{ "best_metric": 4.382836818695068, "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/pp-mod-subj2/lstm/0/checkpoints/checkpoint-228960", "epoch": 0.025000606015738065, "eval_steps": 10, "global_step": 228960, "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": 10.8195, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.999161405248948e-05, "loss": 7.5629, "step": 512 }, { "epoch": 0.0, "learning_rate": 4.998322810497896e-05, "loss": 7.0621, "step": 1024 }, { "epoch": 0.0, "learning_rate": 4.997484215746844e-05, "loss": 6.9908, "step": 1536 }, { "epoch": 0.0, "learning_rate": 4.996645620995792e-05, "loss": 6.9369, "step": 2048 }, { "epoch": 0.0, "learning_rate": 4.99580702624474e-05, "loss": 6.8371, "step": 2560 }, { "epoch": 0.0, "learning_rate": 4.994968431493688e-05, "loss": 6.7009, "step": 3072 }, { "epoch": 0.0, "learning_rate": 4.994129836742636e-05, "loss": 6.5988, "step": 3584 }, { "epoch": 0.0, "learning_rate": 4.993291241991584e-05, "loss": 6.5058, "step": 4096 }, { "epoch": 0.0, "learning_rate": 4.992452647240532e-05, "loss": 6.4276, "step": 4608 }, { "epoch": 0.0, "learning_rate": 4.99161405248948e-05, "loss": 6.3668, "step": 5120 }, { "epoch": 0.0, "learning_rate": 4.990775457738428e-05, "loss": 6.2887, "step": 5632 }, { "epoch": 0.0, "learning_rate": 4.989936862987376e-05, "loss": 6.2278, "step": 6144 }, { "epoch": 0.0, "learning_rate": 4.989098268236324e-05, "loss": 6.1552, "step": 6656 }, { "epoch": 0.0, "learning_rate": 4.988259673485272e-05, "loss": 6.1049, "step": 7168 }, { "epoch": 0.0, "learning_rate": 4.98742107873422e-05, "loss": 6.04, "step": 7680 }, { "epoch": 0.0, "learning_rate": 4.986582483983168e-05, "loss": 6.0126, "step": 8192 }, { "epoch": 0.0, "learning_rate": 4.985743889232116e-05, "loss": 5.9574, "step": 8704 }, { "epoch": 0.0, "learning_rate": 4.984905294481064e-05, "loss": 5.9143, "step": 9216 }, { "epoch": 0.0, "learning_rate": 4.984066699730012e-05, "loss": 5.8808, "step": 9728 }, { "epoch": 0.0, "learning_rate": 4.983229742859333e-05, "loss": 5.8575, "step": 10240 }, { "epoch": 0.0, "learning_rate": 4.982391148108281e-05, "loss": 5.8109, "step": 10752 }, { "epoch": 0.0, "learning_rate": 4.981552553357229e-05, "loss": 5.7867, "step": 11264 }, { "epoch": 0.0, "learning_rate": 4.980713958606178e-05, "loss": 5.7509, "step": 11776 }, { "epoch": 0.0, "learning_rate": 4.9798770017354986e-05, "loss": 5.7184, "step": 12288 }, { "epoch": 0.0, "learning_rate": 4.9790384069844466e-05, "loss": 5.7003, "step": 12800 }, { "epoch": 0.0, "learning_rate": 4.9781998122333946e-05, "loss": 5.6629, "step": 13312 }, { "epoch": 0.0, "learning_rate": 4.9773628553627155e-05, "loss": 5.6507, "step": 13824 }, { "epoch": 0.0, "learning_rate": 4.9765242606116635e-05, "loss": 5.625, "step": 14336 }, { "epoch": 0.0, "learning_rate": 4.9756856658606115e-05, "loss": 5.5976, "step": 14848 }, { "epoch": 0.01, "learning_rate": 4.9748470711095595e-05, "loss": 5.5835, "step": 15360 }, { "epoch": 0.01, "learning_rate": 4.974008476358507e-05, "loss": 5.561, "step": 15872 }, { "epoch": 0.01, "learning_rate": 4.973169881607455e-05, "loss": 5.5482, "step": 16384 }, { "epoch": 0.01, "learning_rate": 4.972331286856403e-05, "loss": 5.529, "step": 16896 }, { "epoch": 0.01, "learning_rate": 4.9714943299857244e-05, "loss": 5.5105, "step": 17408 }, { "epoch": 0.01, "learning_rate": 4.970657373115046e-05, "loss": 5.5025, "step": 17920 }, { "epoch": 0.01, "learning_rate": 4.969818778363994e-05, "loss": 5.495, "step": 18432 }, { "epoch": 0.01, "learning_rate": 4.968980183612942e-05, "loss": 5.4556, "step": 18944 }, { "epoch": 0.01, "learning_rate": 4.968143226742263e-05, "loss": 5.4478, "step": 19456 }, { "epoch": 0.01, "learning_rate": 4.967304631991211e-05, "loss": 5.4339, "step": 19968 }, { "epoch": 0.01, "learning_rate": 4.966466037240159e-05, "loss": 5.4193, "step": 20480 }, { "epoch": 0.01, "learning_rate": 4.965627442489107e-05, "loss": 5.4053, "step": 20992 }, { "epoch": 0.01, "learning_rate": 4.964788847738054e-05, "loss": 5.4073, "step": 21504 }, { "epoch": 0.01, "learning_rate": 4.963950252987002e-05, "loss": 5.3857, "step": 22016 }, { "epoch": 0.01, "learning_rate": 4.96311165823595e-05, "loss": 5.371, "step": 22528 }, { "epoch": 0.01, "learning_rate": 4.962273063484898e-05, "loss": 5.3638, "step": 23040 }, { "epoch": 0.01, "learning_rate": 4.961434468733847e-05, "loss": 5.3546, "step": 23552 }, { "epoch": 0.01, "learning_rate": 4.960595873982795e-05, "loss": 5.3485, "step": 24064 }, { "epoch": 0.01, "learning_rate": 4.959757279231743e-05, "loss": 5.3232, "step": 24576 }, { "epoch": 0.01, "learning_rate": 4.958918684480691e-05, "loss": 5.3127, "step": 25088 }, { "epoch": 0.01, "learning_rate": 4.958080089729639e-05, "loss": 5.3212, "step": 25600 }, { "epoch": 0.01, "learning_rate": 4.95724313285896e-05, "loss": 5.31, "step": 26112 }, { "epoch": 0.01, "learning_rate": 4.956404538107908e-05, "loss": 5.2974, "step": 26624 }, { "epoch": 0.01, "learning_rate": 4.955567581237229e-05, "loss": 5.2747, "step": 27136 }, { "epoch": 0.01, "learning_rate": 4.954728986486177e-05, "loss": 5.2834, "step": 27648 }, { "epoch": 0.01, "learning_rate": 4.953890391735125e-05, "loss": 