{ "best_metric": 4.348172187805176, "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/det-adj-noun/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": 10.8204, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.999161405248948e-05, "loss": 7.5528, "step": 512 }, { "epoch": 0.0, "learning_rate": 4.998322810497896e-05, "loss": 7.0599, "step": 1024 }, { "epoch": 0.0, "learning_rate": 4.997484215746844e-05, "loss": 6.9953, "step": 1536 }, { "epoch": 0.0, "learning_rate": 4.996645620995792e-05, "loss": 6.9468, "step": 2048 }, { "epoch": 0.0, "learning_rate": 4.99580702624474e-05, "loss": 6.9288, "step": 2560 }, { "epoch": 0.0, "learning_rate": 4.994968431493688e-05, "loss": 6.7821, "step": 3072 }, { "epoch": 0.0, "learning_rate": 4.994129836742636e-05, "loss": 6.6636, "step": 3584 }, { "epoch": 0.0, "learning_rate": 4.993291241991584e-05, "loss": 6.5615, "step": 4096 }, { "epoch": 0.0, "learning_rate": 4.992454285120906e-05, "loss": 6.4842, "step": 4608 }, { "epoch": 0.0, "learning_rate": 4.991615690369854e-05, "loss": 6.4163, "step": 5120 }, { "epoch": 0.0, "learning_rate": 4.990777095618801e-05, "loss": 6.3404, "step": 5632 }, { "epoch": 0.0, "learning_rate": 4.989938500867749e-05, "loss": 6.2619, "step": 6144 }, { "epoch": 0.0, "learning_rate": 4.989099906116697e-05, "loss": 6.1988, "step": 6656 }, { "epoch": 0.0, "learning_rate": 4.988261311365645e-05, "loss": 6.1375, "step": 7168 }, { "epoch": 0.0, "learning_rate": 4.987424354494966e-05, "loss": 6.0821, "step": 7680 }, { "epoch": 0.0, "learning_rate": 4.986585759743914e-05, "loss": 6.0391, "step": 8192 }, { "epoch": 0.0, "learning_rate": 4.985747164992862e-05, "loss": 6.0008, "step": 8704 }, { "epoch": 0.0, "learning_rate": 4.98490857024181e-05, "loss": 5.9575, "step": 9216 }, { "epoch": 0.0, "learning_rate": 4.984069975490758e-05, "loss": 5.9175, "step": 9728 }, { "epoch": 0.0, "learning_rate": 4.983231380739706e-05, "loss": 5.884, "step": 10240 }, { "epoch": 0.0, "learning_rate": 4.9823927859886547e-05, "loss": 5.8381, "step": 10752 }, { "epoch": 0.0, "learning_rate": 4.9815541912376026e-05, "loss": 5.8192, "step": 11264 }, { "epoch": 0.0, "learning_rate": 4.9807155964865506e-05, "loss": 5.7822, "step": 11776 }, { "epoch": 0.0, "learning_rate": 4.9798770017354986e-05, "loss": 5.7643, "step": 12288 }, { "epoch": 0.0, "learning_rate": 4.9790384069844466e-05, "loss": 5.7258, "step": 12800 }, { "epoch": 0.0, "learning_rate": 4.9782014501137675e-05, "loss": 5.6998, "step": 13312 }, { "epoch": 0.0, "learning_rate": 4.9773628553627155e-05, "loss": 5.6753, "step": 13824 }, { "epoch": 0.0, "learning_rate": 4.9765242606116635e-05, "loss": 5.6586, "step": 14336 }, { "epoch": 0.0, "learning_rate": 4.9756856658606115e-05, "loss": 5.6354, "step": 14848 }, { "epoch": 0.01, "learning_rate": 4.9748470711095595e-05, "loss": 5.6124, "step": 15360 }, { "epoch": 0.01, "learning_rate": 4.9740101142388804e-05, "loss": 5.6026, "step": 15872 }, { "epoch": 0.01, "learning_rate": 4.9731715194878284e-05, "loss": 5.5661, "step": 16384 }, { "epoch": 0.01, "learning_rate": 4.9723329247367764e-05, "loss": 5.5599, "step": 16896 }, { "epoch": 0.01, "learning_rate": 4.9714943299857244e-05, "loss": 5.5416, "step": 17408 }, { "epoch": 0.01, "learning_rate": 4.9706557352346724e-05, "loss": 5.525, "step": 17920 }, { "epoch": 0.01, "learning_rate": 4.9698171404836204e-05, "loss": 5.5098, "step": 18432 }, { "epoch": 0.01, "learning_rate": 4.968980183612942e-05, "loss": 5.4703, "step": 18944 }, { "epoch": 0.01, "learning_rate": 4.96814158886189e-05, "loss": 5.4769, "step": 19456 }, { "epoch": 0.01, "learning_rate": 4.967302994110837e-05, "loss": 5.4427, "step": 19968 }, { "epoch": 0.01, "learning_rate": 4.966464399359785e-05, "loss": 5.4463, "step": 20480 }, { "epoch": 0.01, "learning_rate": 4.965627442489107e-05, "loss": 5.4259, "step": 20992 }, { "epoch": 0.01, "learning_rate": 4.964788847738054e-05, "loss": 5.4284, "step": 21504 }, { "epoch": 0.01, "learning_rate": 4.963950252987002e-05, "loss": 5.3993, "step": 22016 }, { "epoch": 0.01, "learning_rate": 4.96311165823595e-05, "loss": 5.385, "step": 22528 }, { "epoch": 0.01, "learning_rate": 4.962273063484898e-05, "loss": 5.3903, "step": 23040 }, { "epoch": 0.01, "learning_rate": 4.961434468733847e-05, "loss": 5.3789, "step": 23552 }, { "epoch": 0.01, "learning_rate": 4.960595873982795e-05, "loss": 5.3669, "step": 24064 }, { "epoch": 0.01, "learning_rate": 4.959757279231743e-05, "loss": 5.3498, "step": 24576 }, { "epoch": 0.01, "learning_rate": 4.958920322361064e-05, "loss": 5.3288, "step": 25088 }, { "epoch": 0.01, "learning_rate": 4.958081727610012e-05, "loss": 5.3392, "step": 25600 }, { "epoch": 0.01, "learning_rate": 4.957244770739333e-05, "loss": 5.3211, "step": 26112 }, { "epoch": 0.01, "learning_rate": 4.956406175988281e-05, "loss": 5.3209, "step": 26624 }, { "epoch": 0.01, "learning_rate": 4.955567581237229e-05, "loss": 5.2901, "step": 27136 }, { "epoch": 0.01, "learning_rate": 4.9547306243665496e-05, "loss": 5.2946, "step": 27648 }, { "epoch": 0.01, "learning_rate": 4.9538920296154976e-05, "loss": 5.2708, "step": 28160 }, { "epoch": 0.01, "learning_rate": 4.9530534348644456e-05, "loss": 