": 3.9048, + "step": 1701888 + }, + { + "epoch": 1.01, + "learning_rate": 2.2126715597796854e-05, + "loss": 3.9044, + "step": 1702400 + }, + { + "epoch": 1.01, + "learning_rate": 2.2118346029090066e-05, + "loss": 3.904, + "step": 1702912 + }, + { + "epoch": 1.01, + "learning_rate": 2.2109960081579546e-05, + "loss": 3.8975, + "step": 1703424 + }, + { + "epoch": 1.01, + "learning_rate": 2.210159051287276e-05, + "loss": 3.8786, + "step": 1703936 + }, + { + "epoch": 1.01, + "learning_rate": 2.209320456536224e-05, + "loss": 3.9276, + "step": 1704448 + }, + { + "epoch": 1.01, + "learning_rate": 2.208481861785172e-05, + "loss": 3.9051, + "step": 1704960 + }, + { + "epoch": 1.01, + "learning_rate": 2.20764326703412e-05, + "loss": 3.914, + "step": 1705472 + }, + { + "epoch": 1.01, + "learning_rate": 2.2068046722830675e-05, + "loss": 3.9009, + "step": 1705984 + }, + { + "epoch": 1.01, + "learning_rate": 2.2059660775320155e-05, + "loss": 3.9026, + "step": 1706496 + }, + { + "epoch": 1.01, + "learning_rate": 2.205127482780964e-05, + "loss": 3.8964, + "step": 1707008 + }, + { + "epoch": 1.01, + "learning_rate": 2.204290525910285e-05, + "loss": 3.9031, + "step": 1707520 + }, + { + "epoch": 1.01, + "learning_rate": 2.2034519311592328e-05, + "loss": 3.9041, + "step": 1708032 + }, + { + "epoch": 1.01, + "learning_rate": 2.2026133364081808e-05, + "loss": 3.9004, + "step": 1708544 + }, + { + "epoch": 1.01, + "learning_rate": 2.2017747416571287e-05, + "loss": 3.9238, + "step": 1709056 + }, + { + "epoch": 1.01, + "learning_rate": 2.2009361469060767e-05, + "loss": 3.8941, + "step": 1709568 + }, + { + "epoch": 1.01, + "learning_rate": 2.2000975521550247e-05, + "loss": 3.8919, + "step": 1710080 + }, + { + "epoch": 1.01, + "learning_rate": 2.199260595284346e-05, + "loss": 3.9175, + "step": 1710592 + }, + { + "epoch": 1.01, + "learning_rate": 2.198422000533294e-05, + "loss": 3.8838, + "step": 1711104 + }, + { + "epoch": 1.01, + "learning_rate": 2.197583405782242e-05, + "loss": 3.8899, + "step": 1711616 + }, + { + "epoch": 1.01, + "learning_rate": 2.19674481103119e-05, + "loss": 3.9128, + "step": 1712128 + }, + { + "epoch": 1.01, + "learning_rate": 2.195906216280138e-05, + "loss": 3.9101, + "step": 1712640 + }, + { + "epoch": 1.01, + "learning_rate": 2.1950676215290856e-05, + "loss": 3.8956, + "step": 1713152 + }, + { + "epoch": 1.01, + "learning_rate": 2.194229026778034e-05, + "loss": 3.9018, + "step": 1713664 + }, + { + "epoch": 1.01, + "learning_rate": 2.193390432026982e-05, + "loss": 3.8816, + "step": 1714176 + }, + { + "epoch": 1.01, + "learning_rate": 2.1925534751563032e-05, + "loss": 3.9023, + "step": 1714688 + }, + { + "epoch": 1.01, + "learning_rate": 2.191714880405251e-05, + "loss": 3.9067, + "step": 1715200 + }, + { + "epoch": 1.01, + "learning_rate": 2.190876285654199e-05, + "loss": 3.8952, + "step": 1715712 + }, + { + "epoch": 1.01, + "learning_rate": 2.19003932878352e-05, + "loss": 3.9045, + "step": 1716224 + }, + { + "epoch": 1.01, + "learning_rate": 2.189200734032468e-05, + "loss": 3.9062, + "step": 1716736 + }, + { + "epoch": 1.01, + "learning_rate": 2.188362139281416e-05, + "loss": 3.9185, + "step": 1717248 + }, + { + "epoch": 1.01, + "learning_rate": 2.187523544530364e-05, + "loss": 3.8983, + "step": 1717760 + }, + { + "epoch": 1.01, + "learning_rate": 2.186684949779312e-05, + "loss": 3.8984, + "step": 1718272 + }, + { + "epoch": 1.01, + "learning_rate": 2.18584635502826e-05, + "loss": 3.9127, + "step": 1718784 + }, + { + "epoch": 1.01, + "learning_rate": 2.185007760277208e-05, + "loss": 3.8951, + "step": 1719296 + }, + { + "epoch": 1.01, + "learning_rate": 2.184169165526156e-05, + "loss": 3.9046, + "step": 1719808 + }, + { + "epoch": 1.01, + "learning_rate": 2.183330570775104e-05, + "loss": 3.9076, + "step": 1720320 + }, + { + "epoch": 1.01, + "learning_rate": 2.182491976024052e-05, + "loss": 3.8989, + "step": 1720832 + }, + { + "epoch": 1.01, + "learning_rate": 2.181653381273e-05, + "loss": 3.9118, + "step": 1721344 + }, + { + "epoch": 1.01, + "learning_rate": 2.1808164244023213e-05, + "loss": 3.8908, + "step": 1721856 + }, + { + "epoch": 1.01, + "learning_rate": 2.1799794675316422e-05, + "loss": 3.8932, + "step": 1722368 + }, + { + "epoch": 1.01, + "learning_rate": 2.1791408727805902e-05, + "loss": 3.9025, + "step": 1722880 + }, + { + "epoch": 1.01, + "learning_rate": 2.1783022780295385e-05, + "loss": 3.9072, + "step": 1723392 + }, + { + "epoch": 1.01, + "learning_rate": 2.1774636832784862e-05, + "loss": 3.8984, + "step": 1723904 + }, + { + "epoch": 1.01, + "learning_rate": 2.1766267264078075e-05, + "loss": 3.903, + "step": 1724416 + }, + { + "epoch": 1.02, + "learning_rate": 2.1757881316567554e-05, + "loss": 3.8883, + "step": 1724928 + }, + { + "epoch": 1.02, + "learning_rate": 2.1749495369057034e-05, + "loss": 3.8984, + "step": 1725440 + }, + { + "epoch": 1.02, + "learning_rate": 2.1741109421546514e-05, + "loss": 3.8912, + "step": 1725952 + }, + { + "epoch": 1.02, + "learning_rate": 2.1732723474035994e-05, + "loss": 3.9025, + "step": 1726464 + }, + { + "epoch": 1.02, + "learning_rate": 2.1724337526525474e-05, + "loss": 3.9158, + "step": 1726976 + }, + { + "epoch": 1.02, + "learning_rate": 2.1715967957818687e-05, + "loss": 3.9033, + "step": 1727488 + }, + { + "epoch": 1.02, + "learning_rate": 2.1707582010308163e-05, + "loss": 3.9041, + "step": 1728000 + }, + { + "epoch": 1.02, + "learning_rate": 2.1699196062797643e-05, + "loss": 3.8958, + "step": 1728512 + }, + { + "epoch": 1.02, + "learning_rate": 2.1690810115287123e-05, + "loss": 3.8972, + "step": 1729024 + }, + { + "epoch": 1.02, + "learning_rate": 2.1682424167776603e-05, + "loss": 3.8981, + "step": 1729536 + }, + { + "epoch": 1.02, + "learning_rate": 2.1674038220266087e-05, + "loss": 3.9092, + "step": 1730048 + }, + { + "epoch": 1.02, + "learning_rate": 2.1665652272755567e-05, + "loss": 3.9052, + "step": 1730560 + }, + { + "epoch": 1.02, + "learning_rate": 2.1657266325245043e-05, + "loss": 3.9058, + "step": 1731072 + }, + { + "epoch": 1.02, + "learning_rate": 2.1648896756538256e-05, + "loss": 3.8946, + "step": 1731584 + }, + { + "epoch": 1.02, + "learning_rate": 2.1640510809027736e-05, + "loss": 3.9104, + "step": 1732096 + }, + { + "epoch": 1.02, + "learning_rate": 2.1632124861517216e-05, + "loss": 3.8925, + "step": 1732608 + }, + { + "epoch": 1.02, + "learning_rate": 2.1623738914006695e-05, + "loss": 3.9037, + "step": 1733120 + }, + { + "epoch": 1.02, + "learning_rate": 2.1615369345299908e-05, + "loss": 3.888, + "step": 1733632 + }, + { + "epoch": 1.02, + "learning_rate": 2.1606983397789388e-05, + "loss": 3.9059, + "step": 1734144 + }, + { + "epoch": 1.02, + "learning_rate": 2.1598597450278864e-05, + "loss": 3.894, + "step": 1734656 + }, + { + "epoch": 1.02, + "learning_rate": 2.1590211502768344e-05, + "loss": 3.899, + "step": 1735168 + }, + { + "epoch": 1.02, + "learning_rate": 2.1581825555257824e-05, + "loss": 3.8918, + "step": 1735680 + }, + { + "epoch": 1.02, + "learning_rate": 2.1573439607747308e-05, + "loss": 3.901, + "step": 1736192 + }, + { + "epoch": 1.02, + "learning_rate": 2.1565070039040517e-05, + "loss": 3.897, + "step": 1736704 + }, + { + "epoch": 1.02, + "learning_rate": 2.1556684091529997e-05, + "loss": 3.8943, + "step": 1737216 + }, + { + "epoch": 1.02, + "learning_rate": 2.1548298144019477e-05, + "loss": 3.9103, + "step": 1737728 + }, + { + "epoch": 1.02, + "learning_rate": 2.1539912196508957e-05, + "loss": 3.9153, + "step": 1738240 + }, + { + "epoch": 1.02, + "learning_rate": 2.1531526248998437e-05, + "loss": 3.9057, + "step": 1738752 + }, + { + "epoch": 1.02, + "learning_rate": 2.152315668029165e-05, + "loss": 3.9038, + "step": 1739264 + }, + { + "epoch": 1.02, + "learning_rate": 2.151477073278113e-05, + "loss": 3.8969, + "step": 1739776 + }, + { + "epoch": 1.02, + "learning_rate": 2.150638478527061e-05, + "loss": 3.8965, + "step": 1740288 + }, + { + "epoch": 1.02, + "learning_rate": 2.149799883776009e-05, + "loss": 3.903, + "step": 1740800 + }, + { + "epoch": 1.02, + "learning_rate": 2.148961289024957e-05, + "loss": 3.9073, + "step": 1741312 + }, + { + "epoch": 1.02, + "learning_rate": 2.1481226942739046e-05, + "loss": 3.8847, + "step": 1741824 + }, + { + "epoch": 1.02, + "learning_rate": 2.1472840995228525e-05, + "loss": 3.9054, + "step": 1742336 + }, + { + "epoch": 1.02, + "learning_rate": 2.146445504771801e-05, + "loss": 3.9033, + "step": 1742848 + }, + { + "epoch": 1.02, + "learning_rate": 2.145608547901122e-05, + "loss": 3.9025, + "step": 1743360 + }, + { + "epoch": 1.02, + "learning_rate": 2.1447699531500698e-05, + "loss": 3.8976, + "step": 1743872 + }, + { + "epoch": 1.02, + "learning_rate": 2.143932996279391e-05, + "loss": 3.9022, + "step": 1744384 + }, + { + "epoch": 1.02, + "learning_rate": 2.143094401528339e-05, + "loss": 3.8955, + "step": 1744896 + }, + { + "epoch": 1.02, + "learning_rate": 2.142255806777287e-05, + "loss": 3.9101, + "step": 1745408 + }, + { + "epoch": 1.02, + "learning_rate": 2.141417212026235e-05, + "loss": 3.9057, + "step": 1745920 + }, + { + "epoch": 1.02, + "learning_rate": 2.140578617275183e-05, + "loss": 3.9086, + "step": 1746432 + }, + { + "epoch": 1.02, + "learning_rate": 2.139740022524131e-05, + "loss": 3.8923, + "step": 1746944 + }, + { + "epoch": 1.02, + "learning_rate": 2.138901427773079e-05, + "loss": 3.9074, + "step": 1747456 + }, + { + "epoch": 1.02, + "learning_rate": 2.138062833022027e-05, + "loss": 3.9086, + "step": 1747968 + }, + { + "epoch": 1.02, + "learning_rate": 2.137225876151348e-05, + "loss": 3.8944, + "step": 1748480 + }, + { + "epoch": 1.02, + "learning_rate": 2.1363872814002962e-05, + "loss": 3.9033, + "step": 1748992 + }, + { + "epoch": 1.02, + "learning_rate": 2.1355486866492442e-05, + "loss": 3.9058, + "step": 1749504 + }, + { + "epoch": 1.02, + "learning_rate": 2.134711729778565e-05, + "loss": 3.9048, + "step": 1750016 + }, + { + "epoch": 1.02, + "learning_rate": 2.133873135027513e-05, + "loss": 3.9202, + "step": 1750528 + }, + { + "epoch": 1.02, + "learning_rate": 2.133034540276461e-05, + "loss": 3.8944, + "step": 1751040 + }, + { + "epoch": 1.02, + "learning_rate": 2.132195945525409e-05, + "loss": 3.9088, + "step": 1751552 + }, + { + "epoch": 1.02, + "learning_rate": 2.131357350774357e-05, + "loss": 3.9031, + "step": 1752064 + }, + { + "epoch": 1.02, + "learning_rate": 2.1305187560233055e-05, + "loss": 3.9031, + "step": 1752576 + }, + { + "epoch": 1.02, + "learning_rate": 2.129680161272253e-05, + "loss": 3.8963, + "step": 1753088 + }, + { + "epoch": 1.02, + "learning_rate": 2.128841566521201e-05, + "loss": 3.8966, + "step": 1753600 + }, + { + "epoch": 1.02, + "learning_rate": 2.128002971770149e-05, + "loss": 3.8993, + "step": 1754112 + }, + { + "epoch": 1.02, + "learning_rate": 2.12716601489947e-05, + "loss": 3.8889, + "step": 1754624 + }, + { + "epoch": 1.02, + "learning_rate": 2.126327420148418e-05, + "loss": 3.9073, + "step": 1755136 + }, + { + "epoch": 1.03, + "eval_loss": 3.997260808944702, + "eval_runtime": 287.6392, + "eval_samples_per_second": 1326.631, + "eval_steps_per_second": 41.458, + "step": 1755345 + }, + { + "epoch": 0.0, + "learning_rate": 2.1254888253973664e-05, + "loss": 3.9085, + "step": 1755648 + }, + { + "epoch": 0.0, + "learning_rate": 2.1246502306463144e-05, + "loss": 3.8954, + "step": 1756160 + }, + { + "epoch": 0.0, + "learning_rate": 2.1238116358952623e-05, + "loss": 3.905, + "step": 1756672 + }, + { + "epoch": 0.0, + "learning_rate": 2.1229730411442103e-05, + "loss": 3.9041, + "step": 1757184 + }, + { + "epoch": 0.0, + "learning_rate": 2.1221344463931583e-05, + "loss": 3.9109, + "step": 1757696 + }, + { + "epoch": 0.0, + "learning_rate": 2.121295851642106e-05, + "loss": 3.9017, + "step": 1758208 + }, + { + "epoch": 0.0, + "learning_rate": 2.120457256891054e-05, + "loss": 3.9039, + "step": 1758720 + }, + { + "epoch": 0.0, + "learning_rate": 2.119618662140002e-05, + "loss": 3.8924, + "step": 1759232 + }, + { + "epoch": 0.0, + "learning_rate": 2.1187800673889503e-05, + "loss": 3.8981, + "step": 1759744 + }, + { + "epoch": 0.0, + "learning_rate": 2.1179414726378983e-05, + "loss": 3.9058, + "step": 1760256 + }, + { + "epoch": 0.0, + "learning_rate": 2.1171028778868463e-05, + "loss": 3.8999, + "step": 1760768 + }, + { + "epoch": 0.0, + "learning_rate": 2.1162642831357943e-05, + "loss": 3.8997, + "step": 1761280 + }, + { + "epoch": 0.0, + "learning_rate": 2.1154273262651152e-05, + "loss": 3.903, + "step": 1761792 + }, + { + "epoch": 0.0, + "learning_rate": 2.1145887315140632e-05, + "loss": 3.8949, + "step": 1762304 + }, + { + "epoch": 0.0, + "learning_rate": 2.1137501367630112e-05, + "loss": 3.8931, + "step": 1762816 + }, + { + "epoch": 0.0, + "learning_rate": 2.1129115420119592e-05, + "loss": 3.8914, + "step": 1763328 + }, + { + "epoch": 0.0, + "learning_rate": 2.1120729472609072e-05, + "loss": 3.8918, + "step": 1763840 + }, + { + "epoch": 0.0, + "learning_rate": 2.1112343525098552e-05, + "loss": 3.9044, + "step": 1764352 + }, + { + "epoch": 0.0, + "learning_rate": 2.1103957577588032e-05, + "loss": 3.8966, + "step": 1764864 + }, + { + "epoch": 0.0, + "learning_rate": 2.1095571630077512e-05, + "loss": 3.9152, + "step": 1765376 + }, + { + "epoch": 0.0, + "learning_rate": 2.1087185682566992e-05, + "loss": 3.9076, + "step": 1765888 + }, + { + "epoch": 0.0, + "learning_rate": 2.1078799735056472e-05, + "loss": 3.909, + "step": 1766400 + }, + { + "epoch": 0.0, + "learning_rate": 2.1070413787545952e-05, + "loss": 3.8964, + "step": 1766912 + }, + { + "epoch": 0.0, + "learning_rate": 2.1062044218839164e-05, + "loss": 3.9034, + "step": 1767424 + }, + { + "epoch": 0.0, + "learning_rate": 2.1053658271328644e-05, + "loss": 3.8999, + "step": 1767936 + }, + { + "epoch": 0.0, + "learning_rate": 2.104527232381812e-05, + "loss": 3.9076, + "step": 1768448 + }, + { + "epoch": 0.0, + "learning_rate": 2.10368863763076e-05, + "loss": 3.8956, + "step": 1768960 + }, + { + "epoch": 0.0, + "learning_rate": 2.102850042879708e-05, + "loss": 3.8981, + "step": 1769472 + }, + { + "epoch": 0.0, + "learning_rate": 2.1020114481286564e-05, + "loss": 3.8925, + "step": 1769984 + }, + { + "epoch": 0.0, + "learning_rate": 2.1011728533776044e-05, + "loss": 3.8982, + "step": 1770496 + }, + { + "epoch": 0.01, + "learning_rate": 2.1003358965069253e-05, + "loss": 3.8963, + "step": 1771008 + }, + { + "epoch": 0.01, + "learning_rate": 2.0994989396362466e-05, + "loss": 3.9105, + "step": 1771520 + }, + { + "epoch": 0.01, + "learning_rate": 2.0986603448851945e-05, + "loss": 3.9044, + "step": 1772032 + }, + { + "epoch": 0.01, + "learning_rate": 2.0978217501341425e-05, + "loss": 3.9058, + "step": 1772544 + }, + { + "epoch": 0.01, + "learning_rate": 2.0969831553830905e-05, + "loss": 3.903, + "step": 1773056 + }, + { + "epoch": 0.01, + "learning_rate": 2.0961461985124118e-05, + "loss": 3.9018, + "step": 1773568 + }, + { + "epoch": 0.01, + "learning_rate": 2.0953076037613594e-05, + "loss": 3.9065, + "step": 1774080 + }, + { + "epoch": 0.01, + "learning_rate": 2.0944690090103074e-05, + "loss": 3.891, + "step": 1774592 + }, + { + "epoch": 0.01, + "learning_rate": 2.0936304142592554e-05, + "loss": 3.8897, + "step": 1775104 + }, + { + "epoch": 0.01, + "learning_rate": 2.0927918195082034e-05, + "loss": 3.8849, + "step": 1775616 + }, + { + "epoch": 0.01, + "learning_rate": 2.0919532247571518e-05, + "loss": 3.896, + "step": 1776128 + }, + { + "epoch": 0.01, + "learning_rate": 2.0911146300060998e-05, + "loss": 3.9038, + "step": 1776640 + }, + { + "epoch": 0.01, + "learning_rate": 2.0902760352550478e-05, + "loss": 3.8998, + "step": 1777152 + }, + { + "epoch": 0.01, + "learning_rate": 2.0894390783843687e-05, + "loss": 3.8969, + "step": 1777664 + }, + { + "epoch": 0.01, + "learning_rate": 2.0886004836333167e-05, + "loss": 3.8975, + "step": 1778176 + }, + { + "epoch": 0.01, + "learning_rate": 2.0877618888822647e-05, + "loss": 3.8999, + "step": 1778688 + }, + { + "epoch": 0.01, + "learning_rate": 2.0869232941312127e-05, + "loss": 3.9004, + "step": 1779200 + }, + { + "epoch": 0.01, + "learning_rate": 2.0860846993801606e-05, + "loss": 3.8866, + "step": 1779712 + }, + { + "epoch": 0.01, + "learning_rate": 2.0852461046291086e-05, + "loss": 3.873, + "step": 1780224 + }, + { + "epoch": 0.01, + "learning_rate": 2.0844075098780566e-05, + "loss": 3.913, + "step": 1780736 + }, + { + "epoch": 0.01, + "learning_rate": 2.0835689151270046e-05, + "loss": 3.903, + "step": 1781248 + }, + { + "epoch": 0.01, + "learning_rate": 2.0827319582563255e-05, + "loss": 3.9074, + "step": 1781760 + }, + { + "epoch": 0.01, + "learning_rate": 2.0818933635052735e-05, + "loss": 3.8938, + "step": 1782272 + }, + { + "epoch": 0.01, + "learning_rate": 2.081054768754222e-05, + "loss": 3.8987, + "step": 1782784 + }, + { + "epoch": 0.01, + "learning_rate": 2.08021617400317e-05, + "loss": 3.8897, + "step": 1783296 + }, + { + "epoch": 0.01, + "learning_rate": 2.0793792171324908e-05, + "loss": 3.8942, + "step": 1783808 + }, + { + "epoch": 0.01, + "learning_rate": 2.0785406223814388e-05, + "loss": 3.8988, + "step": 1784320 + }, + { + "epoch": 0.01, + "learning_rate": 2.0777020276303868e-05, + "loss": 3.8913, + "step": 1784832 + }, + { + "epoch": 0.01, + "learning_rate": 2.0768634328793348e-05, + "loss": 3.9184, + "step": 1785344 + }, + { + "epoch": 0.01, + "learning_rate": 2.076026476008656e-05, + "loss": 3.8935, + "step": 1785856 + }, + { + "epoch": 0.01, + "learning_rate": 2.075187881257604e-05, + "loss": 3.8799, + "step": 1786368 + }, + { + "epoch": 0.01, + "learning_rate": 2.074349286506552e-05, + "loss": 3.9097, + "step": 1786880 + }, + { + "epoch": 0.01, + "learning_rate": 2.0735106917555e-05, + "loss": 3.8804, + "step": 1787392 + }, + { + "epoch": 0.01, + "learning_rate": 2.072673734884821e-05, + "loss": 3.8812, + "step": 1787904 + }, + { + "epoch": 0.01, + "learning_rate": 2.071835140133769e-05, + "loss": 3.9079, + "step": 1788416 + }, + { + "epoch": 0.01, + "learning_rate": 2.0709965453827172e-05, + "loss": 3.9044, + "step": 1788928 + }, + { + "epoch": 0.01, + "learning_rate": 2.0701579506316652e-05, + "loss": 3.8868, + "step": 1789440 + }, + { + "epoch": 0.01, + "learning_rate": 2.069320993760986e-05, + "loss": 3.8959, + "step": 1789952 + }, + { + "epoch": 0.01, + "learning_rate": 2.0684840368903074e-05, + "loss": 3.8785, + "step": 1790464 + }, + { + "epoch": 0.01, + "learning_rate": 2.067645442139255e-05, + "loss": 3.8922, + "step": 1790976 + }, + { + "epoch": 0.01, + "learning_rate": 2.0668068473882034e-05, + "loss": 3.8963, + "step": 1791488 + }, + { + "epoch": 0.01, + "learning_rate": 2.0659682526371514e-05, + "loss": 3.8916, + "step": 1792000 + }, + { + "epoch": 0.01, + "learning_rate": 2.0651296578860994e-05, + "loss": 3.9006, + "step": 1792512 + }, + { + "epoch": 0.01, + "learning_rate": 2.0642910631350474e-05, + "loss": 3.8966, + "step": 1793024 + }, + { + "epoch": 0.01, + "learning_rate": 2.0634524683839954e-05, + "loss": 3.9101, + "step": 1793536 + }, + { + "epoch": 0.01, + "learning_rate": 2.062613873632943e-05, + "loss": 3.8937, + "step": 1794048 + }, + { + "epoch": 0.01, + "learning_rate": 2.0617769167622643e-05, + "loss": 3.8944, + "step": 1794560 + }, + { + "epoch": 0.01, + "learning_rate": 2.0609383220112126e-05, + "loss": 3.9021, + "step": 1795072 + }, + { + "epoch": 0.01, + "learning_rate": 2.0600997272601606e-05, + "loss": 3.8907, + "step": 1795584 + }, + { + "epoch": 0.01, + "learning_rate": 2.0592611325091083e-05, + "loss": 3.9004, + "step": 1796096 + }, + { + "epoch": 0.01, + "learning_rate": 2.0584225377580563e-05, + "loss": 3.8961, + "step": 1796608 + }, + { + "epoch": 0.01, + "learning_rate": 2.0575839430070043e-05, + "loss": 3.8926, + "step": 1797120 + }, + { + "epoch": 0.01, + "learning_rate": 2.0567453482559522e-05, + "loss": 3.9078, + "step": 1797632 + }, + { + "epoch": 0.01, + "learning_rate": 2.0559083913852735e-05, + "loss": 3.8784, + "step": 1798144 + }, + { + "epoch": 0.01, + "learning_rate": 2.0550697966342215e-05, + "loss": 3.8887, + "step": 1798656 + }, + { + "epoch": 0.01, + "learning_rate": 2.0542312018831695e-05, + "loss": 3.9005, + "step": 1799168 + }, + { + "epoch": 0.01, + "learning_rate": 2.0533926071321175e-05, + "loss": 3.9024, + "step": 1799680 + }, + { + "epoch": 0.01, + "learning_rate": 2.0525540123810655e-05, + "loss": 3.8914, + "step": 1800192 + }, + { + "epoch": 0.01, + "learning_rate": 2.0517170555103864e-05, + "loss": 3.9001, + "step": 1800704 + }, + { + "epoch": 0.02, + "learning_rate": 2.0508784607593344e-05, + "loss": 3.8782, + "step": 1801216 + }, + { + "epoch": 0.02, + "learning_rate": 2.0500398660082827e-05, + "loss": 3.8932, + "step": 1801728 + }, + { + "epoch": 0.02, + "learning_rate": 2.0492029091376036e-05, + "loss": 3.8886, + "step": 1802240 + }, + { + "epoch": 0.02, + "learning_rate": 2.0483643143865516e-05, + "loss": 3.8944, + "step": 1802752 + }, + { + "epoch": 0.02, + "learning_rate": 2.0475257196354996e-05, + "loss": 3.905, + "step": 1803264 + }, + { + "epoch": 0.02, + "learning_rate": 2.0466871248844476e-05, + "loss": 3.8962, + "step": 1803776 + }, + { + "epoch": 0.02, + "learning_rate": 2.0458485301333956e-05, + "loss": 3.9005, + "step": 1804288 + }, + { + "epoch": 0.02, + "learning_rate": 2.0450099353823436e-05, + "loss": 3.89, + "step": 1804800 + }, + { + "epoch": 0.02, + "learning_rate": 2.0441713406312916e-05, + "loss": 3.8913, + "step": 1805312 + }, + { + "epoch": 0.02, + "learning_rate": 2.0433327458802396e-05, + "loss": 3.8919, + "step": 1805824 + }, + { + "epoch": 0.02, + "learning_rate": 2.042495789009561e-05, + "loss": 3.904, + "step": 1806336 + }, + { + "epoch": 0.02, + "learning_rate": 2.0416571942585085e-05, + "loss": 3.8977, + "step": 1806848 + }, + { + "epoch": 0.02, + "learning_rate": 2.0408185995074565e-05, + "loss": 3.9043, + "step": 1807360 + }, + { + "epoch": 0.02, + "learning_rate": 2.0399800047564045e-05, + "loss": 3.8884, + "step": 1807872 + }, + { + "epoch": 0.02, + "learning_rate": 2.0391414100053528e-05, + "loss": 3.9009, + "step": 1808384 + }, + { + "epoch": 0.02, + "learning_rate": 2.0383044531346737e-05, + "loss": 3.8915, + "step": 1808896 + }, + { + "epoch": 0.02, + "learning_rate": 2.0374658583836217e-05, + "loss": 3.8963, + "step": 1809408 + }, + { + "epoch": 0.02, + "learning_rate": 2.0366272636325697e-05, + "loss": 3.8826, + "step": 1809920 + }, + { + "epoch": 0.02, + "learning_rate": 2.0357886688815177e-05, + "loss": 3.9014, + "step": 1810432 + }, + { + "epoch": 0.02, + "learning_rate": 2.0349500741304657e-05, + "loss": 3.887, + "step": 1810944 + }, + { + "epoch": 0.02, + "learning_rate": 2.034113117259787e-05, + "loss": 3.8949, + "step": 1811456 + }, + { + "epoch": 0.02, + "learning_rate": 2.033274522508735e-05, + "loss": 3.8849, + "step": 1811968 + }, + { + "epoch": 0.02, + "learning_rate": 2.032435927757683e-05, + "loss": 3.8981, + "step": 1812480 + }, + { + "epoch": 0.02, + "learning_rate": 2.031597333006631e-05, + "loss": 3.8904, + "step": 1812992 + }, + { + "epoch": 0.02, + "learning_rate": 2.030758738255579e-05, + "loss": 3.8881, + "step": 1813504 + }, + { + "epoch": 0.02, + "learning_rate": 2.0299201435045266e-05, + "loss": 3.9039, + "step": 1814016 + }, + { + "epoch": 0.02, + "learning_rate": 2.029081548753475e-05, + "loss": 3.9069, + "step": 1814528 + }, + { + "epoch": 0.02, + "learning_rate": 2.028242954002423e-05, + "loss": 3.9041, + "step": 1815040 + }, + { + "epoch": 0.02, + "learning_rate": 2.027407635012117e-05, + "loss": 3.8977, + "step": 1815552 + }, + { + "epoch": 0.02, + "learning_rate": 2.026569040261065e-05, + "loss": 3.8856, + "step": 1816064 + }, + { + "epoch": 0.02, + "learning_rate": 2.025730445510013e-05, + "loss": 3.8904, + "step": 1816576 + }, + { + "epoch": 0.02, + "learning_rate": 2.024891850758961e-05, + "loss": 3.8957, + "step": 1817088 + }, + { + "epoch": 0.02, + "learning_rate": 2.024053256007909e-05, + "loss": 3.8985, + "step": 1817600 + }, + { + "epoch": 0.02, + "learning_rate": 2.023214661256857e-05, + "loss": 3.8837, + "step": 1818112 + }, + { + "epoch": 0.02, + "learning_rate": 2.022376066505805e-05, + "loss": 3.8995, + "step": 1818624 + }, + { + "epoch": 0.02, + "learning_rate": 2.021537471754753e-05, + "loss": 3.8995, + "step": 1819136 + }, + { + "epoch": 0.02, + "learning_rate": 2.020700514884074e-05, + "loss": 3.8943, + "step": 1819648 + }, + { + "epoch": 0.02, + "learning_rate": 2.019861920133022e-05, + "loss": 3.8921, + "step": 1820160 + }, + { + "epoch": 0.02, + "learning_rate": 2.0190249632623436e-05, + "loss": 3.8934, + "step": 1820672 + }, + { + "epoch": 0.02, + "learning_rate": 2.0181863685112916e-05, + "loss": 3.8872, + "step": 1821184 + }, + { + "epoch": 0.02, + "learning_rate": 2.0173477737602392e-05, + "loss": 3.9044, + "step": 1821696 + }, + { + "epoch": 0.02, + "learning_rate": 2.0165091790091872e-05, + "loss": 3.9003, + "step": 1822208 + }, + { + "epoch": 0.02, + "learning_rate": 2.0156705842581352e-05, + "loss": 3.9021, + "step": 1822720 + }, + { + "epoch": 0.02, + "learning_rate": 2.0148336273874565e-05, + "loss": 3.8821, + "step": 1823232 + }, + { + "epoch": 0.02, + "learning_rate": 2.0139950326364044e-05, + "loss": 3.9054, + "step": 1823744 + }, + { + "epoch": 0.02, + "learning_rate": 2.0131564378853524e-05, + "loss": 3.9007, + "step": 1824256 + }, + { + "epoch": 0.02, + "learning_rate": 2.0123178431343004e-05, + "loss": 3.8915, + "step": 1824768 + }, + { + "epoch": 0.02, + "learning_rate": 2.0114808862636213e-05, + "loss": 3.8981, + "step": 1825280 + }, + { + "epoch": 0.02, + "learning_rate": 2.0106422915125693e-05, + "loss": 3.8953, + "step": 1825792 + }, + { + "epoch": 0.02, + "learning_rate": 2.0098036967615173e-05, + "loss": 3.9012, + "step": 1826304 + }, + { + "epoch": 0.02, + "learning_rate": 2.0089651020104657e-05, + "loss": 3.9123, + "step": 1826816 + }, + { + "epoch": 0.02, + "learning_rate": 2.0081265072594137e-05, + "loss": 3.8896, + "step": 1827328 + }, + { + "epoch": 0.02, + "learning_rate": 2.0072879125083617e-05, + "loss": 3.8992, + "step": 1827840 + }, + { + "epoch": 0.02, + "learning_rate": 2.0064493177573097e-05, + "loss": 3.9024, + "step": 1828352 + }, + { + "epoch": 0.02, + "learning_rate": 2.0056107230062573e-05, + "loss": 3.8982, + "step": 1828864 + }, + { + "epoch": 0.02, + "learning_rate": 2.0047737661355786e-05, + "loss": 3.8886, + "step": 1829376 + }, + { + "epoch": 0.02, + "learning_rate": 2.0039351713845266e-05, + "loss": 3.8897, + "step": 1829888 + }, + { + "epoch": 0.02, + "learning_rate": 2.0030965766334746e-05, + "loss": 3.8947, + "step": 1830400 + }, + { + "epoch": 0.02, + "learning_rate": 2.0022596197627958e-05, + "loss": 3.8831, + "step": 1830912 + }, + { + "epoch": 0.02, + "learning_rate": 2.0014210250117438e-05, + "loss": 3.9009, + "step": 1831424 + }, + { + "epoch": 0.03, + "eval_loss": 3.9953536987304688, + "eval_runtime": 302.3836, + "eval_samples_per_second": 1261.944, + "eval_steps_per_second": 39.437, + "step": 1831665 + }, + { + "epoch": 1.0, + "learning_rate": 2.0005824302606918e-05, + "loss": 3.8975, + "step": 1831936 + }, + { + "epoch": 1.0, + "learning_rate": 1.9997438355096395e-05, + "loss": 3.8926, + "step": 1832448 + }, + { + "epoch": 1.0, + "learning_rate": 1.9989052407585874e-05, + "loss": 3.8949, + "step": 1832960 + }, + { + "epoch": 1.0, + "learning_rate": 1.9980666460075358e-05, + "loss": 3.8981, + "step": 1833472 + }, + { + "epoch": 1.0, + "learning_rate": 1.997229689136857e-05, + "loss": 3.9066, + "step": 1833984 + }, + { + "epoch": 1.0, + "learning_rate": 1.9963910943858047e-05, + "loss": 3.8965, + "step": 1834496 + }, + { + "epoch": 1.0, + "learning_rate": 1.9955524996347527e-05, + "loss": 3.9011, + "step": 1835008 + }, + { + "epoch": 1.0, + "learning_rate": 1.9947139048837007e-05, + "loss": 3.8869, + "step": 1835520 + }, + { + "epoch": 1.0, + "learning_rate": 1.9938753101326487e-05, + "loss": 3.8894, + "step": 1836032 + }, + { + "epoch": 1.0, + "learning_rate": 1.99303835326197e-05, + "loss": 3.9025, + "step": 1836544 + }, + { + "epoch": 1.0, + "learning_rate": 1.992199758510918e-05, + "loss": 3.8946, + "step": 1837056 + }, + { + "epoch": 1.0, + "learning_rate": 1.991361163759866e-05, + "loss": 3.8932, + "step": 1837568 + }, + { + "epoch": 1.0, + "learning_rate": 1.9905242068891868e-05, + "loss": 3.902, + "step": 1838080 + }, + { + "epoch": 1.0, + "learning_rate": 1.9896856121381348e-05, + "loss": 3.8865, + "step": 1838592 + }, + { + "epoch": 1.0, + "learning_rate": 1.9888470173870828e-05, + "loss": 3.8875, + "step": 1839104 + }, + { + "epoch": 1.0, + "learning_rate": 1.988008422636031e-05, + "loss": 3.8822, + "step": 1839616 + }, + { + "epoch": 1.0, + "learning_rate": 1.987169827884979e-05, + "loss": 3.8894, + "step": 1840128 + }, + { + "epoch": 1.0, + "learning_rate": 1.986331233133927e-05, + "loss": 3.8965, + "step": 1840640 + }, + { + "epoch": 1.0, + "learning_rate": 1.9854926383828748e-05, + "loss": 3.8902, + "step": 1841152 + }, + { + "epoch": 1.0, + "learning_rate": 1.984655681512196e-05, + "loss": 3.9055, + "step": 1841664 + }, + { + "epoch": 1.0, + "learning_rate": 1.983817086761144e-05, + "loss": 3.9023, + "step": 1842176 + }, + { + "epoch": 1.0, + "learning_rate": 1.982978492010092e-05, + "loss": 3.9011, + "step": 1842688 + }, + { + "epoch": 1.0, + "learning_rate": 1.98213989725904e-05, + "loss": 3.8948, + "step": 1843200 + }, + { + "epoch": 1.0, + "learning_rate": 1.981301302507988e-05, + "loss": 3.8944, + "step": 1843712 + }, + { + "epoch": 1.0, + "learning_rate": 1.980462707756936e-05, + "loss": 3.8962, + "step": 1844224 + }, + { + "epoch": 1.0, + "learning_rate": 1.979624113005884e-05, + "loss": 3.8982, + "step": 1844736 + }, + { + "epoch": 1.0, + "learning_rate": 1.978785518254832e-05, + "loss": 3.892, + "step": 1845248 + }, + { + "epoch": 1.0, + "learning_rate": 1.97794692350378e-05, + "loss": 3.8951, + "step": 1845760 + }, + { + "epoch": 1.0, + "learning_rate": 1.977108328752728e-05, + "loss": 3.8839, + "step": 1846272 + }, + { + "epoch": 1.0, + "learning_rate": 1.976269734001676e-05, + "loss": 3.8971, + "step": 1846784 + }, + { + "epoch": 1.01, + "learning_rate": 1.9754327771309973e-05, + "loss": 3.8878, + "step": 1847296 + }, + { + "epoch": 1.01, + "learning_rate": 1.9745941823799452e-05, + "loss": 3.9017, + "step": 1847808 + }, + { + "epoch": 1.01, + "learning_rate": 1.973757225509266e-05, + "loss": 3.9018, + "step": 1848320 + }, + { + "epoch": 1.01, + "learning_rate": 1.9729202686385874e-05, + "loss": 3.9021, + "step": 1848832 + }, + { + "epoch": 1.01, + "learning_rate": 1.9720816738875354e-05, + "loss": 3.891, + "step": 1849344 + }, + { + "epoch": 1.01, + "learning_rate": 1.9712447170168566e-05, + "loss": 3.8984, + "step": 1849856 + }, + { + "epoch": 1.01, + "learning_rate": 1.9704061222658046e-05, + "loss": 3.9008, + "step": 1850368 + }, + { + "epoch": 1.01, + "learning_rate": 1.9695675275147523e-05, + "loss": 3.8845, + "step": 1850880 + }, + { + "epoch": 1.01, + "learning_rate": 1.9687289327637003e-05, + "loss": 3.8872, + "step": 1851392 + }, + { + "epoch": 