5.26, "step": 28160 }, { "epoch": 0.01, "learning_rate": 4.9530517969840727e-05, "loss": 5.2779, "step": 28672 }, { "epoch": 0.01, "learning_rate": 4.9522132022330207e-05, "loss": 5.243, "step": 29184 }, { "epoch": 0.01, "learning_rate": 4.9513746074819686e-05, "loss": 5.2466, "step": 29696 }, { "epoch": 0.01, "learning_rate": 4.9505360127309166e-05, "loss": 5.2271, "step": 30208 }, { "epoch": 0.01, "learning_rate": 4.9496974179798646e-05, "loss": 5.2167, "step": 30720 }, { "epoch": 0.01, "learning_rate": 4.948858823228813e-05, "loss": 5.2181, "step": 31232 }, { "epoch": 0.01, "learning_rate": 4.948020228477761e-05, "loss": 5.2027, "step": 31744 }, { "epoch": 0.01, "learning_rate": 4.9471816337267086e-05, "loss": 5.1976, "step": 32256 }, { "epoch": 0.01, "learning_rate": 4.94634467685603e-05, "loss": 5.1888, "step": 32768 }, { "epoch": 0.01, "learning_rate": 4.945506082104978e-05, "loss": 5.1986, "step": 33280 }, { "epoch": 0.01, "learning_rate": 4.9446674873539255e-05, "loss": 5.1712, "step": 33792 }, { "epoch": 0.01, "learning_rate": 4.9438288926028735e-05, "loss": 5.1614, "step": 34304 }, { "epoch": 0.01, "learning_rate": 4.9429902978518215e-05, "loss": 5.1454, "step": 34816 }, { "epoch": 0.01, "learning_rate": 4.942153340981143e-05, "loss": 5.1456, "step": 35328 }, { "epoch": 0.01, "learning_rate": 4.9413147462300904e-05, "loss": 5.1594, "step": 35840 }, { "epoch": 0.01, "learning_rate": 4.9404761514790384e-05, "loss": 5.1494, "step": 36352 }, { "epoch": 0.01, "learning_rate": 4.939637556727987e-05, "loss": 5.1299, "step": 36864 }, { "epoch": 0.01, "learning_rate": 4.938798961976935e-05, "loss": 5.1276, "step": 37376 }, { "epoch": 0.01, "learning_rate": 4.937960367225883e-05, "loss": 5.1282, "step": 37888 }, { "epoch": 0.01, "learning_rate": 4.937121772474831e-05, "loss": 5.1223, "step": 38400 }, { "epoch": 0.01, "learning_rate": 4.936284815604152e-05, "loss": 5.099, "step": 38912 }, { "epoch": 0.01, "learning_rate": 4.9354462208531e-05, "loss": 5.1024, "step": 39424 }, { "epoch": 0.01, "learning_rate": 4.934607626102048e-05, "loss": 5.0889, "step": 39936 }, { "epoch": 0.01, "learning_rate": 4.933769031350996e-05, "loss": 5.0915, "step": 40448 }, { "epoch": 0.01, "learning_rate": 4.932932074480317e-05, "loss": 5.0867, "step": 40960 }, { "epoch": 0.01, "learning_rate": 4.932093479729265e-05, "loss": 5.076, "step": 41472 }, { "epoch": 0.01, "learning_rate": 4.931254884978213e-05, "loss": 5.078, "step": 41984 }, { "epoch": 0.01, "learning_rate": 4.930416290227161e-05, "loss": 5.0683, "step": 42496 }, { "epoch": 0.01, "learning_rate": 4.929577695476109e-05, "loss": 5.0516, "step": 43008 }, { "epoch": 0.01, "learning_rate": 4.9287407386054305e-05, "loss": 5.0508, "step": 43520 }, { "epoch": 0.01, "learning_rate": 4.9279021438543785e-05, "loss": 5.0557, "step": 44032 }, { "epoch": 0.01, "learning_rate": 4.9270651869836994e-05, "loss": 5.0489, "step": 44544 }, { "epoch": 0.01, "learning_rate": 4.9262265922326474e-05, "loss": 5.0365, "step": 45056 }, { "epoch": 0.01, "learning_rate": 4.9253879974815954e-05, "loss": 5.0226, "step": 45568 }, { "epoch": 0.02, "learning_rate": 4.9245494027305433e-05, "loss": 5.0284, "step": 46080 }, { "epoch": 0.02, "learning_rate": 4.9237108079794913e-05, "loss": 5.0152, "step": 46592 }, { "epoch": 0.02, "learning_rate": 4.922872213228439e-05, "loss": 5.0247, "step": 47104 }, { "epoch": 0.02, "learning_rate": 4.922033618477387e-05, "loss": 5.0043, "step": 47616 }, { "epoch": 0.02, "learning_rate": 4.921195023726335e-05, "loss": 4.9997, "step": 48128 }, { "epoch": 0.02, "learning_rate": 4.920358066855656e-05, "loss": 5.0168, "step": 48640 }, { "epoch": 0.02, "learning_rate": 4.919519472104604e-05, "loss": 4.9837, "step": 49152 }, { "epoch": 0.02, "learning_rate": 4.918680877353552e-05, "loss": 4.9776, "step": 49664 }, { "epoch": 0.02, "learning_rate": 4.917842282602501e-05, "loss": 4.9822, "step": 50176 }, { "epoch": 0.02, "learning_rate": 4.917003687851449e-05, "loss": 4.9802, "step": 50688 }, { "epoch": 0.02, "learning_rate": 4.91616673098077e-05, "loss": 4.9811, "step": 51200 }, { "epoch": 0.02, "learning_rate": 4.915328136229718e-05, "loss": 4.9687, "step": 51712 }, { "epoch": 0.02, "learning_rate": 4.914489541478666e-05, "loss": 4.963, "step": 52224 }, { "epoch": 0.02, "learning_rate": 4.913650946727614e-05, "loss": 4.9546, "step": 52736 }, { "epoch": 0.02, "learning_rate": 4.912812351976562e-05, "loss": 4.9551, "step": 53248 }, { "epoch": 0.02, "learning_rate": 4.911973757225509e-05, "loss": 4.9382, "step": 53760 }, { "epoch": 0.02, "learning_rate": 4.911136800354831e-05, "loss": 4.9392, "step": 54272 }, { "epoch": 0.02, "learning_rate": 4.910298205603779e-05, "loss": 4.9423, "step": 54784 }, { "epoch": 0.02, "learning_rate": 4.909459610852727e-05, "loss": 4.9314, "step": 55296 }, { "epoch": 0.02, "learning_rate": 4.908621016101675e-05, "loss": 4.9248, "step": 55808 }, { "epoch": 0.02, "learning_rate": 4.907782421350623e-05, "loss": 4.92, "step": 56320 }, { "epoch": 0.02, "learning_rate": 4.906943826599571e-05, "loss": 4.9109, "step": 56832 }, { "epoch": 0.02, "learning_rate": 4.906105231848519e-05, "loss": 4.9207, "step": 57344 }, { "epoch": 