5.2937, "step": 28672 }, { "epoch": 0.01, "learning_rate": 4.9522148401133936e-05, "loss": 5.256, "step": 29184 }, { "epoch": 0.01, "learning_rate": 4.951376245362342e-05, "loss": 5.2591, "step": 29696 }, { "epoch": 0.01, "learning_rate": 4.95053765061129e-05, "loss": 5.2445, "step": 30208 }, { "epoch": 0.01, "learning_rate": 4.949699055860238e-05, "loss": 5.2331, "step": 30720 }, { "epoch": 0.01, "learning_rate": 4.948860461109186e-05, "loss": 5.2339, "step": 31232 }, { "epoch": 0.01, "learning_rate": 4.948021866358134e-05, "loss": 5.2226, "step": 31744 }, { "epoch": 0.01, "learning_rate": 4.947184909487455e-05, "loss": 5.1971, "step": 32256 }, { "epoch": 0.01, "learning_rate": 4.946346314736403e-05, "loss": 5.2012, "step": 32768 }, { "epoch": 0.01, "learning_rate": 4.945507719985351e-05, "loss": 5.2126, "step": 33280 }, { "epoch": 0.01, "learning_rate": 4.944669125234299e-05, "loss": 5.1889, "step": 33792 }, { "epoch": 0.01, "learning_rate": 4.943830530483247e-05, "loss": 5.1798, "step": 34304 }, { "epoch": 0.01, "learning_rate": 4.942991935732195e-05, "loss": 5.1606, "step": 34816 }, { "epoch": 0.01, "learning_rate": 4.942154978861516e-05, "loss": 5.1633, "step": 35328 }, { "epoch": 0.01, "learning_rate": 4.941316384110464e-05, "loss": 5.1802, "step": 35840 }, { "epoch": 0.01, "learning_rate": 4.940477789359412e-05, "loss": 5.1634, "step": 36352 }, { "epoch": 0.01, "learning_rate": 4.93963919460836e-05, "loss": 5.1502, "step": 36864 }, { "epoch": 0.01, "learning_rate": 4.938800599857309e-05, "loss": 5.1533, "step": 37376 }, { "epoch": 0.01, "learning_rate": 4.937962005106256e-05, "loss": 5.1509, "step": 37888 }, { "epoch": 0.01, "learning_rate": 4.937123410355204e-05, "loss": 5.1326, "step": 38400 }, { "epoch": 0.01, "learning_rate": 4.936284815604152e-05, "loss": 5.1162, "step": 38912 }, { "epoch": 0.01, "learning_rate": 4.935447858733473e-05, "loss": 5.1218, "step": 39424 }, { "epoch": 0.01, "learning_rate": 4.934609263982421e-05, "loss": 5.1021, "step": 39936 }, { "epoch": 0.01, "learning_rate": 4.933770669231369e-05, "loss": 5.1041, "step": 40448 }, { "epoch": 0.01, "learning_rate": 4.932932074480317e-05, "loss": 5.1089, "step": 40960 }, { "epoch": 0.01, "learning_rate": 4.932095117609638e-05, "loss": 5.0958, "step": 41472 }, { "epoch": 0.01, "learning_rate": 4.931256522858586e-05, "loss": 5.0974, "step": 41984 }, { "epoch": 0.01, "learning_rate": 4.930417928107534e-05, "loss": 5.0868, "step": 42496 }, { "epoch": 0.01, "learning_rate": 4.9295793333564825e-05, "loss": 5.0668, "step": 43008 }, { "epoch": 0.01, "learning_rate": 4.9287407386054305e-05, "loss": 5.0692, "step": 43520 }, { "epoch": 0.01, "learning_rate": 4.9279021438543785e-05, "loss": 5.0661, "step": 44032 }, { "epoch": 0.01, "learning_rate": 4.9270651869836994e-05, "loss": 5.0718, "step": 44544 }, { "epoch": 0.01, "learning_rate": 4.9262265922326474e-05, "loss": 5.0522, "step": 45056 }, { "epoch": 0.01, "learning_rate": 4.9253879974815954e-05, "loss": 5.0525, "step": 45568 }, { "epoch": 0.02, "learning_rate": 4.9245494027305433e-05, "loss": 5.0502, "step": 46080 }, { "epoch": 0.02, "learning_rate": 4.9237108079794913e-05, "loss": 5.0426, "step": 46592 }, { "epoch": 0.02, "learning_rate": 4.922873851108812e-05, "loss": 5.0402, "step": 47104 }, { "epoch": 0.02, "learning_rate": 4.92203525635776e-05, "loss": 5.0289, "step": 47616 }, { "epoch": 0.02, "learning_rate": 4.921196661606708e-05, "loss": 5.016, "step": 48128 }, { "epoch": 0.02, "learning_rate": 4.920358066855656e-05, "loss": 5.0246, "step": 48640 }, { "epoch": 0.02, "learning_rate": 4.919519472104604e-05, "loss": 5.0119, "step": 49152 }, { "epoch": 0.02, "learning_rate": 4.918682515233926e-05, "loss": 5.0061, "step": 49664 }, { "epoch": 0.02, "learning_rate": 4.917843920482874e-05, "loss": 4.9987, "step": 50176 }, { "epoch": 0.02, "learning_rate": 4.917005325731822e-05, "loss": 4.9984, "step": 50688 }, { "epoch": 0.02, "learning_rate": 4.91616673098077e-05, "loss": 5.002, "step": 51200 }, { "epoch": 0.02, "learning_rate": 4.915329774110091e-05, "loss": 4.9901, "step": 51712 }, { "epoch": 0.02, "learning_rate": 4.914491179359039e-05, "loss": 4.9799, "step": 52224 }, { "epoch": 0.02, "learning_rate": 4.913652584607987e-05, "loss": 4.9762, "step": 52736 }, { "epoch": 0.02, "learning_rate": 4.912813989856935e-05, "loss": 4.9685, "step": 53248 }, { "epoch": 0.02, "learning_rate": 4.911975395105883e-05, "loss": 4.9551, "step": 53760 }, { "epoch": 0.02, "learning_rate": 4.9111384382352036e-05, "loss": 4.9509, "step": 54272 }, { "epoch": 0.02, "learning_rate": 4.9102998434841516e-05, "loss": 4.9552, "step": 54784 }, { "epoch": 0.02, "learning_rate": 4.9094612487330996e-05, "loss": 4.9523, "step": 55296 }, { "epoch": 0.02, "learning_rate": 4.9086226539820476e-05, "loss": 4.9392, "step": 55808 }, { "epoch": 0.02, "learning_rate": 4.907784059230996e-05, "loss": 4.9369, "step": 56320 }, { "epoch": 0.02, "learning_rate": 4.906945464479944e-05, "loss": 4.9364, "step": 56832 }, { "epoch": 0.02, "learning_rate": 4.906106869728892e-05, "loss": 4.9418, "step": 57344 }, { "epoch": 0.02, "learning_rate": 4.905269912858213e-05, "loss": 4.9357, "step": 57856 }, { "epoch": 0.02, "learning_rate": 4.904431318107161e-05, "loss": 4.9279, "step": 58368 }, { "epoch": 0.02, "learning_rate": 4.903592723356109e-05, "loss": 4.9264, "step": 58880 }, { "epoch": 0.02, "learning_rate": 4.9027541286050565e-05, "loss": 4.9251, "step": 59392 }, { "epoch": 0.02, "learning_rate": 4.9019155338540045e-05, "loss": 4.9128, "step": 59904 }, { "epoch": 0.02, "learning_rate": 4.9010769391029525e-05, "loss": 4.9065, "step": 60416 }, { "epoch": 0.02, "learning_rate": 4.9002383443519005e-05, "loss": 4.9114, "step": 60928 }, { "epoch": 0.02, "learning_rate": 4.8994013874812214e-05, "loss": 4.91, "step": 61440 }, { "epoch": 0.02, "learning_rate": 4.89856279273017e-05, "loss": 4.8883, "step": 61952 }, { "epoch": 0.02, "learning_rate": 4.897724197979118e-05, "loss": 4.8903, "step": 62464 }, { "epoch": 0.02, "learning_rate": 4.896885603228066e-05, "loss": 4.8877, "step": 62976 }, { "epoch": 0.02, "learning_rate": 4.896047008477014e-05, "loss": 4.8764, "step": 63488 }, { "epoch": 0.02, "learning_rate": 4.895208413725962e-05, "loss": 4.88, "step": 64000 }, { "epoch": 0.02, "learning_rate": 4.89436981897491e-05, "loss": 4.8853, "step": 64512 }, { "epoch": 0.02, "learning_rate": 4.893531224223858e-05, "loss": 4.8682, "step": 65024 }, { "epoch": 0.02, "learning_rate": 4.892694267353179e-05, "loss": 4.8702, "step": 65536 }, { "epoch": 0.02, "learning_rate": 4.891855672602127e-05, "loss": 4.8594, "step": 66048 }, { "epoch": 0.02, "learning_rate": 4.891018715731448e-05, "loss": 4.874, "step": 66560 }, { "epoch": 0.02, "learning_rate": 4.890180120980396e-05, "loss": 4.8653, "step": 67072 }, { "epoch": 0.02, "learning_rate": 4.889341526229344e-05, "loss": 4.856, "step": 67584 }, { "epoch": 0.02, "learning_rate": 4.888502931478292e-05, "loss": 4.8532, "step": 68096 }, { "epoch": 0.02, "learning_rate": 4.88766433672724e-05, "loss": 4.8541, "step": 68608 }, { "epoch": 0.02, "learning_rate": 4.886825741976188e-05, "loss": 4.8524, "step": 69120 }, { "epoch": 0.02, "learning_rate": 4.8859871472251365e-05, "loss": 4.8423, "step": 69632 }, { "epoch": 0.02, "learning_rate": 4.8851501903544574e-05, "loss": 4.8403, "step": 70144 }, { "epoch": 0.02, "learning_rate": 4.8843115956034054e-05, "loss": 4.8416, "step": 70656 }, { "epoch": 0.02, "learning_rate": 4.883474638732726e-05, "loss": 4.8371, "step": 71168 }, { "epoch": 0.02, "learning_rate": 4.882636043981674e-05, "loss": 4.8359, "step": 71680 }, { "epoch": 0.02, "learning_rate": 4.881797449230622e-05, "loss": 4.816, "step": 72192 }, { "epoch": 0.02, "learning_rate": 4.88095885447957e-05, "loss": 4.8273, "step": 72704 }, { "epoch": 0.02, "learning_rate": 4.880120259728518e-05, "loss": 4.8126, "step": 73216 }, { "epoch": 0.02, "learning_rate": 4.879281664977466e-05, "loss": 4.8023, "step": 73728 }, { "epoch": 0.02, "learning_rate": 4.878443070226414e-05, "loss": 4.8209, "step": 74240 }, { "epoch": 0.02, "learning_rate": 4.877604475475362e-05, "loss": 4.8096, "step": 74752 }, { "epoch": 0.02, "learning_rate": 4.87676588072431e-05, "loss": 4.8104, "step": 75264 }, { "epoch": 0.02, "learning_rate": 4.875927285973258e-05, "loss": 4.8118, "step": 75776 }, { "epoch": 0.02, "learning_rate": 4.875088691222206e-05, "loss": 4.7911, "step": 76288 }, { "epoch": 0.03, "eval_loss": 4.767621994018555, "eval_runtime": 290.1884, "eval_samples_per_second": 1314.977, "eval_steps_per_second": 41.094, "step": 76320 }, { "epoch": 1.0, "learning_rate": 4.874251734351528e-05, "loss": 4.7945, "step": 76800 }, { "epoch": 1.0, "learning_rate": 4.873413139600475e-05, "loss": 4.7912, "step": 77312 }, { "epoch": 1.0, "learning_rate": 4.872574544849423e-05, "loss": 4.807, "step": 77824 }, { "epoch": 1.0, "learning_rate": 4.871735950098371e-05, "loss": 4.7866, "step": 78336 }, { "epoch": 1.0, "learning_rate": 4.870897355347319e-05, "loss": 4.7944, "step": 78848 }, { "epoch": 1.0, "learning_rate": 4.870058760596267e-05, "loss": 4.7739, "step": 79360 }, { "epoch": 1.0, "learning_rate": 4.869220165845215e-05, "loss": 4.78, "step": 79872 }, { "epoch": 1.0, "learning_rate": 4.868381571094163e-05, "loss": 4.7678, "step": 80384 }, { "epoch": 1.0, "learning_rate": 4.867544614223484e-05, "loss": 4.7741, "step": 80896 }, { "epoch": 1.0, "learning_rate": 4.8667076573528056e-05, "loss": 4.7741, "step": 81408 }, { "epoch": 1.0, "learning_rate": 4.8658690626017536e-05, "loss": 4.7761, "step": 81920 }, { "epoch": 1.0, "learning_rate": 4.8650304678507016e-05, "loss": 4.7763, "step": 82432 }, { "epoch": 1.0, "learning_rate": 4.8641918730996496e-05, "loss": 4.7559, "step": 82944 }, { "epoch": 1.0, "learning_rate": 4.8633532783485976e-05, "loss": 4.7565, "step": 83456 }, { "epoch": 1.0, "learning_rate": 4.8625146835975456e-05, "loss": 4.7519, "step": 83968 }, { "epoch": 1.0, "learning_rate": 4.8616760888464936e-05, "loss": 4.7441, "step": 84480 }, { "epoch": 1.0, "learning_rate": 4.8608391319758145e-05, "loss": 4.7567, "step": 84992 }, { "epoch": 1.0, "learning_rate": 4.8600005372247625e-05, "loss": 4.7417, "step": 85504 }, { "epoch": 1.0, "learning_rate": 4.8591619424737105e-05, "loss": 4.7409, "step": 86016 }, { "epoch": 1.0, "learning_rate": 4.8583233477226585e-05, "loss": 