1.01, + "learning_rate": 1.9678903380126483e-05, + "loss": 3.881, + "step": 1851904 + }, + { + "epoch": 1.01, + "learning_rate": 1.9670517432615966e-05, + "loss": 3.8855, + "step": 1852416 + }, + { + "epoch": 1.01, + "learning_rate": 1.9662131485105446e-05, + "loss": 3.8931, + "step": 1852928 + }, + { + "epoch": 1.01, + "learning_rate": 1.9653745537594926e-05, + "loss": 3.8953, + "step": 1853440 + }, + { + "epoch": 1.01, + "learning_rate": 1.9645359590084403e-05, + "loss": 3.8903, + "step": 1853952 + }, + { + "epoch": 1.01, + "learning_rate": 1.9636973642573883e-05, + "loss": 3.8907, + "step": 1854464 + }, + { + "epoch": 1.01, + "learning_rate": 1.9628587695063363e-05, + "loss": 3.8919, + "step": 1854976 + }, + { + "epoch": 1.01, + "learning_rate": 1.9620201747552843e-05, + "loss": 3.8964, + "step": 1855488 + }, + { + "epoch": 1.01, + "learning_rate": 1.9611832178846055e-05, + "loss": 3.8813, + "step": 1856000 + }, + { + "epoch": 1.01, + "learning_rate": 1.9603446231335535e-05, + "loss": 3.8676, + "step": 1856512 + }, + { + "epoch": 1.01, + "learning_rate": 1.9595060283825015e-05, + "loss": 3.9066, + "step": 1857024 + }, + { + "epoch": 1.01, + "learning_rate": 1.9586674336314495e-05, + "loss": 3.8966, + "step": 1857536 + }, + { + "epoch": 1.01, + "learning_rate": 1.9578288388803975e-05, + "loss": 3.9033, + "step": 1858048 + }, + { + "epoch": 1.01, + "learning_rate": 1.9569902441293455e-05, + "loss": 3.888, + "step": 1858560 + }, + { + "epoch": 1.01, + "learning_rate": 1.9561532872586667e-05, + "loss": 3.8902, + "step": 1859072 + }, + { + "epoch": 1.01, + "learning_rate": 1.9553146925076147e-05, + "loss": 3.8868, + "step": 1859584 + }, + { + "epoch": 1.01, + "learning_rate": 1.9544760977565627e-05, + "loss": 3.8877, + "step": 1860096 + }, + { + "epoch": 1.01, + "learning_rate": 1.9536391408858836e-05, + "loss": 3.8965, + "step": 1860608 + }, + { + "epoch": 1.01, + "learning_rate": 1.9528005461348316e-05, + "loss": 3.8807, + "step": 1861120 + }, + { + "epoch": 1.01, + "learning_rate": 1.9519619513837796e-05, + "loss": 3.9134, + "step": 1861632 + }, + { + "epoch": 1.01, + "learning_rate": 1.9511233566327276e-05, + "loss": 3.8873, + "step": 1862144 + }, + { + "epoch": 1.01, + "learning_rate": 1.950284761881676e-05, + "loss": 3.8719, + "step": 1862656 + }, + { + "epoch": 1.01, + "learning_rate": 1.9494461671306236e-05, + "loss": 3.9016, + "step": 1863168 + }, + { + "epoch": 1.01, + "learning_rate": 1.9486075723795716e-05, + "loss": 3.8807, + "step": 1863680 + }, + { + "epoch": 1.01, + "learning_rate": 1.9477689776285196e-05, + "loss": 3.874, + "step": 1864192 + }, + { + "epoch": 1.01, + "learning_rate": 1.9469303828774676e-05, + "loss": 3.9009, + "step": 1864704 + }, + { + "epoch": 1.01, + "learning_rate": 1.9460934260067885e-05, + "loss": 3.8973, + "step": 1865216 + }, + { + "epoch": 1.01, + "learning_rate": 1.94525646913611e-05, + "loss": 3.8834, + "step": 1865728 + }, + { + "epoch": 1.01, + "learning_rate": 1.944417874385058e-05, + "loss": 3.8872, + "step": 1866240 + }, + { + "epoch": 1.01, + "learning_rate": 1.9435792796340058e-05, + "loss": 3.8727, + "step": 1866752 + }, + { + "epoch": 1.01, + "learning_rate": 1.9427406848829537e-05, + "loss": 3.8877, + "step": 1867264 + }, + { + "epoch": 1.01, + "learning_rate": 1.9419020901319017e-05, + "loss": 3.8926, + "step": 1867776 + }, + { + "epoch": 1.01, + "learning_rate": 1.9410634953808497e-05, + "loss": 3.8861, + "step": 1868288 + }, + { + "epoch": 1.01, + "learning_rate": 1.9402249006297977e-05, + "loss": 3.8953, + "step": 1868800 + }, + { + "epoch": 1.01, + "learning_rate": 1.939386305878746e-05, + "loss": 3.8891, + "step": 1869312 + }, + { + "epoch": 1.01, + "learning_rate": 1.938547711127694e-05, + "loss": 3.9016, + "step": 1869824 + }, + { + "epoch": 1.01, + "learning_rate": 1.937710754257015e-05, + "loss": 3.887, + "step": 1870336 + }, + { + "epoch": 1.01, + "learning_rate": 1.936872159505963e-05, + "loss": 3.8925, + "step": 1870848 + }, + { + "epoch": 1.01, + "learning_rate": 1.936033564754911e-05, + "loss": 3.8912, + "step": 1871360 + }, + { + "epoch": 1.01, + "learning_rate": 1.935194970003859e-05, + "loss": 3.8874, + "step": 1871872 + }, + { + "epoch": 1.01, + "learning_rate": 1.934356375252807e-05, + "loss": 3.8949, + "step": 1872384 + }, + { + "epoch": 1.01, + "learning_rate": 1.9335194183821282e-05, + "loss": 3.8894, + "step": 1872896 + }, + { + "epoch": 1.01, + "learning_rate": 1.9326808236310762e-05, + "loss": 3.8866, + "step": 1873408 + }, + { + "epoch": 1.01, + "learning_rate": 1.931842228880024e-05, + "loss": 3.9019, + "step": 1873920 + }, + { + "epoch": 1.01, + "learning_rate": 1.931003634128972e-05, + "loss": 3.8749, + "step": 1874432 + }, + { + "epoch": 1.01, + "learning_rate": 1.93016503937792e-05, + "loss": 3.8819, + "step": 1874944 + }, + { + "epoch": 1.01, + "learning_rate": 1.9293264446268682e-05, + "loss": 3.8927, + "step": 1875456 + }, + { + "epoch": 1.01, + "learning_rate": 1.928489487756189e-05, + "loss": 3.8901, + "step": 1875968 + }, + { + "epoch": 1.01, + "learning_rate": 1.927650893005137e-05, + "loss": 3.8845, + "step": 1876480 + }, + { + "epoch": 1.01, + "learning_rate": 1.926812298254085e-05, + "loss": 3.8973, + "step": 1876992 + }, + { + "epoch": 1.02, + "learning_rate": 1.925973703503033e-05, + "loss": 3.8759, + "step": 1877504 + }, + { + "epoch": 1.02, + "learning_rate": 1.925135108751981e-05, + "loss": 3.8845, + "step": 1878016 + }, + { + "epoch": 1.02, + "learning_rate": 1.924296514000929e-05, + "loss": 3.8823, + "step": 1878528 + }, + { + "epoch": 1.02, + "learning_rate": 1.923457919249877e-05, + "loss": 3.8841, + "step": 1879040 + }, + { + "epoch": 1.02, + "learning_rate": 1.922619324498825e-05, + "loss": 3.9056, + "step": 1879552 + }, + { + "epoch": 1.02, + "learning_rate": 1.9217823676281463e-05, + "loss": 3.889, + "step": 1880064 + }, + { + "epoch": 1.02, + "learning_rate": 1.9209437728770943e-05, + "loss": 3.8938, + "step": 1880576 + }, + { + "epoch": 1.02, + "learning_rate": 1.920105178126042e-05, + "loss": 3.8862, + "step": 1881088 + }, + { + "epoch": 1.02, + "learning_rate": 1.91926658337499e-05, + "loss": 3.886, + "step": 1881600 + }, + { + "epoch": 1.02, + "learning_rate": 1.9184296265043115e-05, + "loss": 3.8855, + "step": 1882112 + }, + { + "epoch": 1.02, + "learning_rate": 1.9175910317532595e-05, + "loss": 3.8961, + "step": 1882624 + }, + { + "epoch": 1.02, + "learning_rate": 1.9167540748825804e-05, + "loss": 3.8912, + "step": 1883136 + }, + { + "epoch": 1.02, + "learning_rate": 1.9159154801315284e-05, + "loss": 3.8974, + "step": 1883648 + }, + { + "epoch": 1.02, + "learning_rate": 1.9150768853804764e-05, + "loss": 3.8883, + "step": 1884160 + }, + { + "epoch": 1.02, + "learning_rate": 1.9142382906294244e-05, + "loss": 3.8905, + "step": 1884672 + }, + { + "epoch": 1.02, + "learning_rate": 1.9133996958783724e-05, + "loss": 3.8888, + "step": 1885184 + }, + { + "epoch": 1.02, + "learning_rate": 1.9125611011273204e-05, + "loss": 3.8887, + "step": 1885696 + }, + { + "epoch": 1.02, + "learning_rate": 1.9117241442566417e-05, + "loss": 3.8769, + "step": 1886208 + }, + { + "epoch": 1.02, + "learning_rate": 1.9108855495055893e-05, + "loss": 3.8976, + "step": 1886720 + }, + { + "epoch": 1.02, + "learning_rate": 1.9100469547545373e-05, + "loss": 3.8813, + "step": 1887232 + }, + { + "epoch": 1.02, + "learning_rate": 1.9092083600034853e-05, + "loss": 3.8875, + "step": 1887744 + }, + { + "epoch": 1.02, + "learning_rate": 1.9083697652524337e-05, + "loss": 3.8785, + "step": 1888256 + }, + { + "epoch": 1.02, + "learning_rate": 1.9075311705013817e-05, + "loss": 3.8881, + "step": 1888768 + }, + { + "epoch": 1.02, + "learning_rate": 1.9066925757503296e-05, + "loss": 3.8883, + "step": 1889280 + }, + { + "epoch": 1.02, + "learning_rate": 1.9058539809992776e-05, + "loss": 3.8822, + "step": 1889792 + }, + { + "epoch": 1.02, + "learning_rate": 1.9050170241285986e-05, + "loss": 3.8993, + "step": 1890304 + }, + { + "epoch": 1.02, + "learning_rate": 1.9041784293775465e-05, + "loss": 3.9027, + "step": 1890816 + }, + { + "epoch": 1.02, + "learning_rate": 1.9033398346264945e-05, + "loss": 3.8985, + "step": 1891328 + }, + { + "epoch": 1.02, + "learning_rate": 1.9025012398754425e-05, + "loss": 3.8891, + "step": 1891840 + }, + { + "epoch": 1.02, + "learning_rate": 1.9016642830047638e-05, + "loss": 3.8829, + "step": 1892352 + }, + { + "epoch": 1.02, + "learning_rate": 1.9008256882537118e-05, + "loss": 3.8851, + "step": 1892864 + }, + { + "epoch": 1.02, + "learning_rate": 1.8999870935026598e-05, + "loss": 3.8911, + "step": 1893376 + }, + { + "epoch": 1.02, + "learning_rate": 1.8991501366319807e-05, + "loss": 3.8917, + "step": 1893888 + }, + { + "epoch": 1.02, + "learning_rate": 1.898311541880929e-05, + "loss": 3.8793, + "step": 1894400 + }, + { + "epoch": 1.02, + "learning_rate": 1.897472947129877e-05, + "loss": 3.8876, + "step": 1894912 + }, + { + "epoch": 1.02, + "learning_rate": 1.896634352378825e-05, + "loss": 3.8945, + "step": 1895424 + }, + { + "epoch": 1.02, + "learning_rate": 1.8957957576277727e-05, + "loss": 3.8878, + "step": 1895936 + }, + { + "epoch": 1.02, + "learning_rate": 1.8949571628767207e-05, + "loss": 3.8864, + "step": 1896448 + }, + { + "epoch": 1.02, + "learning_rate": 1.894120206006042e-05, + "loss": 3.8904, + "step": 1896960 + }, + { + "epoch": 1.02, + "learning_rate": 1.89328161125499e-05, + "loss": 3.8822, + "step": 1897472 + }, + { + "epoch": 1.02, + "learning_rate": 1.892443016503938e-05, + "loss": 3.8982, + "step": 1897984 + }, + { + "epoch": 1.02, + "learning_rate": 1.891604421752886e-05, + "loss": 3.8922, + "step": 1898496 + }, + { + "epoch": 1.02, + "learning_rate": 1.890765827001834e-05, + "loss": 3.9007, + "step": 1899008 + }, + { + "epoch": 1.02, + "learning_rate": 1.889927232250782e-05, + "loss": 3.8807, + "step": 1899520 + }, + { + "epoch": 1.02, + "learning_rate": 1.88908863749973e-05, + "loss": 3.8932, + "step": 1900032 + }, + { + "epoch": 1.02, + "learning_rate": 1.8882516806290508e-05, + "loss": 3.898, + "step": 1900544 + }, + { + "epoch": 1.02, + "learning_rate": 1.887413085877999e-05, + "loss": 3.8879, + "step": 1901056 + }, + { + "epoch": 1.02, + "learning_rate": 1.886574491126947e-05, + "loss": 3.8907, + "step": 1901568 + }, + { + "epoch": 1.02, + "learning_rate": 1.885735896375895e-05, + "loss": 3.8894, + "step": 1902080 + }, + { + "epoch": 1.02, + "learning_rate": 1.884898939505216e-05, + "loss": 3.8925, + "step": 1902592 + }, + { + "epoch": 1.02, + "learning_rate": 1.884060344754164e-05, + "loss": 3.9097, + "step": 1903104 + }, + { + "epoch": 1.02, + "learning_rate": 1.883221750003112e-05, + "loss": 3.8837, + "step": 1903616 + }, + { + "epoch": 1.02, + "learning_rate": 1.88238315525206e-05, + "loss": 3.8953, + "step": 1904128 + }, + { + "epoch": 1.02, + "learning_rate": 1.881544560501008e-05, + "loss": 3.8955, + "step": 1904640 + }, + { + "epoch": 1.02, + "learning_rate": 1.8807076036303293e-05, + "loss": 3.8903, + "step": 1905152 + }, + { + "epoch": 1.02, + "learning_rate": 1.8798690088792773e-05, + "loss": 3.8821, + "step": 1905664 + }, + { + "epoch": 1.02, + "learning_rate": 1.8790304141282253e-05, + "loss": 3.8813, + "step": 1906176 + }, + { + "epoch": 1.02, + "learning_rate": 1.878191819377173e-05, + "loss": 3.8881, + "step": 1906688 + }, + { + "epoch": 1.02, + "learning_rate": 1.8773548625064945e-05, + "loss": 3.884, + "step": 1907200 + }, + { + "epoch": 1.02, + "learning_rate": 1.8765162677554425e-05, + "loss": 3.8922, + "step": 1907712 + }, + { + "epoch": 1.03, + "eval_loss": 3.993293523788452, + "eval_runtime": 299.2151, + "eval_samples_per_second": 1275.307, + "eval_steps_per_second": 39.854, + "step": 1907985 + }, + { + "epoch": 0.0, + "learning_rate": 1.8756793108847634e-05, + "loss": 3.8919, + "step": 1908224 + }, + { + "epoch": 0.0, + "learning_rate": 1.8748407161337114e-05, + "loss": 3.8859, + "step": 1908736 + }, + { + "epoch": 0.0, + "learning_rate": 1.8740021213826594e-05, + "loss": 3.8885, + "step": 1909248 + }, + { + "epoch": 0.0, + "learning_rate": 1.8731635266316074e-05, + "loss": 3.8913, + "step": 1909760 + }, + { + "epoch": 0.0, + "learning_rate": 1.8723249318805554e-05, + "loss": 3.9025, + "step": 1910272 + }, + { + "epoch": 0.0, + "learning_rate": 1.8714863371295034e-05, + "loss": 3.8915, + "step": 1910784 + }, + { + "epoch": 0.0, + "learning_rate": 1.8706477423784514e-05, + "loss": 3.8942, + "step": 1911296 + }, + { + "epoch": 0.0, + "learning_rate": 1.8698091476273994e-05, + "loss": 3.8803, + "step": 1911808 + }, + { + "epoch": 0.0, + "learning_rate": 1.8689705528763474e-05, + "loss": 3.8813, + "step": 1912320 + }, + { + "epoch": 0.0, + "learning_rate": 1.8681319581252954e-05, + "loss": 3.8991, + "step": 1912832 + }, + { + "epoch": 0.0, + "learning_rate": 1.8672933633742434e-05, + "loss": 3.8874, + "step": 1913344 + }, + { + "epoch": 0.0, + "learning_rate": 1.866454768623191e-05, + "loss": 3.8875, + "step": 1913856 + }, + { + "epoch": 0.0, + "learning_rate": 1.8656178117525126e-05, + "loss": 3.8951, + "step": 1914368 + }, + { + "epoch": 0.0, + "learning_rate": 1.8647792170014606e-05, + "loss": 3.8814, + "step": 1914880 + }, + { + "epoch": 0.0, + "learning_rate": 1.8639406222504083e-05, + "loss": 3.8823, + "step": 1915392 + }, + { + "epoch": 0.0, + "learning_rate": 1.8631020274993563e-05, + "loss": 3.8693, + "step": 1915904 + }, + { + "epoch": 0.0, + "learning_rate": 1.8622634327483042e-05, + "loss": 3.8892, + "step": 1916416 + }, + { + "epoch": 0.0, + "learning_rate": 1.8614248379972522e-05, + "loss": 3.8932, + "step": 1916928 + }, + { + "epoch": 0.0, + "learning_rate": 1.8605862432462002e-05, + "loss": 3.8795, + "step": 1917440 + }, + { + "epoch": 0.0, + "learning_rate": 1.8597476484951486e-05, + "loss": 3.8983, + "step": 1917952 + }, + { + "epoch": 0.0, + "learning_rate": 1.8589090537440966e-05, + "loss": 3.898, + "step": 1918464 + }, + { + "epoch": 0.0, + "learning_rate": 1.8580704589930442e-05, + "loss": 3.8954, + "step": 1918976 + }, + { + "epoch": 0.0, + "learning_rate": 1.8572318642419922e-05, + "loss": 3.8885, + "step": 1919488 + }, + { + "epoch": 0.0, + "learning_rate": 1.8563932694909402e-05, + "loss": 3.8924, + "step": 1920000 + }, + { + "epoch": 0.0, + "learning_rate": 1.8555579505006347e-05, + "loss": 3.8913, + "step": 1920512 + }, + { + "epoch": 0.0, + "learning_rate": 1.8547193557495827e-05, + "loss": 3.8915, + "step": 1921024 + }, + { + "epoch": 0.0, + "learning_rate": 1.8538807609985307e-05, + "loss": 3.8816, + "step": 1921536 + }, + { + "epoch": 0.0, + "learning_rate": 1.8530421662474787e-05, + "loss": 3.8886, + "step": 1922048 + }, + { + "epoch": 0.0, + "learning_rate": 1.8522035714964264e-05, + "loss": 3.8822, + "step": 1922560 + }, + { + "epoch": 0.0, + "learning_rate": 1.8513649767453744e-05, + "loss": 3.8861, + "step": 1923072 + }, + { + "epoch": 0.01, + "learning_rate": 1.8505280198746956e-05, + "loss": 3.8897, + "step": 1923584 + }, + { + "epoch": 0.01, + "learning_rate": 1.849689425123644e-05, + "loss": 3.8894, + "step": 1924096 + }, + { + "epoch": 0.01, + "learning_rate": 1.8488508303725916e-05, + "loss": 3.8924, + "step": 1924608 + }, + { + "epoch": 0.01, + "learning_rate": 1.8480122356215396e-05, + "loss": 3.9016, + "step": 1925120 + }, + { + "epoch": 0.01, + "learning_rate": 1.8471736408704876e-05, + "loss": 3.8871, + "step": 1925632 + }, + { + "epoch": 0.01, + "learning_rate": 1.846336683999809e-05, + "loss": 3.8888, + "step": 1926144 + }, + { + "epoch": 0.01, + "learning_rate": 1.845498089248757e-05, + "loss": 3.8975, + "step": 1926656 + }, + { + "epoch": 0.01, + "learning_rate": 1.8446594944977048e-05, + "loss": 3.8774, + "step": 1927168 + }, + { + "epoch": 0.01, + "learning_rate": 1.8438208997466528e-05, + "loss": 3.882, + "step": 1927680 + }, + { + "epoch": 0.01, + "learning_rate": 1.8429839428759737e-05, + "loss": 3.8757, + "step": 1928192 + }, + { + "epoch": 0.01, + "learning_rate": 1.8421453481249217e-05, + "loss": 3.8804, + "step": 1928704 + }, + { + "epoch": 0.01, + "learning_rate": 1.8413067533738697e-05, + "loss": 3.887, + "step": 1929216 + }, + { + "epoch": 0.01, + "learning_rate": 1.8404681586228177e-05, + "loss": 3.8948, + "step": 1929728 + }, + { + "epoch": 0.01, + "learning_rate": 1.839631201752139e-05, + "loss": 3.8822, + "step": 1930240 + }, + { + "epoch": 0.01, + "learning_rate": 1.8387942448814602e-05, + "loss": 3.886, + "step": 1930752 + }, + { + "epoch": 0.01, + "learning_rate": 1.8379556501304082e-05, + "loss": 3.8829, + "step": 1931264 + }, + { + "epoch": 0.01, + "learning_rate": 1.8371170553793562e-05, + "loss": 3.8906, + "step": 1931776 + }, + { + "epoch": 0.01, + "learning_rate": 1.836278460628304e-05, + "loss": 3.8755, + "step": 1932288 + }, + { + "epoch": 0.01, + "learning_rate": 1.8354398658772522e-05, + "loss": 3.8679, + "step": 1932800 + }, + { + "epoch": 0.01, + "learning_rate": 1.8346012711262002e-05, + "loss": 3.8983, + "step": 1933312 + }, + { + "epoch": 0.01, + "learning_rate": 1.833764314255521e-05, + "loss": 3.8912, + "step": 1933824 + }, + { + "epoch": 0.01, + "learning_rate": 1.832925719504469e-05, + "loss": 3.9003, + "step": 1934336 + }, + { + "epoch": 0.01, + "learning_rate": 1.832087124753417e-05, + "loss": 3.8786, + "step": 1934848 + }, + { + "epoch": 0.01, + "learning_rate": 1.831248530002365e-05, + "loss": 3.8837, + "step": 1935360 + }, + { + "epoch": 0.01, + "learning_rate": 1.830409935251313e-05, + "loss": 3.8806, + "step": 1935872 + }, + { + "epoch": 0.01, + "learning_rate": 1.829571340500261e-05, + "loss": 3.8819, + "step": 1936384 + }, + { + "epoch": 0.01, + "learning_rate": 1.8287327457492094e-05, + "loss": 3.8917, + "step": 1936896 + }, + { + "epoch": 0.01, + "learning_rate": 1.827894150998157e-05, + "loss": 3.8724, + "step": 1937408 + }, + { + "epoch": 0.01, + "learning_rate": 1.827055556247105e-05, + "loss": 3.9081, + "step": 1937920 + }, + { + "epoch": 0.01, + "learning_rate": 1.8262202372567992e-05, + "loss": 3.8873, + "step": 1938432 + }, + { + "epoch": 0.01, + "learning_rate": 1.8253816425057472e-05, + "loss": 3.8638, + "step": 1938944 + }, + { + "epoch": 0.01, + "learning_rate": 1.8245430477546956e-05, + "loss": 3.8987, + "step": 1939456 + }, + { + "epoch": 0.01, + "learning_rate": 1.8237044530036436e-05, + "loss": 3.8784, + "step": 1939968 + }, + { + "epoch": 0.01, + "learning_rate": 1.8228658582525916e-05, + "loss": 3.8668, + "step": 1940480 + }, + { + "epoch": 0.01, + "learning_rate": 1.8220272635015392e-05, + "loss": 3.8937, + "step": 1940992 + }, + { + "epoch": 0.01, + "learning_rate": 1.8211886687504872e-05, + "loss": 3.8932, + "step": 1941504 + }, + { + "epoch": 0.01, + "learning_rate": 1.8203500739994352e-05, + "loss": 3.88, + "step": 1942016 + }, + { + "epoch": 0.01, + "learning_rate": 1.8195114792483832e-05, + "loss": 3.8844, + "step": 1942528 + }, + { + "epoch": 0.01, + "learning_rate": 1.8186728844973315e-05, + "loss": 3.865, + "step": 1943040 + }, + { + "epoch": 0.01, + "learning_rate": 1.8178359276266524e-05, + "loss": 3.8801, + "step": 1943552 + }, + { + "epoch": 0.01, + "learning_rate": 1.8169973328756004e-05, + "loss": 3.886, + "step": 1944064 + }, + { + "epoch": 0.01, + "learning_rate": 1.8161587381245484e-05, + "loss": 3.8809, + "step": 1944576 + }, + { + "epoch": 0.01, + "learning_rate": 1.8153217812538693e-05, + "loss": 3.8913, + "step": 1945088 + }, + { + "epoch": 0.01, + "learning_rate": 1.8144831865028177e-05, + "loss": 3.8792, + "step": 1945600 + }, + { + "epoch": 0.01, + "learning_rate": 1.8136445917517657e-05, + "loss": 3.8993, + "step": 1946112 + }, + { + "epoch": 0.01, + "learning_rate": 1.8128059970007137e-05, + "loss": 3.8862, + "step": 1946624 + }, + { + "epoch": 0.01, + "learning_rate": 1.8119674022496617e-05, + "loss": 3.8836, + "step": 1947136 + }, + { + "epoch": 0.01, + "learning_rate": 1.8111288074986097e-05, + "loss": 3.885, + "step": 1947648 + }, + { + "epoch": 0.01, + "learning_rate": 1.8102902127475573e-05, + "loss": 3.8821, + "step": 1948160 + }, + { + "epoch": 0.01, + "learning_rate": 1.8094516179965053e-05, + "loss": 3.8883, + "step": 1948672 + }, + { + "epoch": 0.01, + "learning_rate": 1.8086130232454533e-05, + "loss": 3.8851, + "step": 1949184 + }, + { + "epoch": 0.01, + "learning_rate": 1.8077744284944016e-05, + "loss": 3.8824, + "step": 1949696 + }, + { + "epoch": 0.01, + "learning_rate": 1.8069358337433496e-05, + "loss": 3.8937, + "step": 1950208 + }, + { + "epoch": 0.01, + "learning_rate": 1.8060972389922976e-05, + "loss": 3.8745, + "step": 1950720 + }, + { + "epoch": 0.01, + "learning_rate": 1.8052602821216185e-05, + "loss": 3.8734, + "step": 1951232 + }, + { + "epoch": 0.01, + "learning_rate": 1.8044216873705665e-05, + "loss": 3.8868, + "step": 1951744 + }, + { + "epoch": 0.01, + "learning_rate": 1.8035830926195145e-05, + "loss": 3.8899, + "step": 1952256 + }, + { + "epoch": 0.01, + "learning_rate": 1.8027461357488358e-05, + "loss": 3.8767, + "step": 1952768 + }, + { + "epoch": 0.01, + "learning_rate": 1.8019075409977838e-05, + "loss": 3.8892, + "step": 1953280 + }, + { + "epoch": 0.02, + "learning_rate": 1.8010689462467318e-05, + "loss": 3.873, + "step": 1953792 + }, + { + "epoch": 0.02, + "learning_rate": 1.8002303514956798e-05, + "loss": 3.8778, + "step": 1954304 + }, + { + "epoch": 0.02, + "learning_rate": 1.7993917567446278e-05, + "loss": 3.8763, + "step": 1954816 + }, + { + "epoch": 0.02, + "learning_rate": 1.7985531619935754e-05, + "loss": 3.882, + "step": 1955328 + }, + { + "epoch": 0.02, + "learning_rate": 1.7977145672425234e-05, + "loss": 3.8981, + "step": 1955840 + }, + { + "epoch": 0.02, + "learning_rate": 1.796879248252218e-05, + "loss": 3.8851, + "step": 1956352 + }, + { + "epoch": 0.02, + "learning_rate": 1.796040653501166e-05, + "loss": 3.887, + "step": 1956864 + }, + { + "epoch": 0.02, + "learning_rate": 1.795202058750114e-05, + "loss": 3.88, + "step": 1957376 + }, + { + "epoch": 0.02, + "learning_rate": 1.794363463999062e-05, + "loss": 3.8783, + "step": 1957888 + }, + { + "epoch": 0.02, + "learning_rate": 1.79352486924801e-05, + "loss": 3.882, + "step": 1958400 + }, + { + "epoch": 0.02, + "learning_rate": 1.792686274496958e-05, + "loss": 3.8908, + "step": 1958912 + }, + { + "epoch": 0.02, + "learning_rate": 1.791847679745906e-05, + "loss": 3.8891, + "step": 1959424 + }, + { + "epoch": 0.02, + "learning_rate": 1.791009084994854e-05, + "loss": 3.8891, + "step": 1959936 + }, + { + "epoch": 0.02, + "learning_rate": 1.790172128124175e-05, + "loss": 3.8859, + "step": 1960448 + }, + { + "epoch": 0.02, + "learning_rate": 1.7893335333731228e-05, + "loss": 3.8806, + "step": 1960960 + }, + { + "epoch": 0.02, + "learning_rate": 1.7884949386220708e-05, + "loss": 3.8845, + "step": 1961472 + }, + { + "epoch": 0.02, + "learning_rate": 1.7876563438710188e-05, + "loss": 3.8815, + "step": 1961984 + }, + { + "epoch": 0.02, + "learning_rate": 1.786817749119967e-05, + "loss": 3.8743, + "step": 1962496 + }, + { + "epoch": 0.02, + "learning_rate": 1.785979154368915e-05, + "loss": 3.8835, + "step": 1963008 + }, + { + "epoch": 0.02, + "learning_rate": 1.785140559617863e-05, + "loss": 3.8792, + "step": 1963520 + }, + { + "epoch": 0.02, + "learning_rate": 1.784301964866811e-05, + "loss": 3.8792, + "step": 1964032 + }, + { + "epoch": 0.02, + "learning_rate": 1.7834633701157588e-05, + "loss": 3.8776, + "step": 1964544 + }, + { + "epoch": 0.02, + "learning_rate": 1.78262641324508e-05, + "loss": 3.8773, + "step": 1965056 + }, + { + "epoch": 0.02, + "learning_rate": 1.781787818494028e-05, + "loss": 3.8823, + "step": 1965568 + }, + { + "epoch": 0.02, + "learning_rate": 1.780949223742976e-05, + "loss": 3.8779, + "step": 1966080 + }, + { + "epoch": 0.02, + "learning_rate": 1.780110628991924e-05, + "loss": 3.8927, + "step": 1966592 + }, + { + "epoch": 0.02, + "learning_rate": 1.779272034240872e-05, + "loss": 3.8946, + "step": 1967104 + }, + { + "epoch": 0.02, + "learning_rate": 1.77843343948982e-05, + "loss": 3.8912, + "step": 1967616 + }, + { + "epoch": 0.02, + "learning_rate": 1.777594844738768e-05, + "loss": 3.885, + "step": 1968128 + }, + { + "epoch": 0.02, + "learning_rate": 1.776756249987716e-05, + "loss": 3.8777, + "step": 1968640 + }, + { + "epoch": 0.02, + "learning_rate": 1.7759192931170372e-05, + "loss": 3.8807, + "step": 1969152 + }, + { + "epoch": 0.02, + "learning_rate": 1.7750823362463585e-05, + "loss": 3.8876, + "step": 1969664 + }, + { + "epoch": 0.02, + "learning_rate": 1.774243741495306e-05, + "loss": 3.8919, + "step": 1970176 + }, + { + "epoch": 0.02, + "learning_rate": 1.773405146744254e-05, + "loss": 3.8686, + "step": 1970688 + }, + { + "epoch": 0.02, + "learning_rate": 1.772566551993202e-05, + "loss": 3.883, + "step": 1971200 + }, + { + "epoch": 0.02, + "learning_rate": 1.77172795724215e-05, + "loss": 3.8839, + "step": 1971712 + }, + { + "epoch": 0.02, + "learning_rate": 1.770889362491098e-05, + "loss": 3.8856, + "step": 1972224 + }, + { + "epoch": 0.02, + "learning_rate": 1.7700507677400464e-05, + "loss": 3.8784, + "step": 1972736 + }, + { + "epoch": 0.02, + "learning_rate": 1.7692138108693674e-05, + "loss": 3.8815, + "step": 1973248 + }, + { + "epoch": 0.02, + "learning_rate": 1.7683752161183154e-05, + "loss": 3.8782, + "step": 1973760 + }, + { + "epoch": 0.02, + "learning_rate": 1.7675366213672633e-05, + "loss": 3.8947, + "step": 1974272 + }, + { + "epoch": 0.02, + "learning_rate": 1.7666980266162113e-05, + "loss": 3.8881, + "step": 1974784 + }, + { + "epoch": 0.02, + "learning_rate": 1.7658594318651593e-05, + "loss": 3.8941, + "step": 1975296 + }, + { + "epoch": 0.02, + "learning_rate": 1.7650208371141073e-05, + "loss": 3.8761, + "step": 1975808 + }, + { + "epoch": 0.02, + "learning_rate": 1.7641822423630553e-05, + "loss": 3.8849, + "step": 1976320 + }, + { + "epoch": 0.02, + "learning_rate": 1.7633436476120033e-05, + "loss": 3.8939, + "step": 1976832 + }, + { + "epoch": 0.02, + "learning_rate": 1.7625066907413242e-05, + "loss": 3.8805, + "step": 1977344 + }, + { + "epoch": 0.02, + "learning_rate": 1.7616680959902722e-05, + "loss": 3.8863, + "step": 1977856 + }, + { + "epoch": 0.02, + "learning_rate": 1.7608295012392202e-05, + "loss": 3.8826, + "step": 1978368 + }, + { + "epoch": 0.02, + "learning_rate": 1.7599909064881686e-05, + "loss": 3.8866, + "step": 1978880 + }, + { + "epoch": 0.02, + "learning_rate": 1.7591523117371166e-05, + "loss": 3.8986, + "step": 1979392 + }, + { + "epoch": 0.02, + "learning_rate": 1.7583137169860645e-05, + "loss": 3.8824, + "step": 1979904 + }, + { + "epoch": 0.02, + "learning_rate": 1.7574767601153855e-05, + "loss": 3.8905, + "step": 1980416 + }, + { + "epoch": 0.02, + "learning_rate": 1.7566381653643335e-05, + "loss": 3.8896, + "step": 1980928 + }, + { + "epoch": 0.02, + "learning_rate": 1.7557995706132815e-05, + "loss": 3.8872, + "step": 1981440 + }, + { + "epoch": 0.02, + "learning_rate": 1.7549609758622294e-05, + "loss": 3.8767, + "step": 1981952 + }, + { + "epoch": 0.02, + "learning_rate": 1.7541223811111774e-05, + "loss": 3.8768, + "step": 1982464 + }, + { + "epoch": 0.02, + "learning_rate": 1.7532837863601254e-05, + "loss": 3.879, + "step": 1982976 + }, + { + "epoch": 0.02, + "learning_rate": 1.7524468294894467e-05, + "loss": 3.8812, + "step": 1983488 + }, + { + "epoch": 0.02, + "learning_rate": 1.7516082347383947e-05, + "loss": 3.8829, + "step": 1984000 + }, + { + "epoch": 0.03, + "eval_loss": 3.991048574447632, + "eval_runtime": 452.8149, + "eval_samples_per_second": 842.709, + "eval_steps_per_second": 26.335, + "step": 1984305 + }, + { + "epoch": 1.0, + "learning_rate": 1.7507712778677156e-05, + "loss": 3.8834, + "step": 1984512 + }, + { + "epoch": 1.0, + "learning_rate": 1.7499326831166636e-05, + "loss": 3.8804, + "step": 1985024 + }, + { + "epoch": 1.0, + "learning_rate": 1.749094088365612e-05, + "loss": 3.8847, + "step": 1985536 + }, + { + "epoch": 1.0, + "learning_rate": 1.7482554936145596e-05, + "loss": 3.8876, + "step": 1986048 + }, + { + "epoch": 1.0, + "learning_rate": 1.7474168988635076e-05, + "loss": 3.8967, + "step": 1986560 + }, + { + "epoch": 1.0, + "learning_rate": 1.7465783041124556e-05, + "loss": 3.8846, + "step": 1987072 + }, + { + "epoch": 1.0, + "learning_rate": 1.7457397093614036e-05, + "loss": 3.8887, + "step": 1987584 + }, + { + "epoch": 1.0, + "learning_rate": 1.7449011146103516e-05, + "loss": 3.8766, + "step": 1988096 + }, + { + "epoch": 1.0, + "learning_rate": 1.7440625198592996e-05, + "loss": 3.8783, + "step": 1988608 + }, + { + "epoch": 1.0, + "learning_rate": 1.7432255629886208e-05, + "loss": 3.8863, + "step": 1989120 + }, + { + "epoch": 1.0, + "learning_rate": 1.7423869682375688e-05, + "loss": 3.8908, + "step": 1989632 + }, + { + "epoch": 1.0, + "learning_rate": 1.7415500113668897e-05, + "loss": 3.8821, + "step": 1990144 + }, + { + "epoch": 1.0, + "learning_rate": 1.740713054496211e-05, + "loss": 3.8903, + "step": 1990656 + }, + { + "epoch": 1.0, + "learning_rate": 1.739874459745159e-05, + "loss": 3.8726, + "step": 1991168 + }, + { + "epoch": 1.0, + "learning_rate": 1.739035864994107e-05, + "loss": 3.8802, + "step": 1991680 + }, + { + "epoch": 1.0, + "learning_rate": 1.738197270243055e-05, + "loss": 3.8619, + "step": 1992192 + }, + { + "epoch": 1.0, + "learning_rate": 1.737358675492003e-05, + "loss": 3.886, + "step": 1992704 + }, + { + "epoch": 1.0, + "learning_rate": 1.736520080740951e-05, + "loss": 3.8867, + "step": 1993216 + }, + { + "epoch": 1.0, + "learning_rate": 1.735681485989899e-05, + "loss": 3.8772, + "step": 1993728 + }, + { + "epoch": 1.0, + "learning_rate": 1.734842891238847e-05, + "loss": 3.8901, + "step": 1994240 + }, + { + "epoch": 1.0, + "learning_rate": 1.734004296487795e-05, + "loss": 3.8932, + "step": 1994752 + }, + { + "epoch": 1.0, + "learning_rate": 1.733165701736743e-05, + "loss": 3.8891, + "step": 1995264 + }, + { + "epoch": 1.0, + "learning_rate": 1.732327106985691e-05, + "loss": 3.885, + "step": 1995776 + }, + { + "epoch": 1.0, + "learning_rate": 1.731488512234639e-05, + "loss": 3.8838, + "step": 1996288 + }, + { + "epoch": 1.0, + "learning_rate": 1.7306515553639598e-05, + "loss": 3.8869, + "step": 1996800 + }, + { + "epoch": 1.0, + "learning_rate": 1.729814598493281e-05, + "loss": 3.8898, + "step": 1997312 + }, + { + "epoch": 1.0, + "learning_rate": 1.7289760037422294e-05, + "loss": 3.8736, + "step": 1997824 + }, + { + "epoch": 1.0, + "learning_rate": 1.7281374089911774e-05, + "loss": 3.8835, + "step": 1998336 + }, + { + "epoch": 1.0, + "learning_rate": 1.727298814240125e-05, + "loss": 3.877, + "step": 1998848 + }, + { + "epoch": 1.0, + "learning_rate": 1.726460219489073e-05, + "loss": 3.8792, + "step": 1999360 + }, + { + "epoch": 1.01, + "learning_rate": 1.725621624738021e-05, + "loss": 3.8841, + "step": 1999872 + }, + { + "epoch": 1.01, + "learning_rate": 1.724783029986969e-05, + "loss": 3.8879, + "step": 2000384 + }, + { + "epoch": 1.01, + "learning_rate": 1.723944435235917e-05, + "loss": 3.8847, + "step": 2000896 + }, + { + "epoch": 1.01, + "learning_rate": 