0.02, "learning_rate": 4.9052682749778396e-05, "loss": 4.9065, "step": 57856 }, { "epoch": 0.02, "learning_rate": 4.9044296802267876e-05, "loss": 4.9033, "step": 58368 }, { "epoch": 0.02, "learning_rate": 4.9035910854757356e-05, "loss": 4.9045, "step": 58880 }, { "epoch": 0.02, "learning_rate": 4.9027524907246836e-05, "loss": 4.8985, "step": 59392 }, { "epoch": 0.02, "learning_rate": 4.9019155338540045e-05, "loss": 4.8923, "step": 59904 }, { "epoch": 0.02, "learning_rate": 4.9010769391029525e-05, "loss": 4.8907, "step": 60416 }, { "epoch": 0.02, "learning_rate": 4.9002383443519005e-05, "loss": 4.8915, "step": 60928 }, { "epoch": 0.02, "learning_rate": 4.8993997496008485e-05, "loss": 4.8907, "step": 61440 }, { "epoch": 0.02, "learning_rate": 4.89856279273017e-05, "loss": 4.8697, "step": 61952 }, { "epoch": 0.02, "learning_rate": 4.897724197979118e-05, "loss": 4.8797, "step": 62464 }, { "epoch": 0.02, "learning_rate": 4.896885603228066e-05, "loss": 4.8704, "step": 62976 }, { "epoch": 0.02, "learning_rate": 4.896047008477014e-05, "loss": 4.8727, "step": 63488 }, { "epoch": 0.02, "learning_rate": 4.895208413725962e-05, "loss": 4.8536, "step": 64000 }, { "epoch": 0.02, "learning_rate": 4.89436981897491e-05, "loss": 4.8626, "step": 64512 }, { "epoch": 0.02, "learning_rate": 4.893531224223858e-05, "loss": 4.8477, "step": 65024 }, { "epoch": 0.02, "learning_rate": 4.892694267353179e-05, "loss": 4.8542, "step": 65536 }, { "epoch": 0.02, "learning_rate": 4.891855672602127e-05, "loss": 4.831, "step": 66048 }, { "epoch": 0.02, "learning_rate": 4.891017077851075e-05, "loss": 4.8561, "step": 66560 }, { "epoch": 0.02, "learning_rate": 4.890178483100023e-05, "loss": 4.8589, "step": 67072 }, { "epoch": 0.02, "learning_rate": 4.889341526229344e-05, "loss": 4.8391, "step": 67584 }, { "epoch": 0.02, "learning_rate": 4.888502931478292e-05, "loss": 4.8271, "step": 68096 }, { "epoch": 0.02, "learning_rate": 4.88766433672724e-05, "loss": 4.8383, "step": 68608 }, { "epoch": 0.02, "learning_rate": 4.886825741976188e-05, "loss": 4.831, "step": 69120 }, { "epoch": 0.02, "learning_rate": 4.8859871472251365e-05, "loss": 4.829, "step": 69632 }, { "epoch": 0.02, "learning_rate": 4.8851485524740845e-05, "loss": 4.8236, "step": 70144 }, { "epoch": 0.02, "learning_rate": 4.8843099577230325e-05, "loss": 4.8268, "step": 70656 }, { "epoch": 0.02, "learning_rate": 4.8834713629719805e-05, "loss": 4.8167, "step": 71168 }, { "epoch": 0.02, "learning_rate": 4.882636043981674e-05, "loss": 4.8112, "step": 71680 }, { "epoch": 0.02, "learning_rate": 4.881797449230622e-05, "loss": 4.7965, "step": 72192 }, { "epoch": 0.02, "learning_rate": 4.88095885447957e-05, "loss": 4.808, "step": 72704 }, { "epoch": 0.02, "learning_rate": 4.880120259728518e-05, "loss": 4.8048, "step": 73216 }, { "epoch": 0.02, "learning_rate": 4.879281664977466e-05, "loss": 4.7955, "step": 73728 }, { "epoch": 0.02, "learning_rate": 4.878443070226414e-05, "loss": 4.8082, "step": 74240 }, { "epoch": 0.02, "learning_rate": 4.877604475475362e-05, "loss": 4.7873, "step": 74752 }, { "epoch": 0.02, "learning_rate": 4.87676588072431e-05, "loss": 4.7961, "step": 75264 }, { "epoch": 0.02, "learning_rate": 4.875928923853632e-05, "loss": 4.7969, "step": 75776 }, { "epoch": 0.02, "learning_rate": 4.87509032910258e-05, "loss": 4.7865, "step": 76288 }, { "epoch": 0.03, "eval_loss": 4.7972731590271, "eval_runtime": 313.984, "eval_samples_per_second": 1215.32, "eval_steps_per_second": 37.98, "step": 76320 }, { "epoch": 1.0, "learning_rate": 4.874251734351528e-05, "loss": 4.7745, "step": 76800 }, { "epoch": 1.0, "learning_rate": 4.873413139600475e-05, "loss": 4.7801, "step": 77312 }, { "epoch": 1.0, "learning_rate": 4.872574544849423e-05, "loss": 4.7839, "step": 77824 }, { "epoch": 1.0, "learning_rate": 4.871735950098371e-05, "loss": 4.7728, "step": 78336 }, { "epoch": 1.0, "learning_rate": 4.870897355347319e-05, "loss": 4.7752, "step": 78848 }, { "epoch": 1.0, "learning_rate": 4.870058760596267e-05, "loss": 4.758, "step": 79360 }, { "epoch": 1.0, "learning_rate": 4.8692234416059616e-05, "loss": 4.7629, "step": 79872 }, { "epoch": 1.0, "learning_rate": 4.8683848468549096e-05, "loss": 4.7537, "step": 80384 }, { "epoch": 1.0, "learning_rate": 4.8675462521038576e-05, "loss": 4.7592, "step": 80896 }, { "epoch": 1.0, "learning_rate": 4.8667076573528056e-05, "loss": 4.75, "step": 81408 }, { "epoch": 1.0, "learning_rate": 4.8658690626017536e-05, "loss": 4.7557, "step": 81920 }, { "epoch": 1.0, "learning_rate": 4.865032105731075e-05, "loss": 4.7575, "step": 82432 }, { "epoch": 1.0, "learning_rate": 4.864195148860396e-05, "loss": 4.7424, "step": 82944 }, { "epoch": 1.0, "learning_rate": 4.863356554109344e-05, "loss": 4.7396, "step": 83456 }, { "epoch": 1.0, "learning_rate": 4.862517959358292e-05, "loss": 4.7257, "step": 83968 }, { "epoch": 1.0, "learning_rate": 4.86167936460724e-05, "loss": 4.7365, "step": 84480 }, { "epoch": 1.0, "learning_rate": 4.8608407698561874e-05, "loss": 4.7299, "step": 84992 }, { "epoch": 1.0, "learning_rate": 4.8600021751051354e-05, "loss": 4.7238, "step": 85504 }, { "epoch": 1.0, "learning_rate": 4.8591635803540834e-05, "loss": 4.721, "step": 86016 }, { "epoch": 