4.76, "step": 86528 }, { "epoch": 1.0, "learning_rate": 4.8574847529716065e-05, "loss": 4.7387, "step": 87040 }, { "epoch": 1.0, "learning_rate": 4.8566461582205545e-05, "loss": 4.7389, "step": 87552 }, { "epoch": 1.0, "learning_rate": 4.8558092013498754e-05, "loss": 4.7365, "step": 88064 }, { "epoch": 1.0, "learning_rate": 4.854970606598824e-05, "loss": 4.7422, "step": 88576 }, { "epoch": 1.0, "learning_rate": 4.854132011847772e-05, "loss": 4.7294, "step": 89088 }, { "epoch": 1.0, "learning_rate": 4.85329341709672e-05, "loss": 4.7239, "step": 89600 }, { "epoch": 1.0, "learning_rate": 4.852454822345668e-05, "loss": 4.7181, "step": 90112 }, { "epoch": 1.0, "learning_rate": 4.851616227594616e-05, "loss": 4.7271, "step": 90624 }, { "epoch": 1.0, "learning_rate": 4.850777632843564e-05, "loss": 4.7098, "step": 91136 }, { "epoch": 1.01, "learning_rate": 4.8499390380925114e-05, "loss": 4.7163, "step": 91648 }, { "epoch": 1.01, "learning_rate": 4.849102081221833e-05, "loss": 4.7166, "step": 92160 }, { "epoch": 1.01, "learning_rate": 4.848263486470781e-05, "loss": 4.7107, "step": 92672 }, { "epoch": 1.01, "learning_rate": 4.847424891719729e-05, "loss": 4.7168, "step": 93184 }, { "epoch": 1.01, "learning_rate": 4.846586296968676e-05, "loss": 4.7051, "step": 93696 }, { "epoch": 1.01, "learning_rate": 4.845749340097998e-05, "loss": 4.706, "step": 94208 }, { "epoch": 1.01, "learning_rate": 4.844910745346946e-05, "loss": 4.7077, "step": 94720 }, { "epoch": 1.01, "learning_rate": 4.844072150595894e-05, "loss": 4.6833, "step": 95232 }, { "epoch": 1.01, "learning_rate": 4.843233555844842e-05, "loss": 4.6945, "step": 95744 }, { "epoch": 1.01, "learning_rate": 4.84239496109379e-05, "loss": 4.6777, "step": 96256 }, { "epoch": 1.01, "learning_rate": 4.8415580042231114e-05, "loss": 4.6894, "step": 96768 }, { "epoch": 1.01, "learning_rate": 4.840719409472059e-05, "loss": 4.6893, "step": 97280 }, { "epoch": 1.01, "learning_rate": 4.839880814721007e-05, "loss": 4.6933, "step": 97792 }, { "epoch": 1.01, "learning_rate": 4.839042219969955e-05, "loss": 4.6832, "step": 98304 }, { "epoch": 1.01, "learning_rate": 4.838203625218903e-05, "loss": 4.6821, "step": 98816 }, { "epoch": 1.01, "learning_rate": 4.8373666683482236e-05, "loss": 4.688, "step": 99328 }, { "epoch": 1.01, "learning_rate": 4.8365280735971716e-05, "loss": 4.6818, "step": 99840 }, { "epoch": 1.01, "learning_rate": 4.8356894788461196e-05, "loss": 4.6783, "step": 100352 }, { "epoch": 1.01, "learning_rate": 4.8348508840950676e-05, "loss": 4.6702, "step": 100864 }, { "epoch": 1.01, "learning_rate": 4.834013927224389e-05, "loss": 4.658, "step": 101376 }, { "epoch": 1.01, "learning_rate": 4.833175332473337e-05, "loss": 4.6753, "step": 101888 }, { "epoch": 1.01, "learning_rate": 4.832336737722285e-05, "loss": 4.6692, "step": 102400 }, { "epoch": 1.01, "learning_rate": 4.831498142971233e-05, "loss": 4.6684, "step": 102912 }, { "epoch": 1.01, "learning_rate": 4.830659548220181e-05, "loss": 4.6562, "step": 103424 }, { "epoch": 1.01, "learning_rate": 4.829822591349502e-05, "loss": 4.6629, "step": 103936 }, { "epoch": 1.01, "learning_rate": 4.82898399659845e-05, "loss": 4.6472, "step": 104448 }, { "epoch": 1.01, "learning_rate": 4.828145401847398e-05, "loss": 4.6685, "step": 104960 }, { "epoch": 1.01, "learning_rate": 4.827306807096346e-05, "loss": 4.6491, "step": 105472 }, { "epoch": 1.01, "learning_rate": 4.826468212345294e-05, "loss": 4.6577, "step": 105984 }, { "epoch": 1.01, "learning_rate": 4.825631255474615e-05, "loss": 4.6481, "step": 106496 }, { "epoch": 1.01, "learning_rate": 4.824792660723563e-05, "loss": 4.6395, "step": 107008 }, { "epoch": 1.01, "learning_rate": 4.823954065972511e-05, "loss": 4.643, "step": 107520 }, { "epoch": 1.01, "learning_rate": 4.82311547122146e-05, "loss": 4.6481, "step": 108032 }, { "epoch": 1.01, "learning_rate": 4.8222768764704077e-05, "loss": 4.6225, "step": 108544 }, { "epoch": 1.01, "learning_rate": 4.8214382817193557e-05, "loss": 4.631, "step": 109056 }, { "epoch": 1.01, "learning_rate": 4.8206013248486766e-05, "loss": 4.6438, "step": 109568 }, { "epoch": 1.01, "learning_rate": 4.8197627300976246e-05, "loss": 4.6309, "step": 110080 }, { "epoch": 1.01, "learning_rate": 4.8189241353465726e-05, "loss": 4.6254, "step": 110592 }, { "epoch": 1.01, "learning_rate": 4.8180855405955206e-05, "loss": 4.6214, "step": 111104 }, { "epoch": 1.01, "learning_rate": 4.8172469458444685e-05, "loss": 4.6213, "step": 111616 }, { "epoch": 1.01, "learning_rate": 4.8164083510934165e-05, "loss": 4.6422, "step": 112128 }, { "epoch": 1.01, "learning_rate": 4.8155697563423645e-05, "loss": 4.6325, "step": 112640 }, { "epoch": 1.01, "learning_rate": 4.8147311615913125e-05, "loss": 4.6245, "step": 113152 }, { "epoch": 1.01, "learning_rate": 4.8138942047206334e-05, "loss": 4.6295, "step": 113664 }, { "epoch": 1.01, "learning_rate": 4.813057247849955e-05, "loss": 4.6311, "step": 114176 }, { "epoch": 1.01, "learning_rate": 4.812218653098903e-05, "loss": 4.624, "step": 114688 }, { "epoch": 1.01, "learning_rate": 4.811380058347851e-05, "loss": 4.6068, "step": 115200 }, { "epoch": 1.01, "learning_rate": 4.810541463596799e-05, "loss": 