1.723105840484865e-05, + "loss": 3.8958, + "step": 2001408 + }, + { + "epoch": 1.01, + "learning_rate": 1.7222672457338134e-05, + "loss": 3.8821, + "step": 2001920 + }, + { + "epoch": 1.01, + "learning_rate": 1.7214302888631343e-05, + "loss": 3.8836, + "step": 2002432 + }, + { + "epoch": 1.01, + "learning_rate": 1.7205916941120823e-05, + "loss": 3.8895, + "step": 2002944 + }, + { + "epoch": 1.01, + "learning_rate": 1.7197530993610303e-05, + "loss": 3.8694, + "step": 2003456 + }, + { + "epoch": 1.01, + "learning_rate": 1.718914504609978e-05, + "loss": 3.8757, + "step": 2003968 + }, + { + "epoch": 1.01, + "learning_rate": 1.718075909858926e-05, + "loss": 3.8754, + "step": 2004480 + }, + { + "epoch": 1.01, + "learning_rate": 1.7172373151078743e-05, + "loss": 3.87, + "step": 2004992 + }, + { + "epoch": 1.01, + "learning_rate": 1.7164003582371955e-05, + "loss": 3.8833, + "step": 2005504 + }, + { + "epoch": 1.01, + "learning_rate": 1.715561763486143e-05, + "loss": 3.8916, + "step": 2006016 + }, + { + "epoch": 1.01, + "learning_rate": 1.714723168735091e-05, + "loss": 3.8757, + "step": 2006528 + }, + { + "epoch": 1.01, + "learning_rate": 1.713884573984039e-05, + "loss": 3.8815, + "step": 2007040 + }, + { + "epoch": 1.01, + "learning_rate": 1.7130476171133604e-05, + "loss": 3.8761, + "step": 2007552 + }, + { + "epoch": 1.01, + "learning_rate": 1.7122106602426816e-05, + "loss": 3.8864, + "step": 2008064 + }, + { + "epoch": 1.01, + "learning_rate": 1.7113720654916296e-05, + "loss": 3.8701, + "step": 2008576 + }, + { + "epoch": 1.01, + "learning_rate": 1.7105334707405776e-05, + "loss": 3.8603, + "step": 2009088 + }, + { + "epoch": 1.01, + "learning_rate": 1.7096948759895253e-05, + "loss": 3.8929, + "step": 2009600 + }, + { + "epoch": 1.01, + "learning_rate": 1.7088562812384733e-05, + "loss": 3.8845, + "step": 2010112 + }, + { + "epoch": 1.01, + "learning_rate": 1.7080176864874213e-05, + "loss": 3.8966, + "step": 2010624 + }, + { + "epoch": 1.01, + "learning_rate": 1.7071790917363696e-05, + "loss": 3.8747, + "step": 2011136 + }, + { + "epoch": 1.01, + "learning_rate": 1.7063404969853176e-05, + "loss": 3.8775, + "step": 2011648 + }, + { + "epoch": 1.01, + "learning_rate": 1.7055019022342656e-05, + "loss": 3.8779, + "step": 2012160 + }, + { + "epoch": 1.01, + "learning_rate": 1.7046649453635865e-05, + "loss": 3.8744, + "step": 2012672 + }, + { + "epoch": 1.01, + "learning_rate": 1.7038263506125345e-05, + "loss": 3.8902, + "step": 2013184 + }, + { + "epoch": 1.01, + "learning_rate": 1.7029877558614825e-05, + "loss": 3.8629, + "step": 2013696 + }, + { + "epoch": 1.01, + "learning_rate": 1.7021491611104305e-05, + "loss": 3.9047, + "step": 2014208 + }, + { + "epoch": 1.01, + "learning_rate": 1.701310566359379e-05, + "loss": 3.8832, + "step": 2014720 + }, + { + "epoch": 1.01, + "learning_rate": 1.7004719716083265e-05, + "loss": 3.8574, + "step": 2015232 + }, + { + "epoch": 1.01, + "learning_rate": 1.6996333768572745e-05, + "loss": 3.8875, + "step": 2015744 + }, + { + "epoch": 1.01, + "learning_rate": 1.6987947821062225e-05, + "loss": 3.8748, + "step": 2016256 + }, + { + "epoch": 1.01, + "learning_rate": 1.6979594631159167e-05, + "loss": 3.8601, + "step": 2016768 + }, + { + "epoch": 1.01, + "learning_rate": 1.697120868364865e-05, + "loss": 3.8864, + "step": 2017280 + }, + { + "epoch": 1.01, + "learning_rate": 1.696282273613813e-05, + "loss": 3.8909, + "step": 2017792 + }, + { + "epoch": 1.01, + "learning_rate": 1.695443678862761e-05, + "loss": 3.8714, + "step": 2018304 + }, + { + "epoch": 1.01, + "learning_rate": 1.6946050841117086e-05, + "loss": 3.8822, + "step": 2018816 + }, + { + "epoch": 1.01, + "learning_rate": 1.69376812724103e-05, + "loss": 3.8599, + "step": 2019328 + }, + { + "epoch": 1.01, + "learning_rate": 1.692929532489978e-05, + "loss": 3.8725, + "step": 2019840 + }, + { + "epoch": 1.01, + "learning_rate": 1.692090937738926e-05, + "loss": 3.8821, + "step": 2020352 + }, + { + "epoch": 1.01, + "learning_rate": 1.691253980868247e-05, + "loss": 3.8754, + "step": 2020864 + }, + { + "epoch": 1.01, + "learning_rate": 1.690415386117195e-05, + "loss": 3.8842, + "step": 2021376 + }, + { + "epoch": 1.01, + "learning_rate": 1.689576791366143e-05, + "loss": 3.8765, + "step": 2021888 + }, + { + "epoch": 1.01, + "learning_rate": 1.6887381966150908e-05, + "loss": 3.8959, + "step": 2022400 + }, + { + "epoch": 1.01, + "learning_rate": 1.6878996018640388e-05, + "loss": 3.8765, + "step": 2022912 + }, + { + "epoch": 1.01, + "learning_rate": 1.6870610071129868e-05, + "loss": 3.8865, + "step": 2023424 + }, + { + "epoch": 1.01, + "learning_rate": 1.686222412361935e-05, + "loss": 3.8761, + "step": 2023936 + }, + { + "epoch": 1.01, + "learning_rate": 1.685383817610883e-05, + "loss": 3.8777, + "step": 2024448 + }, + { + "epoch": 1.01, + "learning_rate": 1.684545222859831e-05, + "loss": 3.8828, + "step": 2024960 + }, + { + "epoch": 1.01, + "learning_rate": 1.683708265989152e-05, + "loss": 3.8784, + "step": 2025472 + }, + { + "epoch": 1.01, + "learning_rate": 1.6828696712381e-05, + "loss": 3.8798, + "step": 2025984 + }, + { + "epoch": 1.01, + "learning_rate": 1.682031076487048e-05, + "loss": 3.8888, + "step": 2026496 + }, + { + "epoch": 1.01, + "learning_rate": 1.681192481735996e-05, + "loss": 3.8744, + "step": 2027008 + }, + { + "epoch": 1.01, + "learning_rate": 1.680353886984944e-05, + "loss": 3.8654, + "step": 2027520 + }, + { + "epoch": 1.01, + "learning_rate": 1.6795169301142652e-05, + "loss": 3.878, + "step": 2028032 + }, + { + "epoch": 1.01, + "learning_rate": 1.6786783353632132e-05, + "loss": 3.8837, + "step": 2028544 + }, + { + "epoch": 1.01, + "learning_rate": 1.6778397406121612e-05, + "loss": 3.8737, + "step": 2029056 + }, + { + "epoch": 1.01, + "learning_rate": 1.677001145861109e-05, + "loss": 3.8846, + "step": 2029568 + }, + { + "epoch": 1.01, + "learning_rate": 1.6761625511100572e-05, + "loss": 3.8698, + "step": 2030080 + }, + { + "epoch": 1.02, + "learning_rate": 1.6753255942393785e-05, + "loss": 3.8702, + "step": 2030592 + }, + { + "epoch": 1.02, + "learning_rate": 1.6744869994883265e-05, + "loss": 3.8719, + "step": 2031104 + }, + { + "epoch": 1.02, + "learning_rate": 1.673648404737274e-05, + "loss": 3.8757, + "step": 2031616 + }, + { + "epoch": 1.02, + "learning_rate": 1.672809809986222e-05, + "loss": 3.8938, + "step": 2032128 + }, + { + "epoch": 1.02, + "learning_rate": 1.6719728531155434e-05, + "loss": 3.8796, + "step": 2032640 + }, + { + "epoch": 1.02, + "learning_rate": 1.6711342583644914e-05, + "loss": 3.8807, + "step": 2033152 + }, + { + "epoch": 1.02, + "learning_rate": 1.6702956636134393e-05, + "loss": 3.8746, + "step": 2033664 + }, + { + "epoch": 1.02, + "learning_rate": 1.6694570688623873e-05, + "loss": 3.8724, + "step": 2034176 + }, + { + "epoch": 1.02, + "learning_rate": 1.6686184741113353e-05, + "loss": 3.8768, + "step": 2034688 + }, + { + "epoch": 1.02, + "learning_rate": 1.6677798793602833e-05, + "loss": 3.8844, + "step": 2035200 + }, + { + "epoch": 1.02, + "learning_rate": 1.6669412846092313e-05, + "loss": 3.8867, + "step": 2035712 + }, + { + "epoch": 1.02, + "learning_rate": 1.6661026898581793e-05, + "loss": 3.88, + "step": 2036224 + }, + { + "epoch": 1.02, + "learning_rate": 1.6652657329875006e-05, + "loss": 3.8805, + "step": 2036736 + }, + { + "epoch": 1.02, + "learning_rate": 1.6644271382364486e-05, + "loss": 3.8718, + "step": 2037248 + }, + { + "epoch": 1.02, + "learning_rate": 1.6635885434853966e-05, + "loss": 3.8799, + "step": 2037760 + }, + { + "epoch": 1.02, + "learning_rate": 1.6627499487343446e-05, + "loss": 3.8779, + "step": 2038272 + }, + { + "epoch": 1.02, + "learning_rate": 1.6619129918636655e-05, + "loss": 3.8722, + "step": 2038784 + }, + { + "epoch": 1.02, + "learning_rate": 1.6610743971126135e-05, + "loss": 3.8763, + "step": 2039296 + }, + { + "epoch": 1.02, + "learning_rate": 1.6602358023615615e-05, + "loss": 3.8746, + "step": 2039808 + }, + { + "epoch": 1.02, + "learning_rate": 1.6593972076105095e-05, + "loss": 3.8785, + "step": 2040320 + }, + { + "epoch": 1.02, + "learning_rate": 1.6585586128594575e-05, + "loss": 3.8678, + "step": 2040832 + }, + { + "epoch": 1.02, + "learning_rate": 1.6577216559887787e-05, + "loss": 3.8713, + "step": 2041344 + }, + { + "epoch": 1.02, + "learning_rate": 1.6568830612377267e-05, + "loss": 3.8772, + "step": 2041856 + }, + { + "epoch": 1.02, + "learning_rate": 1.6560444664866744e-05, + "loss": 3.8765, + "step": 2042368 + }, + { + "epoch": 1.02, + "learning_rate": 1.6552058717356227e-05, + "loss": 3.8835, + "step": 2042880 + }, + { + "epoch": 1.02, + "learning_rate": 1.6543672769845707e-05, + "loss": 3.8856, + "step": 2043392 + }, + { + "epoch": 1.02, + "learning_rate": 1.6535286822335187e-05, + "loss": 3.8876, + "step": 2043904 + }, + { + "epoch": 1.02, + "learning_rate": 1.652693363243213e-05, + "loss": 3.8807, + "step": 2044416 + }, + { + "epoch": 1.02, + "learning_rate": 1.651854768492161e-05, + "loss": 3.8751, + "step": 2044928 + }, + { + "epoch": 1.02, + "learning_rate": 1.6510161737411088e-05, + "loss": 3.8705, + "step": 2045440 + }, + { + "epoch": 1.02, + "learning_rate": 1.6501775789900568e-05, + "loss": 3.8854, + "step": 2045952 + }, + { + "epoch": 1.02, + "learning_rate": 1.6493389842390048e-05, + "loss": 3.8791, + "step": 2046464 + }, + { + "epoch": 1.02, + "learning_rate": 1.6485003894879528e-05, + "loss": 3.8702, + "step": 2046976 + }, + { + "epoch": 1.02, + "learning_rate": 1.6476617947369008e-05, + "loss": 3.8726, + "step": 2047488 + }, + { + "epoch": 1.02, + "learning_rate": 1.6468231999858488e-05, + "loss": 3.884, + "step": 2048000 + }, + { + "epoch": 1.02, + "learning_rate": 1.6459846052347968e-05, + "loss": 3.8779, + "step": 2048512 + }, + { + "epoch": 1.02, + "learning_rate": 1.6451460104837448e-05, + "loss": 3.8729, + "step": 2049024 + }, + { + "epoch": 1.02, + "learning_rate": 1.6443074157326928e-05, + "loss": 3.8755, + "step": 2049536 + }, + { + "epoch": 1.02, + "learning_rate": 1.6434688209816408e-05, + "loss": 3.8734, + "step": 2050048 + }, + { + "epoch": 1.02, + "learning_rate": 1.6426302262305888e-05, + "loss": 3.8854, + "step": 2050560 + }, + { + "epoch": 1.02, + "learning_rate": 1.64179326935991e-05, + "loss": 3.8828, + "step": 2051072 + }, + { + "epoch": 1.02, + "learning_rate": 1.6409546746088577e-05, + "loss": 3.8854, + "step": 2051584 + }, + { + "epoch": 1.02, + "learning_rate": 1.6401160798578057e-05, + "loss": 3.8744, + "step": 2052096 + }, + { + "epoch": 1.02, + "learning_rate": 1.6392774851067537e-05, + "loss": 3.8814, + "step": 2052608 + }, + { + "epoch": 1.02, + "learning_rate": 1.638440528236075e-05, + "loss": 3.8883, + "step": 2053120 + }, + { + "epoch": 1.02, + "learning_rate": 1.6376035713653962e-05, + "loss": 3.8734, + "step": 2053632 + }, + { + "epoch": 1.02, + "learning_rate": 1.6367649766143442e-05, + "loss": 3.8796, + "step": 2054144 + }, + { + "epoch": 1.02, + "learning_rate": 1.6359263818632922e-05, + "loss": 3.8823, + "step": 2054656 + }, + { + "epoch": 1.02, + "learning_rate": 1.6350877871122398e-05, + "loss": 3.8811, + "step": 2055168 + }, + { + "epoch": 1.02, + "learning_rate": 1.6342508302415614e-05, + "loss": 3.8942, + "step": 2055680 + }, + { + "epoch": 1.02, + "learning_rate": 1.6334122354905094e-05, + "loss": 3.8794, + "step": 2056192 + }, + { + "epoch": 1.02, + "learning_rate": 1.6325736407394574e-05, + "loss": 3.8819, + "step": 2056704 + }, + { + "epoch": 1.02, + "learning_rate": 1.631735045988405e-05, + "loss": 3.8885, + "step": 2057216 + }, + { + "epoch": 1.02, + "learning_rate": 1.630896451237353e-05, + "loss": 3.8795, + "step": 2057728 + }, + { + "epoch": 1.02, + "learning_rate": 1.630057856486301e-05, + "loss": 3.8717, + "step": 2058240 + }, + { + "epoch": 1.02, + "learning_rate": 1.629219261735249e-05, + "loss": 3.8734, + "step": 2058752 + }, + { + "epoch": 1.02, + "learning_rate": 1.6283806669841974e-05, + "loss": 3.8738, + "step": 2059264 + }, + { + "epoch": 1.02, + "learning_rate": 1.6275437101135183e-05, + "loss": 3.8797, + "step": 2059776 + }, + { + "epoch": 1.02, + "learning_rate": 1.6267051153624663e-05, + "loss": 3.8762, + "step": 2060288 + }, + { + "epoch": 1.03, + "eval_loss": 3.989001750946045, + "eval_runtime": 428.4368, + "eval_samples_per_second": 890.659, + "eval_steps_per_second": 27.834, + "step": 2060625 + }, + { + "epoch": 0.0, + "learning_rate": 1.6258681584917872e-05, + "loss": 3.8743, + "step": 2060800 + }, + { + "epoch": 0.0, + "learning_rate": 1.6250295637407352e-05, + "loss": 3.8767, + "step": 2061312 + }, + { + "epoch": 0.0, + "learning_rate": 1.6241909689896835e-05, + "loss": 3.8798, + "step": 2061824 + }, + { + "epoch": 0.0, + "learning_rate": 1.6233523742386315e-05, + "loss": 3.8809, + "step": 2062336 + }, + { + "epoch": 0.0, + "learning_rate": 1.6225137794875795e-05, + "loss": 3.8899, + "step": 2062848 + }, + { + "epoch": 0.0, + "learning_rate": 1.6216751847365275e-05, + "loss": 3.8804, + "step": 2063360 + }, + { + "epoch": 0.0, + "learning_rate": 1.6208365899854752e-05, + "loss": 3.8831, + "step": 2063872 + }, + { + "epoch": 0.0, + "learning_rate": 1.619997995234423e-05, + "loss": 3.8747, + "step": 2064384 + }, + { + "epoch": 0.0, + "learning_rate": 1.619159400483371e-05, + "loss": 3.8709, + "step": 2064896 + }, + { + "epoch": 0.0, + "learning_rate": 1.618320805732319e-05, + "loss": 3.8779, + "step": 2065408 + }, + { + "epoch": 0.0, + "learning_rate": 1.6174822109812675e-05, + "loss": 3.8885, + "step": 2065920 + }, + { + "epoch": 0.0, + "learning_rate": 1.6166436162302155e-05, + "loss": 3.8727, + "step": 2066432 + }, + { + "epoch": 0.0, + "learning_rate": 1.6158066593595364e-05, + "loss": 3.8844, + "step": 2066944 + }, + { + "epoch": 0.0, + "learning_rate": 1.6149680646084844e-05, + "loss": 3.8682, + "step": 2067456 + }, + { + "epoch": 0.0, + "learning_rate": 1.6141294698574324e-05, + "loss": 3.8721, + "step": 2067968 + }, + { + "epoch": 0.0, + "learning_rate": 1.6132908751063804e-05, + "loss": 3.8636, + "step": 2068480 + }, + { + "epoch": 0.0, + "learning_rate": 1.6124522803553284e-05, + "loss": 3.8796, + "step": 2068992 + }, + { + "epoch": 0.0, + "learning_rate": 1.6116136856042764e-05, + "loss": 3.8816, + "step": 2069504 + }, + { + "epoch": 0.0, + "learning_rate": 1.6107750908532244e-05, + "loss": 3.8745, + "step": 2070016 + }, + { + "epoch": 0.0, + "learning_rate": 1.6099364961021724e-05, + "loss": 3.8794, + "step": 2070528 + }, + { + "epoch": 0.0, + "learning_rate": 1.6090979013511204e-05, + "loss": 3.89, + "step": 2071040 + }, + { + "epoch": 0.0, + "learning_rate": 1.6082593066000684e-05, + "loss": 3.8839, + "step": 2071552 + }, + { + "epoch": 0.0, + "learning_rate": 1.6074207118490164e-05, + "loss": 3.8767, + "step": 2072064 + }, + { + "epoch": 0.0, + "learning_rate": 1.6065837549783376e-05, + "loss": 3.8801, + "step": 2072576 + }, + { + "epoch": 0.0, + "learning_rate": 1.6057451602272856e-05, + "loss": 3.8851, + "step": 2073088 + }, + { + "epoch": 0.0, + "learning_rate": 1.6049065654762336e-05, + "loss": 3.8808, + "step": 2073600 + }, + { + "epoch": 0.0, + "learning_rate": 1.6040679707251816e-05, + "loss": 3.8685, + "step": 2074112 + }, + { + "epoch": 0.0, + "learning_rate": 1.6032293759741292e-05, + "loss": 3.8769, + "step": 2074624 + }, + { + "epoch": 0.0, + "learning_rate": 1.6023907812230772e-05, + "loss": 3.8748, + "step": 2075136 + }, + { + "epoch": 0.0, + "learning_rate": 1.6015521864720252e-05, + "loss": 3.8704, + "step": 2075648 + }, + { + "epoch": 0.01, + "learning_rate": 1.6007135917209736e-05, + "loss": 3.8814, + "step": 2076160 + }, + { + "epoch": 0.01, + "learning_rate": 1.5998766348502945e-05, + "loss": 3.8812, + "step": 2076672 + }, + { + "epoch": 0.01, + "learning_rate": 1.5990380400992425e-05, + "loss": 3.8838, + "step": 2077184 + }, + { + "epoch": 0.01, + "learning_rate": 1.5981994453481905e-05, + "loss": 3.8878, + "step": 2077696 + }, + { + "epoch": 0.01, + "learning_rate": 1.5973624884775114e-05, + "loss": 3.8736, + "step": 2078208 + }, + { + "epoch": 0.01, + "learning_rate": 1.596525531606833e-05, + "loss": 3.8849, + "step": 2078720 + }, + { + "epoch": 0.01, + "learning_rate": 1.595686936855781e-05, + "loss": 3.8853, + "step": 2079232 + }, + { + "epoch": 0.01, + "learning_rate": 1.594848342104729e-05, + "loss": 3.8649, + "step": 2079744 + }, + { + "epoch": 0.01, + "learning_rate": 1.5940097473536766e-05, + "loss": 3.8728, + "step": 2080256 + }, + { + "epoch": 0.01, + "learning_rate": 1.5931711526026246e-05, + "loss": 3.8652, + "step": 2080768 + }, + { + "epoch": 0.01, + "learning_rate": 1.5923325578515726e-05, + "loss": 3.8657, + "step": 2081280 + }, + { + "epoch": 0.01, + "learning_rate": 1.5914939631005206e-05, + "loss": 3.8796, + "step": 2081792 + }, + { + "epoch": 0.01, + "learning_rate": 1.5906553683494686e-05, + "loss": 3.8872, + "step": 2082304 + }, + { + "epoch": 0.01, + "learning_rate": 1.589816773598417e-05, + "loss": 3.8707, + "step": 2082816 + }, + { + "epoch": 0.01, + "learning_rate": 1.588978178847365e-05, + "loss": 3.8708, + "step": 2083328 + }, + { + "epoch": 0.01, + "learning_rate": 1.588141221976686e-05, + "loss": 3.8771, + "step": 2083840 + }, + { + "epoch": 0.01, + "learning_rate": 1.587302627225634e-05, + "loss": 3.8775, + "step": 2084352 + }, + { + "epoch": 0.01, + "learning_rate": 1.5864640324745818e-05, + "loss": 3.869, + "step": 2084864 + }, + { + "epoch": 0.01, + "learning_rate": 1.5856254377235298e-05, + "loss": 3.8557, + "step": 2085376 + }, + { + "epoch": 0.01, + "learning_rate": 1.5847868429724778e-05, + "loss": 3.8834, + "step": 2085888 + }, + { + "epoch": 0.01, + "learning_rate": 1.5839482482214258e-05, + "loss": 3.8822, + "step": 2086400 + }, + { + "epoch": 0.01, + "learning_rate": 1.5831096534703738e-05, + "loss": 3.8906, + "step": 2086912 + }, + { + "epoch": 0.01, + "learning_rate": 1.5822710587193218e-05, + "loss": 3.8721, + "step": 2087424 + }, + { + "epoch": 0.01, + "learning_rate": 1.5814341018486427e-05, + "loss": 3.8695, + "step": 2087936 + }, + { + "epoch": 0.01, + "learning_rate": 1.5805955070975907e-05, + "loss": 3.8742, + "step": 2088448 + }, + { + "epoch": 0.01, + "learning_rate": 1.579756912346539e-05, + "loss": 3.8657, + "step": 2088960 + }, + { + "epoch": 0.01, + "learning_rate": 1.578918317595487e-05, + "loss": 3.885, + "step": 2089472 + }, + { + "epoch": 0.01, + "learning_rate": 1.578079722844435e-05, + "loss": 3.8575, + "step": 2089984 + }, + { + "epoch": 0.01, + "learning_rate": 1.577241128093383e-05, + "loss": 3.8993, + "step": 2090496 + }, + { + "epoch": 0.01, + "learning_rate": 1.5764025333423307e-05, + "loss": 3.8759, + "step": 2091008 + }, + { + "epoch": 0.01, + "learning_rate": 1.5755639385912787e-05, + "loss": 3.8591, + "step": 2091520 + }, + { + "epoch": 0.01, + "learning_rate": 1.5747269817206e-05, + "loss": 3.8791, + "step": 2092032 + }, + { + "epoch": 0.01, + "learning_rate": 1.5738900248499212e-05, + "loss": 3.8699, + "step": 2092544 + }, + { + "epoch": 0.01, + "learning_rate": 1.5730514300988692e-05, + "loss": 3.8559, + "step": 2093056 + }, + { + "epoch": 0.01, + "learning_rate": 1.5722128353478172e-05, + "loss": 3.8784, + "step": 2093568 + }, + { + "epoch": 0.01, + "learning_rate": 1.571374240596765e-05, + "loss": 3.8856, + "step": 2094080 + }, + { + "epoch": 0.01, + "learning_rate": 1.5705356458457128e-05, + "loss": 3.8683, + "step": 2094592 + }, + { + "epoch": 0.01, + "learning_rate": 1.5696970510946608e-05, + "loss": 3.8767, + "step": 2095104 + }, + { + "epoch": 0.01, + "learning_rate": 1.568858456343609e-05, + "loss": 3.853, + "step": 2095616 + }, + { + "epoch": 0.01, + "learning_rate": 1.56802149947293e-05, + "loss": 3.8715, + "step": 2096128 + }, + { + "epoch": 0.01, + "learning_rate": 1.567182904721878e-05, + "loss": 3.8714, + "step": 2096640 + }, + { + "epoch": 0.01, + "learning_rate": 1.566344309970826e-05, + "loss": 3.8776, + "step": 2097152 + }, + { + "epoch": 0.01, + "learning_rate": 1.565505715219774e-05, + "loss": 3.8748, + "step": 2097664 + }, + { + "epoch": 0.01, + "learning_rate": 1.564667120468722e-05, + "loss": 3.8705, + "step": 2098176 + }, + { + "epoch": 0.01, + "learning_rate": 1.56382852571767e-05, + "loss": 3.8903, + "step": 2098688 + }, + { + "epoch": 0.01, + "learning_rate": 1.5629899309666184e-05, + "loss": 3.8744, + "step": 2099200 + }, + { + "epoch": 0.01, + "learning_rate": 1.562151336215566e-05, + "loss": 3.8798, + "step": 2099712 + }, + { + "epoch": 0.01, + "learning_rate": 1.5613143793448873e-05, + "loss": 3.8712, + "step": 2100224 + }, + { + "epoch": 0.01, + "learning_rate": 1.5604757845938353e-05, + "loss": 3.8725, + "step": 2100736 + }, + { + "epoch": 0.01, + "learning_rate": 1.5596371898427833e-05, + "loss": 3.8804, + "step": 2101248 + }, + { + "epoch": 0.01, + "learning_rate": 1.558798595091731e-05, + "loss": 3.8714, + "step": 2101760 + }, + { + "epoch": 0.01, + "learning_rate": 1.5579616382210525e-05, + "loss": 3.8716, + "step": 2102272 + }, + { + "epoch": 0.01, + "learning_rate": 1.5571230434700005e-05, + "loss": 3.8846, + "step": 2102784 + }, + { + "epoch": 0.01, + "learning_rate": 1.5562844487189482e-05, + "loss": 3.8692, + "step": 2103296 + }, + { + "epoch": 0.01, + "learning_rate": 1.555445853967896e-05, + "loss": 3.86, + "step": 2103808 + }, + { + "epoch": 0.01, + "learning_rate": 1.5546088970972174e-05, + "loss": 3.8714, + "step": 2104320 + }, + { + "epoch": 0.01, + "learning_rate": 1.5537703023461654e-05, + "loss": 3.8794, + "step": 2104832 + }, + { + "epoch": 0.01, + "learning_rate": 1.5529333454754867e-05, + "loss": 3.868, + "step": 2105344 + }, + { + "epoch": 0.01, + "learning_rate": 1.5520947507244347e-05, + "loss": 3.8789, + "step": 2105856 + }, + { + "epoch": 0.01, + "learning_rate": 1.5512561559733826e-05, + "loss": 3.8694, + "step": 2106368 + }, + { + "epoch": 0.02, + "learning_rate": 1.5504175612223306e-05, + "loss": 3.8633, + "step": 2106880 + }, + { + "epoch": 0.02, + "learning_rate": 1.5495789664712783e-05, + "loss": 3.8703, + "step": 2107392 + }, + { + "epoch": 0.02, + "learning_rate": 1.5487403717202263e-05, + "loss": 3.8682, + "step": 2107904 + }, + { + "epoch": 0.02, + "learning_rate": 1.5479017769691746e-05, + "loss": 3.887, + "step": 2108416 + }, + { + "epoch": 0.02, + "learning_rate": 1.5470648200984955e-05, + "loss": 3.8749, + "step": 2108928 + }, + { + "epoch": 0.02, + "learning_rate": 1.5462262253474435e-05, + "loss": 3.8812, + "step": 2109440 + }, + { + "epoch": 0.02, + "learning_rate": 1.5453876305963915e-05, + "loss": 3.8684, + "step": 2109952 + }, + { + "epoch": 0.02, + "learning_rate": 1.5445490358453395e-05, + "loss": 3.8679, + "step": 2110464 + }, + { + "epoch": 0.02, + "learning_rate": 1.5437120789746608e-05, + "loss": 3.8728, + "step": 2110976 + }, + { + "epoch": 0.02, + "learning_rate": 1.5428734842236088e-05, + "loss": 3.879, + "step": 2111488 + }, + { + "epoch": 0.02, + "learning_rate": 1.5420348894725568e-05, + "loss": 3.8805, + "step": 2112000 + }, + { + "epoch": 0.02, + "learning_rate": 1.5411962947215048e-05, + "loss": 3.8773, + "step": 2112512 + }, + { + "epoch": 0.02, + "learning_rate": 1.5403593378508257e-05, + "loss": 3.8767, + "step": 2113024 + }, + { + "epoch": 0.02, + "learning_rate": 1.5395207430997737e-05, + "loss": 3.8675, + "step": 2113536 + }, + { + "epoch": 0.02, + "learning_rate": 1.5386821483487217e-05, + "loss": 3.8803, + "step": 2114048 + }, + { + "epoch": 0.02, + "learning_rate": 1.53784355359767e-05, + "loss": 3.8698, + "step": 2114560 + }, + { + "epoch": 0.02, + "learning_rate": 1.537004958846618e-05, + "loss": 3.8671, + "step": 2115072 + }, + { + "epoch": 0.02, + "learning_rate": 1.536166364095566e-05, + "loss": 3.8714, + "step": 2115584 + }, + { + "epoch": 0.02, + "learning_rate": 1.535329407224887e-05, + "loss": 3.8713, + "step": 2116096 + }, + { + "epoch": 0.02, + "learning_rate": 1.534490812473835e-05, + "loss": 3.8719, + "step": 2116608 + }, + { + "epoch": 0.02, + "learning_rate": 1.533653855603156e-05, + "loss": 3.8643, + "step": 2117120 + }, + { + "epoch": 0.02, + "learning_rate": 1.532815260852104e-05, + "loss": 3.866, + "step": 2117632 + }, + { + "epoch": 0.02, + "learning_rate": 1.531976666101052e-05, + "loss": 3.8762, + "step": 2118144 + }, + { + "epoch": 0.02, + "learning_rate": 1.53113807135e-05, + "loss": 3.8728, + "step": 2118656 + }, + { + "epoch": 0.02, + "learning_rate": 1.530299476598948e-05, + "loss": 3.8732, + "step": 2119168 + }, + { + "epoch": 0.02, + "learning_rate": 1.529460881847896e-05, + "loss": 3.8796, + "step": 2119680 + }, + { + "epoch": 0.02, + "learning_rate": 1.5286222870968438e-05, + "loss": 3.8884, + "step": 2120192 + }, + { + "epoch": 0.02, + "learning_rate": 1.5277836923457918e-05, + "loss": 3.873, + "step": 2120704 + }, + { + "epoch": 0.02, + "learning_rate": 1.52694509759474e-05, + "loss": 3.8706, + "step": 2121216 + }, + { + "epoch": 0.02, + "learning_rate": 1.526106502843688e-05, + "loss": 3.8689, + "step": 2121728 + }, + { + "epoch": 0.02, + "learning_rate": 1.525269545973009e-05, + "loss": 3.8755, + "step": 2122240 + }, + { + "epoch": 0.02, + "learning_rate": 1.524430951221957e-05, + "loss": 3.8756, + "step": 2122752 + }, + { + "epoch": 0.02, + "learning_rate": 1.5235923564709052e-05, + "loss": 3.8705, + "step": 2123264 + }, + { + "epoch": 0.02, + "learning_rate": 1.5227537617198532e-05, + "loss": 3.8687, + "step": 2123776 + }, + { + "epoch": 0.02, + "learning_rate": 1.5219151669688012e-05, + "loss": 3.8797, + "step": 2124288 + }, + { + "epoch": 0.02, + "learning_rate": 1.5210782100981222e-05, + "loss": 3.8736, + "step": 2124800 + }, + { + "epoch": 0.02, + "learning_rate": 1.5202412532274435e-05, + "loss": 3.866, + "step": 2125312 + }, + { + "epoch": 0.02, + "learning_rate": 1.5194026584763913e-05, + "loss": 3.8739, + "step": 2125824 + }, + { + "epoch": 0.02, + "learning_rate": 1.5185640637253393e-05, + "loss": 3.8723, + "step": 2126336 + }, + { + "epoch": 0.02, + "learning_rate": 1.5177254689742873e-05, + "loss": 3.8779, + "step": 2126848 + }, + { + "epoch": 0.02, + "learning_rate": 1.5168868742232353e-05, + "loss": 3.8784, + "step": 2127360 + }, + { + "epoch": 0.02, + "learning_rate": 1.5160482794721833e-05, + "loss": 3.8839, + "step": 2127872 + }, + { + "epoch": 0.02, + "learning_rate": 1.5152096847211315e-05, + "loss": 3.8702, + "step": 2128384 + }, + { + "epoch": 0.02, + "learning_rate": 1.5143710899700791e-05, + "loss": 3.8776, + "step": 2128896 + }, + { + "epoch": 0.02, + "learning_rate": 1.5135324952190271e-05, + "loss": 3.8808, + "step": 2129408 + }, + { + "epoch": 0.02, + "learning_rate": 1.5126955383483485e-05, + "loss": 3.8694, + "step": 2129920 + }, + { + "epoch": 0.02, + "learning_rate": 1.5118569435972965e-05, + "loss": 3.8733, + "step": 2130432 + }, + { + "epoch": 0.02, + "learning_rate": 1.5110183488462442e-05, + "loss": 3.8762, + "step": 2130944 + }, + { + "epoch": 0.02, + "learning_rate": 1.5101797540951924e-05, + "loss": 3.8861, + "step": 2131456 + }, + { + "epoch": 0.02, + "learning_rate": 1.5093411593441403e-05, + "loss": 3.8856, + "step": 2131968 + }, + { + "epoch": 0.02, + "learning_rate": 1.5085042024734614e-05, + "loss": 3.8718, + "step": 2132480 + }, + { + "epoch": 0.02, + "learning_rate": 1.5076656077224094e-05, + "loss": 3.8774, + "step": 2132992 + }, + { + "epoch": 0.02, + "learning_rate": 1.5068286508517307e-05, + "loss": 3.885, + "step": 2133504 + }, + { + "epoch": 0.02, + "learning_rate": 1.5059900561006787e-05, + "loss": 3.8742, + "step": 2134016 + }, + { + "epoch": 0.02, + "learning_rate": 1.5051514613496265e-05, + "loss": 3.8697, + "step": 2134528 + }, + { + "epoch": 0.02, + "learning_rate": 1.5043128665985745e-05, + "loss": 3.8698, + "step": 2135040 + }, + { + "epoch": 0.02, + "learning_rate": 1.5034742718475225e-05, + "loss": 3.8669, + "step": 2135552 + }, + { + "epoch": 0.02, + "learning_rate": 1.5026356770964706e-05, + "loss": 3.8723, + "step": 2136064 + }, + { + "epoch": 0.02, + "learning_rate": 1.5017970823454186e-05, + "loss": 3.8746, + "step": 2136576 + }, + { + "epoch": 0.03, + "eval_loss": 3.9877591133117676, + "eval_runtime": 296.4213, + "eval_samples_per_second": 1287.327, + "eval_steps_per_second": 40.23, + "step": 2136945 + }, + { + "epoch": 1.0, + "learning_rate": 1.5009584875943666e-05, + "loss": 3.8752, + "step": 2137088 + }, + { + "epoch": 1.0, + "learning_rate": 1.5001198928433146e-05, + "loss": 3.8714, + "step": 2137600 + }, + { + "epoch": 1.0, + "learning_rate": 1.4992812980922625e-05, + "loss": 3.8716, + "step": 2138112 + }, + { + "epoch": 1.0, + "learning_rate": 1.4984427033412105e-05, + "loss": 3.8839, + "step": 2138624 + }, + { + "epoch": 1.0, + "learning_rate": 1.4976041085901585e-05, + "loss": 3.882, + "step": 2139136 + }, + { + "epoch": 1.0, + "learning_rate": 1.4967671517194795e-05, + "loss": 3.8773, + "step": 2139648 + }, + { + "epoch": 1.0, + "learning_rate": 1.4959285569684275e-05, + "loss": 3.8813, + "step": 2140160 + }, + { + "epoch": 1.0, + "learning_rate": 1.4950916000977488e-05, + "loss": 3.8654, + "step": 2140672 + }, + { + "epoch": 1.0, + "learning_rate": 1.494253005346697e-05, + "loss": 3.8647, + "step": 2141184 + }, + { + "epoch": 1.0, + "learning_rate": 1.4934144105956446e-05, + "loss": 3.8745, + "step": 2141696 + }, + { + "epoch": 1.0, + "learning_rate": 1.4925758158445926e-05, + "loss": 3.8803, + "step": 2142208 + }, + { + "epoch": 1.0, + "learning_rate": 1.491738858973914e-05, + "loss": 3.8713, + "step": 2142720 + }, + { + "epoch": 1.0, + "learning_rate": 1.490901902103235e-05, + "loss": 3.8849, + "step": 2143232 + }, + { + "epoch": 1.0, + "learning_rate": 1.4900633073521831e-05, + "loss": 3.8604, + "step": 2143744 + }, + { + "epoch": 1.0, + "learning_rate": 1.489224712601131e-05, + "loss": 3.8691, + "step": 2144256 + }, + { + "epoch": 1.0, + "learning_rate": 1.488386117850079e-05, + "loss": 3.8554, + "step": 2144768 + }, + { + "epoch": 1.0, + "learning_rate": 1.4875475230990269e-05, + "loss": 3.8765, + "step": 2145280 + }, + { + "epoch": 1.0, + "learning_rate": 1.4867089283479749e-05, + "loss": 3.874, + "step": 2145792 + }, + { + "epoch": 1.0, + "learning_rate": 1.4858719714772961e-05, + "loss": 3.8724, + "step": 2146304 + }, + { + "epoch": 1.0, + "learning_rate": 1.4850333767262441e-05, + "loss": 3.8728, + "step": 2146816 + }, + { + "epoch": 1.0, + "learning_rate": 1.484194781975192e-05, + "loss": 3.8875, + "step": 2147328 + }, + { + "epoch": 1.0, + "learning_rate": 1.48335618722414e-05, + "loss": 3.879, + "step": 2147840 + }, + { + "epoch": 1.0, + "learning_rate": 1.482517592473088e-05, + "loss": 3.8756, + "step": 2148352 + }, + { + "epoch": 1.0, + "learning_rate": 1.4816789977220361e-05, + "loss": 3.8733, + "step": 2148864 + }, + { + "epoch": 1.0, + "learning_rate": 1.480842040851357e-05, + "loss": 3.8831, + "step": 2149376 + }, + { + "epoch": 1.0, + "learning_rate": 1.480003446100305e-05, + "loss": 3.8735, + "step": 2149888 + }, + { + "epoch": 1.0, + "learning_rate": 1.4791648513492532e-05, + "loss": 3.8702, + "step": 2150400 + }, + { + "epoch": 1.0, + "learning_rate": 1.4783262565982012e-05, + "loss": 