1.0, "learning_rate": 4.8583249856030314e-05, "loss": 4.7489, "step": 86528 }, { "epoch": 1.0, "learning_rate": 4.8574863908519794e-05, "loss": 4.7194, "step": 87040 }, { "epoch": 1.0, "learning_rate": 4.8566477961009274e-05, "loss": 4.7224, "step": 87552 }, { "epoch": 1.0, "learning_rate": 4.855810839230249e-05, "loss": 4.7253, "step": 88064 }, { "epoch": 1.0, "learning_rate": 4.854972244479197e-05, "loss": 4.7146, "step": 88576 }, { "epoch": 1.0, "learning_rate": 4.854133649728145e-05, "loss": 4.7129, "step": 89088 }, { "epoch": 1.0, "learning_rate": 4.853295054977093e-05, "loss": 4.7014, "step": 89600 }, { "epoch": 1.0, "learning_rate": 4.852456460226041e-05, "loss": 4.7086, "step": 90112 }, { "epoch": 1.0, "learning_rate": 4.851617865474989e-05, "loss": 4.7065, "step": 90624 }, { "epoch": 1.0, "learning_rate": 4.85078090860431e-05, "loss": 4.6927, "step": 91136 }, { "epoch": 1.01, "learning_rate": 4.849942313853258e-05, "loss": 4.7015, "step": 91648 }, { "epoch": 1.01, "learning_rate": 4.849103719102206e-05, "loss": 4.6906, "step": 92160 }, { "epoch": 1.01, "learning_rate": 4.848265124351154e-05, "loss": 4.7008, "step": 92672 }, { "epoch": 1.01, "learning_rate": 4.847426529600102e-05, "loss": 4.6963, "step": 93184 }, { "epoch": 1.01, "learning_rate": 4.84658793484905e-05, "loss": 4.6844, "step": 93696 }, { "epoch": 1.01, "learning_rate": 4.845750977978371e-05, "loss": 4.6944, "step": 94208 }, { "epoch": 1.01, "learning_rate": 4.8449123832273194e-05, "loss": 4.7025, "step": 94720 }, { "epoch": 1.01, "learning_rate": 4.8440737884762674e-05, "loss": 4.6713, "step": 95232 }, { "epoch": 1.01, "learning_rate": 4.8432351937252154e-05, "loss": 4.6818, "step": 95744 }, { "epoch": 1.01, "learning_rate": 4.8423965989741634e-05, "loss": 4.6775, "step": 96256 }, { "epoch": 1.01, "learning_rate": 4.8415580042231114e-05, "loss": 4.6692, "step": 96768 }, { "epoch": 1.01, "learning_rate": 4.840719409472059e-05, "loss": 4.6715, "step": 97280 }, { "epoch": 1.01, "learning_rate": 4.839880814721007e-05, "loss": 4.6795, "step": 97792 }, { "epoch": 1.01, "learning_rate": 4.839043857850328e-05, "loss": 4.6717, "step": 98304 }, { "epoch": 1.01, "learning_rate": 4.838206900979649e-05, "loss": 4.6668, "step": 98816 }, { "epoch": 1.01, "learning_rate": 4.837368306228597e-05, "loss": 4.6697, "step": 99328 }, { "epoch": 1.01, "learning_rate": 4.836529711477545e-05, "loss": 4.6626, "step": 99840 }, { "epoch": 1.01, "learning_rate": 4.835691116726493e-05, "loss": 4.6672, "step": 100352 }, { "epoch": 1.01, "learning_rate": 4.834852521975441e-05, "loss": 4.6494, "step": 100864 }, { "epoch": 1.01, "learning_rate": 4.834013927224389e-05, "loss": 4.6453, "step": 101376 }, { "epoch": 1.01, "learning_rate": 4.833176970353711e-05, "loss": 4.6637, "step": 101888 }, { "epoch": 1.01, "learning_rate": 4.832338375602659e-05, "loss": 4.6606, "step": 102400 }, { "epoch": 1.01, "learning_rate": 4.831499780851606e-05, "loss": 4.6499, "step": 102912 }, { "epoch": 1.01, "learning_rate": 4.830661186100554e-05, "loss": 4.6413, "step": 103424 }, { "epoch": 1.01, "learning_rate": 4.829822591349502e-05, "loss": 4.6504, "step": 103936 }, { "epoch": 1.01, "learning_rate": 4.828985634478824e-05, "loss": 4.6328, "step": 104448 }, { "epoch": 1.01, "learning_rate": 4.828147039727771e-05, "loss": 4.6605, "step": 104960 }, { "epoch": 1.01, "learning_rate": 4.827308444976719e-05, "loss": 4.6355, "step": 105472 }, { "epoch": 1.01, "learning_rate": 4.826469850225667e-05, "loss": 4.6493, "step": 105984 }, { "epoch": 1.01, "learning_rate": 4.825631255474615e-05, "loss": 4.6324, "step": 106496 }, { "epoch": 1.01, "learning_rate": 4.824792660723563e-05, "loss": 4.6271, "step": 107008 }, { "epoch": 1.01, "learning_rate": 4.823954065972511e-05, "loss": 4.6301, "step": 107520 }, { "epoch": 1.01, "learning_rate": 4.82311547122146e-05, "loss": 4.6351, "step": 108032 }, { "epoch": 1.01, "learning_rate": 4.8222785143507806e-05, "loss": 4.6169, "step": 108544 }, { "epoch": 1.01, "learning_rate": 4.8214415574801015e-05, "loss": 4.6218, "step": 109056 }, { "epoch": 1.01, "learning_rate": 4.8206029627290495e-05, "loss": 4.6346, "step": 109568 }, { "epoch": 1.01, "learning_rate": 4.8197643679779975e-05, "loss": 4.6163, "step": 110080 }, { "epoch": 1.01, "learning_rate": 4.8189257732269455e-05, "loss": 4.6085, "step": 110592 }, { "epoch": 1.01, "learning_rate": 4.8180888163562664e-05, "loss": 4.6067, "step": 111104 }, { "epoch": 1.01, "learning_rate": 4.8172502216052144e-05, "loss": 4.6063, "step": 111616 }, { "epoch": 1.01, "learning_rate": 4.8164116268541624e-05, "loss": 4.6283, "step": 112128 }, { "epoch": 1.01, "learning_rate": 4.8155730321031104e-05, "loss": 4.6201, "step": 112640 }, { "epoch": 1.01, "learning_rate": 4.8147344373520584e-05, "loss": 4.6057, "step": 113152 }, { "epoch": 1.01, "learning_rate": 4.81389748048138e-05, "loss": 4.61, "step": 113664 }, { "epoch": 1.01, "learning_rate": 4.813058885730328e-05, "loss": 4.6136, "step": 114176 }, { "epoch": 1.01, "learning_rate": 4.812220290979276e-05, "loss": 4.6115, "step": 114688 }, { "epoch": 1.01, "learning_rate": 4.811381696228224e-05, "loss": 4.5961, "step": 115200 }, { "epoch": 