4.6206, "step": 115712 }, { "epoch": 1.01, "learning_rate": 4.809702868845747e-05, "loss": 4.6111, "step": 116224 }, { "epoch": 1.01, "learning_rate": 4.808865911975068e-05, "loss": 4.6071, "step": 116736 }, { "epoch": 1.01, "learning_rate": 4.808027317224016e-05, "loss": 4.6199, "step": 117248 }, { "epoch": 1.01, "learning_rate": 4.807190360353337e-05, "loss": 4.6081, "step": 117760 }, { "epoch": 1.01, "learning_rate": 4.806351765602285e-05, "loss": 4.6185, "step": 118272 }, { "epoch": 1.01, "learning_rate": 4.805513170851233e-05, "loss": 4.6115, "step": 118784 }, { "epoch": 1.01, "learning_rate": 4.804674576100181e-05, "loss": 4.5897, "step": 119296 }, { "epoch": 1.01, "learning_rate": 4.803835981349129e-05, "loss": 4.6042, "step": 119808 }, { "epoch": 1.01, "learning_rate": 4.802997386598077e-05, "loss": 4.5977, "step": 120320 }, { "epoch": 1.01, "learning_rate": 4.8021604297273984e-05, "loss": 4.6109, "step": 120832 }, { "epoch": 1.01, "learning_rate": 4.8013218349763464e-05, "loss": 4.5995, "step": 121344 }, { "epoch": 1.01, "learning_rate": 4.8004832402252944e-05, "loss": 4.6026, "step": 121856 }, { "epoch": 1.02, "learning_rate": 4.7996446454742424e-05, "loss": 4.6004, "step": 122368 }, { "epoch": 1.02, "learning_rate": 4.798807688603563e-05, "loss": 4.5972, "step": 122880 }, { "epoch": 1.02, "learning_rate": 4.797969093852511e-05, "loss": 4.5997, "step": 123392 }, { "epoch": 1.02, "learning_rate": 4.797130499101459e-05, "loss": 4.5926, "step": 123904 }, { "epoch": 1.02, "learning_rate": 4.796291904350407e-05, "loss": 4.5803, "step": 124416 }, { "epoch": 1.02, "learning_rate": 4.7954533095993546e-05, "loss": 4.592, "step": 124928 }, { "epoch": 1.02, "learning_rate": 4.7946147148483026e-05, "loss": 4.5898, "step": 125440 }, { "epoch": 1.02, "learning_rate": 4.7937761200972506e-05, "loss": 4.5813, "step": 125952 }, { "epoch": 1.02, "learning_rate": 4.7929375253461986e-05, "loss": 4.5877, "step": 126464 }, { "epoch": 1.02, "learning_rate": 4.792098930595147e-05, "loss": 4.5848, "step": 126976 }, { "epoch": 1.02, "learning_rate": 4.791260335844095e-05, "loss": 4.5954, "step": 127488 }, { "epoch": 1.02, "learning_rate": 4.790421741093043e-05, "loss": 4.5799, "step": 128000 }, { "epoch": 1.02, "learning_rate": 4.789584784222364e-05, "loss": 4.5726, "step": 128512 }, { "epoch": 1.02, "learning_rate": 4.788746189471312e-05, "loss": 4.5754, "step": 129024 }, { "epoch": 1.02, "learning_rate": 4.78790759472026e-05, "loss": 4.5731, "step": 129536 }, { "epoch": 1.02, "learning_rate": 4.787068999969208e-05, "loss": 4.5542, "step": 130048 }, { "epoch": 1.02, "learning_rate": 4.786232043098529e-05, "loss": 4.5696, "step": 130560 }, { "epoch": 1.02, "learning_rate": 4.785393448347477e-05, "loss": 4.5659, "step": 131072 }, { "epoch": 1.02, "learning_rate": 4.784554853596425e-05, "loss": 4.5667, "step": 131584 }, { "epoch": 1.02, "learning_rate": 4.783716258845373e-05, "loss": 4.5606, "step": 132096 }, { "epoch": 1.02, "learning_rate": 4.782877664094321e-05, "loss": 4.552, "step": 132608 }, { "epoch": 1.02, "learning_rate": 4.782040707223642e-05, "loss": 4.5597, "step": 133120 }, { "epoch": 1.02, "learning_rate": 4.7812021124725906e-05, "loss": 4.5707, "step": 133632 }, { "epoch": 1.02, "learning_rate": 4.7803635177215386e-05, "loss": 4.5639, "step": 134144 }, { "epoch": 1.02, "learning_rate": 4.7795249229704866e-05, "loss": 4.5629, "step": 134656 }, { "epoch": 1.02, "learning_rate": 4.7786863282194346e-05, "loss": 4.5598, "step": 135168 }, { "epoch": 1.02, "learning_rate": 4.7778493713487555e-05, "loss": 4.57, "step": 135680 }, { "epoch": 1.02, "learning_rate": 4.7770107765977035e-05, "loss": 4.5527, "step": 136192 }, { "epoch": 1.02, "learning_rate": 4.7761721818466515e-05, "loss": 4.5498, "step": 136704 }, { "epoch": 1.02, "learning_rate": 4.7753335870955995e-05, "loss": 4.556, "step": 137216 }, { "epoch": 1.02, "learning_rate": 4.7744949923445475e-05, "loss": 4.558, "step": 137728 }, { "epoch": 1.02, "learning_rate": 4.7736563975934955e-05, "loss": 4.5402, "step": 138240 }, { "epoch": 1.02, "learning_rate": 4.7728194407228164e-05, "loss": 4.5427, "step": 138752 }, { "epoch": 1.02, "learning_rate": 4.7719808459717644e-05, "loss": 4.5465, "step": 139264 }, { "epoch": 1.02, "learning_rate": 4.7711422512207124e-05, "loss": 4.5389, "step": 139776 }, { "epoch": 1.02, "learning_rate": 4.7703036564696604e-05, "loss": 4.5434, "step": 140288 }, { "epoch": 1.02, "learning_rate": 4.7694650617186084e-05, "loss": 4.5513, "step": 140800 }, { "epoch": 1.02, "learning_rate": 4.7686264669675564e-05, "loss": 4.5296, "step": 141312 }, { "epoch": 1.02, "learning_rate": 4.7677878722165044e-05, "loss": 4.5412, "step": 141824 }, { "epoch": 1.02, "learning_rate": 4.766950915345826e-05, "loss": 4.5298, "step": 142336 }, { "epoch": 1.02, "learning_rate": 4.766113958475147e-05, "loss": 4.553, "step": 142848 }, { "epoch": 1.02, "learning_rate": 4.765275363724095e-05, "loss": 4.5437, "step": 143360 }, { "epoch": 1.02, "learning_rate": 4.764436768973043e-05, "loss": 4.5327, "step": 143872 }, { "epoch": 1.02, "learning_rate": 4.76359817422199e-05, "loss": 4.5354, "step": 144384 }, { "epoch": 1.02, "learning_rate": 