3.8723, + "step": 2150912 + }, + { + "epoch": 1.0, + "learning_rate": 1.4774876618471492e-05, + "loss": 3.8657, + "step": 2151424 + }, + { + "epoch": 1.0, + "learning_rate": 1.4766490670960972e-05, + "loss": 3.8671, + "step": 2151936 + }, + { + "epoch": 1.01, + "learning_rate": 1.475810472345045e-05, + "loss": 3.8782, + "step": 2152448 + }, + { + "epoch": 1.01, + "learning_rate": 1.474971877593993e-05, + "loss": 3.8713, + "step": 2152960 + }, + { + "epoch": 1.01, + "learning_rate": 1.4741349207233143e-05, + "loss": 3.879, + "step": 2153472 + }, + { + "epoch": 1.01, + "learning_rate": 1.4732979638526353e-05, + "loss": 3.883, + "step": 2153984 + }, + { + "epoch": 1.01, + "learning_rate": 1.4724593691015833e-05, + "loss": 3.8696, + "step": 2154496 + }, + { + "epoch": 1.01, + "learning_rate": 1.4716224122309044e-05, + "loss": 3.8803, + "step": 2155008 + }, + { + "epoch": 1.01, + "learning_rate": 1.4707838174798524e-05, + "loss": 3.8803, + "step": 2155520 + }, + { + "epoch": 1.01, + "learning_rate": 1.4699468606091738e-05, + "loss": 3.8646, + "step": 2156032 + }, + { + "epoch": 1.01, + "learning_rate": 1.4691082658581215e-05, + "loss": 3.8672, + "step": 2156544 + }, + { + "epoch": 1.01, + "learning_rate": 1.4682696711070695e-05, + "loss": 3.8598, + "step": 2157056 + }, + { + "epoch": 1.01, + "learning_rate": 1.4674310763560176e-05, + "loss": 3.8602, + "step": 2157568 + }, + { + "epoch": 1.01, + "learning_rate": 1.4665924816049656e-05, + "loss": 3.876, + "step": 2158080 + }, + { + "epoch": 1.01, + "learning_rate": 1.4657538868539136e-05, + "loss": 3.8824, + "step": 2158592 + }, + { + "epoch": 1.01, + "learning_rate": 1.4649152921028616e-05, + "loss": 3.8651, + "step": 2159104 + }, + { + "epoch": 1.01, + "learning_rate": 1.4640766973518096e-05, + "loss": 3.8726, + "step": 2159616 + }, + { + "epoch": 1.01, + "learning_rate": 1.4632381026007574e-05, + "loss": 3.869, + "step": 2160128 + }, + { + "epoch": 1.01, + "learning_rate": 1.4623995078497054e-05, + "loss": 3.8723, + "step": 2160640 + }, + { + "epoch": 1.01, + "learning_rate": 1.4615609130986534e-05, + "loss": 3.863, + "step": 2161152 + }, + { + "epoch": 1.01, + "learning_rate": 1.4607223183476016e-05, + "loss": 3.8531, + "step": 2161664 + }, + { + "epoch": 1.01, + "learning_rate": 1.4598837235965496e-05, + "loss": 3.8763, + "step": 2162176 + }, + { + "epoch": 1.01, + "learning_rate": 1.4590451288454976e-05, + "loss": 3.8829, + "step": 2162688 + }, + { + "epoch": 1.01, + "learning_rate": 1.4582065340944456e-05, + "loss": 3.8846, + "step": 2163200 + }, + { + "epoch": 1.01, + "learning_rate": 1.4573679393433934e-05, + "loss": 3.8641, + "step": 2163712 + }, + { + "epoch": 1.01, + "learning_rate": 1.4565293445923414e-05, + "loss": 3.8693, + "step": 2164224 + }, + { + "epoch": 1.01, + "learning_rate": 1.4556923877216627e-05, + "loss": 3.8675, + "step": 2164736 + }, + { + "epoch": 1.01, + "learning_rate": 1.4548537929706105e-05, + "loss": 3.8606, + "step": 2165248 + }, + { + "epoch": 1.01, + "learning_rate": 1.4540151982195585e-05, + "loss": 3.8847, + "step": 2165760 + }, + { + "epoch": 1.01, + "learning_rate": 1.4531766034685065e-05, + "loss": 3.8556, + "step": 2166272 + }, + { + "epoch": 1.01, + "learning_rate": 1.4523396465978279e-05, + "loss": 3.891, + "step": 2166784 + }, + { + "epoch": 1.01, + "learning_rate": 1.4515010518467756e-05, + "loss": 3.8712, + "step": 2167296 + }, + { + "epoch": 1.01, + "learning_rate": 1.4506624570957235e-05, + "loss": 3.857, + "step": 2167808 + }, + { + "epoch": 1.01, + "learning_rate": 1.4498238623446717e-05, + "loss": 3.8745, + "step": 2168320 + }, + { + "epoch": 1.01, + "learning_rate": 1.4489869054739926e-05, + "loss": 3.8673, + "step": 2168832 + }, + { + "epoch": 1.01, + "learning_rate": 1.4481483107229408e-05, + "loss": 3.852, + "step": 2169344 + }, + { + "epoch": 1.01, + "learning_rate": 1.4473097159718888e-05, + "loss": 3.8723, + "step": 2169856 + }, + { + "epoch": 1.01, + "learning_rate": 1.4464711212208368e-05, + "loss": 3.8811, + "step": 2170368 + }, + { + "epoch": 1.01, + "learning_rate": 1.4456325264697848e-05, + "loss": 3.8655, + "step": 2170880 + }, + { + "epoch": 1.01, + "learning_rate": 1.4447939317187328e-05, + "loss": 3.8741, + "step": 2171392 + }, + { + "epoch": 1.01, + "learning_rate": 1.443955336967681e-05, + "loss": 3.8492, + "step": 2171904 + }, + { + "epoch": 1.01, + "learning_rate": 1.4431167422166286e-05, + "loss": 3.8682, + "step": 2172416 + }, + { + "epoch": 1.01, + "learning_rate": 1.4422781474655766e-05, + "loss": 3.8667, + "step": 2172928 + }, + { + "epoch": 1.01, + "learning_rate": 1.4414395527145247e-05, + "loss": 3.8746, + "step": 2173440 + }, + { + "epoch": 1.01, + "learning_rate": 1.4406009579634727e-05, + "loss": 3.8668, + "step": 2173952 + }, + { + "epoch": 1.01, + "learning_rate": 1.4397623632124207e-05, + "loss": 3.8675, + "step": 2174464 + }, + { + "epoch": 1.01, + "learning_rate": 1.438927044222115e-05, + "loss": 3.885, + "step": 2174976 + }, + { + "epoch": 1.01, + "learning_rate": 1.438088449471063e-05, + "loss": 3.8727, + "step": 2175488 + }, + { + "epoch": 1.01, + "learning_rate": 1.4372514926003841e-05, + "loss": 3.8721, + "step": 2176000 + }, + { + "epoch": 1.01, + "learning_rate": 1.4364128978493321e-05, + "loss": 3.8678, + "step": 2176512 + }, + { + "epoch": 1.01, + "learning_rate": 1.4355743030982801e-05, + "loss": 3.8678, + "step": 2177024 + }, + { + "epoch": 1.01, + "learning_rate": 1.4347357083472281e-05, + "loss": 3.8762, + "step": 2177536 + }, + { + "epoch": 1.01, + "learning_rate": 1.433897113596176e-05, + "loss": 3.8642, + "step": 2178048 + }, + { + "epoch": 1.01, + "learning_rate": 1.433058518845124e-05, + "loss": 3.87, + "step": 2178560 + }, + { + "epoch": 1.01, + "learning_rate": 1.432219924094072e-05, + "loss": 3.8788, + "step": 2179072 + }, + { + "epoch": 1.01, + "learning_rate": 1.4313813293430201e-05, + "loss": 3.8709, + "step": 2179584 + }, + { + "epoch": 1.01, + "learning_rate": 1.4305427345919681e-05, + "loss": 3.8515, + "step": 2180096 + }, + { + "epoch": 1.01, + "learning_rate": 1.4297074156016624e-05, + "loss": 3.8694, + "step": 2180608 + }, + { + "epoch": 1.01, + "learning_rate": 1.4288688208506104e-05, + "loss": 3.8725, + "step": 2181120 + }, + { + "epoch": 1.01, + "learning_rate": 1.4280302260995581e-05, + "loss": 3.8641, + "step": 2181632 + }, + { + "epoch": 1.01, + "learning_rate": 1.4271916313485063e-05, + "loss": 3.8721, + "step": 2182144 + }, + { + "epoch": 1.01, + "learning_rate": 1.4263530365974543e-05, + "loss": 3.866, + "step": 2182656 + }, + { + "epoch": 1.02, + "learning_rate": 1.4255144418464023e-05, + "loss": 3.8576, + "step": 2183168 + }, + { + "epoch": 1.02, + "learning_rate": 1.4246758470953502e-05, + "loss": 3.8633, + "step": 2183680 + }, + { + "epoch": 1.02, + "learning_rate": 1.4238372523442982e-05, + "loss": 3.8664, + "step": 2184192 + }, + { + "epoch": 1.02, + "learning_rate": 1.4229986575932464e-05, + "loss": 3.8792, + "step": 2184704 + }, + { + "epoch": 1.02, + "learning_rate": 1.422160062842194e-05, + "loss": 3.871, + "step": 2185216 + }, + { + "epoch": 1.02, + "learning_rate": 1.4213231059715155e-05, + "loss": 3.8785, + "step": 2185728 + }, + { + "epoch": 1.02, + "learning_rate": 1.4204845112204635e-05, + "loss": 3.8655, + "step": 2186240 + }, + { + "epoch": 1.02, + "learning_rate": 1.4196459164694115e-05, + "loss": 3.8637, + "step": 2186752 + }, + { + "epoch": 1.02, + "learning_rate": 1.4188089595987326e-05, + "loss": 3.8667, + "step": 2187264 + }, + { + "epoch": 1.02, + "learning_rate": 1.4179703648476805e-05, + "loss": 3.8756, + "step": 2187776 + }, + { + "epoch": 1.02, + "learning_rate": 1.4171317700966285e-05, + "loss": 3.8763, + "step": 2188288 + }, + { + "epoch": 1.02, + "learning_rate": 1.4162931753455764e-05, + "loss": 3.8725, + "step": 2188800 + }, + { + "epoch": 1.02, + "learning_rate": 1.4154545805945244e-05, + "loss": 3.871, + "step": 2189312 + }, + { + "epoch": 1.02, + "learning_rate": 1.4146159858434724e-05, + "loss": 3.866, + "step": 2189824 + }, + { + "epoch": 1.02, + "learning_rate": 1.4137773910924204e-05, + "loss": 3.8717, + "step": 2190336 + }, + { + "epoch": 1.02, + "learning_rate": 1.4129387963413685e-05, + "loss": 3.8684, + "step": 2190848 + }, + { + "epoch": 1.02, + "learning_rate": 1.4121002015903165e-05, + "loss": 3.8648, + "step": 2191360 + }, + { + "epoch": 1.02, + "learning_rate": 1.4112616068392645e-05, + "loss": 3.8697, + "step": 2191872 + }, + { + "epoch": 1.02, + "learning_rate": 1.4104246499685856e-05, + "loss": 3.865, + "step": 2192384 + }, + { + "epoch": 1.02, + "learning_rate": 1.4095860552175336e-05, + "loss": 3.8655, + "step": 2192896 + }, + { + "epoch": 1.02, + "learning_rate": 1.4087474604664816e-05, + "loss": 3.8625, + "step": 2193408 + }, + { + "epoch": 1.02, + "learning_rate": 1.4079088657154296e-05, + "loss": 3.8584, + "step": 2193920 + }, + { + "epoch": 1.02, + "learning_rate": 1.4070702709643774e-05, + "loss": 3.8722, + "step": 2194432 + }, + { + "epoch": 1.02, + "learning_rate": 1.4062316762133254e-05, + "loss": 3.871, + "step": 2194944 + }, + { + "epoch": 1.02, + "learning_rate": 1.4053947193426466e-05, + "loss": 3.8692, + "step": 2195456 + }, + { + "epoch": 1.02, + "learning_rate": 1.4045561245915945e-05, + "loss": 3.8755, + "step": 2195968 + }, + { + "epoch": 1.02, + "learning_rate": 1.4037175298405425e-05, + "loss": 3.8818, + "step": 2196480 + }, + { + "epoch": 1.02, + "learning_rate": 1.4028789350894905e-05, + "loss": 3.8659, + "step": 2196992 + }, + { + "epoch": 1.02, + "learning_rate": 1.4020419782188119e-05, + "loss": 3.8672, + "step": 2197504 + }, + { + "epoch": 1.02, + "learning_rate": 1.4012033834677595e-05, + "loss": 3.8648, + "step": 2198016 + }, + { + "epoch": 1.02, + "learning_rate": 1.4003647887167075e-05, + "loss": 3.8724, + "step": 2198528 + }, + { + "epoch": 1.02, + "learning_rate": 1.3995261939656557e-05, + "loss": 3.869, + "step": 2199040 + }, + { + "epoch": 1.02, + "learning_rate": 1.3986875992146037e-05, + "loss": 3.8655, + "step": 2199552 + }, + { + "epoch": 1.02, + "learning_rate": 1.3978490044635517e-05, + "loss": 3.8635, + "step": 2200064 + }, + { + "epoch": 1.02, + "learning_rate": 1.3970104097124997e-05, + "loss": 3.8757, + "step": 2200576 + }, + { + "epoch": 1.02, + "learning_rate": 1.3961734528418208e-05, + "loss": 3.8651, + "step": 2201088 + }, + { + "epoch": 1.02, + "learning_rate": 1.3953348580907688e-05, + "loss": 3.8686, + "step": 2201600 + }, + { + "epoch": 1.02, + "learning_rate": 1.3944962633397168e-05, + "loss": 3.8708, + "step": 2202112 + }, + { + "epoch": 1.02, + "learning_rate": 1.393657668588665e-05, + "loss": 3.8678, + "step": 2202624 + }, + { + "epoch": 1.02, + "learning_rate": 1.3928190738376126e-05, + "loss": 3.8676, + "step": 2203136 + }, + { + "epoch": 1.02, + "learning_rate": 1.3919804790865606e-05, + "loss": 3.8809, + "step": 2203648 + }, + { + "epoch": 1.02, + "learning_rate": 1.391143522215882e-05, + "loss": 3.8732, + "step": 2204160 + }, + { + "epoch": 1.02, + "learning_rate": 1.39030492746483e-05, + "loss": 3.8686, + "step": 2204672 + }, + { + "epoch": 1.02, + "learning_rate": 1.3894663327137778e-05, + "loss": 3.8714, + "step": 2205184 + }, + { + "epoch": 1.02, + "learning_rate": 1.3886277379627258e-05, + "loss": 3.8789, + "step": 2205696 + }, + { + "epoch": 1.02, + "learning_rate": 1.3877891432116738e-05, + "loss": 3.8634, + "step": 2206208 + }, + { + "epoch": 1.02, + "learning_rate": 1.3869505484606218e-05, + "loss": 3.8717, + "step": 2206720 + }, + { + "epoch": 1.02, + "learning_rate": 1.3861119537095698e-05, + "loss": 3.8725, + "step": 2207232 + }, + { + "epoch": 1.02, + "learning_rate": 1.385273358958518e-05, + "loss": 3.8808, + "step": 2207744 + }, + { + "epoch": 1.02, + "learning_rate": 1.3844364020878389e-05, + "loss": 3.878, + "step": 2208256 + }, + { + "epoch": 1.02, + "learning_rate": 1.383597807336787e-05, + "loss": 3.8761, + "step": 2208768 + }, + { + "epoch": 1.02, + "learning_rate": 1.382760850466108e-05, + "loss": 3.8695, + "step": 2209280 + }, + { + "epoch": 1.02, + "learning_rate": 1.381922255715056e-05, + "loss": 3.8787, + "step": 2209792 + }, + { + "epoch": 1.02, + "learning_rate": 1.3810836609640041e-05, + "loss": 3.8729, + "step": 2210304 + }, + { + "epoch": 1.02, + "learning_rate": 1.3802450662129521e-05, + "loss": 3.8626, + "step": 2210816 + }, + { + "epoch": 1.02, + "learning_rate": 1.3794064714619001e-05, + "loss": 3.8696, + "step": 2211328 + }, + { + "epoch": 1.02, + "learning_rate": 1.3785678767108481e-05, + "loss": 3.8582, + "step": 2211840 + }, + { + "epoch": 1.02, + "learning_rate": 1.377729281959796e-05, + "loss": 3.8701, + "step": 2212352 + }, + { + "epoch": 1.02, + "learning_rate": 1.376890687208744e-05, + "loss": 3.8673, + "step": 2212864 + }, + { + "epoch": 1.03, + "eval_loss": 3.986243724822998, + "eval_runtime": 296.7132, + "eval_samples_per_second": 1286.06, + "eval_steps_per_second": 40.19, + "step": 2213265 + }, + { + "epoch": 0.0, + "learning_rate": 1.3760520924576919e-05, + "loss": 3.8748, + "step": 2213376 + }, + { + "epoch": 0.0, + "learning_rate": 1.3752134977066399e-05, + "loss": 3.8656, + "step": 2213888 + }, + { + "epoch": 0.0, + "learning_rate": 1.374374902955588e-05, + "loss": 3.8682, + "step": 2214400 + }, + { + "epoch": 0.0, + "learning_rate": 1.373536308204536e-05, + "loss": 3.8799, + "step": 2214912 + }, + { + "epoch": 0.0, + "learning_rate": 1.3726977134534837e-05, + "loss": 3.8732, + "step": 2215424 + }, + { + "epoch": 0.0, + "learning_rate": 1.3718591187024319e-05, + "loss": 3.8811, + "step": 2215936 + }, + { + "epoch": 0.0, + "learning_rate": 1.3710205239513799e-05, + "loss": 3.8721, + "step": 2216448 + }, + { + "epoch": 0.0, + "learning_rate": 1.3701819292003279e-05, + "loss": 3.8656, + "step": 2216960 + }, + { + "epoch": 0.0, + "learning_rate": 1.3693433344492759e-05, + "loss": 3.8603, + "step": 2217472 + }, + { + "epoch": 0.0, + "learning_rate": 1.368504739698224e-05, + "loss": 3.872, + "step": 2217984 + }, + { + "epoch": 0.0, + "learning_rate": 1.367666144947172e-05, + "loss": 3.8728, + "step": 2218496 + }, + { + "epoch": 0.0, + "learning_rate": 1.3668275501961197e-05, + "loss": 3.8703, + "step": 2219008 + }, + { + "epoch": 0.0, + "learning_rate": 1.3659905933254411e-05, + "loss": 3.8818, + "step": 2219520 + }, + { + "epoch": 0.0, + "learning_rate": 1.3651519985743891e-05, + "loss": 3.8538, + "step": 2220032 + }, + { + "epoch": 0.0, + "learning_rate": 1.3643134038233371e-05, + "loss": 3.8644, + "step": 2220544 + }, + { + "epoch": 0.0, + "learning_rate": 1.363474809072285e-05, + "loss": 3.8566, + "step": 2221056 + }, + { + "epoch": 0.0, + "learning_rate": 1.362636214321233e-05, + "loss": 3.8713, + "step": 2221568 + }, + { + "epoch": 0.0, + "learning_rate": 1.361797619570181e-05, + "loss": 3.8671, + "step": 2222080 + }, + { + "epoch": 0.0, + "learning_rate": 1.3609590248191289e-05, + "loss": 3.8696, + "step": 2222592 + }, + { + "epoch": 0.0, + "learning_rate": 1.36012206794845e-05, + "loss": 3.876, + "step": 2223104 + }, + { + "epoch": 0.0, + "learning_rate": 1.359283473197398e-05, + "loss": 3.8783, + "step": 2223616 + }, + { + "epoch": 0.0, + "learning_rate": 1.358444878446346e-05, + "loss": 3.8753, + "step": 2224128 + }, + { + "epoch": 0.0, + "learning_rate": 1.3576062836952941e-05, + "loss": 3.8678, + "step": 2224640 + }, + { + "epoch": 0.0, + "learning_rate": 1.3567676889442421e-05, + "loss": 3.8725, + "step": 2225152 + }, + { + "epoch": 0.0, + "learning_rate": 1.3559290941931901e-05, + "loss": 3.8794, + "step": 2225664 + }, + { + "epoch": 0.0, + "learning_rate": 1.355090499442138e-05, + "loss": 3.8712, + "step": 2226176 + }, + { + "epoch": 0.0, + "learning_rate": 1.354251904691086e-05, + "loss": 3.8592, + "step": 2226688 + }, + { + "epoch": 0.0, + "learning_rate": 1.3534149478204072e-05, + "loss": 3.8713, + "step": 2227200 + }, + { + "epoch": 0.0, + "learning_rate": 1.3525763530693552e-05, + "loss": 3.8591, + "step": 2227712 + }, + { + "epoch": 0.0, + "learning_rate": 1.351737758318303e-05, + "loss": 3.8655, + "step": 2228224 + }, + { + "epoch": 0.01, + "learning_rate": 1.350899163567251e-05, + "loss": 3.8719, + "step": 2228736 + }, + { + "epoch": 0.01, + "learning_rate": 1.3500622066965723e-05, + "loss": 3.8687, + "step": 2229248 + }, + { + "epoch": 0.01, + "learning_rate": 1.3492236119455201e-05, + "loss": 3.875, + "step": 2229760 + }, + { + "epoch": 0.01, + "learning_rate": 1.3483850171944681e-05, + "loss": 3.8818, + "step": 2230272 + }, + { + "epoch": 0.01, + "learning_rate": 1.3475464224434161e-05, + "loss": 3.8669, + "step": 2230784 + }, + { + "epoch": 0.01, + "learning_rate": 1.3467094655727375e-05, + "loss": 3.8765, + "step": 2231296 + }, + { + "epoch": 0.01, + "learning_rate": 1.3458708708216852e-05, + "loss": 3.8769, + "step": 2231808 + }, + { + "epoch": 0.01, + "learning_rate": 1.3450322760706333e-05, + "loss": 3.8547, + "step": 2232320 + }, + { + "epoch": 0.01, + "learning_rate": 1.3441936813195813e-05, + "loss": 3.8666, + "step": 2232832 + }, + { + "epoch": 0.01, + "learning_rate": 1.3433567244489026e-05, + "loss": 3.8525, + "step": 2233344 + }, + { + "epoch": 0.01, + "learning_rate": 1.3425181296978504e-05, + "loss": 3.8614, + "step": 2233856 + }, + { + "epoch": 0.01, + "learning_rate": 1.3416811728271717e-05, + "loss": 3.8684, + "step": 2234368 + }, + { + "epoch": 0.01, + "learning_rate": 1.3408425780761196e-05, + "loss": 3.8813, + "step": 2234880 + }, + { + "epoch": 0.01, + "learning_rate": 1.3400039833250675e-05, + "loss": 3.857, + "step": 2235392 + }, + { + "epoch": 0.01, + "learning_rate": 1.3391653885740155e-05, + "loss": 3.8671, + "step": 2235904 + }, + { + "epoch": 0.01, + "learning_rate": 1.3383267938229635e-05, + "loss": 3.8665, + "step": 2236416 + }, + { + "epoch": 0.01, + "learning_rate": 1.3374881990719115e-05, + "loss": 3.868, + "step": 2236928 + }, + { + "epoch": 0.01, + "learning_rate": 1.3366496043208596e-05, + "loss": 3.8647, + "step": 2237440 + }, + { + "epoch": 0.01, + "learning_rate": 1.3358110095698076e-05, + "loss": 3.8461, + "step": 2237952 + }, + { + "epoch": 0.01, + "learning_rate": 1.3349724148187556e-05, + "loss": 3.8704, + "step": 2238464 + }, + { + "epoch": 0.01, + "learning_rate": 1.3341338200677034e-05, + "loss": 3.8831, + "step": 2238976 + }, + { + "epoch": 0.01, + "learning_rate": 1.3332952253166514e-05, + "loss": 3.8758, + "step": 2239488 + }, + { + "epoch": 0.01, + "learning_rate": 1.3324566305655994e-05, + "loss": 3.8607, + "step": 2240000 + }, + { + "epoch": 0.01, + "learning_rate": 1.3316180358145474e-05, + "loss": 3.864, + "step": 2240512 + }, + { + "epoch": 0.01, + "learning_rate": 1.3307810789438685e-05, + "loss": 3.8678, + "step": 2241024 + }, + { + "epoch": 0.01, + "learning_rate": 1.3299424841928165e-05, + "loss": 3.8541, + "step": 2241536 + }, + { + "epoch": 0.01, + "learning_rate": 1.3291038894417645e-05, + "loss": 3.8827, + "step": 2242048 + }, + { + "epoch": 0.01, + "learning_rate": 1.3282652946907127e-05, + "loss": 3.8489, + "step": 2242560 + }, + { + "epoch": 0.01, + "learning_rate": 1.3274266999396607e-05, + "loss": 3.8873, + "step": 2243072 + }, + { + "epoch": 0.01, + "learning_rate": 1.3265897430689816e-05, + "loss": 3.8684, + "step": 2243584 + }, + { + "epoch": 0.01, + "learning_rate": 1.3257511483179297e-05, + "loss": 3.8531, + "step": 2244096 + }, + { + "epoch": 0.01, + "learning_rate": 1.3249125535668777e-05, + "loss": 3.8677, + "step": 2244608 + }, + { + "epoch": 0.01, + "learning_rate": 1.3240739588158257e-05, + "loss": 3.8662, + "step": 2245120 + }, + { + "epoch": 0.01, + "learning_rate": 1.3232353640647737e-05, + "loss": 3.8475, + "step": 2245632 + }, + { + "epoch": 0.01, + "learning_rate": 1.3223967693137215e-05, + "loss": 3.8636, + "step": 2246144 + }, + { + "epoch": 0.01, + "learning_rate": 1.3215581745626695e-05, + "loss": 3.88, + "step": 2246656 + }, + { + "epoch": 0.01, + "learning_rate": 1.3207195798116175e-05, + "loss": 3.8628, + "step": 2247168 + }, + { + "epoch": 0.01, + "learning_rate": 1.3198826229409386e-05, + "loss": 3.868, + "step": 2247680 + }, + { + "epoch": 0.01, + "learning_rate": 1.3190440281898866e-05, + "loss": 3.8456, + "step": 2248192 + }, + { + "epoch": 0.01, + "learning_rate": 1.3182054334388346e-05, + "loss": 3.8625, + "step": 2248704 + }, + { + "epoch": 0.01, + "learning_rate": 1.3173668386877828e-05, + "loss": 3.8639, + "step": 2249216 + }, + { + "epoch": 0.01, + "learning_rate": 1.3165282439367308e-05, + "loss": 3.8678, + "step": 2249728 + }, + { + "epoch": 0.01, + "learning_rate": 1.3156912870660518e-05, + "loss": 3.863, + "step": 2250240 + }, + { + "epoch": 0.01, + "learning_rate": 1.3148526923149998e-05, + "loss": 3.8683, + "step": 2250752 + }, + { + "epoch": 0.01, + "learning_rate": 1.3140140975639478e-05, + "loss": 3.8758, + "step": 2251264 + }, + { + "epoch": 0.01, + "learning_rate": 1.3131755028128958e-05, + "loss": 3.8688, + "step": 2251776 + }, + { + "epoch": 0.01, + "learning_rate": 1.3123369080618438e-05, + "loss": 3.8695, + "step": 2252288 + }, + { + "epoch": 0.01, + "learning_rate": 1.311498313310792e-05, + "loss": 3.8663, + "step": 2252800 + }, + { + "epoch": 0.01, + "learning_rate": 1.3106597185597397e-05, + "loss": 3.865, + "step": 2253312 + }, + { + "epoch": 0.01, + "learning_rate": 1.3098211238086876e-05, + "loss": 3.8696, + "step": 2253824 + }, + { + "epoch": 0.01, + "learning_rate": 1.308984166938009e-05, + "loss": 3.8592, + "step": 2254336 + }, + { + "epoch": 0.01, + "learning_rate": 1.3081455721869567e-05, + "loss": 3.8675, + "step": 2254848 + }, + { + "epoch": 0.01, + "learning_rate": 1.3073086153162781e-05, + "loss": 3.8778, + "step": 2255360 + }, + { + "epoch": 0.01, + "learning_rate": 1.3064700205652261e-05, + "loss": 3.8672, + "step": 2255872 + }, + { + "epoch": 0.01, + "learning_rate": 1.3056314258141741e-05, + "loss": 3.8449, + "step": 2256384 + }, + { + "epoch": 0.01, + "learning_rate": 1.304792831063122e-05, + "loss": 3.8691, + "step": 2256896 + }, + { + "epoch": 0.01, + "learning_rate": 1.30395423631207e-05, + "loss": 3.8676, + "step": 2257408 + }, + { + "epoch": 0.01, + "learning_rate": 1.303115641561018e-05, + "loss": 3.8608, + "step": 2257920 + }, + { + "epoch": 0.01, + "learning_rate": 1.302277046809966e-05, + "loss": 3.8676, + "step": 2258432 + }, + { + "epoch": 0.01, + "learning_rate": 1.301440089939287e-05, + "loss": 3.8628, + "step": 2258944 + }, + { + "epoch": 0.02, + "learning_rate": 1.300601495188235e-05, + "loss": 3.8534, + "step": 2259456 + }, + { + "epoch": 0.02, + "learning_rate": 1.299762900437183e-05, + "loss": 3.861, + "step": 2259968 + }, + { + "epoch": 0.02, + "learning_rate": 1.2989243056861312e-05, + "loss": 3.8623, + "step": 2260480 + }, + { + "epoch": 0.02, + "learning_rate": 1.2980857109350792e-05, + "loss": 3.8763, + "step": 2260992 + }, + { + "epoch": 0.02, + "learning_rate": 1.2972471161840272e-05, + "loss": 3.8666, + "step": 2261504 + }, + { + "epoch": 0.02, + "learning_rate": 1.2964101593133482e-05, + "loss": 3.8724, + "step": 2262016 + }, + { + "epoch": 0.02, + "learning_rate": 1.2955715645622962e-05, + "loss": 3.8606, + "step": 2262528 + }, + { + "epoch": 0.02, + "learning_rate": 1.2947329698112442e-05, + "loss": 3.8603, + "step": 2263040 + }, + { + "epoch": 0.02, + "learning_rate": 1.2938943750601922e-05, + "loss": 3.8651, + "step": 2263552 + }, + { + "epoch": 0.02, + "learning_rate": 1.29305578030914e-05, + "loss": 3.8736, + "step": 2264064 + }, + { + "epoch": 0.02, + "learning_rate": 1.292217185558088e-05, + "loss": 3.8698, + "step": 2264576 + }, + { + "epoch": 0.02, + "learning_rate": 1.291378590807036e-05, + "loss": 3.8688, + "step": 2265088 + }, + { + "epoch": 0.02, + "learning_rate": 1.2905399960559842e-05, + "loss": 3.8692, + "step": 2265600 + }, + { + "epoch": 0.02, + "learning_rate": 1.2897030391853051e-05, + "loss": 3.8583, + "step": 2266112 + }, + { + "epoch": 0.02, + "learning_rate": 1.2888660823146265e-05, + "loss": 3.8724, + "step": 2266624 + }, + { + "epoch": 0.02, + "learning_rate": 1.2880274875635745e-05, + "loss": 3.86, + "step": 2267136 + }, + { + "epoch": 0.02, + "learning_rate": 1.2871888928125222e-05, + "loss": 3.8646, + "step": 2267648 + }, + { + "epoch": 0.02, + "learning_rate": 1.2863502980614704e-05, + "loss": 3.8622, + "step": 2268160 + }, + { + "epoch": 0.02, + "learning_rate": 1.2855117033104184e-05, + "loss": 3.8624, + "step": 2268672 + }, + { + "epoch": 0.02, + "learning_rate": 1.2846731085593664e-05, + "loss": 3.8616, + "step": 2269184 + }, + { + "epoch": 0.02, + "learning_rate": 1.2838345138083144e-05, + "loss": 3.8598, + "step": 2269696 + }, + { + "epoch": 0.02, + "learning_rate": 1.2829975569376354e-05, + "loss": 3.8546, + "step": 2270208 + }, + { + "epoch": 0.02, + "learning_rate": 1.2821589621865834e-05, + "loss": 3.8694, + "step": 2270720 + }, + { + "epoch": 0.02, + "learning_rate": 1.2813203674355314e-05, + "loss": 3.8667, + "step": 2271232 + }, + { + "epoch": 0.02, + "learning_rate": 1.2804817726844794e-05, + "loss": 3.8608, + "step": 2271744 + }, + { + "epoch": 0.02, + "learning_rate": 1.2796431779334276e-05, + "loss": 3.8765, + "step": 2272256 + }, + { + "epoch": 0.02, + "learning_rate": 1.2788062210627485e-05, + "loss": 3.88, + "step": 2272768 + }, + { + "epoch": 0.02, + "learning_rate": 1.2779676263116967e-05, + "loss": 3.8628, + "step": 2273280 + }, + { + "epoch": 0.02, + "learning_rate": 1.2771290315606447e-05, + "loss": 3.8632, + "step": 2273792 + }, + { + "epoch": 0.02, + "learning_rate": 1.2762904368095926e-05, + "loss": 3.8621, + "step": 2274304 + }, + { + "epoch": 0.02, + "learning_rate": 1.2754518420585405e-05, + "loss": 3.866, + "step": 2274816 + }, + { + "epoch": 0.02, + "learning_rate": 1.2746148851878617e-05, + "loss": 3.8646, + "step": 2275328 + }, + { + "epoch": 0.02, + "learning_rate": 1.2737762904368097e-05, + "loss": 3.8637, + "step": 2275840 + }, + { + "epoch": 0.02, + "learning_rate": 1.2729376956857577e-05, + "loss": 3.8606, + "step": 2276352 + }, + { + "epoch": 0.02, + "learning_rate": 1.2720991009347055e-05, + "loss": 3.8705, + "step": 2276864 + }, + { + "epoch": 0.02, + "learning_rate": 1.2712605061836535e-05, + "loss": 3.8609, + "step": 2277376 + }, + { + "epoch": 0.02, + "learning_rate": 1.2704219114326015e-05, + "loss": 3.8648, + "step": 2277888 + }, + { + "epoch": 0.02, + "learning_rate": 1.2695849545619226e-05, + "loss": 3.863, + "step": 2278400 + }, + { + "epoch": 0.02, + "learning_rate": 1.2687463598108706e-05, + "loss": 3.8649, + "step": 2278912 + }, + { + "epoch": 0.02, + "learning_rate": 1.2679077650598186e-05, + "loss": 3.8584, + "step": 2279424 + }, + { + "epoch": 0.02, + "learning_rate": 1.2670691703087668e-05, + "loss": 3.88, + "step": 2279936 + }, + { + "epoch": 0.02, + "learning_rate": 1.2662305755577148e-05, + "loss": 3.8713, + "step": 2280448 + }, + { + "epoch": 0.02, + "learning_rate": 1.2653936186870358e-05, + "loss": 3.867, + "step": 2280960 + }, + { + "epoch": 0.02, + "learning_rate": 1.2645550239359838e-05, + "loss": 3.8645, + "step": 2281472 + }, + { + "epoch": 0.02, + "learning_rate": 1.2637164291849318e-05, + "loss": 3.8726, + "step": 2281984 + }, + { + "epoch": 0.02, + "learning_rate": 1.2628778344338798e-05, + "loss": 3.8644, + "step": 2282496 + }, + { + "epoch": 0.02, + "learning_rate": 1.2620392396828278e-05, + "loss": 3.8661, + "step": 2283008 + }, + { + "epoch": 0.02, + "learning_rate": 1.261200644931776e-05, + "loss": 3.8677, + "step": 2283520 + }, + { + "epoch": 0.02, + "learning_rate": 1.2603620501807236e-05, + "loss": 3.8776, + "step": 2284032 + }, + { + "epoch": 0.02, + "learning_rate": 1.259525093310045e-05, + "loss": 3.8726, + "step": 2284544 + }, + { + "epoch": 0.02, + "learning_rate": 1.258686498558993e-05, + "loss": 3.8721, + "step": 2285056 + }, + { + "epoch": 0.02, + "learning_rate": 1.2578479038079407e-05, + "loss": 3.8652, + "step": 2285568 + }, + { + "epoch": 0.02, + "learning_rate": 1.2570093090568889e-05, + "loss": 3.8765, + "step": 2286080 + }, + { + "epoch": 0.02, + "learning_rate": 1.2561707143058369e-05, + "loss": 3.8672, + "step": 2286592 + }, + { + "epoch": 0.02, + "learning_rate": 1.2553337574351581e-05, + "loss": 3.8577, + "step": 2287104 + }, + { + "epoch": 0.02, + "learning_rate": 1.254495162684106e-05, + "loss": 3.8664, + "step": 2287616 + }, + { + "epoch": 0.02, + "learning_rate": 1.253656567933054e-05, + "loss": 3.8527, + "step": 2288128 + }, + { + "epoch": 0.02, + "learning_rate": 1.252817973182002e-05, + "loss": 3.8688, + "step": 2288640 + }, + { + "epoch": 0.02, + "learning_rate": 1.25197937843095e-05, + "loss": 3.8607, + "step": 2289152 + }, + { + "epoch": 0.03, + "eval_loss": 3.985013961791992, + "eval_runtime": 292.1699, + "eval_samples_per_second": 1306.059, + "eval_steps_per_second": 40.815, + "step": 2289585 + }, + { + "epoch": 0.0, + "learning_rate": 1.251140783679898e-05, + "loss": 3.877, + "step": 2289664 + }, + { + "epoch": 0.0, + "learning_rate": 1.2503021889288461e-05, + "loss": 3.8624, + "step": 2290176 + }, + { + "epoch": 0.0, + "learning_rate": 1.249463594177794e-05, + "loss": 3.8626, + "step": 2290688 + }, + { + "epoch": 0.0, + "learning_rate": 1.248624999426742e-05, + "loss": 3.8774, + "step": 2291200 + }, + { + "epoch": 0.0, + "learning_rate": 1.2477864046756899e-05, + "loss": 3.8692, + "step": 2291712 + }, + { + "epoch": 0.0, + "learning_rate": 1.2469478099246379e-05, + "loss": 3.8791, + "step": 2292224 + }, + { + "epoch": 0.0, + "learning_rate": 1.2461092151735859e-05, + "loss": 3.8668, + "step": 2292736 + }, + { + "epoch": 0.0, + "learning_rate": 1.2452706204225339e-05, + "loss": 3.8615, + "step": 2293248 + }, + { + "epoch": 0.0, + "learning_rate": 1.2444320256714819e-05, + "loss": 3.858, + "step": 2293760 + }, + { + "epoch": 0.0, + "learning_rate": 1.2435934309204299e-05, + "loss": 3.8719, + "step": 2294272 + }, + { + "epoch": 0.0, + "learning_rate": 1.2427548361693779e-05, + "loss": 3.8669, + "step": 2294784 + }, + { + "epoch": 0.0, + "learning_rate": 1.2419162414183259e-05, + "loss": 3.865, + "step": 2295296 + }, + { + "epoch": 0.0, + "learning_rate": 1.24108092242802e-05, + "loss": 3.8765, + "step": 2295808 + }, + { + "epoch": 0.0, + "learning_rate": 1.2402423276769682e-05, + "loss": 3.8529, + "step": 2296320 + }, + { + "epoch": 0.0, + "learning_rate": 1.2394037329259162e-05, + "loss": 3.8618, + "step": 2296832 + }, + { + "epoch": 0.0, + "learning_rate": 1.238565138174864e-05, + "loss": 3.8532, + "step": 2297344 + }, + { + "epoch": 0.0, + "learning_rate": 1.237726543423812e-05, + "loss": 3.8626, + "step": 2297856 + }, + { + "epoch": 0.0, + "learning_rate": 1.2368879486727602e-05, + "loss": 3.8682, + "step": 2298368 + }, + { + "epoch": 0.0, + "learning_rate": 1.236049353921708e-05, + "loss": 3.862, + "step": 2298880 + }, + { + "epoch": 0.0, + "learning_rate": 1.235210759170656e-05, + "loss": 3.8696, + "step": 2299392 + }, + { + "epoch": 0.0, + "learning_rate": 1.234372164419604e-05, + "loss": 3.8741, + "step": 2299904 + }, + { + "epoch": 0.0, + "learning_rate": 1.2335335696685522e-05, + "loss": 