1.01, "learning_rate": 4.810544739357545e-05, "loss": 4.6081, "step": 115712 }, { "epoch": 1.01, "learning_rate": 4.809706144606493e-05, "loss": 4.5949, "step": 116224 }, { "epoch": 1.01, "learning_rate": 4.808867549855441e-05, "loss": 4.6039, "step": 116736 }, { "epoch": 1.01, "learning_rate": 4.808028955104389e-05, "loss": 4.5949, "step": 117248 }, { "epoch": 1.01, "learning_rate": 4.80719199823371e-05, "loss": 4.5951, "step": 117760 }, { "epoch": 1.01, "learning_rate": 4.806353403482658e-05, "loss": 4.5975, "step": 118272 }, { "epoch": 1.01, "learning_rate": 4.805514808731606e-05, "loss": 4.5985, "step": 118784 }, { "epoch": 1.01, "learning_rate": 4.804676213980554e-05, "loss": 4.5808, "step": 119296 }, { "epoch": 1.01, "learning_rate": 4.803837619229502e-05, "loss": 4.5902, "step": 119808 }, { "epoch": 1.01, "learning_rate": 4.803000662358823e-05, "loss": 4.5887, "step": 120320 }, { "epoch": 1.01, "learning_rate": 4.802162067607771e-05, "loss": 4.5925, "step": 120832 }, { "epoch": 1.01, "learning_rate": 4.801323472856719e-05, "loss": 4.586, "step": 121344 }, { "epoch": 1.01, "learning_rate": 4.800484878105667e-05, "loss": 4.5727, "step": 121856 }, { "epoch": 1.02, "learning_rate": 4.799646283354615e-05, "loss": 4.583, "step": 122368 }, { "epoch": 1.02, "learning_rate": 4.798807688603563e-05, "loss": 4.5752, "step": 122880 }, { "epoch": 1.02, "learning_rate": 4.797970731732884e-05, "loss": 4.5876, "step": 123392 }, { "epoch": 1.02, "learning_rate": 4.797132136981832e-05, "loss": 4.5671, "step": 123904 }, { "epoch": 1.02, "learning_rate": 4.79629354223078e-05, "loss": 4.5717, "step": 124416 }, { "epoch": 1.02, "learning_rate": 4.795454947479728e-05, "loss": 4.5835, "step": 124928 }, { "epoch": 1.02, "learning_rate": 4.794617990609049e-05, "loss": 4.5721, "step": 125440 }, { "epoch": 1.02, "learning_rate": 4.793779395857997e-05, "loss": 4.5541, "step": 125952 }, { "epoch": 1.02, "learning_rate": 4.792940801106946e-05, "loss": 4.5655, "step": 126464 }, { "epoch": 1.02, "learning_rate": 4.792102206355894e-05, "loss": 4.5702, "step": 126976 }, { "epoch": 1.02, "learning_rate": 4.791263611604842e-05, "loss": 4.5724, "step": 127488 }, { "epoch": 1.02, "learning_rate": 4.790426654734163e-05, "loss": 4.5625, "step": 128000 }, { "epoch": 1.02, "learning_rate": 4.7895896978634836e-05, "loss": 4.5601, "step": 128512 }, { "epoch": 1.02, "learning_rate": 4.7887511031124316e-05, "loss": 4.5552, "step": 129024 }, { "epoch": 1.02, "learning_rate": 4.7879125083613796e-05, "loss": 4.5603, "step": 129536 }, { "epoch": 1.02, "learning_rate": 4.7870739136103276e-05, "loss": 4.5413, "step": 130048 }, { "epoch": 1.02, "learning_rate": 4.7862353188592756e-05, "loss": 4.5502, "step": 130560 }, { "epoch": 1.02, "learning_rate": 4.7853967241082236e-05, "loss": 4.5564, "step": 131072 }, { "epoch": 1.02, "learning_rate": 4.7845581293571715e-05, "loss": 4.5536, "step": 131584 }, { "epoch": 1.02, "learning_rate": 4.7837195346061195e-05, "loss": 4.5373, "step": 132096 }, { "epoch": 1.02, "learning_rate": 4.782884215615814e-05, "loss": 4.5474, "step": 132608 }, { "epoch": 1.02, "learning_rate": 4.782045620864762e-05, "loss": 4.5357, "step": 133120 }, { "epoch": 1.02, "learning_rate": 4.78120702611371e-05, "loss": 4.551, "step": 133632 }, { "epoch": 1.02, "learning_rate": 4.780368431362658e-05, "loss": 4.538, "step": 134144 }, { "epoch": 1.02, "learning_rate": 4.779529836611606e-05, "loss": 4.5305, "step": 134656 }, { "epoch": 1.02, "learning_rate": 4.778691241860554e-05, "loss": 4.5478, "step": 135168 }, { "epoch": 1.02, "learning_rate": 4.777852647109502e-05, "loss": 4.5408, "step": 135680 }, { "epoch": 1.02, "learning_rate": 4.777015690238823e-05, "loss": 4.536, "step": 136192 }, { "epoch": 1.02, "learning_rate": 4.776177095487771e-05, "loss": 4.5379, "step": 136704 }, { "epoch": 1.02, "learning_rate": 4.775338500736719e-05, "loss": 4.5373, "step": 137216 }, { "epoch": 1.02, "learning_rate": 4.774499905985666e-05, "loss": 4.5417, "step": 137728 }, { "epoch": 1.02, "learning_rate": 4.773661311234615e-05, "loss": 4.5177, "step": 138240 }, { "epoch": 1.02, "learning_rate": 4.772822716483563e-05, "loss": 4.5395, "step": 138752 }, { "epoch": 1.02, "learning_rate": 4.771984121732511e-05, "loss": 4.5267, "step": 139264 }, { "epoch": 1.02, "learning_rate": 4.771145526981459e-05, "loss": 4.5308, "step": 139776 }, { "epoch": 1.02, "learning_rate": 4.770306932230407e-05, "loss": 4.5224, "step": 140288 }, { "epoch": 1.02, "learning_rate": 4.769469975359728e-05, "loss": 4.527, "step": 140800 }, { "epoch": 1.02, "learning_rate": 4.768631380608676e-05, "loss": 4.5165, "step": 141312 }, { "epoch": 1.02, "learning_rate": 4.767792785857624e-05, "loss": 4.525, "step": 141824 }, { "epoch": 1.02, "learning_rate": 4.766955828986945e-05, "loss": 4.5027, "step": 142336 }, { "epoch": 1.02, "learning_rate": 4.766117234235893e-05, "loss": 4.5365, "step": 142848 }, { "epoch": 1.02, "learning_rate": 4.765278639484841e-05, "loss": 4.5285, "step": 143360 }, { "epoch": 1.02, "learning_rate": 4.764440044733789e-05, "loss": 4.5216, "step": 143872 }, { "epoch": 1.02, "learning_rate": 4.763601449982737e-05, "loss": 4.5085, "step": 144384 }, { "epoch": 1.02, "learning_rate": 4.762762855231685e-05, "loss": 4.5218, "step": 144896 }, { "epoch": 1.02, "learning_rate": 4.761924260480633e-05, "loss": 4.5228, "step": 145408 }, { "epoch": 1.02, "learning_rate": 4.7610856657295813e-05, "loss": 4.5138, "step": 145920 }, { "epoch": 1.02, "learning_rate": 4.7602470709785293e-05, "loss": 4.5085, "step": 146432 }, { "epoch": 1.02, "learning_rate": 4.7594084762274773e-05, "loss": 4.5184, "step": 146944 }, { "epoch": 1.02, "learning_rate": 4.758571519356798e-05, "loss": 4.5118, "step": 147456 }, { "epoch": 1.02, "learning_rate": 4.757732924605746e-05, "loss": 4.5188, "step": 147968 }, { "epoch": 1.02, "learning_rate": 4.756894329854694e-05, "loss": 4.4935, "step": 148480 }, { "epoch": 1.02, "learning_rate": 4.756055735103642e-05, "loss": 4.51, "step": 148992 }, { "epoch": 1.02, "learning_rate": 4.75521714035259e-05, "loss": 4.5066, "step": 149504 }, { "epoch": 1.02, "learning_rate": 4.7543785456015375e-05, "loss": 4.4998, "step": 150016 }, { "epoch": 1.02, "learning_rate": 4.7535399508504855e-05, "loss": 4.5152, "step": 150528 }, { "epoch": 1.02, "learning_rate": 4.752702993979807e-05, "loss": 4.4959, "step": 151040 }, { "epoch": 1.02, "learning_rate": 4.751864399228755e-05, "loss": 4.504, "step": 151552 }, { "epoch": 1.02, "learning_rate": 4.751025804477703e-05, "loss": 4.5012, "step": 152064 }, { "epoch": 1.02, "learning_rate": 4.750187209726651e-05, "loss": 4.502, "step": 152576 }, { "epoch": 1.03, "eval_loss": 4.516730785369873, "eval_runtime": 330.6418, "eval_samples_per_second": 1154.092, "eval_steps_per_second": 36.066, "step": 152640 }, { "epoch": 0.0, "learning_rate": 4.749348614975599e-05, "loss": 4.4896, "step": 153088 }, { "epoch": 0.0, "learning_rate": 4.748510020224547e-05, "loss": 4.4968, "step": 153600 }, { "epoch": 0.0, "learning_rate": 4.747671425473495e-05, "loss": 4.503, "step": 154112 }, { "epoch": 0.0, "learning_rate": 4.746832830722443e-05, "loss": 4.497, "step": 154624 }, { "epoch": 0.0, "learning_rate": 4.745994235971391e-05, "loss": 4.4973, "step": 155136 }, { "epoch": 0.0, "learning_rate": 4.745155641220339e-05, "loss": 4.4807, "step": 155648 }, { "epoch": 0.0, "learning_rate": 4.744317046469287e-05, "loss": 4.489, "step": 156160 }, { "epoch": 0.0, "learning_rate": 4.743478451718235e-05, "loss": 4.4813, "step": 156672 }, { "epoch": 0.0, "learning_rate": 4.742639856967183e-05, "loss": 4.4915, "step": 157184 }, { "epoch": 0.0, "learning_rate": 4.741801262216131e-05, "loss": 4.4826, "step": 157696 }, { "epoch": 0.0, "learning_rate": 4.740962667465079e-05, "loss": 4.4921, "step": 158208 }, { "epoch": 0.0, "learning_rate": 4.7401240727140264e-05, "loss": 4.492, "step": 158720 }, { "epoch": 0.0, "learning_rate": 4.739285477962975e-05, "loss": 4.4751, "step": 159232 }, { "epoch": 0.0, "learning_rate": 4.738446883211923e-05, "loss": 4.4806, "step": 159744 }, { "epoch": 0.0, "learning_rate": 4.737608288460871e-05, "loss": 4.4626, "step": 160256 }, { "epoch": 0.0, "learning_rate": 4.736769693709819e-05, "loss": 4.4756, "step": 160768 }, { "epoch": 0.0, "learning_rate": 4.735931098958767e-05, "loss": 4.4753, "step": 161280 }, { "epoch": 0.0, "learning_rate": 4.735094142088088e-05, "loss": 4.4683, "step": 161792 }, { "epoch": 0.0, "learning_rate": 4.734255547337036e-05, "loss": 4.468, "step": 162304 }, { "epoch": 0.0, "learning_rate": 4.733416952585984e-05, "loss": 4.4929, "step": 162816 }, { "epoch": 0.0, "learning_rate": 4.732578357834932e-05, "loss": 4.4717, "step": 163328 }, { "epoch": 0.0, "learning_rate": 4.73173976308388e-05, "loss": 4.4712, "step": 163840 }, { "epoch": 0.0, "learning_rate": 4.730902806213201e-05, "loss": 4.4762, "step": 164352 }, { "epoch": 0.0, "learning_rate": 4.730064211462149e-05, "loss": 4.473, "step": 164864 }, { "epoch": 0.0, "learning_rate": 4.729225616711097e-05, "loss": 4.4663, "step": 165376 }, { "epoch": 0.0, "learning_rate": 4.728387021960045e-05, "loss": 4.4575, "step": 165888 }, { "epoch": 0.0, "learning_rate": 4.727548427208993e-05, "loss": 4.4669, "step": 166400 }, { "epoch": 0.0, "learning_rate": 4.7267114703383144e-05, "loss": 4.4624, "step": 166912 }, { "epoch": 0.0, "learning_rate": 4.7258728755872624e-05, "loss": 4.4548, "step": 167424 }, { "epoch": 0.01, "learning_rate": 4.7250342808362104e-05, "loss": 4.4644, "step": 167936 }, { "epoch": 0.01, "learning_rate": 4.7241956860851584e-05, "loss": 4.4542, "step": 168448 }, { "epoch": 0.01, "learning_rate": 4.7233570913341064e-05, "loss": 4.4638, "step": 168960 }, { "epoch": 0.01, "learning_rate": 4.7225184965830544e-05, "loss": 4.4596, "step": 169472 }, { "epoch": 0.01, "learning_rate": 4.721681539712375e-05, "loss": 4.4498, "step": 169984 }, { "epoch": 0.01, "learning_rate": 4.720842944961323e-05, "loss": 4.4623, "step": 170496 }, { "epoch": 0.01, "learning_rate": 4.720004350210271e-05, "loss": 4.4746, "step": 171008 }, { "epoch": 0.01, "learning_rate": 4.719165755459219e-05, "loss": 4.4416, "step": 171520 }, { "epoch": 0.01, "learning_rate": 4.718327160708167e-05, "loss": 4.4574, "step": 172032 }, { "epoch": 0.01, "learning_rate": 4.717488565957115e-05, "loss": 4.4451, "step": 172544 }, { "epoch": 0.01, "learning_rate": 