4.762759579470938e-05, "loss": 4.5334, "step": 144896 }, { "epoch": 1.02, "learning_rate": 4.761920984719886e-05, "loss": 4.5409, "step": 145408 }, { "epoch": 1.02, "learning_rate": 4.761082389968834e-05, "loss": 4.5354, "step": 145920 }, { "epoch": 1.02, "learning_rate": 4.760245433098156e-05, "loss": 4.5315, "step": 146432 }, { "epoch": 1.02, "learning_rate": 4.759406838347104e-05, "loss": 4.5306, "step": 146944 }, { "epoch": 1.02, "learning_rate": 4.758569881476425e-05, "loss": 4.5277, "step": 147456 }, { "epoch": 1.02, "learning_rate": 4.757732924605746e-05, "loss": 4.5314, "step": 147968 }, { "epoch": 1.02, "learning_rate": 4.756894329854694e-05, "loss": 4.5165, "step": 148480 }, { "epoch": 1.02, "learning_rate": 4.756055735103642e-05, "loss": 4.5267, "step": 148992 }, { "epoch": 1.02, "learning_rate": 4.75521714035259e-05, "loss": 4.5146, "step": 149504 }, { "epoch": 1.02, "learning_rate": 4.7543785456015375e-05, "loss": 4.5082, "step": 150016 }, { "epoch": 1.02, "learning_rate": 4.7535399508504855e-05, "loss": 4.5279, "step": 150528 }, { "epoch": 1.02, "learning_rate": 4.7527013560994335e-05, "loss": 4.5164, "step": 151040 }, { "epoch": 1.02, "learning_rate": 4.7518627613483815e-05, "loss": 4.5186, "step": 151552 }, { "epoch": 1.02, "learning_rate": 4.7510241665973295e-05, "loss": 4.522, "step": 152064 }, { "epoch": 1.02, "learning_rate": 4.750185571846278e-05, "loss": 4.5068, "step": 152576 }, { "epoch": 1.03, "eval_loss": 4.4849019050598145, "eval_runtime": 290.2875, "eval_samples_per_second": 1314.528, "eval_steps_per_second": 41.08, "step": 152640 }, { "epoch": 0.0, "learning_rate": 4.749348614975599e-05, "loss": 4.512, "step": 153088 }, { "epoch": 0.0, "learning_rate": 4.748510020224547e-05, "loss": 4.5032, "step": 153600 }, { "epoch": 0.0, "learning_rate": 4.747671425473495e-05, "loss": 4.5266, "step": 154112 }, { "epoch": 0.0, "learning_rate": 4.746832830722443e-05, "loss": 4.5037, "step": 154624 }, { "epoch": 0.0, "learning_rate": 4.745994235971391e-05, "loss": 4.5192, "step": 155136 }, { "epoch": 0.0, "learning_rate": 4.745155641220339e-05, "loss": 4.4993, "step": 155648 }, { "epoch": 0.0, "learning_rate": 4.744317046469287e-05, "loss": 4.5034, "step": 156160 }, { "epoch": 0.0, "learning_rate": 4.743478451718235e-05, "loss": 4.4941, "step": 156672 }, { "epoch": 0.0, "learning_rate": 4.742639856967183e-05, "loss": 4.5058, "step": 157184 }, { "epoch": 0.0, "learning_rate": 4.741801262216131e-05, "loss": 4.5045, "step": 157696 }, { "epoch": 0.0, "learning_rate": 4.740962667465079e-05, "loss": 4.5072, "step": 158208 }, { "epoch": 0.0, "learning_rate": 4.7401240727140264e-05, "loss": 4.51, "step": 158720 }, { "epoch": 0.0, "learning_rate": 4.739287115843348e-05, "loss": 4.4916, "step": 159232 }, { "epoch": 0.0, "learning_rate": 4.7384485210922966e-05, "loss": 4.4956, "step": 159744 }, { "epoch": 0.0, "learning_rate": 4.7376099263412446e-05, "loss": 4.4894, "step": 160256 }, { "epoch": 0.0, "learning_rate": 4.736771331590192e-05, "loss": 4.4864, "step": 160768 }, { "epoch": 0.0, "learning_rate": 4.73593273683914e-05, "loss": 4.4954, "step": 161280 }, { "epoch": 0.0, "learning_rate": 4.735094142088088e-05, "loss": 4.4873, "step": 161792 }, { "epoch": 0.0, "learning_rate": 4.734255547337036e-05, "loss": 4.4851, "step": 162304 }, { "epoch": 0.0, "learning_rate": 4.733416952585984e-05, "loss": 4.5013, "step": 162816 }, { "epoch": 0.0, "learning_rate": 4.732578357834932e-05, "loss": 4.4868, "step": 163328 }, { "epoch": 0.0, "learning_rate": 4.73173976308388e-05, "loss": 4.4878, "step": 163840 }, { "epoch": 0.0, "learning_rate": 4.730901168332828e-05, "loss": 4.4882, "step": 164352 }, { "epoch": 0.0, "learning_rate": 4.730064211462149e-05, "loss": 4.496, "step": 164864 }, { "epoch": 0.0, "learning_rate": 4.729225616711097e-05, "loss": 4.4818, "step": 165376 }, { "epoch": 0.0, "learning_rate": 4.728387021960045e-05, "loss": 4.481, "step": 165888 }, { "epoch": 0.0, "learning_rate": 4.727548427208993e-05, "loss": 4.4774, "step": 166400 }, { "epoch": 0.0, "learning_rate": 4.7267098324579415e-05, "loss": 4.4808, "step": 166912 }, { "epoch": 0.0, "learning_rate": 4.7258728755872624e-05, "loss": 4.4706, "step": 167424 }, { "epoch": 0.01, "learning_rate": 4.7250342808362104e-05, "loss": 4.4764, "step": 167936 }, { "epoch": 0.01, "learning_rate": 4.7241956860851584e-05, "loss": 4.4777, "step": 168448 }, { "epoch": 0.01, "learning_rate": 4.7233570913341064e-05, "loss": 4.4733, "step": 168960 }, { "epoch": 0.01, "learning_rate": 4.722523410224174e-05, "loss": 4.4828, "step": 169472 }, { "epoch": 0.01, "learning_rate": 4.721684815473121e-05, "loss": 4.4691, "step": 169984 }, { "epoch": 0.01, "learning_rate": 4.720846220722069e-05, "loss": 4.4718, "step": 170496 }, { "epoch": 0.01, "learning_rate": 4.720007625971017e-05, "loss": 4.4725, "step": 171008 }, { "epoch": 0.01, "learning_rate": 4.719169031219965e-05, "loss": 4.4557, "step": 171520 }, { "epoch": 0.01, "learning_rate": 4.718330436468914e-05, "loss": 4.4602, "step": 172032 }, { "epoch": 0.01, "learning_rate": 4.717491841717862e-05, "loss": 4.453, "step": 172544 }, { "epoch": 0.01, "learning_rate": 4.71665324696681e-05, "loss": 