3.8682, + "step": 2300416 + }, + { + "epoch": 0.0, + "learning_rate": 1.2326949749175e-05, + "loss": 3.8698, + "step": 2300928 + }, + { + "epoch": 0.0, + "learning_rate": 1.231856380166448e-05, + "loss": 3.8638, + "step": 2301440 + }, + { + "epoch": 0.0, + "learning_rate": 1.231017785415396e-05, + "loss": 3.8787, + "step": 2301952 + }, + { + "epoch": 0.0, + "learning_rate": 1.230179190664344e-05, + "loss": 3.8655, + "step": 2302464 + }, + { + "epoch": 0.0, + "learning_rate": 1.229340595913292e-05, + "loss": 3.8619, + "step": 2302976 + }, + { + "epoch": 0.0, + "learning_rate": 1.22850200116224e-05, + "loss": 3.8663, + "step": 2303488 + }, + { + "epoch": 0.0, + "learning_rate": 1.227663406411188e-05, + "loss": 3.8577, + "step": 2304000 + }, + { + "epoch": 0.0, + "learning_rate": 1.226824811660136e-05, + "loss": 3.8623, + "step": 2304512 + }, + { + "epoch": 0.01, + "learning_rate": 1.225987854789457e-05, + "loss": 3.8634, + "step": 2305024 + }, + { + "epoch": 0.01, + "learning_rate": 1.2251492600384052e-05, + "loss": 3.867, + "step": 2305536 + }, + { + "epoch": 0.01, + "learning_rate": 1.224310665287353e-05, + "loss": 3.8705, + "step": 2306048 + }, + { + "epoch": 0.01, + "learning_rate": 1.223472070536301e-05, + "loss": 3.8751, + "step": 2306560 + }, + { + "epoch": 0.01, + "learning_rate": 1.222633475785249e-05, + "loss": 3.8635, + "step": 2307072 + }, + { + "epoch": 0.01, + "learning_rate": 1.2217981567949434e-05, + "loss": 3.8761, + "step": 2307584 + }, + { + "epoch": 0.01, + "learning_rate": 1.2209595620438914e-05, + "loss": 3.8709, + "step": 2308096 + }, + { + "epoch": 0.01, + "learning_rate": 1.2201209672928394e-05, + "loss": 3.8559, + "step": 2308608 + }, + { + "epoch": 0.01, + "learning_rate": 1.2192823725417873e-05, + "loss": 3.8622, + "step": 2309120 + }, + { + "epoch": 0.01, + "learning_rate": 1.2184437777907352e-05, + "loss": 3.8505, + "step": 2309632 + }, + { + "epoch": 0.01, + "learning_rate": 1.2176051830396833e-05, + "loss": 3.8611, + "step": 2310144 + }, + { + "epoch": 0.01, + "learning_rate": 1.2167665882886313e-05, + "loss": 3.8617, + "step": 2310656 + }, + { + "epoch": 0.01, + "learning_rate": 1.2159279935375792e-05, + "loss": 3.8777, + "step": 2311168 + }, + { + "epoch": 0.01, + "learning_rate": 1.2150910366669004e-05, + "loss": 3.8545, + "step": 2311680 + }, + { + "epoch": 0.01, + "learning_rate": 1.2142524419158484e-05, + "loss": 3.8623, + "step": 2312192 + }, + { + "epoch": 0.01, + "learning_rate": 1.2134138471647964e-05, + "loss": 3.8655, + "step": 2312704 + }, + { + "epoch": 0.01, + "learning_rate": 1.2125752524137442e-05, + "loss": 3.8615, + "step": 2313216 + }, + { + "epoch": 0.01, + "learning_rate": 1.2117366576626924e-05, + "loss": 3.8603, + "step": 2313728 + }, + { + "epoch": 0.01, + "learning_rate": 1.2108980629116404e-05, + "loss": 3.843, + "step": 2314240 + }, + { + "epoch": 0.01, + "learning_rate": 1.2100594681605882e-05, + "loss": 3.8666, + "step": 2314752 + }, + { + "epoch": 0.01, + "learning_rate": 1.2092208734095364e-05, + "loss": 3.8763, + "step": 2315264 + }, + { + "epoch": 0.01, + "learning_rate": 1.2083839165388575e-05, + "loss": 3.8731, + "step": 2315776 + }, + { + "epoch": 0.01, + "learning_rate": 1.2075469596681787e-05, + "loss": 3.8592, + "step": 2316288 + }, + { + "epoch": 0.01, + "learning_rate": 1.2067083649171265e-05, + "loss": 3.857, + "step": 2316800 + }, + { + "epoch": 0.01, + "learning_rate": 1.2058697701660745e-05, + "loss": 3.8671, + "step": 2317312 + }, + { + "epoch": 0.01, + "learning_rate": 1.2050311754150225e-05, + "loss": 3.8479, + "step": 2317824 + }, + { + "epoch": 0.01, + "learning_rate": 1.2041925806639707e-05, + "loss": 3.8789, + "step": 2318336 + }, + { + "epoch": 0.01, + "learning_rate": 1.2033556237932916e-05, + "loss": 3.8466, + "step": 2318848 + }, + { + "epoch": 0.01, + "learning_rate": 1.2025170290422396e-05, + "loss": 3.8843, + "step": 2319360 + }, + { + "epoch": 0.01, + "learning_rate": 1.2016784342911878e-05, + "loss": 3.8654, + "step": 2319872 + }, + { + "epoch": 0.01, + "learning_rate": 1.2008398395401356e-05, + "loss": 3.8482, + "step": 2320384 + }, + { + "epoch": 0.01, + "learning_rate": 1.2000012447890836e-05, + "loss": 3.8613, + "step": 2320896 + }, + { + "epoch": 0.01, + "learning_rate": 1.1991626500380316e-05, + "loss": 3.8652, + "step": 2321408 + }, + { + "epoch": 0.01, + "learning_rate": 1.1983240552869797e-05, + "loss": 3.8474, + "step": 2321920 + }, + { + "epoch": 0.01, + "learning_rate": 1.1974854605359276e-05, + "loss": 3.8568, + "step": 2322432 + }, + { + "epoch": 0.01, + "learning_rate": 1.1966485036652488e-05, + "loss": 3.8746, + "step": 2322944 + }, + { + "epoch": 0.01, + "learning_rate": 1.1958099089141968e-05, + "loss": 3.8629, + "step": 2323456 + }, + { + "epoch": 0.01, + "learning_rate": 1.1949729520435179e-05, + "loss": 3.8614, + "step": 2323968 + }, + { + "epoch": 0.01, + "learning_rate": 1.1941343572924659e-05, + "loss": 3.844, + "step": 2324480 + }, + { + "epoch": 0.01, + "learning_rate": 1.1932957625414139e-05, + "loss": 3.8616, + "step": 2324992 + }, + { + "epoch": 0.01, + "learning_rate": 1.1924571677903619e-05, + "loss": 3.8542, + "step": 2325504 + }, + { + "epoch": 0.01, + "learning_rate": 1.1916185730393099e-05, + "loss": 3.8717, + "step": 2326016 + }, + { + "epoch": 0.01, + "learning_rate": 1.190781616168631e-05, + "loss": 3.8572, + "step": 2326528 + }, + { + "epoch": 0.01, + "learning_rate": 1.189943021417579e-05, + "loss": 3.8638, + "step": 2327040 + }, + { + "epoch": 0.01, + "learning_rate": 1.189104426666527e-05, + "loss": 3.8729, + "step": 2327552 + }, + { + "epoch": 0.01, + "learning_rate": 1.188265831915475e-05, + "loss": 3.8647, + "step": 2328064 + }, + { + "epoch": 0.01, + "learning_rate": 1.187427237164423e-05, + "loss": 3.8628, + "step": 2328576 + }, + { + "epoch": 0.01, + "learning_rate": 1.186588642413371e-05, + "loss": 3.864, + "step": 2329088 + }, + { + "epoch": 0.01, + "learning_rate": 1.185751685542692e-05, + "loss": 3.8609, + "step": 2329600 + }, + { + "epoch": 0.01, + "learning_rate": 1.18491309079164e-05, + "loss": 3.8664, + "step": 2330112 + }, + { + "epoch": 0.01, + "learning_rate": 1.184074496040588e-05, + "loss": 3.8577, + "step": 2330624 + }, + { + "epoch": 0.01, + "learning_rate": 1.1832359012895362e-05, + "loss": 3.8642, + "step": 2331136 + }, + { + "epoch": 0.01, + "learning_rate": 1.182397306538484e-05, + "loss": 3.8696, + "step": 2331648 + }, + { + "epoch": 0.01, + "learning_rate": 1.181558711787432e-05, + "loss": 3.8634, + "step": 2332160 + }, + { + "epoch": 0.01, + "learning_rate": 1.18072011703638e-05, + "loss": 3.8448, + "step": 2332672 + }, + { + "epoch": 0.01, + "learning_rate": 1.179881522285328e-05, + "loss": 3.864, + "step": 2333184 + }, + { + "epoch": 0.01, + "learning_rate": 1.179044565414649e-05, + "loss": 3.8584, + "step": 2333696 + }, + { + "epoch": 0.01, + "learning_rate": 1.1782059706635972e-05, + "loss": 3.8621, + "step": 2334208 + }, + { + "epoch": 0.01, + "learning_rate": 1.1773690137929183e-05, + "loss": 3.8616, + "step": 2334720 + }, + { + "epoch": 0.01, + "learning_rate": 1.1765320569222394e-05, + "loss": 3.8652, + "step": 2335232 + }, + { + "epoch": 0.02, + "learning_rate": 1.1756934621711874e-05, + "loss": 3.8524, + "step": 2335744 + }, + { + "epoch": 0.02, + "learning_rate": 1.1748548674201354e-05, + "loss": 3.8548, + "step": 2336256 + }, + { + "epoch": 0.02, + "learning_rate": 1.1740162726690834e-05, + "loss": 3.8561, + "step": 2336768 + }, + { + "epoch": 0.02, + "learning_rate": 1.1731776779180314e-05, + "loss": 3.8713, + "step": 2337280 + }, + { + "epoch": 0.02, + "learning_rate": 1.1723390831669794e-05, + "loss": 3.8597, + "step": 2337792 + }, + { + "epoch": 0.02, + "learning_rate": 1.1715004884159274e-05, + "loss": 3.8703, + "step": 2338304 + }, + { + "epoch": 0.02, + "learning_rate": 1.1706618936648753e-05, + "loss": 3.8567, + "step": 2338816 + }, + { + "epoch": 0.02, + "learning_rate": 1.1698232989138233e-05, + "loss": 3.8597, + "step": 2339328 + }, + { + "epoch": 0.02, + "learning_rate": 1.1689863420431444e-05, + "loss": 3.8575, + "step": 2339840 + }, + { + "epoch": 0.02, + "learning_rate": 1.1681477472920926e-05, + "loss": 3.8708, + "step": 2340352 + }, + { + "epoch": 0.02, + "learning_rate": 1.1673091525410404e-05, + "loss": 3.8666, + "step": 2340864 + }, + { + "epoch": 0.02, + "learning_rate": 1.1664705577899884e-05, + "loss": 3.8656, + "step": 2341376 + }, + { + "epoch": 0.02, + "learning_rate": 1.1656336009193097e-05, + "loss": 3.8618, + "step": 2341888 + }, + { + "epoch": 0.02, + "learning_rate": 1.1647950061682575e-05, + "loss": 3.8578, + "step": 2342400 + }, + { + "epoch": 0.02, + "learning_rate": 1.1639580492975787e-05, + "loss": 3.873, + "step": 2342912 + }, + { + "epoch": 0.02, + "learning_rate": 1.1631194545465267e-05, + "loss": 3.8541, + "step": 2343424 + }, + { + "epoch": 0.02, + "learning_rate": 1.1622808597954747e-05, + "loss": 3.8581, + "step": 2343936 + }, + { + "epoch": 0.02, + "learning_rate": 1.1614422650444226e-05, + "loss": 3.8603, + "step": 2344448 + }, + { + "epoch": 0.02, + "learning_rate": 1.1606036702933707e-05, + "loss": 3.8556, + "step": 2344960 + }, + { + "epoch": 0.02, + "learning_rate": 1.1597650755423187e-05, + "loss": 3.8611, + "step": 2345472 + }, + { + "epoch": 0.02, + "learning_rate": 1.1589264807912665e-05, + "loss": 3.8574, + "step": 2345984 + }, + { + "epoch": 0.02, + "learning_rate": 1.1580878860402145e-05, + "loss": 3.8492, + "step": 2346496 + }, + { + "epoch": 0.02, + "learning_rate": 1.1572492912891627e-05, + "loss": 3.8633, + "step": 2347008 + }, + { + "epoch": 0.02, + "learning_rate": 1.1564106965381105e-05, + "loss": 3.8619, + "step": 2347520 + }, + { + "epoch": 0.02, + "learning_rate": 1.1555737396674316e-05, + "loss": 3.8575, + "step": 2348032 + }, + { + "epoch": 0.02, + "learning_rate": 1.1547367827967529e-05, + "loss": 3.8722, + "step": 2348544 + }, + { + "epoch": 0.02, + "learning_rate": 1.1538981880457008e-05, + "loss": 3.8756, + "step": 2349056 + }, + { + "epoch": 0.02, + "learning_rate": 1.1530595932946488e-05, + "loss": 3.8616, + "step": 2349568 + }, + { + "epoch": 0.02, + "learning_rate": 1.1522209985435968e-05, + "loss": 3.8595, + "step": 2350080 + }, + { + "epoch": 0.02, + "learning_rate": 1.1513824037925448e-05, + "loss": 3.8562, + "step": 2350592 + }, + { + "epoch": 0.02, + "learning_rate": 1.1505438090414928e-05, + "loss": 3.8622, + "step": 2351104 + }, + { + "epoch": 0.02, + "learning_rate": 1.1497052142904408e-05, + "loss": 3.8583, + "step": 2351616 + }, + { + "epoch": 0.02, + "learning_rate": 1.1488666195393888e-05, + "loss": 3.8618, + "step": 2352128 + }, + { + "epoch": 0.02, + "learning_rate": 1.1480296626687099e-05, + "loss": 3.853, + "step": 2352640 + }, + { + "epoch": 0.02, + "learning_rate": 1.1471910679176579e-05, + "loss": 3.8696, + "step": 2353152 + }, + { + "epoch": 0.02, + "learning_rate": 1.1463524731666059e-05, + "loss": 3.8615, + "step": 2353664 + }, + { + "epoch": 0.02, + "learning_rate": 1.1455138784155539e-05, + "loss": 3.8597, + "step": 2354176 + }, + { + "epoch": 0.02, + "learning_rate": 1.1446752836645019e-05, + "loss": 3.861, + "step": 2354688 + }, + { + "epoch": 0.02, + "learning_rate": 1.143839964674196e-05, + "loss": 3.8592, + "step": 2355200 + }, + { + "epoch": 0.02, + "learning_rate": 1.1430013699231442e-05, + "loss": 3.8551, + "step": 2355712 + }, + { + "epoch": 0.02, + "learning_rate": 1.1421627751720922e-05, + "loss": 3.8732, + "step": 2356224 + }, + { + "epoch": 0.02, + "learning_rate": 1.1413241804210402e-05, + "loss": 3.8714, + "step": 2356736 + }, + { + "epoch": 0.02, + "learning_rate": 1.140485585669988e-05, + "loss": 3.8651, + "step": 2357248 + }, + { + "epoch": 0.02, + "learning_rate": 1.1396469909189362e-05, + "loss": 3.8585, + "step": 2357760 + }, + { + "epoch": 0.02, + "learning_rate": 1.1388100340482573e-05, + "loss": 3.8663, + "step": 2358272 + }, + { + "epoch": 0.02, + "learning_rate": 1.1379714392972051e-05, + "loss": 3.8612, + "step": 2358784 + }, + { + "epoch": 0.02, + "learning_rate": 1.1371328445461533e-05, + "loss": 3.8631, + "step": 2359296 + }, + { + "epoch": 0.02, + "learning_rate": 1.1362942497951013e-05, + "loss": 3.8614, + "step": 2359808 + }, + { + "epoch": 0.02, + "learning_rate": 1.1354556550440493e-05, + "loss": 3.8776, + "step": 2360320 + }, + { + "epoch": 0.02, + "learning_rate": 1.1346170602929972e-05, + "loss": 3.8639, + "step": 2360832 + }, + { + "epoch": 0.02, + "learning_rate": 1.1337784655419452e-05, + "loss": 3.8703, + "step": 2361344 + }, + { + "epoch": 0.02, + "learning_rate": 1.1329398707908932e-05, + "loss": 3.8613, + "step": 2361856 + }, + { + "epoch": 0.02, + "learning_rate": 1.1321029139202143e-05, + "loss": 3.8714, + "step": 2362368 + }, + { + "epoch": 0.02, + "learning_rate": 1.1312643191691623e-05, + "loss": 3.8628, + "step": 2362880 + }, + { + "epoch": 0.02, + "learning_rate": 1.1304257244181103e-05, + "loss": 3.8589, + "step": 2363392 + }, + { + "epoch": 0.02, + "learning_rate": 1.1295871296670583e-05, + "loss": 3.864, + "step": 2363904 + }, + { + "epoch": 0.02, + "learning_rate": 1.1287485349160063e-05, + "loss": 3.8504, + "step": 2364416 + }, + { + "epoch": 0.02, + "learning_rate": 1.1279099401649543e-05, + "loss": 3.8633, + "step": 2364928 + }, + { + "epoch": 0.02, + "learning_rate": 1.1270729832942754e-05, + "loss": 3.8607, + "step": 2365440 + }, + { + "epoch": 0.03, + "eval_loss": 3.9842748641967773, + "eval_runtime": 287.8867, + "eval_samples_per_second": 1325.49, + "eval_steps_per_second": 41.423, + "step": 2365905 + }, + { + "epoch": 0.0, + "learning_rate": 1.1262343885432234e-05, + "loss": 3.857, + "step": 2365952 + }, + { + "epoch": 0.0, + "learning_rate": 1.1253957937921714e-05, + "loss": 3.8642, + "step": 2366464 + }, + { + "epoch": 0.0, + "learning_rate": 1.1245571990411194e-05, + "loss": 3.8575, + "step": 2366976 + }, + { + "epoch": 0.0, + "learning_rate": 1.1237186042900674e-05, + "loss": 3.8713, + "step": 2367488 + }, + { + "epoch": 0.0, + "learning_rate": 1.1228800095390154e-05, + "loss": 3.8671, + "step": 2368000 + }, + { + "epoch": 0.0, + "learning_rate": 1.1220414147879633e-05, + "loss": 3.8724, + "step": 2368512 + }, + { + "epoch": 0.0, + "learning_rate": 1.1212028200369113e-05, + "loss": 3.8632, + "step": 2369024 + }, + { + "epoch": 0.0, + "learning_rate": 1.1203642252858593e-05, + "loss": 3.8619, + "step": 2369536 + }, + { + "epoch": 0.0, + "learning_rate": 1.1195256305348073e-05, + "loss": 3.8511, + "step": 2370048 + }, + { + "epoch": 0.0, + "learning_rate": 1.1186870357837553e-05, + "loss": 3.8665, + "step": 2370560 + }, + { + "epoch": 0.0, + "learning_rate": 1.1178484410327032e-05, + "loss": 3.8649, + "step": 2371072 + }, + { + "epoch": 0.0, + "learning_rate": 1.1170098462816513e-05, + "loss": 3.8642, + "step": 2371584 + }, + { + "epoch": 0.0, + "learning_rate": 1.1161712515305993e-05, + "loss": 3.8697, + "step": 2372096 + }, + { + "epoch": 0.0, + "learning_rate": 1.1153342946599204e-05, + "loss": 3.8519, + "step": 2372608 + }, + { + "epoch": 0.0, + "learning_rate": 1.1144956999088684e-05, + "loss": 3.8558, + "step": 2373120 + }, + { + "epoch": 0.0, + "learning_rate": 1.1136571051578164e-05, + "loss": 3.851, + "step": 2373632 + }, + { + "epoch": 0.0, + "learning_rate": 1.1128185104067644e-05, + "loss": 3.857, + "step": 2374144 + }, + { + "epoch": 0.0, + "learning_rate": 1.1119799156557124e-05, + "loss": 3.8626, + "step": 2374656 + }, + { + "epoch": 0.0, + "learning_rate": 1.1111413209046604e-05, + "loss": 3.8624, + "step": 2375168 + }, + { + "epoch": 0.0, + "learning_rate": 1.1103027261536084e-05, + "loss": 3.8615, + "step": 2375680 + }, + { + "epoch": 0.0, + "learning_rate": 1.1094641314025564e-05, + "loss": 3.875, + "step": 2376192 + }, + { + "epoch": 0.0, + "learning_rate": 1.1086255366515044e-05, + "loss": 3.8651, + "step": 2376704 + }, + { + "epoch": 0.0, + "learning_rate": 1.1077869419004524e-05, + "loss": 3.867, + "step": 2377216 + }, + { + "epoch": 0.0, + "learning_rate": 1.1069483471494004e-05, + "loss": 3.8622, + "step": 2377728 + }, + { + "epoch": 0.0, + "learning_rate": 1.1061113902787214e-05, + "loss": 3.8742, + "step": 2378240 + }, + { + "epoch": 0.0, + "learning_rate": 1.1052727955276694e-05, + "loss": 3.857, + "step": 2378752 + }, + { + "epoch": 0.0, + "learning_rate": 1.1044342007766174e-05, + "loss": 3.8605, + "step": 2379264 + }, + { + "epoch": 0.0, + "learning_rate": 1.1035956060255652e-05, + "loss": 3.857, + "step": 2379776 + }, + { + "epoch": 0.0, + "learning_rate": 1.1027586491548865e-05, + "loss": 3.8591, + "step": 2380288 + }, + { + "epoch": 0.0, + "learning_rate": 1.1019200544038345e-05, + "loss": 3.8576, + "step": 2380800 + }, + { + "epoch": 0.01, + "learning_rate": 1.1010814596527825e-05, + "loss": 3.8624, + "step": 2381312 + }, + { + "epoch": 0.01, + "learning_rate": 1.1002445027821036e-05, + "loss": 3.8637, + "step": 2381824 + }, + { + "epoch": 0.01, + "learning_rate": 1.0994059080310516e-05, + "loss": 3.8653, + "step": 2382336 + }, + { + "epoch": 0.01, + "learning_rate": 1.0985673132799997e-05, + "loss": 3.8692, + "step": 2382848 + }, + { + "epoch": 0.01, + "learning_rate": 1.0977287185289477e-05, + "loss": 3.8598, + "step": 2383360 + }, + { + "epoch": 0.01, + "learning_rate": 1.0968917616582686e-05, + "loss": 3.8732, + "step": 2383872 + }, + { + "epoch": 0.01, + "learning_rate": 1.0960531669072168e-05, + "loss": 3.8661, + "step": 2384384 + }, + { + "epoch": 0.01, + "learning_rate": 1.0952145721561648e-05, + "loss": 3.8526, + "step": 2384896 + }, + { + "epoch": 0.01, + "learning_rate": 1.0943759774051126e-05, + "loss": 3.8601, + "step": 2385408 + }, + { + "epoch": 0.01, + "learning_rate": 1.0935373826540606e-05, + "loss": 3.8446, + "step": 2385920 + }, + { + "epoch": 0.01, + "learning_rate": 1.0927004257833819e-05, + "loss": 3.8573, + "step": 2386432 + }, + { + "epoch": 0.01, + "learning_rate": 1.0918618310323299e-05, + "loss": 3.8567, + "step": 2386944 + }, + { + "epoch": 0.01, + "learning_rate": 1.0910232362812779e-05, + "loss": 3.8761, + "step": 2387456 + }, + { + "epoch": 0.01, + "learning_rate": 1.0901846415302259e-05, + "loss": 3.8527, + "step": 2387968 + }, + { + "epoch": 0.01, + "learning_rate": 1.0893460467791738e-05, + "loss": 3.8578, + "step": 2388480 + }, + { + "epoch": 0.01, + "learning_rate": 1.0885074520281217e-05, + "loss": 3.8586, + "step": 2388992 + }, + { + "epoch": 0.01, + "learning_rate": 1.0876688572770698e-05, + "loss": 3.8633, + "step": 2389504 + }, + { + "epoch": 0.01, + "learning_rate": 1.086831900406391e-05, + "loss": 3.8591, + "step": 2390016 + }, + { + "epoch": 0.01, + "learning_rate": 1.0859933056553389e-05, + "loss": 3.839, + "step": 2390528 + }, + { + "epoch": 0.01, + "learning_rate": 1.0851547109042869e-05, + "loss": 3.8591, + "step": 2391040 + }, + { + "epoch": 0.01, + "learning_rate": 1.0843161161532349e-05, + "loss": 3.8737, + "step": 2391552 + }, + { + "epoch": 0.01, + "learning_rate": 1.0834775214021829e-05, + "loss": 3.8675, + "step": 2392064 + }, + { + "epoch": 0.01, + "learning_rate": 1.0826405645315041e-05, + "loss": 3.8584, + "step": 2392576 + }, + { + "epoch": 0.01, + "learning_rate": 1.081801969780452e-05, + "loss": 3.8516, + "step": 2393088 + }, + { + "epoch": 0.01, + "learning_rate": 1.0809633750294e-05, + "loss": 3.8655, + "step": 2393600 + }, + { + "epoch": 0.01, + "learning_rate": 1.080124780278348e-05, + "loss": 3.8476, + "step": 2394112 + }, + { + "epoch": 0.01, + "learning_rate": 1.079286185527296e-05, + "loss": 3.8742, + "step": 2394624 + }, + { + "epoch": 0.01, + "learning_rate": 1.078447590776244e-05, + "loss": 3.8433, + "step": 2395136 + }, + { + "epoch": 0.01, + "learning_rate": 1.077608996025192e-05, + "loss": 3.8782, + "step": 2395648 + }, + { + "epoch": 0.01, + "learning_rate": 1.07677040127414e-05, + "loss": 3.8661, + "step": 2396160 + }, + { + "epoch": 0.01, + "learning_rate": 1.075931806523088e-05, + "loss": 3.8441, + "step": 2396672 + }, + { + "epoch": 0.01, + "learning_rate": 1.075093211772036e-05, + "loss": 3.8564, + "step": 2397184 + }, + { + "epoch": 0.01, + "learning_rate": 1.0742562549013572e-05, + "loss": 3.8692, + "step": 2397696 + }, + { + "epoch": 0.01, + "learning_rate": 1.0734192980306781e-05, + "loss": 3.8363, + "step": 2398208 + }, + { + "epoch": 0.01, + "learning_rate": 1.0725807032796263e-05, + "loss": 3.8518, + "step": 2398720 + }, + { + "epoch": 0.01, + "learning_rate": 1.0717421085285743e-05, + "loss": 3.8747, + "step": 2399232 + }, + { + "epoch": 0.01, + "learning_rate": 1.0709035137775223e-05, + "loss": 3.859, + "step": 2399744 + }, + { + "epoch": 0.01, + "learning_rate": 1.07006491902647e-05, + "loss": 3.8568, + "step": 2400256 + }, + { + "epoch": 0.01, + "learning_rate": 1.0692263242754182e-05, + "loss": 3.8434, + "step": 2400768 + }, + { + "epoch": 0.01, + "learning_rate": 1.0683877295243662e-05, + "loss": 3.8556, + "step": 2401280 + }, + { + "epoch": 0.01, + "learning_rate": 1.067549134773314e-05, + "loss": 3.8512, + "step": 2401792 + }, + { + "epoch": 0.01, + "learning_rate": 1.066710540022262e-05, + "loss": 3.8696, + "step": 2402304 + }, + { + "epoch": 0.01, + "learning_rate": 1.0658719452712102e-05, + "loss": 3.8518, + "step": 2402816 + }, + { + "epoch": 0.01, + "learning_rate": 1.0650349884005313e-05, + "loss": 3.863, + "step": 2403328 + }, + { + "epoch": 0.01, + "learning_rate": 1.0641963936494791e-05, + "loss": 3.8633, + "step": 2403840 + }, + { + "epoch": 0.01, + "learning_rate": 1.0633577988984273e-05, + "loss": 3.8628, + "step": 2404352 + }, + { + "epoch": 0.01, + "learning_rate": 1.0625192041473753e-05, + "loss": 3.8565, + "step": 2404864 + }, + { + "epoch": 0.01, + "learning_rate": 1.0616806093963231e-05, + "loss": 3.8572, + "step": 2405376 + }, + { + "epoch": 0.01, + "learning_rate": 1.0608420146452711e-05, + "loss": 3.8628, + "step": 2405888 + }, + { + "epoch": 0.01, + "learning_rate": 1.0600034198942193e-05, + "loss": 3.8632, + "step": 2406400 + }, + { + "epoch": 0.01, + "learning_rate": 1.0591664630235404e-05, + "loss": 3.8553, + "step": 2406912 + }, + { + "epoch": 0.01, + "learning_rate": 1.0583278682724884e-05, + "loss": 3.8592, + "step": 2407424 + }, + { + "epoch": 0.01, + "learning_rate": 1.0574892735214363e-05, + "loss": 3.8617, + "step": 2407936 + }, + { + "epoch": 0.01, + "learning_rate": 1.0566506787703843e-05, + "loss": 3.8635, + "step": 2408448 + }, + { + "epoch": 0.01, + "learning_rate": 1.0558120840193322e-05, + "loss": 3.8406, + "step": 2408960 + }, + { + "epoch": 0.01, + "learning_rate": 1.0549751271486534e-05, + "loss": 3.8592, + "step": 2409472 + }, + { + "epoch": 0.01, + "learning_rate": 1.0541365323976014e-05, + "loss": 3.856, + "step": 2409984 + }, + { + "epoch": 0.01, + "learning_rate": 1.0532979376465494e-05, + "loss": 3.8604, + "step": 2410496 + }, + { + "epoch": 0.01, + "learning_rate": 1.0524593428954974e-05, + "loss": 3.859, + "step": 2411008 + }, + { + "epoch": 0.01, + "learning_rate": 1.0516207481444454e-05, + "loss": 3.8566, + "step": 2411520 + }, + { + "epoch": 0.02, + "learning_rate": 1.0507821533933934e-05, + "loss": 3.8522, + "step": 2412032 + }, + { + "epoch": 0.02, + "learning_rate": 1.0499451965227145e-05, + "loss": 3.8471, + "step": 2412544 + }, + { + "epoch": 0.02, + "learning_rate": 1.0491066017716625e-05, + "loss": 3.8515, + "step": 2413056 + }, + { + "epoch": 0.02, + "learning_rate": 1.0482680070206105e-05, + "loss": 3.8686, + "step": 2413568 + }, + { + "epoch": 0.02, + "learning_rate": 1.0474294122695585e-05, + "loss": 3.8555, + "step": 2414080 + }, + { + "epoch": 0.02, + "learning_rate": 1.0465908175185065e-05, + "loss": 3.8673, + "step": 2414592 + }, + { + "epoch": 0.02, + "learning_rate": 1.0457538606478275e-05, + "loss": 3.8559, + "step": 2415104 + }, + { + "epoch": 0.02, + "learning_rate": 1.0449152658967757e-05, + "loss": 3.857, + "step": 2415616 + }, + { + "epoch": 0.02, + "learning_rate": 1.0440783090260966e-05, + "loss": 3.8539, + "step": 2416128 + }, + { + "epoch": 0.02, + "learning_rate": 1.0432397142750446e-05, + "loss": 3.8691, + "step": 2416640 + }, + { + "epoch": 0.02, + "learning_rate": 1.0424011195239928e-05, + "loss": 3.861, + "step": 2417152 + }, + { + "epoch": 0.02, + "learning_rate": 1.0415625247729408e-05, + "loss": 3.8637, + "step": 2417664 + }, + { + "epoch": 0.02, + "learning_rate": 1.0407239300218886e-05, + "loss": 3.8577, + "step": 2418176 + }, + { + "epoch": 0.02, + "learning_rate": 1.0398853352708368e-05, + "loss": 3.8525, + "step": 2418688 + }, + { + "epoch": 0.02, + "learning_rate": 1.0390467405197848e-05, + "loss": 3.872, + "step": 2419200 + }, + { + "epoch": 0.02, + "learning_rate": 1.0382081457687326e-05, + "loss": 3.8487, + "step": 2419712 + }, + { + "epoch": 0.02, + "learning_rate": 1.0373711888980538e-05, + "loss": 3.8544, + "step": 2420224 + }, + { + "epoch": 0.02, + "learning_rate": 1.0365342320273749e-05, + "loss": 3.8605, + "step": 2420736 + }, + { + "epoch": 0.02, + "learning_rate": 1.0356956372763229e-05, + "loss": 3.8544, + "step": 2421248 + }, + { + "epoch": 0.02, + "learning_rate": 1.0348570425252709e-05, + "loss": 3.8521, + "step": 2421760 + }, + { + "epoch": 0.02, + "learning_rate": 1.0340184477742189e-05, + "loss": 3.856, + "step": 2422272 + }, + { + "epoch": 0.02, + "learning_rate": 1.0331798530231669e-05, + "loss": 3.8451, + "step": 2422784 + }, + { + "epoch": 0.02, + "learning_rate": 1.0323412582721149e-05, + "loss": 3.8613, + "step": 2423296 + }, + { + "epoch": 0.02, + "learning_rate": 1.0315026635210629e-05, + "loss": 3.8546, + "step": 2423808 + }, + { + "epoch": 0.02, + "learning_rate": 1.0306640687700109e-05, + "loss": 3.8583, + "step": 2424320 + }, + { + "epoch": 0.02, + "learning_rate": 1.0298254740189589e-05, + "loss": 3.863, + "step": 2424832 + }, + { + "epoch": 0.02, + "learning_rate": 1.0289868792679069e-05, + "loss": 3.8741, + "step": 2425344 + }, + { + "epoch": 0.02, + "learning_rate": 1.028149922397228e-05, + "loss": 3.8606, + "step": 2425856 + }, + { + "epoch": 0.02, + "learning_rate": 1.027311327646176e-05, + "loss": 3.8543, + "step": 2426368 + }, + { + "epoch": 0.02, + "learning_rate": 1.026472732895124e-05, + "loss": 3.8547, + "step": 2426880 + }, + { + "epoch": 0.02, + "learning_rate": 1.025634138144072e-05, + "loss": 3.8561, + "step": 2427392 + }, + { + "epoch": 0.02, + "learning_rate": 1.02479554339302e-05, + "loss": 3.8576, + "step": 2427904 + }, + { + "epoch": 0.02, + "learning_rate": 1.023956948641968e-05, + "loss": 3.8581, + "step": 2428416 + }, + { + "epoch": 0.02, + "learning_rate": 1.023118353890916e-05, + "loss": 3.8526, + "step": 2428928 + }, + { + "epoch": 0.02, + "learning_rate": 1.022281397020237e-05, + "loss": 3.8647, + "step": 2429440 + }, + { + "epoch": 0.02, + "learning_rate": 1.021442802269185e-05, + "loss": 3.8573, + "step": 2429952 + }, + { + "epoch": 0.02, + "learning_rate": 1.020604207518133e-05, + "loss": 3.8604, + "step": 2430464 + }, + { + "epoch": 0.02, + "learning_rate": 1.019765612767081e-05, + "loss": 3.8537, + "step": 2430976 + }, + { + "epoch": 0.02, + "learning_rate": 1.0189286558964022e-05, + "loss": 3.8557, + "step": 2431488 + }, + { + "epoch": 0.02, + "learning_rate": 1.0180916990257233e-05, + "loss": 3.8516, + "step": 2432000 + }, + { + "epoch": 0.02, + "learning_rate": 1.0172531042746711e-05, + "loss": 3.8698, + "step": 2432512 + }, + { + "epoch": 0.02, + "learning_rate": 1.0164145095236193e-05, + "loss": 3.8681, + "step": 2433024 + }, + { + "epoch": 0.02, + "learning_rate": 1.0155759147725673e-05, + "loss": 3.8651, + "step": 2433536 + }, + { + "epoch": 0.02, + "learning_rate": 1.0147373200215153e-05, + "loss": 3.8543, + "step": 2434048 + }, + { + "epoch": 0.02, + "learning_rate": 1.0138987252704631e-05, + "loss": 3.8572, + "step": 2434560 + }, + { + "epoch": 0.02, + "learning_rate": 1.0130601305194113e-05, + "loss": 3.8624, + "step": 2435072 + }, + { + "epoch": 0.02, + "learning_rate": 1.0122215357683593e-05, + "loss": 3.8606, + "step": 2435584 + }, + { + "epoch": 0.02, + "learning_rate": 1.0113829410173071e-05, + "loss": 3.8626, + "step": 2436096 + }, + { + "epoch": 0.02, + "learning_rate": 1.0105459841466284e-05, + "loss": 3.868, + "step": 2436608 + }, + { + "epoch": 0.02, + "learning_rate": 1.0097073893955764e-05, + "loss": 3.8616, + "step": 2437120 + }, + { + "epoch": 0.02, + "learning_rate": 1.0088687946445243e-05, + "loss": 3.8686, + "step": 2437632 + }, + { + "epoch": 0.02, + "learning_rate": 1.0080301998934723e-05, + "loss": 3.8582, + "step": 2438144 + }, + { + "epoch": 0.02, + "learning_rate": 1.0071916051424203e-05, + "loss": 3.867, + "step": 2438656 + }, + { + "epoch": 0.02, + "learning_rate": 1.0063530103913683e-05, + "loss": 3.8603, + "step": 2439168 + }, + { + "epoch": 0.02, + "learning_rate": 1.0055144156403162e-05, + "loss": 3.8529, + "step": 2439680 + }, + { + "epoch": 0.02, + "learning_rate": 1.0046774587696374e-05, + "loss": 3.8628, + "step": 2440192 + }, + { + "epoch": 0.02, + "learning_rate": 1.0038388640185854e-05, + "loss": 3.8473, + "step": 2440704 + }, + { + "epoch": 0.02, + "learning_rate": 1.0030002692675334e-05, + "loss": 3.8582, + "step": 2441216 + }, + { + "epoch": 0.02, + "learning_rate": 1.0021633123968545e-05, + "loss": 3.8592, + "step": 2441728 + }, + { + "epoch": 0.03, + "eval_loss": 3.983052968978882, + "eval_runtime": 327.9586, + "eval_samples_per_second": 1163.534, + "eval_steps_per_second": 36.361, + "step": 2442225 + }, + { + "epoch": 1.0, + "learning_rate": 1.0013247176458025e-05, + "loss": 3.856, + "step": 2442240 + }, + { + "epoch": 1.0, + "learning_rate": 1.0004861228947505e-05, + "loss": 3.859, + "step": 2442752 + }, + { + "epoch": 1.0, + "learning_rate": 9.996475281436985e-06, + "loss": 3.8538, + "step": 2443264 + }, + { + "epoch": 1.0, + "learning_rate": 9.988089333926465e-06, + "loss": 3.8665, + "step": 2443776 + }, + { + "epoch": 1.0, + "learning_rate": 9.979703386415945e-06, + "loss": 3.865, + "step": 2444288 + }, + { + "epoch": 1.0, + "learning_rate": 9.971317438905425e-06, + "loss": 3.8717, + "step": 2444800 + }, + { + "epoch": 1.0, + "learning_rate": 9.962947870198635e-06, + "loss": 3.8585, + "step": 2445312 + }, + { + "epoch": 1.0, + "learning_rate": 9.954561922688115e-06, + "loss": 3.8549, + "step": 2445824 + }, + { + "epoch": 1.0, + "learning_rate": 9.946175975177597e-06, + "loss": 3.8514, + "step": 2446336 + }, + { + "epoch": 1.0, + "learning_rate": 9.937790027667075e-06, + "loss": 3.8628, + "step": 2446848 + }, + { + "epoch": 1.0, + "learning_rate": 9.929404080156555e-06, + "loss": 3.8587, + "step": 2447360 + }, + { + "epoch": 1.0, + "learning_rate": 9.921034511449768e-06, + "loss": 3.8626, + "step": 2447872 + }, + { + "epoch": 1.0, + "learning_rate": 9.912664942742978e-06, + "loss": 3.8621, + "step": 2448384 + }, + { + "epoch": 1.0, + "learning_rate": 9.904295374036191e-06, + "loss": 3.8517, + "step": 2448896 + }, + { + "epoch": 1.0, + "learning_rate": 9.89590942652567e-06, + "loss": 3.8548, + "step": 2449408 + }, + { + "epoch": 1.0, + "learning_rate": 9.887523479015149e-06, + "loss": 