4.716649971206063e-05, "loss": 4.4473, "step": 173056 }, { "epoch": 0.01, "learning_rate": 4.715813014335385e-05, "loss": 4.4441, "step": 173568 }, { "epoch": 0.01, "learning_rate": 4.714974419584333e-05, "loss": 4.459, "step": 174080 }, { "epoch": 0.01, "learning_rate": 4.714135824833281e-05, "loss": 4.4508, "step": 174592 }, { "epoch": 0.01, "learning_rate": 4.713298867962602e-05, "loss": 4.4475, "step": 175104 }, { "epoch": 0.01, "learning_rate": 4.71246027321155e-05, "loss": 4.4489, "step": 175616 }, { "epoch": 0.01, "learning_rate": 4.711621678460498e-05, "loss": 4.4411, "step": 176128 }, { "epoch": 0.01, "learning_rate": 4.710783083709446e-05, "loss": 4.4533, "step": 176640 }, { "epoch": 0.01, "learning_rate": 4.709944488958393e-05, "loss": 4.4337, "step": 177152 }, { "epoch": 0.01, "learning_rate": 4.709105894207341e-05, "loss": 4.4319, "step": 177664 }, { "epoch": 0.01, "learning_rate": 4.708267299456289e-05, "loss": 4.4475, "step": 178176 }, { "epoch": 0.01, "learning_rate": 4.707428704705237e-05, "loss": 4.4452, "step": 178688 }, { "epoch": 0.01, "learning_rate": 4.706590109954185e-05, "loss": 4.4367, "step": 179200 }, { "epoch": 0.01, "learning_rate": 4.705751515203134e-05, "loss": 4.4327, "step": 179712 }, { "epoch": 0.01, "learning_rate": 4.7049145583324546e-05, "loss": 4.441, "step": 180224 }, { "epoch": 0.01, "learning_rate": 4.7040759635814026e-05, "loss": 4.4234, "step": 180736 }, { "epoch": 0.01, "learning_rate": 4.7032373688303506e-05, "loss": 4.4468, "step": 181248 }, { "epoch": 0.01, "learning_rate": 4.7023987740792986e-05, "loss": 4.4291, "step": 181760 }, { "epoch": 0.01, "learning_rate": 4.7015601793282466e-05, "loss": 4.4421, "step": 182272 }, { "epoch": 0.01, "learning_rate": 4.7007232224575675e-05, "loss": 4.4341, "step": 182784 }, { "epoch": 0.01, "learning_rate": 4.6998846277065155e-05, "loss": 4.4246, "step": 183296 }, { "epoch": 0.01, "learning_rate": 4.6990460329554635e-05, "loss": 4.4266, "step": 183808 }, { "epoch": 0.01, "learning_rate": 4.6982090760847844e-05, "loss": 4.4346, "step": 184320 }, { "epoch": 0.01, "learning_rate": 4.6973704813337324e-05, "loss": 4.4147, "step": 184832 }, { "epoch": 0.01, "learning_rate": 4.6965318865826804e-05, "loss": 4.4239, "step": 185344 }, { "epoch": 0.01, "learning_rate": 4.695693291831629e-05, "loss": 4.4323, "step": 185856 }, { "epoch": 0.01, "learning_rate": 4.694854697080577e-05, "loss": 4.4228, "step": 186368 }, { "epoch": 0.01, "learning_rate": 4.694017740209898e-05, "loss": 4.4081, "step": 186880 }, { "epoch": 0.01, "learning_rate": 4.693179145458846e-05, "loss": 4.4107, "step": 187392 }, { "epoch": 0.01, "learning_rate": 4.692342188588167e-05, "loss": 4.4085, "step": 187904 }, { "epoch": 0.01, "learning_rate": 4.691503593837115e-05, "loss": 4.4356, "step": 188416 }, { "epoch": 0.01, "learning_rate": 4.690664999086063e-05, "loss": 4.4241, "step": 188928 }, { "epoch": 0.01, "learning_rate": 4.689826404335011e-05, "loss": 4.4143, "step": 189440 }, { "epoch": 0.01, "learning_rate": 4.688987809583959e-05, "loss": 4.4232, "step": 189952 }, { "epoch": 0.01, "learning_rate": 4.688149214832907e-05, "loss": 4.4203, "step": 190464 }, { "epoch": 0.01, "learning_rate": 4.687310620081855e-05, "loss": 4.4238, "step": 190976 }, { "epoch": 0.01, "learning_rate": 4.686472025330803e-05, "loss": 4.4063, "step": 191488 }, { "epoch": 0.01, "learning_rate": 4.6856350684601245e-05, "loss": 4.4204, "step": 192000 }, { "epoch": 0.01, "learning_rate": 4.6847964737090725e-05, "loss": 4.4128, "step": 192512 }, { "epoch": 0.01, "learning_rate": 4.6839578789580204e-05, "loss": 4.4141, "step": 193024 }, { "epoch": 0.01, "learning_rate": 4.6831192842069684e-05, "loss": 4.4087, "step": 193536 }, { "epoch": 0.01, "learning_rate": 4.6822806894559164e-05, "loss": 4.4122, "step": 194048 }, { "epoch": 0.01, "learning_rate": 4.6814420947048644e-05, "loss": 4.4114, "step": 194560 }, { "epoch": 0.01, "learning_rate": 4.6806051378341853e-05, "loss": 4.416, "step": 195072 }, { "epoch": 0.01, "learning_rate": 4.6797665430831333e-05, "loss": 4.4011, "step": 195584 }, { "epoch": 0.01, "learning_rate": 4.678927948332081e-05, "loss": 4.407, "step": 196096 }, { "epoch": 0.01, "learning_rate": 4.6780893535810287e-05, "loss": 4.405, "step": 196608 }, { "epoch": 0.01, "learning_rate": 4.6772507588299766e-05, "loss": 4.4108, "step": 197120 }, { "epoch": 0.01, "learning_rate": 4.6764121640789246e-05, "loss": 4.4101, "step": 197632 }, { "epoch": 0.01, "learning_rate": 4.6755735693278726e-05, "loss": 4.3914, "step": 198144 }, { "epoch": 0.02, "learning_rate": 4.6747349745768206e-05, "loss": 4.405, "step": 198656 }, { "epoch": 0.02, "learning_rate": 4.673898017706142e-05, "loss": 4.4005, "step": 199168 }, { "epoch": 0.02, "learning_rate": 4.673061060835464e-05, "loss": 4.4095, "step": 199680 }, { "epoch": 0.02, "learning_rate": 4.672222466084412e-05, "loss": 4.3884, "step": 200192 }, { "epoch": 0.02, "learning_rate": 4.671383871333359e-05, "loss": 4.3997, "step": 200704 }, { "epoch": 0.02, "learning_rate": 4.670545276582307e-05, "loss": 4.4112, "step": 201216 }, { "epoch": 0.02, "learning_rate": 4.669706681831255e-05, "loss": 4.3986, "step": 201728 }, { "epoch": 0.02, "learning_rate": 4.668868087080203e-05, "loss": 4.3808, "step": 202240 }, { "epoch": 0.02, "learning_rate": 4.668029492329151e-05, "loss": 4.3931, "step": 202752 }, { "epoch": 0.02, "learning_rate": 4.667190897578099e-05, "loss": 4.4011, "step": 203264 }, { "epoch": 0.02, "learning_rate": 4.66635394070742e-05, "loss": 4.4009, "step": 203776 }, { "epoch": 0.02, "learning_rate": 4.665515345956368e-05, "loss": 4.3951, "step": 204288 }, { "epoch": 0.02, "learning_rate": 4.664676751205316e-05, "loss": 4.3869, "step": 204800 }, { "epoch": 0.02, "learning_rate": 4.663838156454265e-05, "loss": 4.3858, "step": 205312 }, { "epoch": 0.02, "learning_rate": 4.6630011995835856e-05, "loss": 4.3924, "step": 205824 }, { "epoch": 0.02, "learning_rate": 4.6621626048325336e-05, "loss": 4.3776, "step": 206336 }, { "epoch": 0.02, "learning_rate": 4.6613240100814816e-05, "loss": 4.3843, "step": 206848 }, { "epoch": 0.02, "learning_rate": 4.6604870532108025e-05, "loss": 4.3903, "step": 207360 }, { "epoch": 0.02, "learning_rate": 4.6596484584597505e-05, "loss": 4.3872, "step": 207872 }, { "epoch": 0.02, "learning_rate": 4.6588098637086985e-05, "loss": 4.37, "step": 208384 }, { "epoch": 0.02, "learning_rate": 4.6579729068380194e-05, "loss": 4.3873, "step": 208896 }, { "epoch": 0.02, "learning_rate": 4.6571343120869674e-05, "loss": 4.372, "step": 209408 }, { "epoch": 0.02, "learning_rate": 4.6562957173359154e-05, "loss": 4.3851, "step": 209920 }, { "epoch": 0.02, "learning_rate": 4.6554571225848634e-05, "loss": 4.3793, "step": 210432 }, { "epoch": 0.02, "learning_rate": 4.6546185278338114e-05, "loss": 4.3704, "step": 210944 }, { "epoch": 0.02, "learning_rate": 4.65377993308276e-05, "loss": 4.3859, "step": 211456 }, { "epoch": 0.02, "learning_rate": 4.652942976212081e-05, "loss": 4.3835, "step": 211968 }, { "epoch": 0.02, "learning_rate": 4.652104381461029e-05, "loss": 4.3729, "step": 212480 }, { "epoch": 0.02, "learning_rate": 4.651265786709977e-05, "loss": 4.3792, "step": 212992 }, { "epoch": 0.02, "learning_rate": 4.650427191958925e-05, "loss": 4.3796, "step": 213504 }, { "epoch": 0.02, "learning_rate": 4.649588597207873e-05, "loss": 4.38, "step": 214016 }, { "epoch": 0.02, "learning_rate": 4.648750002456821e-05, "loss": 4.3655, "step": 214528 }, { "epoch": 0.02, "learning_rate": 4.647911407705769e-05, "loss": 4.3806, "step": 215040 }, { "epoch": 0.02, "learning_rate": 4.647072812954717e-05, "loss": 4.3719, "step": 215552 }, { "epoch": 0.02, "learning_rate": 4.646234218203665e-05, "loss": 4.374, "step": 216064 }, { "epoch": 0.02, "learning_rate": 4.645397261332986e-05, "loss": 4.3696, "step": 216576 }, { "epoch": 0.02, "learning_rate": 4.644558666581934e-05, "loss": 4.3774, "step": 217088 }, { "epoch": 0.02, "learning_rate": 4.643720071830882e-05, "loss": 4.358, "step": 217600 }, { "epoch": 0.02, "learning_rate": 4.64288147707983e-05, "loss": 4.3723, "step": 218112 }, { "epoch": 0.02, "learning_rate": 4.6420445202091514e-05, "loss": 4.3515, "step": 218624 }, { "epoch": 0.02, "learning_rate": 4.6412059254580994e-05, "loss": 4.3823, "step": 219136 }, { "epoch": 0.02, "learning_rate": 4.6403673307070474e-05, "loss": 4.3735, "step": 219648 }, { "epoch": 0.02, "learning_rate": 4.639528735955995e-05, "loss": 4.3706, "step": 220160 }, { "epoch": 0.02, "learning_rate": 4.638691779085316e-05, "loss": 4.3576, "step": 220672 }, { "epoch": 0.02, "learning_rate": 4.637853184334264e-05, "loss": 4.37, "step": 221184 }, { "epoch": 0.02, "learning_rate": 4.637014589583212e-05, "loss": 4.3752, "step": 221696 }, { "epoch": 0.02, "learning_rate": 4.6361759948321596e-05, "loss": 4.3628, "step": 222208 }, { "epoch": 0.02, "learning_rate": 4.6353374000811076e-05, "loss": 4.3619, "step": 222720 }, { "epoch": 0.02, "learning_rate": 4.6344988053300556e-05, "loss": 4.3697, "step": 223232 }, { "epoch": 0.02, "learning_rate": 4.6336602105790036e-05, "loss": 4.364, "step": 223744 }, { "epoch": 0.02, "learning_rate": 4.632823253708325e-05, "loss": 4.3716, "step": 224256 }, { "epoch": 0.02, "learning_rate": 4.631984658957273e-05, "loss": 4.3515, "step": 224768 }, { "epoch": 0.02, "learning_rate": 4.631146064206221e-05, "loss": 4.3638, "step": 225280 }, { "epoch": 0.02, "learning_rate": 4.630307469455169e-05, "loss": 4.3611, "step": 225792 }, { "epoch": 0.02, "learning_rate": 4.629468874704117e-05, "loss": 4.3539, "step": 226304 }, { "epoch": 0.02, "learning_rate": 4.628630279953065e-05, "loss": 4.3727, "step": 226816 }, { "epoch": 0.02, "learning_rate": 4.627791685202013e-05, "loss": 4.3511, "step": 227328 }, { "epoch": 0.02, "learning_rate": 4.626953090450961e-05, "loss": 4.3601, "step": 227840 }, { "epoch": 0.02, "learning_rate": 4.626116133580282e-05, "loss": 4.3617, "step": 228352 }, { "epoch": 0.02, "learning_rate": 4.625279176709603e-05, "loss": 4.3567, "step": 228864 }, { "epoch": 0.03, "eval_loss": 4.382836818695068, "eval_runtime": 295.0935, "eval_samples_per_second": 1293.119, "eval_steps_per_second": 40.411, "step": 228960 } ], "logging_steps": 512, "max_steps": 3052726, "num_train_epochs": 9223372036854775807, "save_steps": 10, "total_flos": 8.818453998502757e+16, "trial_name": null, "trial_params": null }