4.4609, "step": 173056 }, { "epoch": 0.01, "learning_rate": 4.715814652215758e-05, "loss": 4.4673, "step": 173568 }, { "epoch": 0.01, "learning_rate": 4.714976057464706e-05, "loss": 4.4679, "step": 174080 }, { "epoch": 0.01, "learning_rate": 4.714137462713654e-05, "loss": 4.4597, "step": 174592 }, { "epoch": 0.01, "learning_rate": 4.713298867962602e-05, "loss": 4.4596, "step": 175104 }, { "epoch": 0.01, "learning_rate": 4.7124635489722956e-05, "loss": 4.4684, "step": 175616 }, { "epoch": 0.01, "learning_rate": 4.7116265921016165e-05, "loss": 4.4554, "step": 176128 }, { "epoch": 0.01, "learning_rate": 4.7107879973505645e-05, "loss": 4.4599, "step": 176640 }, { "epoch": 0.01, "learning_rate": 4.7099494025995125e-05, "loss": 4.4544, "step": 177152 }, { "epoch": 0.01, "learning_rate": 4.7091108078484605e-05, "loss": 4.4383, "step": 177664 }, { "epoch": 0.01, "learning_rate": 4.708272213097409e-05, "loss": 4.4603, "step": 178176 }, { "epoch": 0.01, "learning_rate": 4.7074368941071037e-05, "loss": 4.4497, "step": 178688 }, { "epoch": 0.01, "learning_rate": 4.7065982993560517e-05, "loss": 4.4548, "step": 179200 }, { "epoch": 0.01, "learning_rate": 4.705759704604999e-05, "loss": 4.4455, "step": 179712 }, { "epoch": 0.01, "learning_rate": 4.704921109853947e-05, "loss": 4.4493, "step": 180224 }, { "epoch": 0.01, "learning_rate": 4.704082515102895e-05, "loss": 4.4338, "step": 180736 }, { "epoch": 0.01, "learning_rate": 4.703243920351843e-05, "loss": 4.4606, "step": 181248 }, { "epoch": 0.01, "learning_rate": 4.702405325600791e-05, "loss": 4.4365, "step": 181760 }, { "epoch": 0.01, "learning_rate": 4.701566730849739e-05, "loss": 4.454, "step": 182272 }, { "epoch": 0.01, "learning_rate": 4.700728136098687e-05, "loss": 4.4393, "step": 182784 }, { "epoch": 0.01, "learning_rate": 4.699889541347635e-05, "loss": 4.4394, "step": 183296 }, { "epoch": 0.01, "learning_rate": 4.699050946596583e-05, "loss": 4.4337, "step": 183808 }, { "epoch": 0.01, "learning_rate": 4.698212351845531e-05, "loss": 4.4453, "step": 184320 }, { "epoch": 0.01, "learning_rate": 4.6973753949748525e-05, "loss": 4.4201, "step": 184832 }, { "epoch": 0.01, "learning_rate": 4.6965368002238005e-05, "loss": 4.4314, "step": 185344 }, { "epoch": 0.01, "learning_rate": 4.6956982054727485e-05, "loss": 4.4382, "step": 185856 }, { "epoch": 0.01, "learning_rate": 4.6948596107216965e-05, "loss": 4.4328, "step": 186368 }, { "epoch": 0.01, "learning_rate": 4.6940210159706445e-05, "loss": 4.427, "step": 186880 }, { "epoch": 0.01, "learning_rate": 4.6931824212195925e-05, "loss": 4.4221, "step": 187392 }, { "epoch": 0.01, "learning_rate": 4.69234382646854e-05, "loss": 4.4246, "step": 187904 }, { "epoch": 0.01, "learning_rate": 4.691505231717488e-05, "loss": 4.4484, "step": 188416 }, { "epoch": 0.01, "learning_rate": 4.690666636966436e-05, "loss": 4.4348, "step": 188928 }, { "epoch": 0.01, "learning_rate": 4.689828042215384e-05, "loss": 4.4281, "step": 189440 }, { "epoch": 0.01, "learning_rate": 4.688989447464332e-05, "loss": 4.4386, "step": 189952 }, { "epoch": 0.01, "learning_rate": 4.688152490593653e-05, "loss": 4.437, "step": 190464 }, { "epoch": 0.01, "learning_rate": 4.6873138958426014e-05, "loss": 4.4343, "step": 190976 }, { "epoch": 0.01, "learning_rate": 4.686476938971923e-05, "loss": 4.4141, "step": 191488 }, { "epoch": 0.01, "learning_rate": 4.68563834422087e-05, "loss": 4.4297, "step": 192000 }, { "epoch": 0.01, "learning_rate": 4.684799749469818e-05, "loss": 4.4272, "step": 192512 }, { "epoch": 0.01, "learning_rate": 4.683961154718766e-05, "loss": 4.4155, "step": 193024 }, { "epoch": 0.01, "learning_rate": 4.683122559967714e-05, "loss": 4.4334, "step": 193536 }, { "epoch": 0.01, "learning_rate": 4.682283965216662e-05, "loss": 4.4216, "step": 194048 }, { "epoch": 0.01, "learning_rate": 4.68144537046561e-05, "loss": 4.4278, "step": 194560 }, { "epoch": 0.01, "learning_rate": 4.680606775714558e-05, "loss": 4.4284, "step": 195072 }, { "epoch": 0.01, "learning_rate": 4.679769818843879e-05, "loss": 4.4091, "step": 195584 }, { "epoch": 0.01, "learning_rate": 4.6789328619732e-05, "loss": 4.4195, "step": 196096 }, { "epoch": 0.01, "learning_rate": 4.678094267222148e-05, "loss": 4.4133, "step": 196608 }, { "epoch": 0.01, "learning_rate": 4.6772573103514697e-05, "loss": 4.4245, "step": 197120 }, { "epoch": 0.01, "learning_rate": 4.6764187156004177e-05, "loss": 4.4239, "step": 197632 }, { "epoch": 0.01, "learning_rate": 4.6755801208493657e-05, "loss": 4.4194, "step": 198144 }, { "epoch": 0.02, "learning_rate": 4.6747415260983136e-05, "loss": 4.4216, "step": 198656 }, { "epoch": 0.02, "learning_rate": 4.6739029313472616e-05, "loss": 4.4189, "step": 199168 }, { "epoch": 0.02, "learning_rate": 4.6730643365962096e-05, "loss": 4.4178, "step": 199680 }, { "epoch": 0.02, "learning_rate": 4.6722257418451576e-05, "loss": 4.4134, "step": 200192 }, { "epoch": 0.02, "learning_rate": 4.6713871470941056e-05, "loss": 4.4108, "step": 200704 }, { "epoch": 0.02, "learning_rate": 4.6705485523430536e-05, "loss": 4.4173, "step": 201216 }, { "epoch": 0.02, "learning_rate": 4.6697099575920016e-05, "loss": 4.4087, "step": 201728 }, { "epoch": 