3.851, + "step": 2449920 + }, + { + "epoch": 1.0, + "learning_rate": 9.87913753150463e-06, + "loss": 3.8478, + "step": 2450432 + }, + { + "epoch": 1.0, + "learning_rate": 9.870751583994109e-06, + "loss": 3.8598, + "step": 2450944 + }, + { + "epoch": 1.0, + "learning_rate": 9.862365636483589e-06, + "loss": 3.8566, + "step": 2451456 + }, + { + "epoch": 1.0, + "learning_rate": 9.853979688973069e-06, + "loss": 3.8593, + "step": 2451968 + }, + { + "epoch": 1.0, + "learning_rate": 9.845593741462549e-06, + "loss": 3.8739, + "step": 2452480 + }, + { + "epoch": 1.0, + "learning_rate": 9.837207793952029e-06, + "loss": 3.8579, + "step": 2452992 + }, + { + "epoch": 1.0, + "learning_rate": 9.828821846441509e-06, + "loss": 3.8648, + "step": 2453504 + }, + { + "epoch": 1.0, + "learning_rate": 9.820435898930989e-06, + "loss": 3.8589, + "step": 2454016 + }, + { + "epoch": 1.0, + "learning_rate": 9.812049951420469e-06, + "loss": 3.8704, + "step": 2454528 + }, + { + "epoch": 1.0, + "learning_rate": 9.803664003909949e-06, + "loss": 3.8555, + "step": 2455040 + }, + { + "epoch": 1.0, + "learning_rate": 9.795294435203161e-06, + "loss": 3.8577, + "step": 2455552 + }, + { + "epoch": 1.0, + "learning_rate": 9.78690848769264e-06, + "loss": 3.8508, + "step": 2456064 + }, + { + "epoch": 1.0, + "learning_rate": 9.77853891898585e-06, + "loss": 3.8597, + "step": 2456576 + }, + { + "epoch": 1.0, + "learning_rate": 9.770152971475332e-06, + "loss": 3.8527, + "step": 2457088 + }, + { + "epoch": 1.01, + "learning_rate": 9.761767023964812e-06, + "loss": 3.8555, + "step": 2457600 + }, + { + "epoch": 1.01, + "learning_rate": 9.75338107645429e-06, + "loss": 3.8651, + "step": 2458112 + }, + { + "epoch": 1.01, + "learning_rate": 9.74499512894377e-06, + "loss": 3.8623, + "step": 2458624 + }, + { + "epoch": 1.01, + "learning_rate": 9.736609181433252e-06, + "loss": 3.8657, + "step": 2459136 + }, + { + "epoch": 1.01, + "learning_rate": 9.72822323392273e-06, + "loss": 3.8612, + "step": 2459648 + }, + { + "epoch": 1.01, + "learning_rate": 9.719853665215942e-06, + "loss": 3.8595, + "step": 2460160 + }, + { + "epoch": 1.01, + "learning_rate": 9.711467717705422e-06, + "loss": 3.8659, + "step": 2460672 + }, + { + "epoch": 1.01, + "learning_rate": 9.703098148998633e-06, + "loss": 3.848, + "step": 2461184 + }, + { + "epoch": 1.01, + "learning_rate": 9.694712201488113e-06, + "loss": 3.8586, + "step": 2461696 + }, + { + "epoch": 1.01, + "learning_rate": 9.686326253977593e-06, + "loss": 3.842, + "step": 2462208 + }, + { + "epoch": 1.01, + "learning_rate": 9.677940306467073e-06, + "loss": 3.854, + "step": 2462720 + }, + { + "epoch": 1.01, + "learning_rate": 9.669554358956551e-06, + "loss": 3.8518, + "step": 2463232 + }, + { + "epoch": 1.01, + "learning_rate": 9.661168411446033e-06, + "loss": 3.8687, + "step": 2463744 + }, + { + "epoch": 1.01, + "learning_rate": 9.652782463935513e-06, + "loss": 3.8546, + "step": 2464256 + }, + { + "epoch": 1.01, + "learning_rate": 9.644396516424993e-06, + "loss": 3.8522, + "step": 2464768 + }, + { + "epoch": 1.01, + "learning_rate": 9.636010568914471e-06, + "loss": 3.8573, + "step": 2465280 + }, + { + "epoch": 1.01, + "learning_rate": 9.627624621403953e-06, + "loss": 3.8578, + "step": 2465792 + }, + { + "epoch": 1.01, + "learning_rate": 9.619238673893433e-06, + "loss": 3.8518, + "step": 2466304 + }, + { + "epoch": 1.01, + "learning_rate": 9.610852726382911e-06, + "loss": 3.8404, + "step": 2466816 + }, + { + "epoch": 1.01, + "learning_rate": 9.602466778872393e-06, + "loss": 3.8502, + "step": 2467328 + }, + { + "epoch": 1.01, + "learning_rate": 9.594080831361873e-06, + "loss": 3.8699, + "step": 2467840 + }, + { + "epoch": 1.01, + "learning_rate": 9.585694883851351e-06, + "loss": 3.8681, + "step": 2468352 + }, + { + "epoch": 1.01, + "learning_rate": 9.57730893634083e-06, + "loss": 3.8542, + "step": 2468864 + }, + { + "epoch": 1.01, + "learning_rate": 9.568955746437774e-06, + "loss": 3.8522, + "step": 2469376 + }, + { + "epoch": 1.01, + "learning_rate": 9.560569798927254e-06, + "loss": 3.8616, + "step": 2469888 + }, + { + "epoch": 1.01, + "learning_rate": 9.552183851416734e-06, + "loss": 3.8412, + "step": 2470400 + }, + { + "epoch": 1.01, + "learning_rate": 9.543797903906214e-06, + "loss": 3.8704, + "step": 2470912 + }, + { + "epoch": 1.01, + "learning_rate": 9.535411956395694e-06, + "loss": 3.8401, + "step": 2471424 + }, + { + "epoch": 1.01, + "learning_rate": 9.527042387688906e-06, + "loss": 3.8708, + "step": 2471936 + }, + { + "epoch": 1.01, + "learning_rate": 9.518656440178385e-06, + "loss": 3.864, + "step": 2472448 + }, + { + "epoch": 1.01, + "learning_rate": 9.510270492667865e-06, + "loss": 3.8419, + "step": 2472960 + }, + { + "epoch": 1.01, + "learning_rate": 9.501884545157345e-06, + "loss": 3.851, + "step": 2473472 + }, + { + "epoch": 1.01, + "learning_rate": 9.493514976450557e-06, + "loss": 3.8636, + "step": 2473984 + }, + { + "epoch": 1.01, + "learning_rate": 9.485129028940035e-06, + "loss": 3.834, + "step": 2474496 + }, + { + "epoch": 1.01, + "learning_rate": 9.476743081429517e-06, + "loss": 3.8487, + "step": 2475008 + }, + { + "epoch": 1.01, + "learning_rate": 9.468357133918997e-06, + "loss": 3.8675, + "step": 2475520 + }, + { + "epoch": 1.01, + "learning_rate": 9.459971186408475e-06, + "loss": 3.8588, + "step": 2476032 + }, + { + "epoch": 1.01, + "learning_rate": 9.451585238897955e-06, + "loss": 3.8536, + "step": 2476544 + }, + { + "epoch": 1.01, + "learning_rate": 9.443215670191168e-06, + "loss": 3.8412, + "step": 2477056 + }, + { + "epoch": 1.01, + "learning_rate": 9.434829722680648e-06, + "loss": 3.8523, + "step": 2477568 + }, + { + "epoch": 1.01, + "learning_rate": 9.426443775170128e-06, + "loss": 3.8446, + "step": 2478080 + }, + { + "epoch": 1.01, + "learning_rate": 9.418057827659608e-06, + "loss": 3.8658, + "step": 2478592 + }, + { + "epoch": 1.01, + "learning_rate": 9.409671880149087e-06, + "loss": 3.8477, + "step": 2479104 + }, + { + "epoch": 1.01, + "learning_rate": 9.401285932638566e-06, + "loss": 3.8611, + "step": 2479616 + }, + { + "epoch": 1.01, + "learning_rate": 9.392899985128047e-06, + "loss": 3.8578, + "step": 2480128 + }, + { + "epoch": 1.01, + "learning_rate": 9.384514037617527e-06, + "loss": 3.8626, + "step": 2480640 + }, + { + "epoch": 1.01, + "learning_rate": 9.376144468910738e-06, + "loss": 3.8548, + "step": 2481152 + }, + { + "epoch": 1.01, + "learning_rate": 9.367758521400218e-06, + "loss": 3.8537, + "step": 2481664 + }, + { + "epoch": 1.01, + "learning_rate": 9.359372573889698e-06, + "loss": 3.8605, + "step": 2482176 + }, + { + "epoch": 1.01, + "learning_rate": 9.351003005182909e-06, + "loss": 3.8578, + "step": 2482688 + }, + { + "epoch": 1.01, + "learning_rate": 9.342633436476121e-06, + "loss": 3.8547, + "step": 2483200 + }, + { + "epoch": 1.01, + "learning_rate": 9.3342474889656e-06, + "loss": 3.852, + "step": 2483712 + }, + { + "epoch": 1.01, + "learning_rate": 9.325861541455081e-06, + "loss": 3.8611, + "step": 2484224 + }, + { + "epoch": 1.01, + "learning_rate": 9.317475593944561e-06, + "loss": 3.8609, + "step": 2484736 + }, + { + "epoch": 1.01, + "learning_rate": 9.30908964643404e-06, + "loss": 3.8427, + "step": 2485248 + }, + { + "epoch": 1.01, + "learning_rate": 9.30070369892352e-06, + "loss": 3.8527, + "step": 2485760 + }, + { + "epoch": 1.01, + "learning_rate": 9.292317751413001e-06, + "loss": 3.8527, + "step": 2486272 + }, + { + "epoch": 1.01, + "learning_rate": 9.28393180390248e-06, + "loss": 3.8581, + "step": 2486784 + }, + { + "epoch": 1.01, + "learning_rate": 9.27554585639196e-06, + "loss": 3.8542, + "step": 2487296 + }, + { + "epoch": 1.01, + "learning_rate": 9.267176287685172e-06, + "loss": 3.8559, + "step": 2487808 + }, + { + "epoch": 1.02, + "learning_rate": 9.258790340174652e-06, + "loss": 3.8478, + "step": 2488320 + }, + { + "epoch": 1.02, + "learning_rate": 9.25040439266413e-06, + "loss": 3.8435, + "step": 2488832 + }, + { + "epoch": 1.02, + "learning_rate": 9.24201844515361e-06, + "loss": 3.8489, + "step": 2489344 + }, + { + "epoch": 1.02, + "learning_rate": 9.233632497643092e-06, + "loss": 3.8624, + "step": 2489856 + }, + { + "epoch": 1.02, + "learning_rate": 9.22524655013257e-06, + "loss": 3.8534, + "step": 2490368 + }, + { + "epoch": 1.02, + "learning_rate": 9.216876981425782e-06, + "loss": 3.8647, + "step": 2490880 + }, + { + "epoch": 1.02, + "learning_rate": 9.208491033915262e-06, + "loss": 3.8559, + "step": 2491392 + }, + { + "epoch": 1.02, + "learning_rate": 9.200105086404742e-06, + "loss": 3.8511, + "step": 2491904 + }, + { + "epoch": 1.02, + "learning_rate": 9.19171913889422e-06, + "loss": 3.8507, + "step": 2492416 + }, + { + "epoch": 1.02, + "learning_rate": 9.183333191383702e-06, + "loss": 3.8641, + "step": 2492928 + }, + { + "epoch": 1.02, + "learning_rate": 9.174963622676913e-06, + "loss": 3.8572, + "step": 2493440 + }, + { + "epoch": 1.02, + "learning_rate": 9.166577675166391e-06, + "loss": 3.8647, + "step": 2493952 + }, + { + "epoch": 1.02, + "learning_rate": 9.158208106459604e-06, + "loss": 3.8497, + "step": 2494464 + }, + { + "epoch": 1.02, + "learning_rate": 9.149822158949084e-06, + "loss": 3.8478, + "step": 2494976 + }, + { + "epoch": 1.02, + "learning_rate": 9.141436211438564e-06, + "loss": 3.8699, + "step": 2495488 + }, + { + "epoch": 1.02, + "learning_rate": 9.133050263928044e-06, + "loss": 3.8433, + "step": 2496000 + }, + { + "epoch": 1.02, + "learning_rate": 9.124664316417524e-06, + "loss": 3.8519, + "step": 2496512 + }, + { + "epoch": 1.02, + "learning_rate": 9.116278368907003e-06, + "loss": 3.855, + "step": 2497024 + }, + { + "epoch": 1.02, + "learning_rate": 9.107908800200216e-06, + "loss": 3.8551, + "step": 2497536 + }, + { + "epoch": 1.02, + "learning_rate": 9.099522852689694e-06, + "loss": 3.8476, + "step": 2498048 + }, + { + "epoch": 1.02, + "learning_rate": 9.091136905179174e-06, + "loss": 3.8499, + "step": 2498560 + }, + { + "epoch": 1.02, + "learning_rate": 9.082750957668656e-06, + "loss": 3.847, + "step": 2499072 + }, + { + "epoch": 1.02, + "learning_rate": 9.074365010158134e-06, + "loss": 3.855, + "step": 2499584 + }, + { + "epoch": 1.02, + "learning_rate": 9.065979062647614e-06, + "loss": 3.8553, + "step": 2500096 + }, + { + "epoch": 1.02, + "learning_rate": 9.057593115137094e-06, + "loss": 3.8525, + "step": 2500608 + }, + { + "epoch": 1.02, + "learning_rate": 9.049207167626574e-06, + "loss": 3.8641, + "step": 2501120 + }, + { + "epoch": 1.02, + "learning_rate": 9.040821220116054e-06, + "loss": 3.8705, + "step": 2501632 + }, + { + "epoch": 1.02, + "learning_rate": 9.032451651409265e-06, + "loss": 3.8586, + "step": 2502144 + }, + { + "epoch": 1.02, + "learning_rate": 9.024065703898746e-06, + "loss": 3.8531, + "step": 2502656 + }, + { + "epoch": 1.02, + "learning_rate": 9.015679756388225e-06, + "loss": 3.8494, + "step": 2503168 + }, + { + "epoch": 1.02, + "learning_rate": 9.007293808877705e-06, + "loss": 3.8501, + "step": 2503680 + }, + { + "epoch": 1.02, + "learning_rate": 8.998907861367186e-06, + "loss": 3.8566, + "step": 2504192 + }, + { + "epoch": 1.02, + "learning_rate": 8.990538292660397e-06, + "loss": 3.8575, + "step": 2504704 + }, + { + "epoch": 1.02, + "learning_rate": 8.982152345149875e-06, + "loss": 3.8488, + "step": 2505216 + }, + { + "epoch": 1.02, + "learning_rate": 8.973766397639357e-06, + "loss": 3.8594, + "step": 2505728 + }, + { + "epoch": 1.02, + "learning_rate": 8.965380450128837e-06, + "loss": 3.8553, + "step": 2506240 + }, + { + "epoch": 1.02, + "learning_rate": 8.956994502618315e-06, + "loss": 3.8552, + "step": 2506752 + }, + { + "epoch": 1.02, + "learning_rate": 8.948608555107795e-06, + "loss": 3.8541, + "step": 2507264 + }, + { + "epoch": 1.02, + "learning_rate": 8.940238986401008e-06, + "loss": 3.8497, + "step": 2507776 + }, + { + "epoch": 1.02, + "learning_rate": 8.931869417694218e-06, + "loss": 3.8501, + "step": 2508288 + }, + { + "epoch": 1.02, + "learning_rate": 8.923483470183698e-06, + "loss": 3.8666, + "step": 2508800 + }, + { + "epoch": 1.02, + "learning_rate": 8.915097522673178e-06, + "loss": 3.8645, + "step": 2509312 + }, + { + "epoch": 1.02, + "learning_rate": 8.906711575162658e-06, + "loss": 3.8633, + "step": 2509824 + }, + { + "epoch": 1.02, + "learning_rate": 8.898325627652138e-06, + "loss": 3.8464, + "step": 2510336 + }, + { + "epoch": 1.02, + "learning_rate": 8.889956058945349e-06, + "loss": 3.8591, + "step": 2510848 + }, + { + "epoch": 1.02, + "learning_rate": 8.881570111434829e-06, + "loss": 3.8558, + "step": 2511360 + }, + { + "epoch": 1.02, + "learning_rate": 8.87318416392431e-06, + "loss": 3.8595, + "step": 2511872 + }, + { + "epoch": 1.02, + "learning_rate": 8.864798216413789e-06, + "loss": 3.8536, + "step": 2512384 + }, + { + "epoch": 1.02, + "learning_rate": 8.856412268903269e-06, + "loss": 3.87, + "step": 2512896 + }, + { + "epoch": 1.02, + "learning_rate": 8.848026321392749e-06, + "loss": 3.8524, + "step": 2513408 + }, + { + "epoch": 1.02, + "learning_rate": 8.839640373882229e-06, + "loss": 3.8698, + "step": 2513920 + }, + { + "epoch": 1.02, + "learning_rate": 8.831254426371709e-06, + "loss": 3.8517, + "step": 2514432 + }, + { + "epoch": 1.02, + "learning_rate": 8.822868478861189e-06, + "loss": 3.8675, + "step": 2514944 + }, + { + "epoch": 1.02, + "learning_rate": 8.814482531350669e-06, + "loss": 3.8562, + "step": 2515456 + }, + { + "epoch": 1.02, + "learning_rate": 8.80611296264388e-06, + "loss": 3.8487, + "step": 2515968 + }, + { + "epoch": 1.02, + "learning_rate": 8.79772701513336e-06, + "loss": 3.8607, + "step": 2516480 + }, + { + "epoch": 1.02, + "learning_rate": 8.789341067622841e-06, + "loss": 3.8472, + "step": 2516992 + }, + { + "epoch": 1.02, + "learning_rate": 8.78095512011232e-06, + "loss": 3.8539, + "step": 2517504 + }, + { + "epoch": 1.02, + "learning_rate": 8.77258555140553e-06, + "loss": 3.8569, + "step": 2518016 + }, + { + "epoch": 1.02, + "learning_rate": 8.764199603895012e-06, + "loss": 3.8521, + "step": 2518528 + }, + { + "epoch": 1.03, + "eval_loss": 3.9822466373443604, + "eval_runtime": 312.2034, + "eval_samples_per_second": 1222.251, + "eval_steps_per_second": 38.196, + "step": 2518545 + }, + { + "epoch": 0.0, + "learning_rate": 8.755813656384492e-06, + "loss": 3.8561, + "step": 2519040 + }, + { + "epoch": 0.0, + "learning_rate": 8.74742770887397e-06, + "loss": 3.8541, + "step": 2519552 + }, + { + "epoch": 0.0, + "learning_rate": 8.73904176136345e-06, + "loss": 3.86, + "step": 2520064 + }, + { + "epoch": 0.0, + "learning_rate": 8.730655813852931e-06, + "loss": 3.8585, + "step": 2520576 + }, + { + "epoch": 0.0, + "learning_rate": 8.72226986634241e-06, + "loss": 3.8747, + "step": 2521088 + }, + { + "epoch": 0.0, + "learning_rate": 8.71388391883189e-06, + "loss": 3.8519, + "step": 2521600 + }, + { + "epoch": 0.0, + "learning_rate": 8.705497971321371e-06, + "loss": 3.8535, + "step": 2522112 + }, + { + "epoch": 0.0, + "learning_rate": 8.69711202381085e-06, + "loss": 3.8449, + "step": 2522624 + }, + { + "epoch": 0.0, + "learning_rate": 8.68872607630033e-06, + "loss": 3.8609, + "step": 2523136 + }, + { + "epoch": 0.0, + "learning_rate": 8.68034012878981e-06, + "loss": 3.8581, + "step": 2523648 + }, + { + "epoch": 0.0, + "learning_rate": 8.671954181279291e-06, + "loss": 3.8562, + "step": 2524160 + }, + { + "epoch": 0.0, + "learning_rate": 8.66356823376877e-06, + "loss": 3.8607, + "step": 2524672 + }, + { + "epoch": 0.0, + "learning_rate": 8.65519866506198e-06, + "loss": 3.8457, + "step": 2525184 + }, + { + "epoch": 0.0, + "learning_rate": 8.646812717551462e-06, + "loss": 3.8589, + "step": 2525696 + }, + { + "epoch": 0.0, + "learning_rate": 8.63842677004094e-06, + "loss": 3.8453, + "step": 2526208 + }, + { + "epoch": 0.0, + "learning_rate": 8.63004082253042e-06, + "loss": 3.8463, + "step": 2526720 + }, + { + "epoch": 0.0, + "learning_rate": 8.6216548750199e-06, + "loss": 3.8534, + "step": 2527232 + }, + { + "epoch": 0.0, + "learning_rate": 8.613268927509382e-06, + "loss": 3.8525, + "step": 2527744 + }, + { + "epoch": 0.0, + "learning_rate": 8.60489935880259e-06, + "loss": 3.8566, + "step": 2528256 + }, + { + "epoch": 0.0, + "learning_rate": 8.596513411292072e-06, + "loss": 3.8757, + "step": 2528768 + }, + { + "epoch": 0.0, + "learning_rate": 8.588127463781552e-06, + "loss": 3.8531, + "step": 2529280 + }, + { + "epoch": 0.0, + "learning_rate": 8.57974151627103e-06, + "loss": 3.8613, + "step": 2529792 + }, + { + "epoch": 0.0, + "learning_rate": 8.57135556876051e-06, + "loss": 3.8561, + "step": 2530304 + }, + { + "epoch": 0.0, + "learning_rate": 8.562969621249992e-06, + "loss": 3.8654, + "step": 2530816 + }, + { + "epoch": 0.0, + "learning_rate": 8.554583673739472e-06, + "loss": 3.8528, + "step": 2531328 + }, + { + "epoch": 0.0, + "learning_rate": 8.54619772622895e-06, + "loss": 3.8581, + "step": 2531840 + }, + { + "epoch": 0.0, + "learning_rate": 8.537828157522163e-06, + "loss": 3.847, + "step": 2532352 + }, + { + "epoch": 0.0, + "learning_rate": 8.529442210011643e-06, + "loss": 3.8567, + "step": 2532864 + }, + { + "epoch": 0.0, + "learning_rate": 8.521056262501121e-06, + "loss": 3.8477, + "step": 2533376 + }, + { + "epoch": 0.01, + "learning_rate": 8.512670314990603e-06, + "loss": 3.8553, + "step": 2533888 + }, + { + "epoch": 0.01, + "learning_rate": 8.504284367480083e-06, + "loss": 3.8584, + "step": 2534400 + }, + { + "epoch": 0.01, + "learning_rate": 8.495914798773294e-06, + "loss": 3.8569, + "step": 2534912 + }, + { + "epoch": 0.01, + "learning_rate": 8.487528851262774e-06, + "loss": 3.8649, + "step": 2535424 + }, + { + "epoch": 0.01, + "learning_rate": 8.479159282555984e-06, + "loss": 3.8552, + "step": 2535936 + }, + { + "epoch": 0.01, + "learning_rate": 8.470789713849197e-06, + "loss": 3.8624, + "step": 2536448 + }, + { + "epoch": 0.01, + "learning_rate": 8.462403766338677e-06, + "loss": 3.8585, + "step": 2536960 + }, + { + "epoch": 0.01, + "learning_rate": 8.454017818828155e-06, + "loss": 3.85, + "step": 2537472 + }, + { + "epoch": 0.01, + "learning_rate": 8.445631871317635e-06, + "loss": 3.8517, + "step": 2537984 + }, + { + "epoch": 0.01, + "learning_rate": 8.437245923807117e-06, + "loss": 3.8392, + "step": 2538496 + }, + { + "epoch": 0.01, + "learning_rate": 8.428859976296595e-06, + "loss": 3.8517, + "step": 2539008 + }, + { + "epoch": 0.01, + "learning_rate": 8.420474028786075e-06, + "loss": 3.851, + "step": 2539520 + }, + { + "epoch": 0.01, + "learning_rate": 8.412088081275555e-06, + "loss": 3.8649, + "step": 2540032 + }, + { + "epoch": 0.01, + "learning_rate": 8.403702133765036e-06, + "loss": 3.8505, + "step": 2540544 + }, + { + "epoch": 0.01, + "learning_rate": 8.395316186254515e-06, + "loss": 3.8527, + "step": 2541056 + }, + { + "epoch": 0.01, + "learning_rate": 8.386930238743995e-06, + "loss": 3.8531, + "step": 2541568 + }, + { + "epoch": 0.01, + "learning_rate": 8.378544291233476e-06, + "loss": 3.8537, + "step": 2542080 + }, + { + "epoch": 0.01, + "learning_rate": 8.370158343722955e-06, + "loss": 3.8512, + "step": 2542592 + }, + { + "epoch": 0.01, + "learning_rate": 8.361788775016165e-06, + "loss": 3.8391, + "step": 2543104 + }, + { + "epoch": 0.01, + "learning_rate": 8.353402827505647e-06, + "loss": 3.8456, + "step": 2543616 + }, + { + "epoch": 0.01, + "learning_rate": 8.345016879995127e-06, + "loss": 3.8665, + "step": 2544128 + }, + { + "epoch": 0.01, + "learning_rate": 8.336630932484605e-06, + "loss": 3.8639, + "step": 2544640 + }, + { + "epoch": 0.01, + "learning_rate": 8.328261363777818e-06, + "loss": 3.8478, + "step": 2545152 + }, + { + "epoch": 0.01, + "learning_rate": 8.319875416267298e-06, + "loss": 3.8507, + "step": 2545664 + }, + { + "epoch": 0.01, + "learning_rate": 8.311489468756776e-06, + "loss": 3.8615, + "step": 2546176 + }, + { + "epoch": 0.01, + "learning_rate": 8.303103521246258e-06, + "loss": 3.8392, + "step": 2546688 + }, + { + "epoch": 0.01, + "learning_rate": 8.294717573735738e-06, + "loss": 3.8624, + "step": 2547200 + }, + { + "epoch": 0.01, + "learning_rate": 8.286348005028948e-06, + "loss": 3.8379, + "step": 2547712 + }, + { + "epoch": 0.01, + "learning_rate": 8.277962057518428e-06, + "loss": 3.87, + "step": 2548224 + }, + { + "epoch": 0.01, + "learning_rate": 8.269576110007908e-06, + "loss": 3.8588, + "step": 2548736 + }, + { + "epoch": 0.01, + "learning_rate": 8.261190162497388e-06, + "loss": 3.8463, + "step": 2549248 + }, + { + "epoch": 0.01, + "learning_rate": 8.252804214986866e-06, + "loss": 3.8431, + "step": 2549760 + }, + { + "epoch": 0.01, + "learning_rate": 8.244418267476348e-06, + "loss": 3.8616, + "step": 2550272 + }, + { + "epoch": 0.01, + "learning_rate": 8.236065077573291e-06, + "loss": 3.8323, + "step": 2550784 + }, + { + "epoch": 0.01, + "learning_rate": 8.227679130062771e-06, + "loss": 3.8461, + "step": 2551296 + }, + { + "epoch": 0.01, + "learning_rate": 8.21929318255225e-06, + "loss": 3.8619, + "step": 2551808 + }, + { + "epoch": 0.01, + "learning_rate": 8.21090723504173e-06, + "loss": 3.8574, + "step": 2552320 + }, + { + "epoch": 0.01, + "learning_rate": 8.202521287531211e-06, + "loss": 3.8503, + "step": 2552832 + }, + { + "epoch": 0.01, + "learning_rate": 8.194135340020691e-06, + "loss": 3.8355, + "step": 2553344 + }, + { + "epoch": 0.01, + "learning_rate": 8.18574939251017e-06, + "loss": 3.8494, + "step": 2553856 + }, + { + "epoch": 0.01, + "learning_rate": 8.17736344499965e-06, + "loss": 3.847, + "step": 2554368 + }, + { + "epoch": 0.01, + "learning_rate": 8.168977497489131e-06, + "loss": 3.8588, + "step": 2554880 + }, + { + "epoch": 0.01, + "learning_rate": 8.16059154997861e-06, + "loss": 3.8479, + "step": 2555392 + }, + { + "epoch": 0.01, + "learning_rate": 8.15222198127182e-06, + "loss": 3.8555, + "step": 2555904 + }, + { + "epoch": 0.01, + "learning_rate": 8.143836033761302e-06, + "loss": 3.8574, + "step": 2556416 + }, + { + "epoch": 0.01, + "learning_rate": 8.135450086250782e-06, + "loss": 3.8638, + "step": 2556928 + }, + { + "epoch": 0.01, + "learning_rate": 8.12706413874026e-06, + "loss": 3.8484, + "step": 2557440 + }, + { + "epoch": 0.01, + "learning_rate": 8.11867819122974e-06, + "loss": 3.8531, + "step": 2557952 + }, + { + "epoch": 0.01, + "learning_rate": 8.110308622522952e-06, + "loss": 3.8562, + "step": 2558464 + }, + { + "epoch": 0.01, + "learning_rate": 8.10192267501243e-06, + "loss": 3.8553, + "step": 2558976 + }, + { + "epoch": 0.01, + "learning_rate": 8.093536727501912e-06, + "loss": 3.8519, + "step": 2559488 + }, + { + "epoch": 0.01, + "learning_rate": 8.085150779991392e-06, + "loss": 3.849, + "step": 2560000 + }, + { + "epoch": 0.01, + "learning_rate": 8.07676483248087e-06, + "loss": 3.8578, + "step": 2560512 + }, + { + "epoch": 0.01, + "learning_rate": 8.06837888497035e-06, + "loss": 3.8537, + "step": 2561024 + }, + { + "epoch": 0.01, + "learning_rate": 8.059992937459832e-06, + "loss": 3.841, + "step": 2561536 + }, + { + "epoch": 0.01, + "learning_rate": 8.051606989949312e-06, + "loss": 3.8453, + "step": 2562048 + }, + { + "epoch": 0.01, + "learning_rate": 8.04322104243879e-06, + "loss": 3.852, + "step": 2562560 + }, + { + "epoch": 0.01, + "learning_rate": 8.034867852535734e-06, + "loss": 3.8556, + "step": 2563072 + }, + { + "epoch": 0.01, + "learning_rate": 8.026481905025214e-06, + "loss": 3.8546, + "step": 2563584 + }, + { + "epoch": 0.01, + "learning_rate": 8.018095957514694e-06, + "loss": 3.8478, + "step": 2564096 + }, + { + "epoch": 0.02, + "learning_rate": 8.009726388807904e-06, + "loss": 3.8452, + "step": 2564608 + }, + { + "epoch": 0.02, + "learning_rate": 8.001340441297384e-06, + "loss": 3.8412, + "step": 2565120 + }, + { + "epoch": 0.02, + "learning_rate": 7.992954493786866e-06, + "loss": 3.8462, + "step": 2565632 + }, + { + "epoch": 0.02, + "learning_rate": 7.984568546276344e-06, + "loss": 3.8579, + "step": 2566144 + }, + { + "epoch": 0.02, + "learning_rate": 7.976182598765824e-06, + "loss": 3.8534, + "step": 2566656 + }, + { + "epoch": 0.02, + "learning_rate": 7.967796651255304e-06, + "loss": 3.8575, + "step": 2567168 + }, + { + "epoch": 0.02, + "learning_rate": 7.959410703744786e-06, + "loss": 3.8583, + "step": 2567680 + }, + { + "epoch": 0.02, + "learning_rate": 7.951024756234264e-06, + "loss": 3.8465, + "step": 2568192 + }, + { + "epoch": 0.02, + "learning_rate": 7.942638808723744e-06, + "loss": 3.8482, + "step": 2568704 + }, + { + "epoch": 0.02, + "learning_rate": 7.934252861213224e-06, + "loss": 3.8572, + "step": 2569216 + }, + { + "epoch": 0.02, + "learning_rate": 7.925883292506435e-06, + "loss": 3.8604, + "step": 2569728 + }, + { + "epoch": 0.02, + "learning_rate": 7.917497344995915e-06, + "loss": 3.8592, + "step": 2570240 + }, + { + "epoch": 0.02, + "learning_rate": 7.909111397485396e-06, + "loss": 3.8497, + "step": 2570752 + }, + { + "epoch": 0.02, + "learning_rate": 7.900725449974876e-06, + "loss": 3.8457, + "step": 2571264 + }, + { + "epoch": 0.02, + "learning_rate": 7.892339502464355e-06, + "loss": 3.8645, + "step": 2571776 + }, + { + "epoch": 0.02, + "learning_rate": 7.883969933757567e-06, + "loss": 3.8412, + "step": 2572288 + }, + { + "epoch": 0.02, + "learning_rate": 7.875583986247047e-06, + "loss": 3.8501, + "step": 2572800 + }, + { + "epoch": 0.02, + "learning_rate": 7.867198038736525e-06, + "loss": 3.8513, + "step": 2573312 + }, + { + "epoch": 0.02, + "learning_rate": 7.858812091226005e-06, + "loss": 3.8516, + "step": 2573824 + }, + { + "epoch": 0.02, + "learning_rate": 7.850442522519218e-06, + "loss": 3.8422, + "step": 2574336 + }, + { + "epoch": 0.02, + "learning_rate": 7.842056575008698e-06, + "loss": 3.8496, + "step": 2574848 + }, + { + "epoch": 0.02, + "learning_rate": 7.833670627498178e-06, + "loss": 3.8435, + "step": 2575360 + }, + { + "epoch": 0.02, + "learning_rate": 7.825284679987658e-06, + "loss": 3.8526, + "step": 2575872 + }, + { + "epoch": 0.02, + "learning_rate": 7.816898732477138e-06, + "loss": 3.8493, + "step": 2576384 + }, + { + "epoch": 0.02, + "learning_rate": 7.808512784966616e-06, + "loss": 3.8457, + "step": 2576896 + }, + { + "epoch": 0.02, + "learning_rate": 7.800126837456098e-06, + "loss": 3.8641, + "step": 2577408 + }, + { + "epoch": 0.02, + "learning_rate": 7.791757268749308e-06, + "loss": 3.8663, + "step": 2577920 + }, + { + "epoch": 0.02, + "learning_rate": 7.783371321238788e-06, + "loss": 3.8544, + "step": 2578432 + }, + { + "epoch": 0.02, + "learning_rate": 7.774985373728268e-06, + "loss": 3.8503, + "step": 2578944 + }, + { + "epoch": 0.02, + "learning_rate": 7.766599426217748e-06, + "loss": 3.847, + "step": 2579456 + }, + { + "epoch": 0.02, + "learning_rate": 7.758213478707228e-06, + "loss": 3.8482, + "step": 2579968 + }, + { + "epoch": 0.02, + "learning_rate": 7.74984391000044e-06, + "loss": 3.8495, + "step": 2580480 + }, + { + "epoch": 0.02, + "learning_rate": 7.741457962489919e-06, + "loss": 3.8591, + "step": 2580992 + }, + { + "epoch": 0.02, + "learning_rate": 7.733072014979399e-06, + "loss": 3.8419, + "step": 2581504 + }, + { + "epoch": 0.02, + "learning_rate": 7.724686067468879e-06, + "loss": 3.8546, + "step": 2582016 + }, + { + "epoch": 0.02, + "learning_rate": 7.716300119958359e-06, + "loss": 3.8517, + "step": 2582528 + }, + { + "epoch": 0.02, + "learning_rate": 7.707914172447839e-06, + "loss": 3.8565, + "step": 2583040 + }, + { + "epoch": 0.02, + "learning_rate": 7.699544603741051e-06, + "loss": 3.8429, + "step": 2583552 + }, + { + "epoch": 0.02, + "learning_rate": 7.691158656230531e-06, + "loss": 3.8489, + "step": 2584064 + }, + { + "epoch": 0.02, + "learning_rate": 7.68277270872001e-06, + "loss": 3.8511, + "step": 2584576 + }, + { + "epoch": 0.02, + "learning_rate": 7.67438676120949e-06, + "loss": 3.8565, + "step": 2585088 + }, + { + "epoch": 0.02, + "learning_rate": 7.666000813698971e-06, + "loss": 3.8614, + "step": 2585600 + }, + { + "epoch": 0.02, + "learning_rate": 7.65761486618845e-06, + "loss": 3.8636, + "step": 2586112 + }, + { + "epoch": 0.02, + "learning_rate": 7.64922891867793e-06, + "loss": 3.8443, + "step": 2586624 + }, + { + "epoch": 0.02, + "learning_rate": 7.640859349971142e-06, + "loss": 3.8556, + "step": 2587136 + }, + { + "epoch": 0.02, + "learning_rate": 7.632473402460622e-06, + "loss": 3.8572, + "step": 2587648 + }, + { + "epoch": 0.02, + "learning_rate": 7.624087454950101e-06, + "loss": 3.853, + "step": 2588160 + }, + { + "epoch": 0.02, + "learning_rate": 7.615701507439581e-06, + "loss": 3.8518, + "step": 2588672 + }, + { + "epoch": 0.02, + "learning_rate": 7.607331938732792e-06, + "loss": 3.8622, + "step": 2589184 + }, + { + "epoch": 0.02, + "learning_rate": 7.5989459912222715e-06, + "loss": 3.8517, + "step": 2589696 + }, + { + "epoch": 0.02, + "learning_rate": 7.590560043711751e-06, + "loss": 3.8698, + "step": 2590208 + }, + { + "epoch": 0.02, + "learning_rate": 7.582174096201232e-06, + "loss": 3.8468, + "step": 2590720 + }, + { + "epoch": 0.02, + "learning_rate": 7.573788148690712e-06, + "loss": 3.8639, + "step": 2591232 + }, + { + "epoch": 0.02, + "learning_rate": 7.565402201180191e-06, + "loss": 3.8545, + "step": 2591744 + }, + { + "epoch": 0.02, + "learning_rate": 7.557016253669671e-06, + "loss": 3.8469, + "step": 2592256 + }, + { + "epoch": 0.02, + "learning_rate": 7.548630306159152e-06, + "loss": 3.8556, + "step": 2592768 + }, + { + "epoch": 0.02, + "learning_rate": 7.540260737452362e-06, + "loss": 3.8474, + "step": 2593280 + }, + { + "epoch": 0.02, + "learning_rate": 7.531874789941842e-06, + "loss": 3.8479, + "step": 2593792 + }, + { + "epoch": 0.02, + "learning_rate": 7.5235052212350544e-06, + "loss": 3.8532, + "step": 2594304 + }, + { + "epoch": 0.02, + "learning_rate": 7.515119273724534e-06, + "loss": 3.8487, + "step": 2594816 + }, + { + "epoch": 0.03, + "eval_loss": 3.9816415309906006, + "eval_runtime": 289.5827, + "eval_samples_per_second": 1317.727, + "eval_steps_per_second": 41.18, + "step": 2594865 + }, + { + "epoch": 1.0, + "learning_rate": 7.5067333262140135e-06, + "loss": 3.8501, + "step": 2595328 + }, + { + "epoch": 1.0, + "learning_rate": 7.4983473787034935e-06, + "loss": 3.8497, + "step": 2595840 + }, + { + "epoch": 1.0, + "learning_rate": 7.489961431192974e-06, + "loss": 3.8627, + "step": 2596352 + }, + { + "epoch": 1.0, + "learning_rate": 7.481591862486184e-06, + "loss": 3.8552, + "step": 2596864 + }, + { + "epoch": 1.0, + "learning_rate": 7.473205914975664e-06, + "loss": 3.8687, + "step": 2597376 + }, + { + "epoch": 1.0, + "learning_rate": 7.464819967465145e-06, + "loss": 3.8491, + "step": 2597888 + }, + { + "epoch": 1.0, + "learning_rate": 7.456434019954625e-06, + "loss": 3.8532, + "step": 2598400 + }, + { + "epoch": 1.0, + "learning_rate": 7.448048072444104e-06, + "loss": 3.8427, + "step": 2598912 + }, + { + "epoch": 1.0, + "learning_rate": 7.439678503737316e-06, + "loss": 3.861, + "step": 2599424 + }, + { + "epoch": 1.0, + "learning_rate": 7.431292556226796e-06, + "loss": 3.8516, + "step": 2599936 + }, + { + "epoch": 1.0, + "learning_rate": 7.422906608716275e-06, + "loss": 3.8506, + "step": 2600448 + }, + { + "epoch": 1.0, + "learning_rate": 7.4145206612057555e-06, + "loss": 3.8642, + "step": 2600960 + }, + { + "epoch": 1.0, + "learning_rate": 7.406151092498967e-06, + "loss": 3.8427, + "step": 2601472 + }, + { + "epoch": 1.0, + "learning_rate": 7.397765144988447e-06, + "loss": 3.8572, + "step": 2601984 + }, + { + "epoch": 1.0, + "learning_rate": 7.389379197477926e-06, + "loss": 3.8388, + "step": 2602496 + }, + { + "epoch": 1.0, + "learning_rate": 7.380993249967406e-06, + "loss": 3.8481, + "step": 2603008 + }, + { + "epoch": 1.0, + "learning_rate": 7.372607302456887e-06, + "loss": 3.8473, + "step": 2603520 + }, + { + "epoch": 1.0, + "learning_rate": 7.364221354946365e-06, + "loss": 3.853, + "step": 2604032 + }, + { + "epoch": 1.0, + "learning_rate": 7.355835407435846e-06, + "loss": 3.8498, + "step": 2604544 + }, + { + "epoch": 1.0, + "learning_rate": 7.347465838729058e-06, + "loss": 3.8742, + "step": 2605056 + }, + { + "epoch": 1.0, + "learning_rate": 7.339079891218538e-06, + "loss": 3.8484, + "step": 2605568 + }, + { + "epoch": 1.0, + "learning_rate": 7.330693943708017e-06, + "loss": 3.8572, + "step": 2606080 + }, + { + "epoch": 1.0, + "learning_rate": 7.3223079961974976e-06, + "loss": 3.8546, + "step": 2606592 + }, + { + "epoch": 1.0, + "learning_rate": 7.3139220486869775e-06, + "loss": 3.8636, + "step": 2607104 + }, + { + "epoch": 1.0, + "learning_rate": 7.305536101176457e-06, + "loss": 3.849, + "step": 2607616 + }, + { + "epoch": 1.0, + "learning_rate": 7.2971501536659366e-06, + "loss": 3.8588, + "step": 2608128 + }, + { + "epoch": 1.0, + "learning_rate": 7.288764206155417e-06, + "loss": 3.8441, + "step": 2608640 + }, + { + "epoch": 1.0, + "learning_rate": 7.280394637448629e-06, + "loss": 3.8518, + "step": 2609152 + }, + { + "epoch": 1.0, + "learning_rate": 7.272008689938107e-06, + "loss": 3.8435, + "step": 2609664 + }, + { + "epoch": 1.01, + "learning_rate": 7.263622742427588e-06, + "loss": 3.8542, + "step": 2610176 + }, + { + "epoch": 1.01, + "learning_rate": 7.255236794917068e-06, + "loss": 3.8554, + "step": 2610688 + }, + { + "epoch": 1.01, + "learning_rate": 7.24686722621028e-06, + "loss": 3.8553, + "step": 2611200 + }, + { + "epoch": 1.01, + "learning_rate": 7.238481278699759e-06, + "loss": 3.8661, + "step": 2611712 + }, + { + "epoch": 1.01, + "learning_rate": 7.230095331189239e-06, + "loss": 3.8502, + "step": 2612224 + }, + { + "epoch": 1.01, + "learning_rate": 7.221742141286182e-06, + "loss": 3.858, + "step": 2612736 + }, + { + "epoch": 1.01, + "learning_rate": 7.213356193775663e-06, + "loss": 3.8559, + "step": 2613248 + }, + { + "epoch": 1.01, + "learning_rate": 7.204970246265141e-06, + "loss": 3.849, + "step": 2613760 + }, + { + "epoch": 1.01, + "learning_rate": 7.196584298754622e-06, + "loss": 3.847, + "step": 2614272 + }, + { + "epoch": 1.01, + "learning_rate": 7.188198351244102e-06, + "loss": 3.8388, + "step": 2614784 + }, + { + "epoch": 1.01, + "learning_rate": 7.179812403733581e-06, + "loss": 3.8425, + "step": 2615296 + }, + { + "epoch": 1.01, + "learning_rate": 7.171426456223061e-06, + "loss": 3.8488, + "step": 2615808 + }, + { + "epoch": 1.01, + "learning_rate": 7.163040508712542e-06, + "loss": 3.8602, + "step": 2616320 + }, + { + "epoch": 1.01, + "learning_rate": 7.154654561202021e-06, + "loss": 3.8516, + "step": 2616832 + }, + { + "epoch": 1.01, + "learning_rate": 7.146268613691501e-06, + "loss": 3.8445, + "step": 2617344 + }, + { + "epoch": 1.01, + "learning_rate": 7.137882666180981e-06, + "loss": 3.8563, + "step": 2617856 + }, + { + "epoch": 1.01, + "learning_rate": 7.1294967186704616e-06, + "loss": 3.8469, + "step": 2618368 + }, + { + "epoch": 1.01, + "learning_rate": 7.1211271499636715e-06, + "loss": 3.847, + "step": 2618880 + }, + { + "epoch": 1.01, + "learning_rate": 7.112741202453152e-06, + "loss": 3.8418, + "step": 2619392 + }, + { + "epoch": 1.01, + "learning_rate": 7.104355254942632e-06, + "loss": 3.8403, + "step": 2619904 + }, + { + "epoch": 1.01, + "learning_rate": 7.095985686235844e-06, + "loss": 3.8658, + "step": 2620416 + }, + { + "epoch": 1.01, + "learning_rate": 7.087599738725323e-06, + "loss": 3.8615, + "step": 2620928 + }, + { + "epoch": 1.01, + "learning_rate": 7.079213791214803e-06, + "loss": 3.8447, + "step": 2621440 + }, + { + "epoch": 1.01, + "learning_rate": 7.070827843704284e-06, + "loss": 3.8476, + "step": 2621952 + }, + { + "epoch": 1.01, + "learning_rate": 7.062441896193762e-06, + "loss": 3.8562, + "step": 2622464 + }, + { + "epoch": 1.01, + "learning_rate": 7.054055948683243e-06, + "loss": 3.8374, + "step": 2622976 + }, + { + "epoch": 1.01, + "learning_rate": 7.045670001172723e-06, + "loss": 3.8578, + "step": 2623488 + }, + { + "epoch": 1.01, + "learning_rate": 7.037284053662202e-06, + "loss": 3.835, + "step": 2624000 + }, + { + "epoch": 1.01, + "learning_rate": 7.028898106151683e-06, + "loss": 3.8641, + "step": 2624512 + }, + { + "epoch": 1.01, + "learning_rate": 7.020512158641163e-06, + "loss": 3.8601, + "step": 2625024 + }, + { + "epoch": 1.01, + "learning_rate": 7.012126211130643e-06, + "loss": 3.8431, + "step": 2625536 + }, + { + "epoch": 1.01, + "learning_rate": 7.003740263620122e-06, + "loss": 3.8397, + "step": 2626048 + }, + { + "epoch": 1.01, + "learning_rate": 6.9953543161096025e-06, + "loss": 3.8605, + "step": 2626560 + }, + { + "epoch": 1.01, + "learning_rate": 6.986984747402814e-06, + "loss": 3.828, + "step": 2627072 + }, + { + "epoch": 1.01, + "learning_rate": 6.9785987998922924e-06, + "loss": 3.8453, + "step": 2627584 + }, + { + "epoch": 1.01, + "learning_rate": 6.970212852381773e-06, + "loss": 3.8589, + "step": 2628096 + }, + { + "epoch": 1.01, + "learning_rate": 6.961826904871253e-06, + "loss": 3.8547, + "step": 2628608 + }, + { + "epoch": 1.01, + "learning_rate": 6.953440957360732e-06, + "loss": 3.8473, + "step": 2629120 + }, + { + "epoch": 1.01, + "learning_rate": 6.945055009850212e-06, + "loss": 3.8345, + "step": 2629632 + }, + { + "epoch": 1.01, + "learning_rate": 6.936685441143424e-06, + "loss": 3.8441, + "step": 2630144 + }, + { + "epoch": 1.01, + "learning_rate": 6.928299493632905e-06, + "loss": 3.8443, + "step": 2630656 + }, + { + "epoch": 1.01, + "learning_rate": 6.919913546122384e-06, + "loss": 3.8548, + "step": 2631168 + }, + { + "epoch": 1.01, + "learning_rate": 6.911527598611864e-06, + "loss": 3.8444, + "step": 2631680 + }, + { + "epoch": 1.01, + "learning_rate": 6.9031416511013446e-06, + "loss": 3.8542, + "step": 2632192 + }, + { + "epoch": 1.01, + "learning_rate": 6.894755703590823e-06, + "loss": 3.8546, + "step": 2632704 + }, + { + "epoch": 1.01, + "learning_rate": 6.8863861348840345e-06, + "loss": 3.8575, + "step": 2633216 + }, + { + "epoch": 1.01, + "learning_rate": 6.878016566177246e-06, + "loss": 3.8461, + "step": 2633728 + }, + { + "epoch": 1.01, + "learning_rate": 6.869630618666727e-06, + "loss": 3.8492, + "step": 2634240 + }, + { + "epoch": 1.01, + "learning_rate": 6.861244671156206e-06, + "loss": 3.857, + "step": 2634752 + }, + { + "epoch": 1.01, + "learning_rate": 6.852858723645686e-06, + "loss": 3.8499, + "step": 2635264 + }, + { + "epoch": 1.01, + "learning_rate": 6.844489154938898e-06, + "loss": 3.8517, + "step": 2635776 + }, + { + "epoch": 1.01, + "learning_rate": 6.8361032074283776e-06, + "loss": 3.8498, + "step": 2636288 + }, + { + "epoch": 1.01, + "learning_rate": 6.827717259917857e-06, + "loss": 3.8511, + "step": 2636800 + }, + { + "epoch": 1.01, + "learning_rate": 6.8193313124073375e-06, + "loss": 3.8521, + "step": 2637312 + }, + { + "epoch": 1.01, + "learning_rate": 6.810945364896817e-06, + "loss": 3.8368, + "step": 2637824 + }, + { + "epoch": 1.01, + "learning_rate": 6.8025594173862965e-06, + "loss": 3.8485, + "step": 2638336 + }, + { + "epoch": 1.01, + "learning_rate": 6.7941734698757765e-06, + "loss": 3.8458, + "step": 2638848 + }, + { + "epoch": 1.01, + "learning_rate": 6.785787522365257e-06, + "loss": 3.8544, + "step": 2639360 + }, + { + "epoch": 1.01, + "learning_rate": 6.777401574854737e-06, + "loss": 3.8523, + "step": 2639872 + }, + { + "epoch": 1.01, + "learning_rate": 6.769015627344216e-06, + "loss": 3.8466, + "step": 2640384 + }, + { + "epoch": 1.02, + "learning_rate": 6.760646058637428e-06, + "loss": 3.8417, + "step": 2640896 + }, + { + "epoch": 1.02, + "learning_rate": 6.752260111126908e-06, + "loss": 3.8364, + "step": 2641408 + }, + { + "epoch": 1.02, + "learning_rate": 6.74389054242012e-06, + "loss": 3.8431, + "step": 2641920 + }, + { + "epoch": 1.02, + "learning_rate": 6.735504594909599e-06, + "loss": 3.8492, + "step": 2642432 + }, + { + "epoch": 1.02, + "learning_rate": 6.7271186473990795e-06, + "loss": 3.8559, + "step": 2642944 + }, + { + "epoch": 1.02, + "learning_rate": 6.7187326998885594e-06, + "loss": 3.857, + "step": 2643456 + }, + { + "epoch": 1.02, + "learning_rate": 6.7103467523780386e-06, + "loss": 3.8511, + "step": 2643968 + }, + { + "epoch": 1.02, + "learning_rate": 6.7019608048675185e-06, + "loss": 3.8454, + "step": 2644480 + }, + { + "epoch": 1.02, + "learning_rate": 6.693574857356999e-06, + "loss": 3.8447, + "step": 2644992 + }, + { + "epoch": 1.02, + "learning_rate": 6.6851889098464776e-06, + "loss": 3.8511, + "step": 2645504 + }, + { + "epoch": 1.02, + "learning_rate": 6.676802962335958e-06, + "loss": 3.8582, + "step": 2646016 + }, + { + "epoch": 1.02, + "learning_rate": 6.668417014825438e-06, + "loss": 3.8555, + "step": 2646528 + }, + { + "epoch": 1.02, + "learning_rate": 6.660031067314919e-06, + "loss": 3.8477, + "step": 2647040 + }, + { + "epoch": 1.02, + "learning_rate": 6.651645119804397e-06, + "loss": 3.8413, + "step": 2647552 + }, + { + "epoch": 1.02, + "learning_rate": 6.643275551097609e-06, + "loss": 3.8605, + "step": 2648064 + }, + { + "epoch": 1.02, + "learning_rate": 6.63488960358709e-06, + "loss": 3.8391, + "step": 2648576 + }, + { + "epoch": 1.02, + "learning_rate": 6.626503656076569e-06, + "loss": 3.8485, + "step": 2649088 + }, + { + "epoch": 1.02, + "learning_rate": 6.618117708566049e-06, + "loss": 3.8467, + "step": 2649600 + }, + { + "epoch": 1.02, + "learning_rate": 6.609731761055529e-06, + "loss": 3.8503, + "step": 2650112 + }, + { + "epoch": 1.02, + "learning_rate": 6.60134581354501e-06, + "loss": 3.8406, + "step": 2650624 + }, + { + "epoch": 1.02, + "learning_rate": 6.592959866034489e-06, + "loss": 3.8496, + "step": 2651136 + }, + { + "epoch": 1.02, + "learning_rate": 6.5845902973277e-06, + "loss": 3.8393, + "step": 2651648 + }, + { + "epoch": 1.02, + "learning_rate": 6.576220728620912e-06, + "loss": 3.8464, + "step": 2652160 + }, + { + "epoch": 1.02, + "learning_rate": 6.567834781110392e-06, + "loss": 3.8518, + "step": 2652672 + }, + { + "epoch": 1.02, + "learning_rate": 6.559448833599871e-06, + "loss": 3.84, + "step": 2653184 + }, + { + "epoch": 1.02, + "learning_rate": 6.551062886089351e-06, + "loss": 3.8617, + "step": 2653696 + }, + { + "epoch": 1.02, + "learning_rate": 6.542676938578832e-06, + "loss": 3.8638, + "step": 2654208 + }, + { + "epoch": 1.02, + "learning_rate": 6.534290991068311e-06, + "loss": 3.8557, + "step": 2654720 + }, + { + "epoch": 1.02, + "learning_rate": 6.525905043557791e-06, + "loss": 3.8467, + "step": 2655232 + }, + { + "epoch": 1.02, + "learning_rate": 6.5175354748510026e-06, + "loss": 3.8424, + "step": 2655744 + }, + { + "epoch": 1.02, + "learning_rate": 6.5091495273404825e-06, + "loss": 3.8472, + "step": 2656256 + }, + { + "epoch": 1.02, + "learning_rate": 6.500763579829962e-06, + "loss": 3.8441, + "step": 2656768 + }, + { + "epoch": 1.02, + "learning_rate": 6.4923776323194424e-06, + "loss": 3.8564, + "step": 2657280 + }, + { + "epoch": 1.02, + "learning_rate": 6.483991684808922e-06, + "loss": 3.8369, + "step": 2657792 + }, + { + "epoch": 1.02, + "learning_rate": 6.4756057372984015e-06, + "loss": 3.8532, + "step": 2658304 + }, + { + "epoch": 1.02, + "learning_rate": 6.4672197897878815e-06, + "loss": 3.8543, + "step": 2658816 + }, + { + "epoch": 1.02, + "learning_rate": 6.458833842277362e-06, + "loss": 3.8513, + "step": 2659328 + }, + { + "epoch": 1.02, + "learning_rate": 6.450464273570574e-06, + "loss": 3.8423, + "step": 2659840 + }, + { + "epoch": 1.02, + "learning_rate": 6.442094704863784e-06, + "loss": 3.8461, + "step": 2660352 + }, + { + "epoch": 1.02, + "learning_rate": 6.433708757353265e-06, + "loss": 3.8503, + "step": 2660864 + }, + { + "epoch": 1.02, + "learning_rate": 6.425322809842745e-06, + "loss": 3.8544, + "step": 2661376 + }, + { + "epoch": 1.02, + "learning_rate": 6.416936862332224e-06, + "loss": 3.8549, + "step": 2661888 + }, + { + "epoch": 1.02, + "learning_rate": 6.408567293625435e-06, + "loss": 3.8618, + "step": 2662400 + }, + { + "epoch": 1.02, + "learning_rate": 6.400181346114915e-06, + "loss": 3.8412, + "step": 2662912 + }, + { + "epoch": 1.02, + "learning_rate": 6.391795398604396e-06, + "loss": 3.8525, + "step": 2663424 + }, + { + "epoch": 1.02, + "learning_rate": 6.383409451093874e-06, + "loss": 3.8572, + "step": 2663936 + }, + { + "epoch": 1.02, + "learning_rate": 6.375023503583355e-06, + "loss": 3.8483, + "step": 2664448 + }, + { + "epoch": 1.02, + "learning_rate": 6.366637556072835e-06, + "loss": 3.8501, + "step": 2664960 + }, + { + "epoch": 1.02, + "learning_rate": 6.358251608562314e-06, + "loss": 3.8593, + "step": 2665472 + }, + { + "epoch": 1.02, + "learning_rate": 6.349865661051794e-06, + "loss": 3.8512, + "step": 2665984 + }, + { + "epoch": 1.02, + "learning_rate": 6.341479713541275e-06, + "loss": 3.8665, + "step": 2666496 + }, + { + "epoch": 1.02, + "learning_rate": 6.333093766030754e-06, + "loss": 3.8483, + "step": 2667008 + }, + { + "epoch": 1.02, + "learning_rate": 6.324707818520234e-06, + "loss": 3.8593, + "step": 2667520 + }, + { + "epoch": 1.02, + "learning_rate": 6.316321871009714e-06, + "loss": 3.8523, + "step": 2668032 + }, + { + "epoch": 1.02, + "learning_rate": 6.307952302302926e-06, + "loss": 3.8479, + "step": 2668544 + }, + { + "epoch": 1.02, + "learning_rate": 6.299566354792405e-06, + "loss": 3.8499, + "step": 2669056 + }, + { + "epoch": 1.02, + "learning_rate": 6.2911804072818856e-06, + "loss": 3.8466, + "step": 2669568 + }, + { + "epoch": 1.02, + "learning_rate": 6.282810838575097e-06, + "loss": 3.8451, + "step": 2670080 + }, + { + "epoch": 1.02, + "learning_rate": 6.274424891064577e-06, + "loss": 3.8491, + "step": 2670592 + }, + { + "epoch": 1.02, + "learning_rate": 6.266038943554056e-06, + "loss": 3.8455, + "step": 2671104 + }, + { + "epoch": 1.03, + "eval_loss": 3.981106996536255, + "eval_runtime": 289.8883, + "eval_samples_per_second": 1316.338, + "eval_steps_per_second": 41.137, + "step": 2671185 + }, + { + "epoch": 0.0, + "learning_rate": 6.257652996043536e-06, + "loss": 3.851, + "step": 2671616 + }, + { + "epoch": 0.0, + "learning_rate": 6.249267048533016e-06, + "loss": 3.8462, + "step": 2672128 + }, + { + "epoch": 0.0, + "learning_rate": 6.240881101022496e-06, + "loss": 3.8597, + "step": 2672640 + }, + { + "epoch": 0.0, + "learning_rate": 6.232495153511976e-06, + "loss": 3.8486, + "step": 2673152 + }, + { + "epoch": 0.0, + "learning_rate": 6.224109206001456e-06, + "loss": 3.8702, + "step": 2673664 + }, + { + "epoch": 0.0, + "learning_rate": 6.215723258490936e-06, + "loss": 3.843, + "step": 2674176 + }, + { + "epoch": 0.0, + "learning_rate": 6.207337310980416e-06, + "loss": 3.8509, + "step": 2674688 + }, + { + "epoch": 0.0, + "learning_rate": 6.198951363469896e-06, + "loss": 3.8405, + "step": 2675200 + }, + { + "epoch": 0.0, + "learning_rate": 6.190565415959376e-06, + "loss": 3.8546, + "step": 2675712 + }, + { + "epoch": 0.0, + "learning_rate": 6.182179468448856e-06, + "loss": 3.8515, + "step": 2676224 + }, + { + "epoch": 0.0, + "learning_rate": 6.173793520938336e-06, + "loss": 3.8469, + "step": 2676736 + }, + { + "epoch": 0.0, + "learning_rate": 6.165407573427815e-06, + "loss": 3.8592, + "step": 2677248 + }, + { + "epoch": 0.0, + "learning_rate": 6.1570380047210265e-06, + "loss": 3.8409, + "step": 2677760 + }, + { + "epoch": 0.0, + "learning_rate": 6.1486520572105065e-06, + "loss": 3.8549, + "step": 2678272 + }, + { + "epoch": 0.0, + "learning_rate": 6.1402661096999864e-06, + "loss": 3.8412, + "step": 2678784 + }, + { + "epoch": 0.0, + "learning_rate": 6.131880162189466e-06, + "loss": 3.8422, + "step": 2679296 + }, + { + "epoch": 0.0, + "learning_rate": 6.123494214678946e-06, + "loss": 3.8448, + "step": 2679808 + }, + { + "epoch": 0.0, + "learning_rate": 6.115108267168426e-06, + "loss": 3.8495, + "step": 2680320 + }, + { + "epoch": 0.0, + "learning_rate": 6.106722319657906e-06, + "loss": 3.8511, + "step": 2680832 + }, + { + "epoch": 0.0, + "learning_rate": 6.098336372147386e-06, + "loss": 3.8671, + "step": 2681344 + }, + { + "epoch": 0.0, + "learning_rate": 6.089950424636866e-06, + "loss": 3.8458, + "step": 2681856 + }, + { + "epoch": 0.0, + "learning_rate": 6.081597234733809e-06, + "loss": 3.8557, + "step": 2682368 + }, + { + "epoch": 0.0, + "learning_rate": 6.073211287223289e-06, + "loss": 3.8538, + "step": 2682880 + }, + { + "epoch": 0.0, + "learning_rate": 6.0648253397127685e-06, + "loss": 3.8574, + "step": 2683392 + }, + { + "epoch": 0.0, + "learning_rate": 6.0564393922022485e-06, + "loss": 3.8473, + "step": 2683904 + }, + { + "epoch": 0.0, + "learning_rate": 6.048053444691728e-06, + "loss": 3.8526, + "step": 2684416 + }, + { + "epoch": 0.0, + "learning_rate": 6.039667497181208e-06, + "loss": 3.8451, + "step": 2684928 + }, + { + "epoch": 0.0, + "learning_rate": 6.0312815496706875e-06, + "loss": 3.8466, + "step": 2685440 + }, + { + "epoch": 0.0, + "learning_rate": 6.022895602160168e-06, + "loss": 3.8394, + "step": 2685952 + }, + { + "epoch": 0.01, + "learning_rate": 6.0145096546496474e-06, + "loss": 3.8515, + "step": 2686464 + }, + { + "epoch": 0.01, + "learning_rate": 6.006123707139127e-06, + "loss": 3.857, + "step": 2686976 + }, + { + "epoch": 0.01, + "learning_rate": 5.997754138432339e-06, + "loss": 3.8496, + "step": 2687488 + }, + { + "epoch": 0.01, + "learning_rate": 5.989368190921819e-06, + "loss": 3.8644, + "step": 2688000 + }, + { + "epoch": 0.01, + "learning_rate": 5.980982243411299e-06, + "loss": 3.8494, + "step": 2688512 + }, + { + "epoch": 0.01, + "learning_rate": 5.97261267470451e-06, + "loss": 3.8513, + "step": 2689024 + }, + { + "epoch": 0.01, + "learning_rate": 5.9642267271939905e-06, + "loss": 3.8535, + "step": 2689536 + }, + { + "epoch": 0.01, + "learning_rate": 5.955857158487201e-06, + "loss": 3.8506, + "step": 2690048 + }, + { + "epoch": 0.01, + "learning_rate": 5.947471210976681e-06, + "loss": 3.8413, + "step": 2690560 + }, + { + "epoch": 0.01, + "learning_rate": 5.939085263466161e-06, + "loss": 3.8345, + "step": 2691072 + }, + { + "epoch": 0.01, + "learning_rate": 5.930699315955641e-06, + "loss": 3.8432, + "step": 2691584 + }, + { + "epoch": 0.01, + "learning_rate": 5.922313368445121e-06, + "loss": 3.8492, + "step": 2692096 + }, + { + "epoch": 0.01, + "learning_rate": 5.913927420934601e-06, + "loss": 3.856, + "step": 2692608 + }, + { + "epoch": 0.01, + "learning_rate": 5.905541473424081e-06, + "loss": 3.8473, + "step": 2693120 + }, + { + "epoch": 0.01, + "learning_rate": 5.897155525913561e-06, + "loss": 3.8422, + "step": 2693632 + }, + { + "epoch": 0.01, + "learning_rate": 5.888769578403041e-06, + "loss": 3.8558, + "step": 2694144 + }, + { + "epoch": 0.01, + "learning_rate": 5.880383630892521e-06, + "loss": 3.8451, + "step": 2694656 + }, + { + "epoch": 0.01, + "learning_rate": 5.871997683382e-06, + "loss": 3.8455, + "step": 2695168 + }, + { + "epoch": 0.01, + "learning_rate": 5.863611735871481e-06, + "loss": 3.8387, + "step": 2695680 + }, + { + "epoch": 0.01, + "learning_rate": 5.85522578836096e-06, + "loss": 3.8299, + "step": 2696192 + }, + { + "epoch": 0.01, + "learning_rate": 5.846839840850441e-06, + "loss": 3.8659, + "step": 2696704 + }, + { + "epoch": 0.01, + "learning_rate": 5.83845389333992e-06, + "loss": 3.8549, + "step": 2697216 + }, + { + "epoch": 0.01, + "learning_rate": 5.8300679458294e-06, + "loss": 3.8471, + "step": 2697728 + }, + { + "epoch": 0.01, + "learning_rate": 5.8216983771226114e-06, + "loss": 3.8479, + "step": 2698240 + }, + { + "epoch": 0.01, + "learning_rate": 5.8133124296120905e-06, + "loss": 3.8502, + "step": 2698752 + }, + { + "epoch": 0.01, + "learning_rate": 5.804926482101571e-06, + "loss": 3.8335, + "step": 2699264 + }, + { + "epoch": 0.01, + "learning_rate": 5.796556913394782e-06, + "loss": 3.8556, + "step": 2699776 + }, + { + "epoch": 0.01, + "learning_rate": 5.788170965884263e-06, + "loss": 3.8389, + "step": 2700288 + }, + { + "epoch": 0.01, + "learning_rate": 5.779785018373742e-06, + "loss": 3.854, + "step": 2700800 + }, + { + "epoch": 0.01, + "learning_rate": 5.771399070863223e-06, + "loss": 3.8606, + "step": 2701312 + }, + { + "epoch": 0.01, + "learning_rate": 5.763029502156434e-06, + "loss": 3.8422, + "step": 2701824 + }, + { + "epoch": 0.01, + "learning_rate": 5.754643554645914e-06, + "loss": 3.8347, + "step": 2702336 + }, + { + "epoch": 0.01, + "learning_rate": 5.7462576071353936e-06, + "loss": 3.8608, + "step": 2702848 + }, + { + "epoch": 0.01, + "learning_rate": 5.737871659624873e-06, + "loss": 3.8255, + "step": 2703360 + }, + { + "epoch": 0.01, + "learning_rate": 5.7294857121143535e-06, + "loss": 3.8403, + "step": 2703872 + }, + { + "epoch": 0.01, + "learning_rate": 5.721099764603833e-06, + "loss": 3.8569, + "step": 2704384 + }, + { + "epoch": 0.01, + "learning_rate": 5.712730195897044e-06, + "loss": 3.8535, + "step": 2704896 + }, + { + "epoch": 0.01, + "learning_rate": 5.704344248386524e-06, + "loss": 3.847, + "step": 2705408 + }, + { + "epoch": 0.01, + "learning_rate": 5.695958300876004e-06, + "loss": 3.8363, + "step": 2705920 + }, + { + "epoch": 0.01, + "learning_rate": 5.687572353365484e-06, + "loss": 3.8353, + "step": 2706432 + }, + { + "epoch": 0.01, + "learning_rate": 5.679186405854964e-06, + "loss": 3.8384, + "step": 2706944 + }, + { + "epoch": 0.01, + "learning_rate": 5.670800458344444e-06, + "loss": 3.8556, + "step": 2707456 + }, + { + "epoch": 0.01, + "learning_rate": 5.662430889637655e-06, + "loss": 3.8384, + "step": 2707968 + }, + { + "epoch": 0.01, + "learning_rate": 5.654044942127136e-06, + "loss": 3.8519, + "step": 2708480 + }, + { + "epoch": 0.01, + "learning_rate": 5.645658994616615e-06, + "loss": 3.851, + "step": 2708992 + }, + { + "epoch": 0.01, + "learning_rate": 5.6372730471060955e-06, + "loss": 3.8584, + "step": 2709504 + }, + { + "epoch": 0.01, + "learning_rate": 5.628887099595575e-06, + "loss": 3.8435, + "step": 2710016 + }, + { + "epoch": 0.01, + "learning_rate": 5.620517530888786e-06, + "loss": 3.8434, + "step": 2710528 + }, + { + "epoch": 0.01, + "learning_rate": 5.612131583378266e-06, + "loss": 3.8548, + "step": 2711040 + }, + { + "epoch": 0.01, + "learning_rate": 5.603745635867746e-06, + "loss": 3.8445, + "step": 2711552 + }, + { + "epoch": 0.01, + "learning_rate": 5.595359688357226e-06, + "loss": 3.8513, + "step": 2712064 + }, + { + "epoch": 0.01, + "learning_rate": 5.586973740846706e-06, + "loss": 3.8471, + "step": 2712576 + }, + { + "epoch": 0.01, + "learning_rate": 5.578587793336186e-06, + "loss": 3.8455, + "step": 2713088 + }, + { + "epoch": 0.01, + "learning_rate": 5.570218224629397e-06, + "loss": 3.8529, + "step": 2713600 + }, + { + "epoch": 0.01, + "learning_rate": 5.561832277118878e-06, + "loss": 3.831, + "step": 2714112 + }, + { + "epoch": 0.01, + "learning_rate": 5.553462708412088e-06, + "loss": 3.846, + "step": 2714624 + }, + { + "epoch": 0.01, + "learning_rate": 5.545076760901568e-06, + "loss": 3.8431, + "step": 2715136 + }, + { + "epoch": 0.01, + "learning_rate": 5.536690813391048e-06, + "loss": 3.8529, + "step": 2715648 + }, + { + "epoch": 0.01, + "learning_rate": 5.5283048658805274e-06, + "loss": 3.847, + "step": 2716160 + }, + { + "epoch": 0.01, + "learning_rate": 5.519918918370008e-06, + "loss": 3.8458, + "step": 2716672 + }, + { + "epoch": 0.02, + "learning_rate": 5.511532970859487e-06, + "loss": 3.8373, + "step": 2717184 + }, + { + "epoch": 0.02, + "learning_rate": 5.503147023348967e-06, + "loss": 3.8406, + "step": 2717696 + }, + { + "epoch": 0.02, + "learning_rate": 5.494761075838448e-06, + "loss": 3.8379, + "step": 2718208 + }, + { + "epoch": 0.02, + "learning_rate": 5.486375128327927e-06, + "loss": 3.8483, + "step": 2718720 + }, + { + "epoch": 0.02, + "learning_rate": 5.477989180817408e-06, + "loss": 3.8574, + "step": 2719232 + }, + { + "epoch": 0.02, + "learning_rate": 5.469603233306887e-06, + "loss": 3.8543, + "step": 2719744 + }, + { + "epoch": 0.02, + "learning_rate": 5.461217285796367e-06, + "loss": 3.8498, + "step": 2720256 + }, + { + "epoch": 0.02, + "learning_rate": 5.4528640958933095e-06, + "loss": 3.8375, + "step": 2720768 + }, + { + "epoch": 0.02, + "learning_rate": 5.44447814838279e-06, + "loss": 3.8442, + "step": 2721280 + }, + { + "epoch": 0.02, + "learning_rate": 5.4360922008722695e-06, + "loss": 3.847, + "step": 2721792 + }, + { + "epoch": 0.02, + "learning_rate": 5.427706253361749e-06, + "loss": 3.8559, + "step": 2722304 + }, + { + "epoch": 0.02, + "learning_rate": 5.419336684654961e-06, + "loss": 3.8539, + "step": 2722816 + }, + { + "epoch": 0.02, + "learning_rate": 5.410950737144441e-06, + "loss": 3.845, + "step": 2723328 + }, + { + "epoch": 0.02, + "learning_rate": 5.402564789633921e-06, + "loss": 3.8396, + "step": 2723840 + }, + { + "epoch": 0.02, + "learning_rate": 5.394178842123401e-06, + "loss": 3.8606, + "step": 2724352 + }, + { + "epoch": 0.02, + "learning_rate": 5.385792894612881e-06, + "loss": 3.8354, + "step": 2724864 + }, + { + "epoch": 0.02, + "learning_rate": 5.377423325906092e-06, + "loss": 3.8484, + "step": 2725376 + }, + { + "epoch": 0.02, + "learning_rate": 5.3690373783955725e-06, + "loss": 3.839, + "step": 2725888 + }, + { + "epoch": 0.02, + "learning_rate": 5.3606514308850516e-06, + "loss": 3.8508, + "step": 2726400 + }, + { + "epoch": 0.02, + "learning_rate": 5.3522654833745315e-06, + "loss": 3.8378, + "step": 2726912 + }, + { + "epoch": 0.02, + "learning_rate": 5.3438795358640115e-06, + "loss": 3.8473, + "step": 2727424 + }, + { + "epoch": 0.02, + "learning_rate": 5.3354935883534914e-06, + "loss": 3.8418, + "step": 2727936 + }, + { + "epoch": 0.02, + "learning_rate": 5.327107640842971e-06, + "loss": 3.8395, + "step": 2728448 + }, + { + "epoch": 0.02, + "learning_rate": 5.318721693332451e-06, + "loss": 3.8497, + "step": 2728960 + }, + { + "epoch": 0.02, + "learning_rate": 5.310335745821931e-06, + "loss": 3.8377, + "step": 2729472 + }, + { + "epoch": 0.02, + "learning_rate": 5.301966177115143e-06, + "loss": 3.8584, + "step": 2729984 + }, + { + "epoch": 0.02, + "learning_rate": 5.293580229604622e-06, + "loss": 3.8631, + "step": 2730496 + }, + { + "epoch": 0.02, + "learning_rate": 5.285194282094103e-06, + "loss": 3.854, + "step": 2731008 + }, + { + "epoch": 0.02, + "learning_rate": 5.276808334583582e-06, + "loss": 3.8407, + "step": 2731520 + }, + { + "epoch": 0.02, + "learning_rate": 5.268422387073063e-06, + "loss": 3.8417, + "step": 2732032 + }, + { + "epoch": 0.02, + "learning_rate": 5.2600528183662736e-06, + "loss": 3.842, + "step": 2732544 + }, + { + "epoch": 0.02, + "learning_rate": 5.2516668708557535e-06, + "loss": 3.8438, + "step": 2733056 + }, + { + "epoch": 0.02, + "learning_rate": 5.2432809233452335e-06, + "loss": 3.8538, + "step": 2733568 + }, + { + "epoch": 0.02, + "learning_rate": 5.2348949758347126e-06, + "loss": 3.8332, + "step": 2734080 + }, + { + "epoch": 0.02, + "learning_rate": 5.226509028324193e-06, + "loss": 3.8539, + "step": 2734592 + }, + { + "epoch": 0.02, + "learning_rate": 5.218139459617404e-06, + "loss": 3.8506, + "step": 2735104 + }, + { + "epoch": 0.02, + "learning_rate": 5.209753512106884e-06, + "loss": 3.8486, + "step": 2735616 + }, + { + "epoch": 0.02, + "learning_rate": 5.201367564596364e-06, + "loss": 3.8434, + "step": 2736128 + }, + { + "epoch": 0.02, + "learning_rate": 5.192981617085845e-06, + "loss": 3.8469, + "step": 2736640 + }, + { + "epoch": 0.02, + "learning_rate": 5.184612048379056e-06, + "loss": 3.8456, + "step": 2737152 + }, + { + "epoch": 0.02, + "learning_rate": 5.176226100868536e-06, + "loss": 3.8519, + "step": 2737664 + }, + { + "epoch": 0.02, + "learning_rate": 5.167840153358016e-06, + "loss": 3.8519, + "step": 2738176 + }, + { + "epoch": 0.02, + "learning_rate": 5.159454205847495e-06, + "loss": 3.8591, + "step": 2738688 + }, + { + "epoch": 0.02, + "learning_rate": 5.1510682583369755e-06, + "loss": 3.8401, + "step": 2739200 + }, + { + "epoch": 0.02, + "learning_rate": 5.142682310826455e-06, + "loss": 3.8513, + "step": 2739712 + }, + { + "epoch": 0.02, + "learning_rate": 5.134312742119666e-06, + "loss": 3.852, + "step": 2740224 + }, + { + "epoch": 0.02, + "learning_rate": 5.125926794609146e-06, + "loss": 3.8455, + "step": 2740736 + }, + { + "epoch": 0.02, + "learning_rate": 5.117540847098626e-06, + "loss": 3.8483, + "step": 2741248 + }, + { + "epoch": 0.02, + "learning_rate": 5.109154899588106e-06, + "loss": 3.8594, + "step": 2741760 + }, + { + "epoch": 0.02, + "learning_rate": 5.100785330881318e-06, + "loss": 3.8493, + "step": 2742272 + }, + { + "epoch": 0.02, + "learning_rate": 5.092399383370798e-06, + "loss": 3.8632, + "step": 2742784 + }, + { + "epoch": 0.02, + "learning_rate": 5.084029814664009e-06, + "loss": 3.843, + "step": 2743296 + }, + { + "epoch": 0.02, + "learning_rate": 5.0756438671534884e-06, + "loss": 3.8565, + "step": 2743808 + }, + { + "epoch": 0.02, + "learning_rate": 5.067257919642968e-06, + "loss": 3.8493, + "step": 2744320 + }, + { + "epoch": 0.02, + "learning_rate": 5.058871972132448e-06, + "loss": 3.8468, + "step": 2744832 + }, + { + "epoch": 0.02, + "learning_rate": 5.050486024621928e-06, + "loss": 3.8423, + "step": 2745344 + }, + { + "epoch": 0.02, + "learning_rate": 5.042100077111408e-06, + "loss": 3.8478, + "step": 2745856 + }, + { + "epoch": 0.02, + "learning_rate": 5.033714129600888e-06, + "loss": 3.8408, + "step": 2746368 + }, + { + "epoch": 0.02, + "learning_rate": 5.025328182090368e-06, + "loss": 3.8432, + "step": 2746880 + }, + { + "epoch": 0.02, + "learning_rate": 5.016942234579848e-06, + "loss": 3.846, + "step": 2747392 + }, + { + "epoch": 0.03, + "eval_loss": 3.980264902114868, + "eval_runtime": 297.5705, + "eval_samples_per_second": 1282.355, + "eval_steps_per_second": 40.075, + "step": 2747505 + }, + { + "epoch": 1.0, + "learning_rate": 5.0085890446767915e-06, + "loss": 3.8486, + "step": 2747904 + }, + { + "epoch": 1.0, + "learning_rate": 5.0002030971662706e-06, + "loss": 3.843, + "step": 2748416 + }, + { + "epoch": 1.0, + "learning_rate": 4.9918171496557505e-06, + "loss": 3.8556, + "step": 2748928 + }, + { + "epoch": 1.0, + "learning_rate": 4.9834312021452305e-06, + "loss": 3.8481, + "step": 2749440 + }, + { + "epoch": 1.0, + "learning_rate": 4.9750452546347104e-06, + "loss": 3.8667, + "step": 2749952 + }, + { + "epoch": 1.0, + "learning_rate": 4.96665930712419e-06, + "loss": 3.8473, + "step": 2750464 + }, + { + "epoch": 1.0, + "learning_rate": 4.95827335961367e-06, + "loss": 