0.02, "learning_rate": 4.6688713628409496e-05, "loss": 4.4074, "step": 202240 }, { "epoch": 0.02, "learning_rate": 4.6680327680898976e-05, "loss": 4.4124, "step": 202752 }, { "epoch": 0.02, "learning_rate": 4.6671941733388456e-05, "loss": 4.412, "step": 203264 }, { "epoch": 0.02, "learning_rate": 4.6663572164681665e-05, "loss": 4.4242, "step": 203776 }, { "epoch": 0.02, "learning_rate": 4.6655186217171145e-05, "loss": 4.4048, "step": 204288 }, { "epoch": 0.02, "learning_rate": 4.664680026966063e-05, "loss": 4.4028, "step": 204800 }, { "epoch": 0.02, "learning_rate": 4.663841432215011e-05, "loss": 4.4049, "step": 205312 }, { "epoch": 0.02, "learning_rate": 4.663002837463959e-05, "loss": 4.4032, "step": 205824 }, { "epoch": 0.02, "learning_rate": 4.66216588059328e-05, "loss": 4.3829, "step": 206336 }, { "epoch": 0.02, "learning_rate": 4.661327285842228e-05, "loss": 4.3983, "step": 206848 }, { "epoch": 0.02, "learning_rate": 4.660488691091176e-05, "loss": 4.4001, "step": 207360 }, { "epoch": 0.02, "learning_rate": 4.6596500963401234e-05, "loss": 4.3946, "step": 207872 }, { "epoch": 0.02, "learning_rate": 4.6588115015890714e-05, "loss": 4.3991, "step": 208384 }, { "epoch": 0.02, "learning_rate": 4.6579729068380194e-05, "loss": 4.3864, "step": 208896 }, { "epoch": 0.02, "learning_rate": 4.6571343120869674e-05, "loss": 4.3897, "step": 209408 }, { "epoch": 0.02, "learning_rate": 4.6562957173359154e-05, "loss": 4.4105, "step": 209920 }, { "epoch": 0.02, "learning_rate": 4.655458760465237e-05, "loss": 4.3964, "step": 210432 }, { "epoch": 0.02, "learning_rate": 4.654620165714185e-05, "loss": 4.4033, "step": 210944 }, { "epoch": 0.02, "learning_rate": 4.653781570963133e-05, "loss": 4.3954, "step": 211456 }, { "epoch": 0.02, "learning_rate": 4.652942976212081e-05, "loss": 4.4075, "step": 211968 }, { "epoch": 0.02, "learning_rate": 4.652106019341402e-05, "loss": 4.3941, "step": 212480 }, { "epoch": 0.02, "learning_rate": 4.65126742459035e-05, "loss": 4.3905, "step": 212992 }, { "epoch": 0.02, "learning_rate": 4.650428829839298e-05, "loss": 4.3924, "step": 213504 }, { "epoch": 0.02, "learning_rate": 4.649590235088246e-05, "loss": 4.3992, "step": 214016 }, { "epoch": 0.02, "learning_rate": 4.648751640337194e-05, "loss": 4.3814, "step": 214528 }, { "epoch": 0.02, "learning_rate": 4.647913045586142e-05, "loss": 4.3834, "step": 215040 }, { "epoch": 0.02, "learning_rate": 4.64707445083509e-05, "loss": 4.3836, "step": 215552 }, { "epoch": 0.02, "learning_rate": 4.646235856084038e-05, "loss": 4.3875, "step": 216064 }, { "epoch": 0.02, "learning_rate": 4.645400537093732e-05, "loss": 4.3869, "step": 216576 }, { "epoch": 0.02, "learning_rate": 4.64456194234268e-05, "loss": 4.3944, "step": 217088 }, { "epoch": 0.02, "learning_rate": 4.643723347591628e-05, "loss": 4.3733, "step": 217600 }, { "epoch": 0.02, "learning_rate": 4.642884752840576e-05, "loss": 4.3841, "step": 218112 }, { "epoch": 0.02, "learning_rate": 4.642046158089524e-05, "loss": 4.3739, "step": 218624 }, { "epoch": 0.02, "learning_rate": 4.641209201218845e-05, "loss": 4.3972, "step": 219136 }, { "epoch": 0.02, "learning_rate": 4.640370606467793e-05, "loss": 4.3955, "step": 219648 }, { "epoch": 0.02, "learning_rate": 4.639532011716741e-05, "loss": 4.3816, "step": 220160 }, { "epoch": 0.02, "learning_rate": 4.638693416965689e-05, "loss": 4.3792, "step": 220672 }, { "epoch": 0.02, "learning_rate": 4.637854822214637e-05, "loss": 4.384, "step": 221184 }, { "epoch": 0.02, "learning_rate": 4.637016227463585e-05, "loss": 4.387, "step": 221696 }, { "epoch": 0.02, "learning_rate": 4.636177632712533e-05, "loss": 4.3882, "step": 222208 }, { "epoch": 0.02, "learning_rate": 4.635339037961481e-05, "loss": 4.3775, "step": 222720 }, { "epoch": 0.02, "learning_rate": 4.634502081090802e-05, "loss": 4.382, "step": 223232 }, { "epoch": 0.02, "learning_rate": 4.633663486339751e-05, "loss": 4.3793, "step": 223744 }, { "epoch": 0.02, "learning_rate": 4.632824891588699e-05, "loss": 4.3847, "step": 224256 }, { "epoch": 0.02, "learning_rate": 4.6319895725983926e-05, "loss": 4.37, "step": 224768 }, { "epoch": 0.02, "learning_rate": 4.6311509778473406e-05, "loss": 4.3758, "step": 225280 }, { "epoch": 0.02, "learning_rate": 4.6303123830962886e-05, "loss": 4.373, "step": 225792 }, { "epoch": 0.02, "learning_rate": 4.6294754262256095e-05, "loss": 4.3587, "step": 226304 }, { "epoch": 0.02, "learning_rate": 4.6286368314745575e-05, "loss": 4.3835, "step": 226816 }, { "epoch": 0.02, "learning_rate": 4.6277982367235055e-05, "loss": 4.3712, "step": 227328 }, { "epoch": 0.02, "learning_rate": 4.6269596419724535e-05, "loss": 4.3743, "step": 227840 }, { "epoch": 0.02, "learning_rate": 4.6261210472214015e-05, "loss": 4.3772, "step": 228352 }, { "epoch": 0.02, "learning_rate": 4.6252824524703495e-05, "loss": 4.3624, "step": 228864 }, { "epoch": 0.03, "eval_loss": 4.348172187805176, "eval_runtime": 292.808, "eval_samples_per_second": 1303.212, "eval_steps_per_second": 40.726, "step": 228960 } ], "logging_steps": 512, "max_steps": 3052726, "num_train_epochs": 9223372036854775807, "save_steps": 10, "total_flos": 9.40925261274002e+16, "trial_name": null, "trial_params": null }