3.8489, + "step": 2750976 + }, + { + "epoch": 1.0, + "learning_rate": 4.9498874121031495e-06, + "loss": 3.8351, + "step": 2751488 + }, + { + "epoch": 1.0, + "learning_rate": 4.941517843396361e-06, + "loss": 3.8499, + "step": 2752000 + }, + { + "epoch": 1.0, + "learning_rate": 4.933131895885841e-06, + "loss": 3.8494, + "step": 2752512 + }, + { + "epoch": 1.0, + "learning_rate": 4.924745948375321e-06, + "loss": 3.8467, + "step": 2753024 + }, + { + "epoch": 1.0, + "learning_rate": 4.916360000864801e-06, + "loss": 3.8518, + "step": 2753536 + }, + { + "epoch": 1.0, + "learning_rate": 4.907990432158013e-06, + "loss": 3.8421, + "step": 2754048 + }, + { + "epoch": 1.0, + "learning_rate": 4.8996044846474926e-06, + "loss": 3.8528, + "step": 2754560 + }, + { + "epoch": 1.0, + "learning_rate": 4.891218537136972e-06, + "loss": 3.8359, + "step": 2755072 + }, + { + "epoch": 1.0, + "learning_rate": 4.8828325896264525e-06, + "loss": 3.8389, + "step": 2755584 + }, + { + "epoch": 1.0, + "learning_rate": 4.8744466421159316e-06, + "loss": 3.8399, + "step": 2756096 + }, + { + "epoch": 1.0, + "learning_rate": 4.866060694605412e-06, + "loss": 3.8484, + "step": 2756608 + }, + { + "epoch": 1.0, + "learning_rate": 4.8576747470948915e-06, + "loss": 3.848, + "step": 2757120 + }, + { + "epoch": 1.0, + "learning_rate": 4.8492887995843714e-06, + "loss": 3.8639, + "step": 2757632 + }, + { + "epoch": 1.0, + "learning_rate": 4.840902852073851e-06, + "loss": 3.848, + "step": 2758144 + }, + { + "epoch": 1.0, + "learning_rate": 4.832516904563331e-06, + "loss": 3.8563, + "step": 2758656 + }, + { + "epoch": 1.0, + "learning_rate": 4.824130957052811e-06, + "loss": 3.8465, + "step": 2759168 + }, + { + "epoch": 1.0, + "learning_rate": 4.815745009542291e-06, + "loss": 3.8527, + "step": 2759680 + }, + { + "epoch": 1.0, + "learning_rate": 4.807375440835503e-06, + "loss": 3.8461, + "step": 2760192 + }, + { + "epoch": 1.0, + "learning_rate": 4.798989493324983e-06, + "loss": 3.8546, + "step": 2760704 + }, + { + "epoch": 1.0, + "learning_rate": 4.7906199246181945e-06, + "loss": 3.8399, + "step": 2761216 + }, + { + "epoch": 1.0, + "learning_rate": 4.782233977107674e-06, + "loss": 3.846, + "step": 2761728 + }, + { + "epoch": 1.0, + "learning_rate": 4.7738480295971536e-06, + "loss": 3.839, + "step": 2762240 + }, + { + "epoch": 1.0, + "learning_rate": 4.7654620820866335e-06, + "loss": 3.846, + "step": 2762752 + }, + { + "epoch": 1.01, + "learning_rate": 4.757092513379844e-06, + "loss": 3.852, + "step": 2763264 + }, + { + "epoch": 1.01, + "learning_rate": 4.748706565869325e-06, + "loss": 3.8497, + "step": 2763776 + }, + { + "epoch": 1.01, + "learning_rate": 4.740320618358804e-06, + "loss": 3.8614, + "step": 2764288 + }, + { + "epoch": 1.01, + "learning_rate": 4.731934670848285e-06, + "loss": 3.8464, + "step": 2764800 + }, + { + "epoch": 1.01, + "learning_rate": 4.723565102141496e-06, + "loss": 3.85, + "step": 2765312 + }, + { + "epoch": 1.01, + "learning_rate": 4.715179154630977e-06, + "loss": 3.8528, + "step": 2765824 + }, + { + "epoch": 1.01, + "learning_rate": 4.706809585924187e-06, + "loss": 3.8516, + "step": 2766336 + }, + { + "epoch": 1.01, + "learning_rate": 4.698423638413667e-06, + "loss": 3.8373, + "step": 2766848 + }, + { + "epoch": 1.01, + "learning_rate": 4.690037690903147e-06, + "loss": 3.8353, + "step": 2767360 + }, + { + "epoch": 1.01, + "learning_rate": 4.681651743392626e-06, + "loss": 3.8352, + "step": 2767872 + }, + { + "epoch": 1.01, + "learning_rate": 4.673265795882107e-06, + "loss": 3.847, + "step": 2768384 + }, + { + "epoch": 1.01, + "learning_rate": 4.664879848371586e-06, + "loss": 3.8565, + "step": 2768896 + }, + { + "epoch": 1.01, + "learning_rate": 4.656493900861067e-06, + "loss": 3.8461, + "step": 2769408 + }, + { + "epoch": 1.01, + "learning_rate": 4.648107953350546e-06, + "loss": 3.8372, + "step": 2769920 + }, + { + "epoch": 1.01, + "learning_rate": 4.639738384643758e-06, + "loss": 3.8476, + "step": 2770432 + }, + { + "epoch": 1.01, + "learning_rate": 4.631352437133238e-06, + "loss": 3.8448, + "step": 2770944 + }, + { + "epoch": 1.01, + "learning_rate": 4.6229828684264495e-06, + "loss": 3.8425, + "step": 2771456 + }, + { + "epoch": 1.01, + "learning_rate": 4.6145969209159294e-06, + "loss": 3.8371, + "step": 2771968 + }, + { + "epoch": 1.01, + "learning_rate": 4.6062109734054085e-06, + "loss": 3.8243, + "step": 2772480 + }, + { + "epoch": 1.01, + "learning_rate": 4.597825025894889e-06, + "loss": 3.8682, + "step": 2772992 + }, + { + "epoch": 1.01, + "learning_rate": 4.5894390783843684e-06, + "loss": 3.8512, + "step": 2773504 + }, + { + "epoch": 1.01, + "learning_rate": 4.581053130873849e-06, + "loss": 3.8486, + "step": 2774016 + }, + { + "epoch": 1.01, + "learning_rate": 4.57268356216706e-06, + "loss": 3.8462, + "step": 2774528 + }, + { + "epoch": 1.01, + "learning_rate": 4.56429761465654e-06, + "loss": 3.8441, + "step": 2775040 + }, + { + "epoch": 1.01, + "learning_rate": 4.55591166714602e-06, + "loss": 3.8352, + "step": 2775552 + }, + { + "epoch": 1.01, + "learning_rate": 4.5475257196355e-06, + "loss": 3.8481, + "step": 2776064 + }, + { + "epoch": 1.01, + "learning_rate": 4.53913977212498e-06, + "loss": 3.8394, + "step": 2776576 + }, + { + "epoch": 1.01, + "learning_rate": 4.53075382461446e-06, + "loss": 3.8464, + "step": 2777088 + }, + { + "epoch": 1.01, + "learning_rate": 4.52236787710394e-06, + "loss": 3.8624, + "step": 2777600 + }, + { + "epoch": 1.01, + "learning_rate": 4.51398192959342e-06, + "loss": 3.8385, + "step": 2778112 + }, + { + "epoch": 1.01, + "learning_rate": 4.505595982082899e-06, + "loss": 3.8345, + "step": 2778624 + }, + { + "epoch": 1.01, + "learning_rate": 4.49721003457238e-06, + "loss": 3.857, + "step": 2779136 + }, + { + "epoch": 1.01, + "learning_rate": 4.488824087061859e-06, + "loss": 3.8267, + "step": 2779648 + }, + { + "epoch": 1.01, + "learning_rate": 4.48045451835507e-06, + "loss": 3.8374, + "step": 2780160 + }, + { + "epoch": 1.01, + "learning_rate": 4.47206857084455e-06, + "loss": 3.8529, + "step": 2780672 + }, + { + "epoch": 1.01, + "learning_rate": 4.46368262333403e-06, + "loss": 3.8479, + "step": 2781184 + }, + { + "epoch": 1.01, + "learning_rate": 4.45529667582351e-06, + "loss": 3.8451, + "step": 2781696 + }, + { + "epoch": 1.01, + "learning_rate": 4.446910728312989e-06, + "loss": 3.8364, + "step": 2782208 + }, + { + "epoch": 1.01, + "learning_rate": 4.43852478080247e-06, + "loss": 3.8271, + "step": 2782720 + }, + { + "epoch": 1.01, + "learning_rate": 4.430138833291949e-06, + "loss": 3.8384, + "step": 2783232 + }, + { + "epoch": 1.01, + "learning_rate": 4.42175288578143e-06, + "loss": 3.8539, + "step": 2783744 + }, + { + "epoch": 1.01, + "learning_rate": 4.413383317074641e-06, + "loss": 3.8382, + "step": 2784256 + }, + { + "epoch": 1.01, + "learning_rate": 4.404997369564122e-06, + "loss": 3.8485, + "step": 2784768 + }, + { + "epoch": 1.01, + "learning_rate": 4.396611422053601e-06, + "loss": 3.8495, + "step": 2785280 + }, + { + "epoch": 1.01, + "learning_rate": 4.388225474543081e-06, + "loss": 3.8612, + "step": 2785792 + }, + { + "epoch": 1.01, + "learning_rate": 4.379872284640023e-06, + "loss": 3.8412, + "step": 2786304 + }, + { + "epoch": 1.01, + "learning_rate": 4.371486337129504e-06, + "loss": 3.8394, + "step": 2786816 + }, + { + "epoch": 1.01, + "learning_rate": 4.363100389618983e-06, + "loss": 3.8519, + "step": 2787328 + }, + { + "epoch": 1.01, + "learning_rate": 4.354714442108463e-06, + "loss": 3.8369, + "step": 2787840 + }, + { + "epoch": 1.01, + "learning_rate": 4.346328494597943e-06, + "loss": 3.8554, + "step": 2788352 + }, + { + "epoch": 1.01, + "learning_rate": 4.337942547087423e-06, + "loss": 3.8434, + "step": 2788864 + }, + { + "epoch": 1.01, + "learning_rate": 4.329556599576903e-06, + "loss": 3.8411, + "step": 2789376 + }, + { + "epoch": 1.01, + "learning_rate": 4.321170652066383e-06, + "loss": 3.8508, + "step": 2789888 + }, + { + "epoch": 1.01, + "learning_rate": 4.3128010833595945e-06, + "loss": 3.8323, + "step": 2790400 + }, + { + "epoch": 1.01, + "learning_rate": 4.3044151358490745e-06, + "loss": 3.8416, + "step": 2790912 + }, + { + "epoch": 1.01, + "learning_rate": 4.296029188338554e-06, + "loss": 3.8412, + "step": 2791424 + }, + { + "epoch": 1.01, + "learning_rate": 4.287643240828034e-06, + "loss": 3.8513, + "step": 2791936 + }, + { + "epoch": 1.01, + "learning_rate": 4.279273672121245e-06, + "loss": 3.8413, + "step": 2792448 + }, + { + "epoch": 1.01, + "learning_rate": 4.270904103414457e-06, + "loss": 3.8452, + "step": 2792960 + }, + { + "epoch": 1.02, + "learning_rate": 4.262518155903936e-06, + "loss": 3.8344, + "step": 2793472 + }, + { + "epoch": 1.02, + "learning_rate": 4.254132208393417e-06, + "loss": 3.8392, + "step": 2793984 + }, + { + "epoch": 1.02, + "learning_rate": 4.245746260882897e-06, + "loss": 3.8348, + "step": 2794496 + }, + { + "epoch": 1.02, + "learning_rate": 4.237360313372376e-06, + "loss": 3.843, + "step": 2795008 + }, + { + "epoch": 1.02, + "learning_rate": 4.228974365861857e-06, + "loss": 3.8557, + "step": 2795520 + }, + { + "epoch": 1.02, + "learning_rate": 4.220621175958799e-06, + "loss": 3.8495, + "step": 2796032 + }, + { + "epoch": 1.02, + "learning_rate": 4.212235228448279e-06, + "loss": 3.8482, + "step": 2796544 + }, + { + "epoch": 1.02, + "learning_rate": 4.203849280937758e-06, + "loss": 3.8386, + "step": 2797056 + }, + { + "epoch": 1.02, + "learning_rate": 4.195463333427239e-06, + "loss": 3.8428, + "step": 2797568 + }, + { + "epoch": 1.02, + "learning_rate": 4.187077385916718e-06, + "loss": 3.8433, + "step": 2798080 + }, + { + "epoch": 1.02, + "learning_rate": 4.178691438406199e-06, + "loss": 3.8535, + "step": 2798592 + }, + { + "epoch": 1.02, + "learning_rate": 4.170305490895678e-06, + "loss": 3.8502, + "step": 2799104 + }, + { + "epoch": 1.02, + "learning_rate": 4.161919543385158e-06, + "loss": 3.8436, + "step": 2799616 + }, + { + "epoch": 1.02, + "learning_rate": 4.153533595874638e-06, + "loss": 3.8373, + "step": 2800128 + }, + { + "epoch": 1.02, + "learning_rate": 4.145147648364118e-06, + "loss": 3.8597, + "step": 2800640 + }, + { + "epoch": 1.02, + "learning_rate": 4.1367780796573295e-06, + "loss": 3.833, + "step": 2801152 + }, + { + "epoch": 1.02, + "learning_rate": 4.1283921321468094e-06, + "loss": 3.8459, + "step": 2801664 + }, + { + "epoch": 1.02, + "learning_rate": 4.120006184636289e-06, + "loss": 3.8386, + "step": 2802176 + }, + { + "epoch": 1.02, + "learning_rate": 4.111620237125769e-06, + "loss": 3.8434, + "step": 2802688 + }, + { + "epoch": 1.02, + "learning_rate": 4.103250668418981e-06, + "loss": 3.8359, + "step": 2803200 + }, + { + "epoch": 1.02, + "learning_rate": 4.094881099712192e-06, + "loss": 3.8487, + "step": 2803712 + }, + { + "epoch": 1.02, + "learning_rate": 4.086495152201672e-06, + "loss": 3.837, + "step": 2804224 + }, + { + "epoch": 1.02, + "learning_rate": 4.078109204691152e-06, + "loss": 3.8399, + "step": 2804736 + }, + { + "epoch": 1.02, + "learning_rate": 4.069723257180632e-06, + "loss": 3.8437, + "step": 2805248 + }, + { + "epoch": 1.02, + "learning_rate": 4.061337309670112e-06, + "loss": 3.8411, + "step": 2805760 + }, + { + "epoch": 1.02, + "learning_rate": 4.0529513621595915e-06, + "loss": 3.8526, + "step": 2806272 + }, + { + "epoch": 1.02, + "learning_rate": 4.0445654146490715e-06, + "loss": 3.8603, + "step": 2806784 + }, + { + "epoch": 1.02, + "learning_rate": 4.0361794671385515e-06, + "loss": 3.8486, + "step": 2807296 + }, + { + "epoch": 1.02, + "learning_rate": 4.0277935196280306e-06, + "loss": 3.8425, + "step": 2807808 + }, + { + "epoch": 1.02, + "learning_rate": 4.019407572117511e-06, + "loss": 3.8427, + "step": 2808320 + }, + { + "epoch": 1.02, + "learning_rate": 4.011038003410722e-06, + "loss": 3.8416, + "step": 2808832 + }, + { + "epoch": 1.02, + "learning_rate": 4.002652055900202e-06, + "loss": 3.8405, + "step": 2809344 + }, + { + "epoch": 1.02, + "learning_rate": 3.994266108389682e-06, + "loss": 3.8521, + "step": 2809856 + }, + { + "epoch": 1.02, + "learning_rate": 3.985880160879162e-06, + "loss": 3.8292, + "step": 2810368 + }, + { + "epoch": 1.02, + "learning_rate": 3.977494213368642e-06, + "loss": 3.8487, + "step": 2810880 + }, + { + "epoch": 1.02, + "learning_rate": 3.969124644661854e-06, + "loss": 3.8427, + "step": 2811392 + }, + { + "epoch": 1.02, + "learning_rate": 3.960738697151333e-06, + "loss": 3.8495, + "step": 2811904 + }, + { + "epoch": 1.02, + "learning_rate": 3.952352749640813e-06, + "loss": 3.8382, + "step": 2812416 + }, + { + "epoch": 1.02, + "learning_rate": 3.943983180934024e-06, + "loss": 3.8456, + "step": 2812928 + }, + { + "epoch": 1.02, + "learning_rate": 3.935597233423504e-06, + "loss": 3.8394, + "step": 2813440 + }, + { + "epoch": 1.02, + "learning_rate": 3.927211285912984e-06, + "loss": 3.855, + "step": 2813952 + }, + { + "epoch": 1.02, + "learning_rate": 3.918825338402464e-06, + "loss": 3.8479, + "step": 2814464 + }, + { + "epoch": 1.02, + "learning_rate": 3.910439390891944e-06, + "loss": 3.8567, + "step": 2814976 + }, + { + "epoch": 1.02, + "learning_rate": 3.902053443381424e-06, + "loss": 3.836, + "step": 2815488 + }, + { + "epoch": 1.02, + "learning_rate": 3.893667495870903e-06, + "loss": 3.8483, + "step": 2816000 + }, + { + "epoch": 1.02, + "learning_rate": 3.885281548360384e-06, + "loss": 3.8534, + "step": 2816512 + }, + { + "epoch": 1.02, + "learning_rate": 3.876895600849863e-06, + "loss": 3.842, + "step": 2817024 + }, + { + "epoch": 1.02, + "learning_rate": 3.868526032143075e-06, + "loss": 3.8446, + "step": 2817536 + }, + { + "epoch": 1.02, + "learning_rate": 3.860140084632555e-06, + "loss": 3.8545, + "step": 2818048 + }, + { + "epoch": 1.02, + "learning_rate": 3.851754137122035e-06, + "loss": 3.847, + "step": 2818560 + }, + { + "epoch": 1.02, + "learning_rate": 3.843368189611515e-06, + "loss": 3.8648, + "step": 2819072 + }, + { + "epoch": 1.02, + "learning_rate": 3.834982242100995e-06, + "loss": 3.8381, + "step": 2819584 + }, + { + "epoch": 1.02, + "learning_rate": 3.8265962945904745e-06, + "loss": 3.852, + "step": 2820096 + }, + { + "epoch": 1.02, + "learning_rate": 3.8182103470799545e-06, + "loss": 3.8457, + "step": 2820608 + }, + { + "epoch": 1.02, + "learning_rate": 3.8098243995694344e-06, + "loss": 3.8453, + "step": 2821120 + }, + { + "epoch": 1.02, + "learning_rate": 3.8014548308626457e-06, + "loss": 3.8464, + "step": 2821632 + }, + { + "epoch": 1.02, + "learning_rate": 3.7930852621558573e-06, + "loss": 3.8406, + "step": 2822144 + }, + { + "epoch": 1.02, + "learning_rate": 3.784699314645337e-06, + "loss": 3.8404, + "step": 2822656 + }, + { + "epoch": 1.02, + "learning_rate": 3.776313367134817e-06, + "loss": 3.8382, + "step": 2823168 + }, + { + "epoch": 1.02, + "learning_rate": 3.7679274196242967e-06, + "loss": 3.846, + "step": 2823680 + }, + { + "epoch": 1.03, + "eval_loss": 3.979618549346924, + "eval_runtime": 293.0612, + "eval_samples_per_second": 1302.087, + "eval_steps_per_second": 40.691, + "step": 2823825 + }, + { + "epoch": 0.0, + "learning_rate": 3.7595414721137763e-06, + "loss": 3.8455, + "step": 2824192 + }, + { + "epoch": 0.0, + "learning_rate": 3.7511555246032567e-06, + "loss": 3.8435, + "step": 2824704 + }, + { + "epoch": 0.0, + "learning_rate": 3.742769577092736e-06, + "loss": 3.8511, + "step": 2825216 + }, + { + "epoch": 0.0, + "learning_rate": 3.7343836295822166e-06, + "loss": 3.8497, + "step": 2825728 + }, + { + "epoch": 0.0, + "learning_rate": 3.725997682071696e-06, + "loss": 3.859, + "step": 2826240 + }, + { + "epoch": 0.0, + "learning_rate": 3.7176117345611756e-06, + "loss": 3.8485, + "step": 2826752 + }, + { + "epoch": 0.0, + "learning_rate": 3.709225787050656e-06, + "loss": 3.8425, + "step": 2827264 + }, + { + "epoch": 0.0, + "learning_rate": 3.7008398395401355e-06, + "loss": 3.838, + "step": 2827776 + }, + { + "epoch": 0.0, + "learning_rate": 3.6924538920296155e-06, + "loss": 3.8474, + "step": 2828288 + }, + { + "epoch": 0.0, + "learning_rate": 3.6840679445190954e-06, + "loss": 3.847, + "step": 2828800 + }, + { + "epoch": 0.0, + "learning_rate": 3.6756819970085754e-06, + "loss": 3.848, + "step": 2829312 + }, + { + "epoch": 0.0, + "learning_rate": 3.6672960494980558e-06, + "loss": 3.847, + "step": 2829824 + }, + { + "epoch": 0.0, + "learning_rate": 3.6589264807912666e-06, + "loss": 3.8413, + "step": 2830336 + }, + { + "epoch": 0.0, + "learning_rate": 3.650540533280747e-06, + "loss": 3.8443, + "step": 2830848 + }, + { + "epoch": 0.0, + "learning_rate": 3.6421545857702265e-06, + "loss": 3.838, + "step": 2831360 + }, + { + "epoch": 0.0, + "learning_rate": 3.633768638259706e-06, + "loss": 3.8328, + "step": 2831872 + }, + { + "epoch": 0.0, + "learning_rate": 3.6253826907491864e-06, + "loss": 3.84, + "step": 2832384 + }, + { + "epoch": 0.0, + "learning_rate": 3.616996743238666e-06, + "loss": 3.8456, + "step": 2832896 + }, + { + "epoch": 0.0, + "learning_rate": 3.6086107957281463e-06, + "loss": 3.8458, + "step": 2833408 + }, + { + "epoch": 0.0, + "learning_rate": 3.600241227021357e-06, + "loss": 3.8616, + "step": 2833920 + }, + { + "epoch": 0.0, + "learning_rate": 3.5918552795108375e-06, + "loss": 3.8477, + "step": 2834432 + }, + { + "epoch": 0.0, + "learning_rate": 3.5834693320003174e-06, + "loss": 3.8541, + "step": 2834944 + }, + { + "epoch": 0.0, + "learning_rate": 3.575083384489797e-06, + "loss": 3.8412, + "step": 2835456 + }, + { + "epoch": 0.0, + "learning_rate": 3.5666974369792773e-06, + "loss": 3.8516, + "step": 2835968 + }, + { + "epoch": 0.0, + "learning_rate": 3.558311489468757e-06, + "loss": 3.8435, + "step": 2836480 + }, + { + "epoch": 0.0, + "learning_rate": 3.5499255419582373e-06, + "loss": 3.8519, + "step": 2836992 + }, + { + "epoch": 0.0, + "learning_rate": 3.5415395944477168e-06, + "loss": 3.8439, + "step": 2837504 + }, + { + "epoch": 0.0, + "learning_rate": 3.5331536469371963e-06, + "loss": 3.8392, + "step": 2838016 + }, + { + "epoch": 0.0, + "learning_rate": 3.524784078230408e-06, + "loss": 3.8385, + "step": 2838528 + }, + { + "epoch": 0.0, + "learning_rate": 3.5163981307198875e-06, + "loss": 3.8444, + "step": 2839040 + }, + { + "epoch": 0.01, + "learning_rate": 3.508012183209368e-06, + "loss": 3.8483, + "step": 2839552 + }, + { + "epoch": 0.01, + "learning_rate": 3.4996262356988474e-06, + "loss": 3.8493, + "step": 2840064 + }, + { + "epoch": 0.01, + "learning_rate": 3.4912402881883278e-06, + "loss": 3.8552, + "step": 2840576 + }, + { + "epoch": 0.01, + "learning_rate": 3.4828543406778073e-06, + "loss": 3.8433, + "step": 2841088 + }, + { + "epoch": 0.01, + "learning_rate": 3.4744683931672873e-06, + "loss": 3.8442, + "step": 2841600 + }, + { + "epoch": 0.01, + "learning_rate": 3.466098824460499e-06, + "loss": 3.8561, + "step": 2842112 + }, + { + "epoch": 0.01, + "learning_rate": 3.4577128769499784e-06, + "loss": 3.8488, + "step": 2842624 + }, + { + "epoch": 0.01, + "learning_rate": 3.449326929439459e-06, + "loss": 3.8332, + "step": 2843136 + }, + { + "epoch": 0.01, + "learning_rate": 3.4409573607326696e-06, + "loss": 3.8358, + "step": 2843648 + }, + { + "epoch": 0.01, + "learning_rate": 3.43257141322215e-06, + "loss": 3.8331, + "step": 2844160 + }, + { + "epoch": 0.01, + "learning_rate": 3.4241854657116295e-06, + "loss": 3.8388, + "step": 2844672 + }, + { + "epoch": 0.01, + "learning_rate": 3.41579951820111e-06, + "loss": 3.8565, + "step": 2845184 + }, + { + "epoch": 0.01, + "learning_rate": 3.4074135706905894e-06, + "loss": 3.8439, + "step": 2845696 + }, + { + "epoch": 0.01, + "learning_rate": 3.399044001983801e-06, + "loss": 3.8367, + "step": 2846208 + }, + { + "epoch": 0.01, + "learning_rate": 3.3906580544732806e-06, + "loss": 3.8484, + "step": 2846720 + }, + { + "epoch": 0.01, + "learning_rate": 3.3822721069627606e-06, + "loss": 3.838, + "step": 2847232 + }, + { + "epoch": 0.01, + "learning_rate": 3.3738861594522405e-06, + "loss": 3.8428, + "step": 2847744 + }, + { + "epoch": 0.01, + "learning_rate": 3.3655165907454517e-06, + "loss": 3.8351, + "step": 2848256 + }, + { + "epoch": 0.01, + "learning_rate": 3.357130643234932e-06, + "loss": 3.8205, + "step": 2848768 + }, + { + "epoch": 0.01, + "learning_rate": 3.3487446957244116e-06, + "loss": 3.8699, + "step": 2849280 + }, + { + "epoch": 0.01, + "learning_rate": 3.340358748213892e-06, + "loss": 3.8473, + "step": 2849792 + }, + { + "epoch": 0.01, + "learning_rate": 3.331989179507103e-06, + "loss": 3.849, + "step": 2850304 + }, + { + "epoch": 0.01, + "learning_rate": 3.323603231996583e-06, + "loss": 3.8408, + "step": 2850816 + }, + { + "epoch": 0.01, + "learning_rate": 3.3152172844860627e-06, + "loss": 3.8444, + "step": 2851328 + }, + { + "epoch": 0.01, + "learning_rate": 3.3068313369755422e-06, + "loss": 3.8356, + "step": 2851840 + }, + { + "epoch": 0.01, + "learning_rate": 3.2984453894650226e-06, + "loss": 3.8417, + "step": 2852352 + }, + { + "epoch": 0.01, + "learning_rate": 3.290059441954502e-06, + "loss": 3.8391, + "step": 2852864 + }, + { + "epoch": 0.01, + "learning_rate": 3.2816734944439825e-06, + "loss": 3.8451, + "step": 2853376 + }, + { + "epoch": 0.01, + "learning_rate": 3.2732875469334625e-06, + "loss": 3.8586, + "step": 2853888 + }, + { + "epoch": 0.01, + "learning_rate": 3.264917978226674e-06, + "loss": 3.8365, + "step": 2854400 + }, + { + "epoch": 0.01, + "learning_rate": 3.2565320307161537e-06, + "loss": 3.8315, + "step": 2854912 + }, + { + "epoch": 0.01, + "learning_rate": 3.248146083205633e-06, + "loss": 3.8545, + "step": 2855424 + }, + { + "epoch": 0.01, + "learning_rate": 3.239776514498845e-06, + "loss": 3.8242, + "step": 2855936 + }, + { + "epoch": 0.01, + "learning_rate": 3.2313905669883244e-06, + "loss": 3.8311, + "step": 2856448 + }, + { + "epoch": 0.01, + "learning_rate": 3.2230046194778047e-06, + "loss": 3.8502, + "step": 2856960 + }, + { + "epoch": 0.01, + "learning_rate": 3.2146186719672843e-06, + "loss": 3.8541, + "step": 2857472 + }, + { + "epoch": 0.01, + "learning_rate": 3.206249103260496e-06, + "loss": 3.8387, + "step": 2857984 + }, + { + "epoch": 0.01, + "learning_rate": 3.197863155749976e-06, + "loss": 3.8392, + "step": 2858496 + }, + { + "epoch": 0.01, + "learning_rate": 3.189477208239456e-06, + "loss": 3.8228, + "step": 2859008 + }, + { + "epoch": 0.01, + "learning_rate": 3.1810912607289358e-06, + "loss": 3.839, + "step": 2859520 + }, + { + "epoch": 0.01, + "learning_rate": 3.1727380708258787e-06, + "loss": 3.8508, + "step": 2860032 + }, + { + "epoch": 0.01, + "learning_rate": 3.164352123315358e-06, + "loss": 3.836, + "step": 2860544 + }, + { + "epoch": 0.01, + "learning_rate": 3.1559661758048386e-06, + "loss": 3.8451, + "step": 2861056 + }, + { + "epoch": 0.01, + "learning_rate": 3.147580228294318e-06, + "loss": 3.8484, + "step": 2861568 + }, + { + "epoch": 0.01, + "learning_rate": 3.1391942807837977e-06, + "loss": 3.8587, + "step": 2862080 + }, + { + "epoch": 0.01, + "learning_rate": 3.130808333273278e-06, + "loss": 3.8392, + "step": 2862592 + }, + { + "epoch": 0.01, + "learning_rate": 3.1224387645664892e-06, + "loss": 3.841, + "step": 2863104 + }, + { + "epoch": 0.01, + "learning_rate": 3.114052817055969e-06, + "loss": 3.8501, + "step": 2863616 + }, + { + "epoch": 0.01, + "learning_rate": 3.105666869545449e-06, + "loss": 3.8337, + "step": 2864128 + }, + { + "epoch": 0.01, + "learning_rate": 3.097280922034929e-06, + "loss": 3.8513, + "step": 2864640 + }, + { + "epoch": 0.01, + "learning_rate": 3.088894974524409e-06, + "loss": 3.8412, + "step": 2865152 + }, + { + "epoch": 0.01, + "learning_rate": 3.080509027013889e-06, + "loss": 3.8417, + "step": 2865664 + }, + { + "epoch": 0.01, + "learning_rate": 3.0721230795033686e-06, + "loss": 3.851, + "step": 2866176 + }, + { + "epoch": 0.01, + "learning_rate": 3.0637371319928485e-06, + "loss": 3.8309, + "step": 2866688 + }, + { + "epoch": 0.01, + "learning_rate": 3.0553511844823285e-06, + "loss": 3.8341, + "step": 2867200 + }, + { + "epoch": 0.01, + "learning_rate": 3.0469652369718084e-06, + "loss": 3.8409, + "step": 2867712 + }, + { + "epoch": 0.01, + "learning_rate": 3.0385792894612884e-06, + "loss": 3.8496, + "step": 2868224 + }, + { + "epoch": 0.01, + "learning_rate": 3.0302097207544996e-06, + "loss": 3.8374, + "step": 2868736 + }, + { + "epoch": 0.01, + "learning_rate": 3.0218237732439795e-06, + "loss": 3.846, + "step": 2869248 + }, + { + "epoch": 0.02, + "learning_rate": 3.013437825733459e-06, + "loss": 3.8321, + "step": 2869760 + }, + { + "epoch": 0.02, + "learning_rate": 3.005051878222939e-06, + "loss": 3.8383, + "step": 2870272 + }, + { + "epoch": 0.02, + "learning_rate": 2.996665930712419e-06, + "loss": 3.832, + "step": 2870784 + }, + { + "epoch": 0.02, + "learning_rate": 2.988279983201899e-06, + "loss": 3.8431, + "step": 2871296 + }, + { + "epoch": 0.02, + "learning_rate": 2.979894035691379e-06, + "loss": 3.8535, + "step": 2871808 + }, + { + "epoch": 0.02, + "learning_rate": 2.971508088180859e-06, + "loss": 3.8448, + "step": 2872320 + }, + { + "epoch": 0.02, + "learning_rate": 2.96313851947407e-06, + "loss": 3.8469, + "step": 2872832 + }, + { + "epoch": 0.02, + "learning_rate": 2.9547689507672817e-06, + "loss": 3.8349, + "step": 2873344 + }, + { + "epoch": 0.02, + "learning_rate": 2.9463830032567612e-06, + "loss": 3.8387, + "step": 2873856 + }, + { + "epoch": 0.02, + "learning_rate": 2.937997055746241e-06, + "loss": 3.8412, + "step": 2874368 + }, + { + "epoch": 0.02, + "learning_rate": 2.929611108235721e-06, + "loss": 3.8526, + "step": 2874880 + }, + { + "epoch": 0.02, + "learning_rate": 2.921225160725201e-06, + "loss": 3.8444, + "step": 2875392 + }, + { + "epoch": 0.02, + "learning_rate": 2.9128555920184123e-06, + "loss": 3.8419, + "step": 2875904 + }, + { + "epoch": 0.02, + "learning_rate": 2.9044696445078923e-06, + "loss": 3.8385, + "step": 2876416 + }, + { + "epoch": 0.02, + "learning_rate": 2.8960836969973727e-06, + "loss": 3.8546, + "step": 2876928 + }, + { + "epoch": 0.02, + "learning_rate": 2.887697749486852e-06, + "loss": 3.8279, + "step": 2877440 + }, + { + "epoch": 0.02, + "learning_rate": 2.879328180780064e-06, + "loss": 3.844, + "step": 2877952 + }, + { + "epoch": 0.02, + "learning_rate": 2.8709422332695434e-06, + "loss": 3.8338, + "step": 2878464 + }, + { + "epoch": 0.02, + "learning_rate": 2.8625562857590233e-06, + "loss": 3.8469, + "step": 2878976 + }, + { + "epoch": 0.02, + "learning_rate": 2.8541703382485033e-06, + "loss": 3.8341, + "step": 2879488 + }, + { + "epoch": 0.02, + "learning_rate": 2.8457843907379832e-06, + "loss": 3.8436, + "step": 2880000 + }, + { + "epoch": 0.02, + "learning_rate": 2.837398443227463e-06, + "loss": 3.8342, + "step": 2880512 + }, + { + "epoch": 0.02, + "learning_rate": 2.8290124957169427e-06, + "loss": 3.8396, + "step": 2881024 + }, + { + "epoch": 0.02, + "learning_rate": 2.8206265482064227e-06, + "loss": 3.8395, + "step": 2881536 + }, + { + "epoch": 0.02, + "learning_rate": 2.8122569794996343e-06, + "loss": 3.8352, + "step": 2882048 + }, + { + "epoch": 0.02, + "learning_rate": 2.8038710319891143e-06, + "loss": 3.8505, + "step": 2882560 + }, + { + "epoch": 0.02, + "learning_rate": 2.7955014632823255e-06, + "loss": 3.8572, + "step": 2883072 + }, + { + "epoch": 0.02, + "learning_rate": 2.7871155157718054e-06, + "loss": 3.8464, + "step": 2883584 + }, + { + "epoch": 0.02, + "learning_rate": 2.7787295682612854e-06, + "loss": 3.8444, + "step": 2884096 + }, + { + "epoch": 0.02, + "learning_rate": 2.7703436207507653e-06, + "loss": 3.8353, + "step": 2884608 + }, + { + "epoch": 0.02, + "learning_rate": 2.7619576732402453e-06, + "loss": 3.8383, + "step": 2885120 + }, + { + "epoch": 0.02, + "learning_rate": 2.753571725729725e-06, + "loss": 3.8406, + "step": 2885632 + }, + { + "epoch": 0.02, + "learning_rate": 2.7452021570229365e-06, + "loss": 3.8532, + "step": 2886144 + }, + { + "epoch": 0.02, + "learning_rate": 2.736816209512416e-06, + "loss": 3.8262, + "step": 2886656 + }, + { + "epoch": 0.02, + "learning_rate": 2.728430262001896e-06, + "loss": 3.8459, + "step": 2887168 + }, + { + "epoch": 0.02, + "learning_rate": 2.720044314491376e-06, + "loss": 3.8433, + "step": 2887680 + }, + { + "epoch": 0.02, + "learning_rate": 2.711658366980856e-06, + "loss": 3.8445, + "step": 2888192 + }, + { + "epoch": 0.02, + "learning_rate": 2.7032887982740675e-06, + "loss": 3.8394, + "step": 2888704 + }, + { + "epoch": 0.02, + "learning_rate": 2.6949192295672787e-06, + "loss": 3.8411, + "step": 2889216 + }, + { + "epoch": 0.02, + "learning_rate": 2.6865332820567587e-06, + "loss": 3.8389, + "step": 2889728 + }, + { + "epoch": 0.02, + "learning_rate": 2.6781473345462386e-06, + "loss": 3.8491, + "step": 2890240 + }, + { + "epoch": 0.02, + "learning_rate": 2.669761387035718e-06, + "loss": 3.8495, + "step": 2890752 + }, + { + "epoch": 0.02, + "learning_rate": 2.661375439525198e-06, + "loss": 3.853, + "step": 2891264 + }, + { + "epoch": 0.02, + "learning_rate": 2.652989492014678e-06, + "loss": 3.8348, + "step": 2891776 + }, + { + "epoch": 0.02, + "learning_rate": 2.644603544504158e-06, + "loss": 3.8472, + "step": 2892288 + }, + { + "epoch": 0.02, + "learning_rate": 2.636217596993638e-06, + "loss": 3.8492, + "step": 2892800 + }, + { + "epoch": 0.02, + "learning_rate": 2.6278316494831175e-06, + "loss": 3.8411, + "step": 2893312 + }, + { + "epoch": 0.02, + "learning_rate": 2.619462080776329e-06, + "loss": 3.8446, + "step": 2893824 + }, + { + "epoch": 0.02, + "learning_rate": 2.611092512069541e-06, + "loss": 3.8502, + "step": 2894336 + }, + { + "epoch": 0.02, + "learning_rate": 2.6027065645590207e-06, + "loss": 3.8438, + "step": 2894848 + }, + { + "epoch": 0.02, + "learning_rate": 2.5943206170485003e-06, + "loss": 3.8631, + "step": 2895360 + }, + { + "epoch": 0.02, + "learning_rate": 2.5859346695379802e-06, + "loss": 3.8399, + "step": 2895872 + }, + { + "epoch": 0.02, + "learning_rate": 2.57754872202746e-06, + "loss": 3.8467, + "step": 2896384 + }, + { + "epoch": 0.02, + "learning_rate": 2.5691791533206714e-06, + "loss": 3.8416, + "step": 2896896 + }, + { + "epoch": 0.02, + "learning_rate": 2.5607932058101514e-06, + "loss": 3.8456, + "step": 2897408 + }, + { + "epoch": 0.02, + "learning_rate": 2.5524072582996313e-06, + "loss": 3.8404, + "step": 2897920 + }, + { + "epoch": 0.02, + "learning_rate": 2.5440213107891113e-06, + "loss": 3.8386, + "step": 2898432 + }, + { + "epoch": 0.02, + "learning_rate": 2.535635363278591e-06, + "loss": 3.843, + "step": 2898944 + }, + { + "epoch": 0.02, + "learning_rate": 2.5272494157680708e-06, + "loss": 3.83, + "step": 2899456 + }, + { + "epoch": 0.02, + "learning_rate": 2.5188634682575507e-06, + "loss": 3.846, + "step": 2899968 + }, + { + "epoch": 0.03, + "eval_loss": 3.979407548904419, + "eval_runtime": 296.6452, + "eval_samples_per_second": 1286.355, + "eval_steps_per_second": 40.2, + "step": 2900145 + } + ], + "logging_steps": 512, + "max_steps": 3052726, + "num_train_epochs": 9223372036854775807, + "save_steps": 10, + "total_flos": 1.1751459017557048e+18, + "trial_name": null, + "trial_params": null +}